UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Postdoctoral researcher Jani Peräntie

University Lecturer Anne Tuomisto

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

University Lecturer Anu Soikkeli

University Lecturer Santeri Palviainen

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-2846-4 (Paperback)ISBN 978-952-62-2847-1 (PDF)ISSN 0355-3191 (Print)ISSN 1796-220X (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAA

SCIENTIAE RERUM NATURALIUM

U N I V E R S I TAT I S O U L U E N S I SACTAA

SCIENTIAE RERUM NATURALIUM

OULU 2021

A 754

Antti Moilanen

NOVEL REGULATORY MECHANISMS AND STRUCTURAL ASPECTS OF OXIDATIVE PROTEIN FOLDING

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF BIOCHEMISTRY AND MOLECULAR MEDICINE

A 754

AC

TAA

ntti Moilanen

ACTA UNIVERS ITAT I S OULUENS I SA S c i e n t i a e R e r u m N a t u r a l i u m 7 5 4

ANTTI MOILANEN

NOVEL REGULATORY MECHANISMS AND STRUCTURAL ASPECTS OF OXIDATIVE PROTEIN FOLDING

Academic dissertation to be presented with the assentof the Doctoral Training Committee of Health andBiosciences of the University of Oulu for publicdefence in the Leena Palotie auditorium (101A) of theFaculty of Medicine (Aapistie 5 A), on 15 February2021, at 12 noon

UNIVERSITY OF OULU, OULU 2021

Copyright © 2021Acta Univ. Oul. A 754, 2021

Supervised byProfessor Lloyd Ruddock

Reviewed byProfessor Carolyn SevierDoctor Adam Benham

ISBN 978-952-62-2846-4 (Paperback)ISBN 978-952-62-2847-1 (PDF)

ISSN 0355-3191 (Printed)ISSN 1796-220X (Online)

Cover DesignRaimo Ahonen

PUNAMUSTATAMPERE 2021

OpponentProfessor Neil Bulleid

Moilanen, Antti, Novel regulatory mechanisms and structural aspects of oxidativeprotein folding. University of Oulu Graduate School; University of Oulu, Faculty of Biochemistry andMolecular MedicineActa Univ. Oul. A 754, 2021University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland

Abstract

Efficient maturation of disulfide-containing proteins in the endoplasmic reticulum (ER) requiresfacilitated folding, aided by various folding catalysts and chaperones. As protein folding is acomplex process, proteins may, even in the presence of endogenous folding factors, formmisfolded states. Therefore, the cell has acquired several layers of quality control mechanisms tokeep the folding load in the ER in balance with the folding capacity. The activity of foldingcatalysts is regulated to avoid stress connected to hyperoxidation. Other quality controlmechanisms include the unfolded protein response (UPR) and ER-associated protein degradation(ERAD). The UPR consists of signaling cascades that sense unfolded protein stress and upregulatefolding factors and other components to relieve the stress. ERAD directly removes non-nativeproteins.

The main folding catalysts that introduce disulfide bonds to folding proteins in the ER aremembers of the ER oxidoreductin 1 (Ero1) family and protein disulfide isomerases (PDI). Ero1 isa sulfhydryl oxidase that reduces molecular oxygen to hydrogen peroxide to form disulfides. Thenewly generated disulfides are then transferred during the catalytic cycle to PDI which catalyzesdisulfide formation in folding proteins. The Ero1-PDI pathway has been studied extensively bycell-based techniques, and its activity is regulated in a redox dependent manner by rearrangingregulatory disulfides. However, the kinetic aspects of these regulatory functions have remainedpoorly characterized.

In this study, we report comprehensive kinetic measurements for the Ero1-PDI pathway thatreveal novel regulatory mechanisms for oxidative protein folding (OPF). These include amechanism consisting of high affinity and apparent cooperativity for substrate oxygen to allowEro1 to function efficiently at hypoxic conditions. We also show that different human Ero1isoforms have differential dependence for substrate PDI with one of the isoforms, Ero1α, showingstrong dependence for PDI for both redox activation as well as the catalytic cycle. This observationallowed us to describe a potential crosstalk between OPF and ERAD, in which OPF by the Ero1α-PDI pathway is downregulated, to potentially increase the efficiency of ERAD which requiresreduced protein substrates. This work is concluded by a structural study of a novel chaperone,pERp1, that is crucial for our immunity, as it is involved in the folding of some immunoglobulins.

Keywords: disulfide bond, endoplasmic reticulum, ERAD, Ero1, oxidative proteinfolding, PDI, UPR

Moilanen, Antti, Disulfidisidoksia sisältävien proteiinien laskostumisen uudetsäätelymekanismit ja rakenteelliset ominaisuudet. Oulun yliopiston tutkijakoulu; Oulun yliopisto, Biokemian ja molekyylilääketieteen tiedekuntaActa Univ. Oul. A 754, 2021Oulun yliopisto, PL 8000, 90014 Oulun yliopisto

Tiivistelmä

Disulfidisidoksia sisältävien proteiinien laskostumista endoplasmakalvostossa avustavat erilai-set laskostumiskatalyytit ja kaperonit. Koska laskostuminen on monimutkaista, voivat proteiinitlaskostumistekijöistä huolimatta laskostua väärin. Endoplasmakalvostoon onkin kehittynyt usei-ta laadunvarmistusmekanismeja, jotka pitävät laskostumistaakan ja -kapasiteetin tasapainossa.Laskostumiskatalyyttien aktiivisuutta voidaan säädellä, minkä avulla vältetään ylihapettumiseenliittyvä stressi. Muihin laadunvarmistuskeinoihin kuuluvat UPR (unfolded protein response) jaERAD (endoplasmic reticulum-associated degradation). UPR tunnistaa laskostumattomien pro-teiinien aiheuttaman stressin ja saa aikaan mm. laskostumistekijöiden lisätuotantoa stressinvähentämiseksi. ERAD poistaa suoraan laskostumattomia proteiineja hajottamalla niitä.

Ero1- ja PDI-proteiiniperheiden jäsenet toimivat pääasiallisina laskostumiskatalyytteinä, jot-ka muodostavat disulfideja laskostuviin proteiineihin. Ero1 on sulfhydryylioksidaasi, joka pel-kistää hapen veryperoksidiksi muodostaessaan disulfidin. Uuden disulfidin Ero1 siirtää sittenPDI:lle, joka puolestaan katalysoi disulfidien muodostuksen laskostuvissa proteiineissa. Ero1-PDI-systeemiä on tutkittu paljon solutasolla, ja sen aktiivisuutta voidaan säädellä järjestelemälläsen omia säätelydisulfideja. Tämän säätelyjärjestelmän entsyymikinetiikka tunnetaan kuitenkinhuonosti.

Tässä työssä esittelemme Ero1-PDI-systeemin laajat entsyymikinetiikan mittaukset, jotkapaljastavat uusia säätelymenetelmiä proteiinien laskostumiseen. Näihin menetelmiin kuuluvatmm. tehokkuus hypoksisissa olosuhteissa ja erilaiset kineettiset ominaisuudet ihmisen Ero1-per-heen jäsenten välillä. Tulokset osoittavat, että Ero1α vaatii korkeat pitoisuudet PDI-substraattiasekä aktivoituakseen että disulfidien siirtoon PDI:lle. Näiden tulosten perusteella havaitsimmemahdollisen laskostumisen ja ERAD:n välisen vuorovaikutuksen, jossa laskostumisen tehok-kuutta lasketaan ERAD:n toiminnan tehostamiseksi. Lisäksi esittelemme kiderakenteen uudellekaperonille, pERp1:lle, joka vasta-aineiden laskostumistekijänä on tärkeä immuniteetillemme.

Asiasanat: disulfidisidos, endoplasmakalvosto, ERAD, Ero1, PDI, proteiinienlaskostuminen, UPR

To family, to friends, to colleagues; You made this possible.

8

9

Acknowledgements

I gratefully acknowledge the Faculty of Biochemistry and Molecular Medicine and

Biocenter Oulu for the expertise and permission to use of their facilities. I would

like to also thank the staff of both organizations who carry out excellent science in

many different fields and educate future scientists. My work was funded by the

Academy of Finland, The Finnish Cultural Foundation, and The University of Oulu

Scholarship Foundation.

Special thanks go to Prof. Lloyd Ruddock, for your excellent ideas that led me

here, for supervision, and for creating the wonderful research group and atmosphere;

to Dr. Mirva Saaranen, who made this work possible with science and help and

support; Kati Korhonen, without whom I would probably not be here, as it was you

who pulled me into this fascinating world of Ero1; to Heli and Yuko, you were also

there every day helping and guiding; Johanna and Katja, it was a pleasure to share

minds with you, and I already miss our brainstorming sessions about science and

families; and to the other group members, George, Lisette, Rosario, Angel, Aatir,

you all are part of this work.

I am thankful to the deans Prof. Kalervo Hiltunen and Prof. Peppi Karppinen,

the dean of education Adj. Prof. Tuomo Glumoff, all the professors and group

leaders, as well as the administration personnel Pia Askonen, Auli Kinnunen, Jari

Heikkinen, and Kaisa Tuominen, who managed our scientific community; you did

excellent work. It was a privilege working with you.

The final acknowledgements go to my wonderful family, who patiently

supported me through this process; to the members of my follow-up group, Dr.

Kristian Koski and Doc. Ulrich Bergmann; to the reviewers of this dissertation,

Assoc. Prof. Carolyn Sevier and Dr. Adam Benham. Creating this work was a long

and eventful adventure. Your inspiring feedback guided me here.

Oulu, 7.10.2020 Antti Moilanen

10

11

Abbreviations

-COO- deprotonated carboxyl (-COOH) group

-NH3+ protonated amine (-NH2) group

-S-S- disulfide bond

-SH thiol group

-SOH sulfenic acid group

1SS folding intermediate with one disulfide

2SS folding intermediate with two disulfides

Ala alanine

ALR augmenter of liver regeneration

Arg arginine

ATF6 cyclic AMP-dependent transcription factor ATF-6

ATP adenosine triphosphate

BiP binding immunoglobulin protein

BLI bilayer interferometry

BPA bisphenol A

BPTI bovine pancreatic trypsin inhibitor

C-terminus carboxyl-terminus of a polypeptide

CD circular dichroism

CGHC amino acid sequence motif; C, cysteine; G, glycine; H,

histidine

CNPY canopy (protein family)

CP peroxidatic cysteine

CPXC amino acid sequence motif; C, cysteine; P, proline; X, any

amino acid

CR resolving cysteine

CRELD cysteine-rich with EGF-like domain protein

CS citrate synthase

CyDisCo cytoplasmic disulfide bond formation in Escherichia coli

Cys cysteine

CXXC amino acid sequence motif; C, cysteine; X, any amino acid

COX cytochrome c oxidase

DHA dehydroascorbate

DNA deoxyribonucleic acid

DTT dithiothreitol

Dsb thiol:disulfide interchange protein

12

E2 estradiol

EDEM ER degradation-enhancing alpha-mannosidase-like protein 1

ER endoplasmic reticulum

ERAD endoplasmic reticulum-associated degradation

ERdj5 endoplasmic reticulum DNA J domain-containing protein 5

Ero1 endoplasmic reticulum oxidoreductin 1

Ero1p endoplasmic reticulum oxidoreductin 1 from Saccharomyces

cerevisiae

ERp27 endoplasmic reticulum resident protein 27

ERp44 endoplasmic reticulum resident protein 44

ERp46 thioredoxin domain-containing protein 5

ERp57 protein disulfide isomerase A3

ERp72 protein disulfide isomerase A4

Erv1p mitochondrial FAD-linked sulfhydryl oxidase ERV1, or

essential for respiration and vegetative growth protein 1, from

Saccharomyces cerevisiae

FAD flavin adenine dinucleotide

FAM20C extracellular serine/threonine protein kinase

FLAG amino acid sequence DYKDDDDK

Glu glutamate

GPx glutathione peroxidase

GR glutathione reductase

GRP94 endoplasmin, or 94 kDa glucose-regulated protein

GSH glutathione (reduced)

GSSG glutathione (oxidized)

HEK human embryo kidney cell line

HeLa human cervical cancer cell line

HSP heat shock protein

HSP70 heat shock 70 kDa protein

I folding intermediate

ID intrinsically disordered

Ig immunoglobulin

IMS intermembrane space

IRE1 serine/threonine-protein kinase/endoribonuclease IRE1

ITC isothermal titration calorimetry

k reaction rate constant

kcat catalytic rate constant

13

KFWWFS peptide consisting of lysine (K), phenylalanine (F), tryptophan

(W), and serine (S)

KM Michaelis constant, or substrate concentration at which the

reaction rate is half its maximal value

LMWO low-molecular-weight oxidant

M misfolded state

MANF mesencephalic astrocyte-derived neurotrophic factor, or

ARMET

Mia40 mitochondrial intermembrane space import and assembly

protein 40

MP membrane protein

mRNA messenger ribonucleic acid

MTP microsomal transfer protein

N native state

N-terminus amino-terminus of a polypeptide

N* two-disulfide folding intermediate of the folding pathway of

bovine pancreatic trypsin inhibitor

NADPH nicotinamide adenine dinucleotide phosphate

NSS

HH the penultimate, two-disulfide folding intermediate of the

folding pathway of bovine pancreatic trypsin inhibitor

OPF oxidative protein folding

ox oxidized

P5 protein disulfide isomerase A6

PDB Protein Data Bank

PDI protein disulfide isomerase

PDI3FLAG protein disulfide isomerase with three inserted FLAG

(DYKDDDDK) sequences

PDIp protein disulfide isomerase A2, or pancreas specific protein

disulfide isomerase

PERK eukaryotic translation initiation factor 2-alpha kinase 3

pERp1 plasma cell-induced resident endoplasmic reticulum protein

Phe phenylalanine

pKa negative logarithm of the acid dissociation constant (Ka)

Prx peroxiredoxin

QC quality control

QSOX quiescin sulfhydryl oxidase

rBPTI reduced bovine pancreatic trypsin inhibitor

14

red reduced

RNA ribonucleic acid

RNase ribonuclease

ROS reactive oxygen species

rpHPLC reversed phase high-performance liquid chromatography

RS- thiolate, where R denotes an alkyl or other organic substituent

RS-SR disulfide bond, where R denotes an alkyl or other organic

substituent

SAPLIP saposin-like protein

SDS-PAGE sodium dodecyl sulfate-polyacryl amide gel electrophoresis

Ser serine

SN2 bimolecular nucleophilic substitution reaction

SRP signal recognition particle

SRPR signal recognition particle receptor

SSS transition state with three sulfur atoms arranged linearly

t1/2 half-time

Trp tryptophan

U unfolded state

UPR unfolded protein response

Val valine

VKOR vitamin K epoxide reductase

15

Original publications

This thesis is based on the following publications, which are referred throughout

the text by their Roman numerals:

I Moilanen, A., Korhonen, K., Saaranen, M. J., & Ruddock, L. W. (2018). Molecular analysis of human Ero1 reveals novel regulatory mechanisms for oxidative protein folding. Life Science Alliance, 1(3), e201800090. doi:10.26508/lsa.201800090

II Moilanen, A., & Ruddock, L. W. (2020). Non-native proteins inhibit the ER oxidoreductin 1 (Ero1)-protein disulfide-isomerase relay when protein folding capacity is exceeded. The Journal of Biological Chemistry, 295(26), 8647-8655. doi:10.1074/jbc.RA119.011766

III Sowa, S. T., Moilanen, A., Biterova, E., Saaranen, M. J., Lehtiö, L., & Ruddock L. W. High-resolution crystal structure of human pERp1, a saposin-like protein crucial for IgA, IgM and integrin maturation in the endoplasmic reticulum. In press.

16

17

Contents

Abstract

Tiivistelmä

Acknowledgements 9 

Abbreviations 11 

Original publications 15 

Contents 17 

1  Introduction 19 

2  Review of the literature 23 

2.1  Protein folding phenomenon ................................................................... 23 

2.1.1  From primary to quaternary structure ........................................... 23 

2.1.2  Protein refolding in vitro and in silico .......................................... 24 

2.1.3  Co-translational protein folding in vivo ....................................... 29 

2.1.4  Molecular chaperones ................................................................... 31 

2.2  Biochemistry of disulfide formation ....................................................... 32 

2.2.1  Disulfide formation is catalyzed redox chemistry ........................ 33 

2.2.2  The pKa value determines the reactivity of a thiol group ............. 36 

2.2.3  Thiol-disulfide exchange reactions ............................................... 38 

2.3  Oxidative protein folding in vivo ............................................................ 41 

2.3.1  Disulfide formation in prokaryotes............................................... 41 

2.3.2  Mitochondrial relay for disulfide bond formation ........................ 43 

2.3.3  Multiple ways to form disulfides in the ER .................................. 44 

2.4  Structural biology of the Ero1-PDI relay ................................................ 54 

2.4.1  Structure of human Ero1 isoforms................................................ 54 

2.4.2  Architecture of PDI ...................................................................... 56 

2.5  Structure-function relationship of Ero1-PDI ........................................... 58 

2.5.1  Substrate oxidation by PDI ........................................................... 58 

2.5.2  PDI reoxidation by Ero1 ............................................................... 60 

2.5.3  Ero1 reoxidation ........................................................................... 60 

2.6  Quality control of oxidative protein folding ........................................... 62 

2.6.1  Modulation of Ero1 activity ......................................................... 62 

2.6.2  Unfolded protein response (UPR) ................................................ 64 

2.6.3  Endoplasmic reticulum-associated protein degradation

(ERAD) ........................................................................................ 66 

3  Aims of the present work 69 

4  Methods 71 

18

5  Results 73 

5.1  Molecular analysis of human Ero1 .......................................................... 73 

5.1.1  Human Ero1 and PDI form a novel Ero1-PDI complex ............... 73 

5.1.2  Generation of a non-linear regression method to analyze

Ero1 kinetics ................................................................................. 74 

5.1.3  Differential substrate PDI dependence between the Ero1

isoforms ........................................................................................ 75 

5.1.4  A novel regulatory mechanism allows efficient operation

in hypoxic conditions ................................................................... 77 

5.2  Feedback inhibition of the Ero1α-PDI system ........................................ 78 

5.2.1  The Ero1α-PDI interaction is inhibited by non-native

proteins ......................................................................................... 79 

5.2.2  Quantification of the inhibition .................................................... 79 

5.2.3  The inhibition is reversible ........................................................... 80 

5.3  Structure-function relationship of human pERp1 ................................... 81 

5.3.1  High-resolution crystal structure of pERp1 .................................. 82 

5.3.2  Mapping function onto the structure ............................................. 82 

5.3.3  Biochemical assays to elucidate function ..................................... 83 

6  Discussion 85 

7  Future prospects 93 

List of references 95 

Original publications 117 

19

1 Introduction

The smallest living unit, the cell, is a factory of a multitude of molecules, which

help the cell to carry out intra- and intercellular physiological processes - from

maintaining and translating the genetic code to signaling with neighboring cells to

ultimately organizing into tissues, organs, and organisms - to sustain life.

Macromolecules deoxyribonucleic acid (DNA), ribonucleic acid (RNA), and

proteins, which are all large polymeric molecules formed from repeating

nucleotides (DNA and RNA) or amino acids (proteins), create a cellular workflow

from blueprint (DNA) to message (RNA) to structure (proteins) (Fig. 1).

The human genome has over three billion nucleotide base pairs and contains

20,000 to 25,000 protein-coding genes (International Human Genome Sequencing

Consortium, 2004). A gene is transcribed into message RNA (mRNA) which

translates to a protein (Fig. 1). The number of proteins available for a human cell

is greater than that the number of protein-coding genes, as one gene can potentially

create several different RNA transcripts, for example, by alternative splicing,

creating different forms (isoforms) of the same protein with different functions and

attributes. The total count of proteins, including isoforms, in human cells is

therefore ~100,000 (Pan et al., 2008). The proteins are sorted during (co-) and after

translation (post-translationally) to different organelles, including the nucleus, the

cytosol, the mitochondrion, and the endoplasmic reticulum (ER). One method to

traffic proteins to specific organelles is the inclusion of a targeting signal, for

example, a signal sequence (also called signal peptide) or a transmembrane

segment that may target proteins to the ER (Fig. 1) (Blobel, 1980). A targeting

signal can also be used to retain a protein in the ER preventing its secretion

(Raykhel et al., 2007).

20

Fig. 1. From genetic code to correctly folded protein. The genetic material (DNA) in the

nucleus is transcribed to mRNA and transported to the cytosol, where it binds

ribosomal subunits to create a mRNA-ribosome complex. The ribosome translates the

mRNA and produces an amino acid sequence based on the mRNA code. If a specific

amino acid sequence (a signal peptide) is translated, it will be recognized by the signal

recognition particle (SRP) which halts translation and targets the mRNA-ribosome-

nascent-polypeptide-SRP complex to an ER membrane receptor (SRPR). Otherwise,

translation continues in the cytoplasm. The SRP-SRPR interaction allows the ribosome

to contact with a membrane channel complex (translocase), SRP dissociates,

translation continues, and co-translational translocation of the growing polypeptide to

the lumen of the ER initiates (Rapoport, 2007). The signal peptide is cleaved during this

process by a signal peptidase complex (Ellgaard et al., 2016). The depicted co-

translational translocation is a common event in all cell types, but ribosome-

independent translocation is also possible, typically utilized by simpler organisms such

as bacteria and yeast (Rapoport, 2007). In the ER, proteins fold and acquire a specific

biological function defined by their native structure. As protein folding is a complicated

process, misfolded states may form that lead to aggregate formation. To avoid this, the

ER has elaborate quality control (QC) features that maintain the balance between the

folded and misfolded proteins, to allow efficient export of natively folded proteins from

the ER (dashed arrow).

21

As many as 40% of all human proteins contain a signal sequence or a

transmembrane region and are translocated co-translationally to the ER (Uhlen et

al., 2015; Fass & Thorpe, 2018), the starting point of the secretory pathway that

guides new proteins to the extracellular space. There they complete translation,

acquire a three-dimensional structure (a native fold), and gain a biological function

(Fig. 1). The protein folding process is complicated, and often proteins may form

unwanted functionless misfolded states. To avoid this, the ER has evolved a

multitude of different folding factors and quality control measures that either

prevent the nascent protein from misfolding, rescue from misfolded states, or

remove terminally misfolded proteins, to promote native folding and export from

the ER (Fig. 1). Folding in the ER is governed also by many co- and post-

translational protein modifications that are important for proper folding and for

regulation of protein function (Ellgaard et al., 2016).

This work provides a basis for understanding the protein folding phenomenon,

in particular, the folding process that includes also disulfide bond formation, a

crucial modification of native folded state of most proteins of the ER and the

subsequent destinations of the secretory pathway. The key folding factors that aid

in disulfide formation are introduced, including protein disulfide isomerase (PDI),

whose molecular function will be shown by systematic review of literature to be

under the control of another ER-resident enzyme, endoplasmic reticulum

oxidoreductin 1 (Ero1). The main quality control features of the ER are reviewed

to help us understand, how cells adapt to different stress situations that are prone to

happen in the ER. Original publications will reveal novel functional, regulatory,

and structural features of the Ero1-PDI folding relay and other folding factors to

gain new insight into the important cellular decisions that influence the

folding/misfolding partitioning in the ER (Fig. 1).

22

23

2 Review of the literature

2.1 Protein folding phenomenon

Proteins gain specific biological functions only when they acquire a native three-

dimensional fold. This review will first provide the basic physicochemical

principles of conformational protein folding, a process that governs the unfolded to

native state transition. The overall protein folding process of many proteins is

influenced by additional co- and post-translational modifications. The composite

process of conformational folding and co- and post-translational disulfide bond

formation - also termed oxidative protein folding (OPF) - is discussed next.

Disulfide bonds are the main topic of this study, and the later chapters will

concentrate on the biochemical aspects of their formation. Other co- and post-

translational modifications are out of the scope for this study, but for an overview

of the most common modifications, see Ellgaard et al. (2016).

2.1.1 From primary to quaternary structure

Protein structure is hierarchical, and it consists of four levels (Finkelstein & Ptitsyn,

2016):

1. The primary structure of a protein is defined by all covalent bonds in an amino

acid sequence. These include peptide bonds, covalent bonds of the side chains,

and any covalent co- and post-translational modifications between amino acids.

2. Secondary structure elements are then formed that define local folding of the

polypeptide chain. The secondary structure is defined by rotation around three

covalent bonds within an amino acid residue corresponding to three dihedral

angles, ω, φ, and ψ. As the peptide bond has a partial double-bond

characteristic, ω is nominally restricted to two angles, 180° (trans

configuration) or 0° (cis configuration). The trans configuration is sterically

less hindered and energetically more favored and consequently occurs more

commonly in natural proteins (the cis is limited to proline residues only, and

the trans-cis interconversion is usually actively assisted by folding factors,

which are described later). The other two main chain angles, φ and ψ, have

more degrees of freedom, and they control most of the folding processes of a

polypeptide chain. Regular secondary structures, defined by repeating dihedral

angles in a linear stretch of amino acids, are especially important for protein

24

folding; for example, alpha helices and beta sheets that are stabilized by inter-

residue hydrogen bonds.

3. Compaction of the polypeptide chain and regular secondary structure elements

defines the tertiary structure of a protein. Several models have been proposed

to characterize the initial steps of this event, for example, the framework and

the hydrophobic collapse models. In the former model, compaction happens

via packing of secondary structure elements that interact, for example, via

hydrogen bonds and electrostatic interactions (Udgaonkar & Baldwin, 1988).

During hydrophobic collapse, interactions between hydrophobic residues drive

the chain to a compact state, and the hydrophobic residues become buried to

the core of the protein (Tanford, 1962; Dill, 1990). A single polypeptide chain

may utilize one or several different compaction methods and may have several

distinct tertiary structure modules, also called domains.

4. Occasionally, a quaternary structure may form when two or more polypeptide

chains (subunits) interact with each other. Such a stable protein complex is

known also as an oligomer.

Most proteins follow these fundamentals. However, there are always exceptions to

the rule: a special group of proteins called intrinsically disordered (ID) proteins (or

proteins containing ID regions) may form biologically functional entities without

regular secondary or tertiary structure elements. They are out of scope for this study,

but for a thoughtful review, see Fonin et al. (2018).

2.1.2 Protein refolding in vitro and in silico

Anfinsen and colleagues (1961) revealed more than fifty years ago that unfolded

proteins can refold in the test tube rapidly and spontaneously to their native states.

This efficient maturation suggested that a protein chain avoids a slow systematic

search of a single native state from countless possible chain configurations (a 100-

residue chain has at least 10100 potential structures, as each residue has at least ten

different combinations of backbone dihedral angles) and matures via a limited

number of folding pathways (Anfinsen et al., 1961; Finkelstein, 2018). The

conformational folding phenomenon is typically modelled by various funnel

models that conceptualize protein maturation as allowed trajectories on multi-

dimensional “rugged” energy landscapes (Fig. 2) (Leopold et al., 1992; Wolynes et

al., 1995; Dinner et al., 2000). A common feature for all these models is that

transition from the unfolded state to the native state proceeds towards intermediate

25

states with higher stability and lower thermodynamic free energy, with the native

state residing at a global free energy minimum (Fig. 2). In productive folding, the

intermediate states utilize mainly native intramolecular interactions (that are found

in the native state), that on average are more stable than non-native ones (Anfinsen,

1973; Abkevich et al., 1994; Sali et al., 1994; Dobson, 2003).

Non-catalyzed protein folding is often inefficient. Productive folding

intermediate states with low energy minima, also called kinetic traps, as well as

non-native intermolecular interactions between folding intermediates that lead to

stable low-energy non-native states, for example, irreversible aggregates, may

decrease the efficiency of folding and slow down the folding process significantly.

Therefore, during the search for the native state, the folding chain needs to also

avoid and escape any traps and non-productive energy minima (Fig. 2). This is a

matter of life and death: Inefficient folding of proteins that are essential for the

development and critical functions of an organism will terminate life. Even non-

essential proteins, if folded incorrectly, may cause protein aggregation diseases that

ultimately lead to death (Dobson, 2003; Jahn & Radford, 2005). Cells have

developed special folding factors, which are described later in this review, to avoid

these potentially dangerous off-pathway reactions.

26

Fig. 2. Schematic of a multidimensional protein folding funnel. The funnel includes all

possible chain configurations, the arrows depict allowed trajectories. Depicted is a

refolding process [adapted from Stigler et al. (2011)] of a two-domain protein which may

fold from unfolded (U) to native (N) state via two potential on-pathway trajectories

containing folding intermediates (I), or form one off-pathway misfolded state (M). M

must unfold to U to return to the productive pathways. The I states are defined by local

energy minima with varying energy depts and activation energy barriers to proceed to

the next state. Deep local minima may lead to very slowly folding intermediates, also

known as kinetic traps (see inset). Although the allowed trajectories are defined, the

search for N is stochastic, i.e., reversible U ↔ I, U ↔ M, I ↔ M, I ↔ N transitions are

possible. Even reversible U ↔ N transition is possible for simple one-domain proteins

with smooth energy landscapes (Finkelstein, 2018). Inset energy diagram: A free energy

diagram adapted from Finkelstein & Ptitsyn (2016) which depicts a three-state folding

process from an initial state (1) to a final state (2) via a kinetic trap (X). X is a trap, as

transition X → 2 has higher activation energy barrier (grey arrow) than the transition 1

→ X. It is also an “on-pathway” intermediate, as it is more stable (has lower free energy)

than the original state (1) but less stable than the final state (2). The conversion of

kinetic traps and the rescue of off-pathway species may be accelerated significantly by

folding factors.

27

The seminal work by Anfinsen et al. (1961) studied OPF of ribonuclease A

(RNase A), a small protein with eight cysteines that theoretically can adopt 105

different four-disulfide patterns (Anfinsen, 1973). Remarkably, only one native

disulfide pattern was observed in the experiments. This leads to a fundamental

question: are disulfides rearranged randomly until the native pattern is reached, or

does OPF utilize limited pathways – analogous to the conformational folding

described above - and favor folding intermediates with native disulfides? Two

models with stark differences have been experimentally demonstrated to answer

this question. One model is based on refolding of mature bovine pancreatic trypsin

inhibitor (BPTI) and the other on hirudin, both small proteins with three disulfides

in the native state. BPTI has an anti-parallel β-sheet core surrounded by alpha

helices that are stabilized by disulfides [Cys5-Cys55, Cys30-Cys51, Cys14-Cys38]

(Fig. 3A). A consensus view on the sequential folding events of BPTI (Fig. 3B)

(Weissman & Kim, 1991) suggests that the reduced state can first adopt any of two

different intermediates with one native disulfide, either [Cys5-Cys55] or [Cys30-

Cys51]. Subsequently, they form two-disulfide states, [Cys5-Cys55, Cys14-Cys38]

or [Cys30-Cys51, Cys14-Cys38], respectively. Attaining the final disulfide is,

counterintuitively, a complex and slow process that requires, in both pathways,

disulfide rearrangements to form a third two-disulfide intermediate, [Cys5-55,

Cys30-51]. Only this state can then efficiently fold to the three-disulfide native state.

The consensus refolding mechanism dictates that maturation of BPTI utilizes

predominantly native disulfides. In contrast, hirudin has been demonstrated to

refold with a so-called “trial and error” method, i.e., it matures with a heterogenous

mixture of folding intermediates, containing both native and non-native disulfides

(Chatrenet & Chang, 1992; Chatrenet & Chang, 1993). These two extreme in vitro

models, and other cases in between them [reviewed by Chang (2011)], suggest that

the diversity of oxidative refolding of an isolated protein molecule in the test tube

is wide and cannot be answered by a single model. Note, however, that little

evidence currently exists that non-native disulfides are required for productive

protein folding in vivo (Fass & Thorpe, 2018), and potential factors, that may

stabilize protein structure in vivo and eliminate formation of non-native disulfides,

are discussed later in Chapter 2.1.3. Therefore, this work will mainly concentrate

on the BPTI-type refolding model that utilizes native interactions.

28

Fig. 3. The BPTI fold. (A) Secondary structure representation of wild-type BPTI (Protein

Data Bank code: 1BPI). Alpha helices (α, magenta), beta strands (β, red), and the three

native disulfides ([Cys-Cys], yellow) are marked. Regions without regular secondary

structure are grey. N, N-terminus; C, C-terminus. (B) Schematic of the consensus BPTI

oxidative folding pathways (Weissman & Kim, 1991). Conversion of the reduced

unfolded state (R) via one-disulfide intermediates (1SS) to the 2SS kinetic trap states

(N* and [30-51, 14-38]) is kinetically fast. The interconversion between the 2SS kinetic

traps and conversion of the 2SS kinetic traps to the NSS

HH are slow processes, as local

structural unfolding is required to expose the free thiols for disulfide rearrangements.

Folding from NSS

HH to N is fast.

29

One fundamental question remains unanswered: Does disulfide formation

drive conformational folding or vice versa? Although the history of OPF spans

more than fifty years, the numerous “wet lab” folding studies have been unable to

answer this question explicitly. Lately, computational methods, especially

molecular simulations, have become indispensable tools for studying folding of

simple proteins. For example, Qin and coworkers (2015) demonstrated that well-

calibrated molecular simulations can refold BPTI consistent with the folding

pathways observed in the seminal biochemical studies of Weissman & Kim (1991).

Their simulations suggested that substantial compaction and formation of the

central β-sheet precedes formation of any disulfides. This hints at conformational

folding being the driving force for disulfide formation. Furthermore, BPTI has been

demonstrated to possess native subdomains, and their sequential formation as well

as reductive unfolding dictates the sequential oxidative folding pattern (Staley &

Kim, 1990; Mendoza et al., 1994). In contrast, reductive unfolding of hirudin shows

an “all-or-none” scheme, in which all three native disulfides are reduced

simultaneously, consistent with hirudin lacking any stable subdomains (Chang,

1997). The causal factor for deciding whether a protein adopts the BPTI or the

hirudin-like refolding model may therefore simply be relative structural stability of

protein subdomains in the in vitro experiments (Chang, 2011).

Collectively, the computational and in vitro biochemical evidences discussed

here suggest that conformational folding of stable (sub)domains may guide native

disulfide formation by defining the accessibility, proximity, and reactivity of thiols

(which are discussed in more detail later in this study) for efficient thiol-disulfide

chemistry.

2.1.3 Co-translational protein folding in vivo

Protein folding in vivo occurs in environments and conditions drastically different

from the in vitro refolding experiments described above. Whereas the unstructured

isolated protein at start of the refolding experiment may access all backbone

residues, a nascent polypeptide in vivo will start folding co-translationally – both

conformationally and oxidatively - while still ribosome-bound and only partially

synthesized. In other words, in vivo folding of some proteins is vectorial (N-

terminus folds first, C-terminus last) and a domain-by-domain folding process

(Bulleid & Freedman, 1988; Kolb et al., 1994; Komar et al., 1997; Holtkamp et al.,

2015; Thommen et al., 2017). The main factors that delineate in vivo folding from

test tube experiments are listed below:

30

– The reported refolding times that range from sub-seconds to tens of hours (for

example, Anfinsen et al., 1961; Schuler & Hofmann, 2013) are often

physiologically unrealistic. Minimal rates of in vivo folding are defined by

ribosomal translation rate (ktranslation), which in human cells is circa four amino

acids incorporated to a nascent chain per second. It takes two minutes to

synthesize an average-sized protein of 50 kDa (Braakman et al., 1991; Yan et

al., 2016). Also, the folding process in vivo rarely takes hours, as cells need to

be able to respond to various physiological stimuli rapidly and efficiently

(Dintzis, 1961; Braakman et al., 1991).

– ktranslation can be modulated on the mRNA and nascent chain levels (Kaiser et

al., 2011, Waudby et al., 2019) to allow controlled partial folding of

polypeptides. For example, rare codons between domain boundaries and

certain amino acid sequences that interact with the ribosome exit tunnel may

cause transient ribosomal arrests (Zhang et al., 2009; Tsai et al., 2014), favoring

the domain-by-domain folding process. Such elegant mechanisms may

significantly impact the folding energy landscape (Fig. 2) by transiently

limiting the number of allowed polypeptide configurations and preventing off-

pathway interactions that may lead to aggregation. This is especially important

for proteins (or protein domains) with high folding kinetics (kfolding). If kfolding

<< ktranslation, folding finishes post-translationally, which can be directly

observed for many proteins, as their N- and C-termini are in close proximity in

their native structures (Ellgaard et al., 2016; see also Fig. 3A).

– In vitro refolding experiments are typically carried out in mild buffer

conditions with high free volume that the folding chain can utilize during

folding. This is in stark contrast with cellular environments which are crowded

with various macromolecules, including proteins and nucleic acids.

Macromolecules may occupy up to 30% of the cytoplasmic volume and can

reach as high as 400 mg/mL concentration (Zimmerman & Trach, 1991; Fonin

et al., 2018). This is likely true also in the ER: certain folding factors in the ER

exist at very high concentrations, for instance, chaperone BiP and the folding

catalyst PDI alone reach millimolar (circa 50 mg/mL) levels (Hillson et al.,

1984; Lyles & Gilbert, 1991; Weitzmann et al., 2007). The crowding effect has

both positive and negative influences on protein folding (Rivas et al., 1999;

Ellis, 2001): on one hand, crowding may enhance early-stage folding, for

example, the intramolecular hydrophobic collapse, but also the late-stage

intermolecular associations leading to quaternary structure formation; on the

other hand, crowding increases the probability for non-native intermolecular

31

contacts that may lead to aggregation. However, the net effect is likely positive,

as cells have developed ways to minimize aggregation, specifically, molecular

chaperones (Martin & Hartl, 1997). Similarly, crowding may also stabilize

formation of the discussed protein subdomains, which may influence the

decision whether a protein adopts the scrambled hirudin-like folding scheme

or the BPTI-like model with native interaction only.

All in all, while in vitro folding studies have provided invaluable information from

thermodynamics to molecular details for protein folding in general, the co-

translational folding behavior inside the crowded cell introduces additional

complexities which are less well understood. Direct comparisons between refolding

and co-translational folding pathways of a given protein are lacking. A promising

study by Samelson and colleagues (2018) compared co-translational folding and

test tube folding of the same protein, and their results demonstrated that ribosome-

mediated folding process lacked an aggregation-prone non-native intermediate.

Thus, cellular conditions provide factors that maximize folding efficiency and

minimize non-native interactions. This review will next concentrate on two

structure-stabilizing factors that cells utilize, molecular chaperones and folding

catalysts for OPF.

2.1.4 Molecular chaperones

A general description (Hartl et al., 2011) defines a molecular chaperone as

any protein that interacts with, stabilizes, or helps another protein to acquire

its functionally active conformation, without being present in its final structure.

Molecular chaperones bind to hydrophobic regions of folding substrates and

prevent their aggregation (Hartl et al., 2011) in a way that was classically

considered not to “appreciably accelerate folding” (Puig & Gilbert, 1994). However,

the latter aspect is a matter of debate, as certain chaperones under some

circumstances seem to considerably accelerate folding (for example, Brinker et al.,

2001; Tang et al., 2006; Apetri & Horwich, 2008; Libich et al., 2015a; Libich et al.,

2015b). To avoid ambiguity, this study draws the line between “folding factor” to

describe any molecule that promotes protein folding and “folding catalyst” that

significantly accelerates the kinetics of the rate-limiting steps of peptide bond

isomerization (peptidyl-prolyl isomerases) or formation of a new covalent bond

(e.g., disulfide formation). “General chaperones” are then used in the classical way

32

to define any factors such as the heat-shock proteins (HSP) and chaperonins

(described below) that bind to hydrophobic residues, prevent aggregation, and

assist folding – regardless of whether they also accelerate folding. “Co-chaperones”

are also a class of folding factors, but they typically bind general chaperones and

enhance their activity.

General chaperones form two main classes based on their utilization of

adenosine triphosphate (ATP). The first class can bind and hydrolyze ATP to cycle

between low or high affinity states to allow controlled binding, release, and

rebinding of folding substrates (Hartl & Hayer-Hartl, 2009). ATP-dependent

chaperones include members of the HSP family (Kim et al., 2013) and chaperonins

that form a cylindrical cage for confined substrate folding (Hartl et al., 2011). The

two systems (HSP and chaperonins) are not redundant, as they likely operate in a

sequential manner: HSP family members interact with early-stage folding

intermediates and the chaperonins downstream when the substrate fails to attain

native state (Langer et al., 1992). The second class consists of ATP-independent

chaperones which mainly “buffer aggregation” and solubilize substrates by binding

and releasing hydrophobic regions in an ATP-independent manner (Hartl et al.,

2011). These proteins include many chaperones important for co- and post-

translational folding processes such as the bacterial trigger factor or the eukaryotic

nascent polypeptide-associated complex, that bind nascent polypeptides in the

ribosomal exit tunnel or as soon as they emerge from the ribosome, delaying chain

collapse until sufficient number of residues are available for productive folding

(Ferbitz et al., 2004; Kaiser et al., 2006; Gamerdinger et al., 2019). Another peculiar

molecular chaperone involved in OPF and oligomeric assembly of

immunoglobulins is introduced later in this study (Chapter 2.6.2).

2.2 Biochemistry of disulfide formation

A disulfide is a covalent bond linking the sulfur atoms of two thiol groups. A protein

disulfide connects cysteines either within the same polypeptide chain

(intramolecular), or between different chains (intermolecular or mixed disulfide).

Disulfide formation (and reduction) deals with electron flows, from an electron

donor (nucleophile) to an electron acceptor (electrophile). Spontaneous disulfide

formation by a nucleophilic attack of a free thiol on another thiol is kinetically a

slow process, with high activation energies. Disulfide formation is therefore often

the rate-limiting process for OPF (Nagy, 2013). This part of the review will

introduce the most important chemistries and mechanisms that enhance the

33

reactivity of thiols and kinetics of disulfide bond formation, to allow disulfides to

form in a physiologically relevant efficiency and time scale.

2.2.1 Disulfide formation is catalyzed redox chemistry

Disulfide bond formation is a reduction-oxidation (redox) reaction that transfers

two electrons from the dithiol to an external electron acceptor. As a result, the

dithiol is oxidized to disulfide and the electron acceptor gets reduced. The electron

acceptor can be a low-molecular-weight oxidant (LMWO) such as molecular

oxygen (O2) or hydrogen peroxide (H2O2), but often another pair of thiols acts as

the acceptor. The underlying physicochemical property that drives biological redox

reactions forward and describes the direction of the electron flux is called biological

standard reduction potential (measured in millivolts). A comprehensive review on

redox reactions and how to derive reduction potentials for a given redox pair are

reviewed by Hatahet & Ruddock (2009). The reduction potential describes only

whether a certain reaction is thermodynamically favorable to take place, but not

about kinetics of the reaction. A species with a higher reduction potential prefers to

gain electrons, i.e., oxidizes the species with a lower reduction potential (Fig. 4A).

Similarly, the redox state of an intracellular organelle can be derived by measuring

the reduction potential of a buffer component in the given compartment. Typically,

the redox potential of glutathione, which exists in various organelles as a mixture

of reduced glutathione (GSH) and its oxidized form (GSSG), is measured. In such

measurements, a cytosolic redox state is usually described to be reducing and

unfavorable for disulfide formation, and that of the ER is more oxidizing favoring

disulfide bond formation and OPF (Hwang et al., 1992; Bass et al., 2004; Dixon et

al., 2008). See Fig. 4B for examples of experimentally derived reduction potentials.

Glutathione and its buffering-capabilities are reviewed later in Chapter 2.3.3.

34

Fig. 4. Redox chemistry of disulfide formation. (A) Schematic of a reduction-oxidation

reaction with electron transfer from a redox couple with a lower reduction potential

(thiol/disulfide) to one with a higher reduction potential (O2/H2O2). -SH, thiol; -S-S-,

disulfide; O2, molecular oxygen; H2O2, hydrogen peroxide. (B) Experimentally derived

reduction potentials of low-molecular-weight oxidants and proteins as well as estimated

reduction potential ranges for selected organelles relevant for this work (Cleland, 1964;

Rost & Rapoport, 1964; Rakauskiene et al., 1989; Hwang et al., 1992; Lundstrom &

Holmgren, 1993; Huber-Wunderlich & Glockshuber, 1998; Dixon et al., 2008; Hatahet &

Ruddock, 2009; Ruddock, 2012; Birk et al., 2013; Romero et al., 2018). DTT, dithiotreitol;

DsbA, Escherichia coli thiol:disulfide interchange protein; ER, endoplasmic reticulum;

GR, glutathione reductase; NADPH, reduced nicotinamide adenine dinucleotide

phosphate.

The reduction potential of O2 is very high (Fig. 4B) which suggests that direct O2-

driven disulfide formation could form a major OPF pathway in aerobic cellular

conditions. However, this is probably not the case, as the kinetics of electron

transfer between a pair of thiols and free O2 is too slow to maintain life alone

(Ruddock, 2012). Therefore, in addition to thermodynamics, relative rates of

disulfide formation of different oxidants must be considered before their

physiological relevance can be assessed. Reported disulfide formation rates (Table

1) suggest that LMWOs oxidize thiols to disulfides rather inefficiently, and they

likely have a supporting role in disulfide formation in vivo. Catalysis can

significantly improve their oxidative effects: for example, disulfide formation by

35

O2 can be improved by transition metal ions and enzyme-bound redox cofactors

(Fass & Thorpe 2018; Romero et al., 2018). The concentration of free transition

metals is low in vivo, making them insignificant disulfide catalysts in the cell

(Ruddock, 2012). Flavin adenine dinucleotide (FAD), while inefficient free catalyst

in solution (Table 1), is a common bound cofactor in many oxidases (for example,

flavin-dependent sulfhydryl oxidases) and accelerates disulfide formation

significantly. Another important LMWO that greatly benefits from catalysis is H2O2

(Table 1). Flavin-dependent sulfhydryl oxidases and peroxidases are discussed in

more detail in Chapter 2.3.3.

Table 1. Comparison of disulfide formation rates (reported second order rate constants)

of selected oxidative systems relevant for this work.

Thiol/cofactor Electron acceptor Rate (M-1s-1) Reference

Free FAD O2 250 Massey (1994)

Enzyme-bound FAD O2 104 - 106 Mattevi (2006),

Romero et al. (2018),

and references

therein

BPTI H2O2 5 Karala et al. (2009)

Cys, DTT, GSH H2O2 18 - 26 Winterbourn &

Metodiewa (1999)

PDIa H2O2 9.2 Karala et al. (2009)

PrxIV H2O2 2.2 x 107 Wang et al. (2012)

GPx7 H2O2 2.6 x 103 Wang et al. (2014)

PDI DHA 12.5 Saaranen et al.

(2010)

BPTI DHA 163 Saaranen et al.

(2010)

BPTI Glutathione 7.3 Karala et al. (2009)

DsbA DsbB 2.7 x 105 Grauschopf et al.

(2003)

There are multiple factors influencing the reported higher enzymatic disulfide

formation rates. These include enhanced thiol nucleophilicity, optimal substrate

orientation and proximity relative to the catalytic center, and thiol-disulfide

exchange reactions, each of which are discussed next. All in all, correct disulfide

bond formation in vivo is a catalyzed redox-sensitive reaction. This sensitivity to

the redox conditions results in different organelles having low to high propensity

for disulfide formation.

36

2.2.2 The pKa value determines the reactivity of a thiol group

The rate of disulfide formation is first and foremost a function of three structural

and chemical factors: i) reactivity, ii) accessibility, iii) and proximity of thiol sulfur

atoms (Wedemeyer et al., 2000; Qin et al., 2015). Reactivity is discussed first and

accessibility and proximity later in Chapter 2.2.3. A disulfide is rarely formed by

non-charged thiol groups. A thiol can form a negatively charged state, a thiolate,

which is more efficient nucleophile (up to 1010-fold), and almost always the

preferred state during a nucleophilic attack (Fass & Thorpe, 2018). The thiol-

thiolate equilibrium is determined by the pKa value which specifies the pH of the

solution at which the thiol has a probability of 50% to exist either as a thiol or a

thiolate [see Hatahet & Ruddock (2009) for instructions on how to derive pKa

values]. If pKa is higher than pH, the equilibrium is driven towards the protonated

non-charged thiol state, while pKa less than pH will drive the group towards the

more reactive thiolate state (Fig. 5A). Thus, in general, lower pKa translates to

higher reactivity at physiological pH, but some exceptions to this rule exist, and the

nucleophilicity of each case must be carefully evaluated (Fass & Thorpe, 2018).

pKa values may vary between thiol groups. At physiological conditions (pH

circa 7), certain thiols are mainly in the reactive thiolate state (low pKa), whereas

most adopt the less-reactive thiol form (high pKa). An important modulator of pKa

is an adjacent charged group: a positive charge drives the equilibrium towards the

thiolate state, and a negative charge towards the thiol state (Fig. 5A) (Hatahet &

Ruddock, 2009). A given thiol may interconvert between the two states due to

protein dynamics, i.e., conformational changes and movement of other charged

residues can cause pKa fluctuations in proteins. This is especially important in the

catalytic mechanism of PDI (Lappi et al., 2004) which is described in more detail

in Chapter 2.5.

37

Fig. 5. The effect of pKa on the thiol:thiolate conversion. If the pKa value of the thiol

group is near the pH of the solution (the middle reaction), an equilibrium is reached

between the thiol and the thiolate states. pKa lower than pH will drive the equilibrium

towards the more reactive thiolate (upper reaction). pKa can be lowered by stabilization

of the thiolate state by a nearby positive charge. Inversely, pKa higher than pH favors

the uncharged thiol and a negatively charged group stabilizes the uncharged thiol

group to avoid repulsion between charged groups. The loss or gain of a proton during

the interconversion reactions is not depicted. (B) Thiol-disulfide exchange is a

bimolecular nucleophilic substitution (SN2) reaction. A nucleophilic sulfur atom

(cysteine residues are yellow, sulfur atoms are spheres) attacks a disulfide bond (in this

hypothetical case, a Cys53-Cys56 oxidized active site of PDI a domain) in line forming

a linear SSS transition state (Nagy et al., 2013). The product is a mixed disulfide between

Cys53 and the nucleophile and a free Cys56 thiol. The mixed disulfide is resolved by

another nucleophile with similar linear coordination chemistry (not depicted). Of note,

the chemistry of disulfide oxidation by low-molecular-weight oxidants is under debate

and may not proceed with the described mechanisms (Zeida et al., 2012). The illustration

has been drawn using PDI from PDB structure 6I7S.

38

2.2.3 Thiol-disulfide exchange reactions

A disulfide can form by a coordinated thiol-disulfide exchange reaction. Instead of

passing the electrons to an LMWO, here, electrons are transferred between two

pairs of thiols. The consensus disulfide exchange reaction proceeds with an SN2

substitution mechanism, in which the attacking sulfur atom approaches in line to

the disulfide (Fig. 5B; Fass & Thorpe, 2018). This mechanism is utilized by

oxidoreductases (but is not restricted to them) such as PDI family members, which

contain a solvent exposed thiol and a buried thiol in the active site (in the reduced

state) aligned optimally for an incoming in-line nucleophilic attack (Fig. 5B).

Therefore, efficient disulfide exchange is a series of coordinated chemistries,

including the structural orientation of the involved thiol sulfur atoms that likely

optimizes the steric requirements (orientation and proximity) of the disulfide

transfer and active modulation of pKa/nucleophilicity of the participating thiols,

that together explain the increased enzymatic disulfide transfer rates (Table 1).

The exchange occurs in two substitution steps (equations 1 and 2): in the first

step, thiolate 3 (-SR3) resolves the original disulfide between thiols 1 and 2 and a

mixed disulfide between thiols 2 and 3 is formed; the mixed disulfide is then

resolved in the second step by thiolate 4, creating a new disulfide between thiols 3

and 4.

𝑅 𝑆˗𝑆𝑅 𝑆𝑅 ↔ 𝑅 𝑆 𝑅 𝑆˗𝑆𝑅 (1) 𝑅 𝑆˗𝑆𝑅 𝑆𝑅 ↔ 𝑅 𝑆 𝑅 𝑆˗𝑆𝑅 (2)

No net disulfide formation takes places (as was described already, de novo

generation of disulfides generally requires direct or indirect participation of a

LMWO). Note also that both reactions are drawn bidirectional, as the intermediate

mixed disulfide and the transferred disulfide could theoretically be attacked by the

product thiolates. However, as already discussed in Chapter 2.2.2, the reactivity of

the participating thiols is often actively modulated to prevent the reverse reactions.

Disulfide exchange is an important mechanism for productive OPF (Fig. 6),

but exchange reactions can result in non-productive reactions which may lead to

misfolding and even aggregate formation (see the off-pathway reaction in Fig. 6).

Non-native disulfides can be corrected by reduction-oxidation cycles and by

intramolecular or intermolecular isomerization. Catalyzed intramolecular

isomerization is a physiologically important function to correct non-native

disulfides and is typically carried out by isomerases such as PDI. It is both a crucial

part of productive OPF as well as a countermeasure against disulfide-linked

39

aggregation and protein aggregation diseases (Mossuto, 2013). Intermolecular

rearrangement is not limited to isomerases, though, as it may modulate the

dynamicity of some protein assemblies, for example, disulfide-rich hair keratin and

hemostasis factors to combat mechanical stretching and shearing (Ganderton et al.,

2007). All intermolecular exchange mechanisms utilize a mixed disulfide

intermediate complex, which is an important indicator of disulfide transfer between

proteins, as these intermediates can often be trapped and identified (for example,

Molinari & Helenius, 1999; Benham et al., 2000; Bass et al., 2004).

Thiol-disulfide exchange is an important mechanism for several other

physiological processes. For example, peroxidases couple antioxidant defense

(reduction of H2O2 to water) to exchange reactions (Chapter 2.3.3). Transient

formation and resolution of disulfides may modulate enzyme activity (Chapter

2.6.1) as well as redox signaling processes (Li & Camacho, 2004; Ushioda et al.,

2016).

40

Fig. 6. Thiol-disulfide exchange mechanisms adapted from Hatahet & Ruddock (2009).

Disulfide transfer to a reduced dithiol starts with a nucleophilic attack on the disulfide

by a thiol in the reduced species, forming a mixed disulfide intermediate complex. The

productive oxidative pathway (box with an intact border) continues by resolution of the

mixed disulfide by a free thiol. A second similar reaction leads to a native state with two

disulfides. If the initial mixed disulfide is resolved by a non-native thiol, one of two non-

productive reactions may happen: A reverse reaction (not depicted) may regenerate the

starting material, or a non-native disulfide may form. The former reaction would be futile

and can be controlled by modulating the pKa of the nucleophilic thiol. The latter can be

corrected by isomerization or by reduction-oxidation cycles to first break the non-native

disulfides and then return to the productive oxidative pathway. Isomerization can either

be an intramolecular rearrangement, which requires a free thiol in the misfolded protein

or an intermolecular reaction which utilizes an external thiol. The latter can potentially

rearrange a fully oxidized misfolded system to the native state.

41

2.3 Oxidative protein folding in vivo

OPF is an enzyme-catalyzed process in vivo. At a macroscopic level, the catalytic

systems between species of the kingdoms of life seem similar (Fig. 7). Disulfides

are generated de novo by oxidoreductases, that pass electrons from their active site

dithiol ultimately to O2. The newly generated disulfides are then transferred to a

central oxidoreductase that catalyzes OPF of a nascent polypeptide. On a

microscopic level, however, the systems have stark differences, from differing

electron transfer pathways to diversity of OPF components. This chapter provides

an overview of disulfide formation in prokaryotes, eukaryotic mitochondria, and

finally the main topic of this study, the complex disulfide bond formation network

of the human ER.

2.3.1 Disulfide formation in prokaryotes

Bacteria generate disulfides mainly in the more oxidizing extracytoplasmic space,

specifically, the intermembrane space called periplasm in Gram-negative bacteria

or within the boundaries of the cell wall and the cytoplasmic membrane of Gram-

positive bacteria (Kadokura & Beckwith, 2010; Sarvas et al., 2004). After

translation in the cytoplasm and translocation to the extracytoplasmic space, OPF

of a nascent polypeptide is carried out by two potential pathways: 1) the Dsb family

in the periplasm of Gram-negative bacteria, or 2) the vitamin K system found in

both the Gram-negative and Gram-positive bacteria. DsbA, DsbB, DsbC, and DsbD

are the best characterized Dsb family members, and together they form the main

pathway for prokaryotic OPF. The DsbA-DsbB pair transfers disulfides to nascent

polypeptides, and DsbC-DsbD forms an isomerase unit to correct non-native

disulfides (Fig. 7). DsbA, which has one of the strongest reported redox potentials

of an oxidoreductase, -120 mV (Fig. 4B), is the central oxidoreductase that transfers

disulfides to folding substrates. It is re-oxidized by the membrane protein DsbB

which generates disulfides de novo, transfers them to DsbA, and regenerates its

oxidized form by passing the electrons to an electron transport chain consisting of

a quinone molecule, cytochrome c oxidase (COX), and O2 as the terminal electron

acceptor (Fig. 7). The DsbD-DsbC pair acts inversely so that DsbD reduces DsbC

which in the reduced free-thiol form may carry out reduction and isomerization of

non-native disulfides. DsbD is then reduced by cytoplasmic thioredoxin system for

a new catalytic cycle (Rietsch et al., 1997).

42

Fig. 7. The major pathways for OPF in the periplasm, the mitochondrial intermembrane

space, and the endoplasmic reticulum. Vertical and tilted arrows depict electron transfer

reactions, while horizontal arrows convert one chemical species to another. The middle

panel depicts substrate folding from unfolded to native state. This is achieved either by

direct oxidation by the oxidative pathways (lower panel) or by the corrective pathways

that rearrange (isomerization) or reduce non-native disulfides (upper panel). The lower

panels show electron transfer reactions from a central oxidoreductase (DsbA, Mia40, or

PDI) to terminal electron acceptors (O2 or H2O2). The cytoplasmic reducing pathways

that are connected to the ER are shown at the top. The reductive system of the ER is

currently under investigation with recent reports suggesting that a cytoplasmic

thioredoxin reduces an unknown ER membrane protein (MP) that may control the redox

state of the PDI reductases (Poet et al., 2017; Cao et al., 2020).

43

The vitamin K system was characterized independently by two groups (Dutton et

al., 2008; Singh et al., 2008). They showed that an atypical membrane protein was

able to catalyze disulfide formation in the extracytoplasmic space of a Dsb-

deficient prokaryotic species as well as E. coli artificially made devoid of the Dsb

pathways. The vitamin K system seems to have evolved independently of the Dsb

system (Dutton et al., 2008), but shares similarities in the catalyzed reactions: the

membrane protein vitamin K epoxide reductase (VKOR) oxidizes a DsbA-like

protein and subsequently passes the electrons to the electron transport chain via

vitamin K (a quinone derivative). VKOR has peculiar homologs that may catalyze

disulfide bond formation in the cytoplasm, e.g., in extremophiles that require

disulfides for thermal adaptation. Their catalytic components reside on the

cytoplasmic side (Hatahet & Ruddock, 2013). These rare cases of cytoplasmic

disulfide bond formation are out of scope for this study; see Saaranen & Ruddock

(2013) for more information.

2.3.2 Mitochondrial relay for disulfide bond formation

The best-characterized mitochondrial system for disulfide bond formation is found

in the intermembrane space (IMS) (Fischer et al., 2013). The central oxidoreductase

is Mia40, which introduces disulfides to imported nascent polypeptides (Fig. 7).

Mia40 is re-oxidized by essential for respiration and vegetative growth protein 1

(Erv1, in yeast) or augmenter of liver regeneration (ALR, in human), both flavin-

dependent sulfhydryl oxidases. Erv1 and ALR then pass electrons to the already

described cytochrome c/COX/O2 electron transport chain. Whereas the bacterial

Dsb system has a well-characterized pathway for disulfide isomerization, the

proofreading system of the mitochondrial IMS is still under debate. Glutathione,

which is possibly kept in the reduced state by glutathione reductase (Kojer et al.,

2012), has been shown to increase the efficiency of the Mia40-Erv1 relay,

suggesting that it may directly reduce non-native disulfides in folding intermediates

(Bien et al., 2010). On the other hand, Mia40 shows isomerase activity in vitro, and

it exists as a mixture of redox states in vivo, which together suggest that Mia40 may

carry out both oxidation and isomerization in the IMS (Kojer et al., 2012; Koch &

Schmid, 2014). Possibly, both systems contribute to and are crucial for the quality

control of disulfide formation in the IMS, similarly to the Ero1-PDI-glutathione

system found in the ER (Fig. 7; see Chapter 2.3.3.).

44

2.3.3 Multiple ways to form disulfides in the ER

40% of all human genes encode for proteins that are translocated to start folding in

the ER (Uhlen et al., 2015). This large proteome includes secreted hormones,

immunoglobulins, plasma membrane proteins, extracellular matrix proteins, and

other proteins that permanently reside within the organelles of the secretory

pathway (the ER, the Golgi apparatus, lysosomes, and the vesicles travelling

between them). A recent genome-wide analysis combined quantitative

transcriptomics to immunohistochemistry and demonstrated that as many as two-

thirds of all transcripts of the pancreas and the salivary glands may encode for

secreted proteins (Uhlen et al., 2015). The burden of protein synthesis and disulfide

formation in the ER may therefore reach extremely high levels, and it comes with

no surprise that the ER has developed an elaborate protein folding network of

multitude of folding factors.

The prokaryotic extracytoplasmic spaces and the mitochondrial IMS each had

clear major pathways for OPF. Dissecting the ER for major and minor pathways is

more complicated, however. GSSG was originally suggested to be the main source

of disulfides in the ER (Hwang et al., 1992). This view was later challenged when

a novel sulfhydryl oxidase, Ero1, was discovered that seemed to fuel PDI-driven

OPF (Frand & Kaiser, 1998; Cabibbo et al., 2000; Pagani et al., 2000). The

consensus main oxidative pathway shifted to the Ero1-PDI relay, while glutathione

was given an indirect role in disulfide formation by controlling the redox state of

other proteins (Fig. 8A). The plausibility of this new model was supported by

essentiality of the single Ero1 gene in yeast (Frand & Kaiser, 1998) and by

biochemical experiments carried out in human cells, demonstrating that Ero1 and

glutathione constitute the central element for ER redox homeostasis (Appenzeller-

Herzog et al., 2010). The story is not that straightforward though: the single gene

for Ero1 is non-essential in Drosophila melanogaster (Tien et al., 2008), and

deletion of both Ero1 isoforms creates viable double mutant mice (Zito et al.,

2010a). These observations argue that the eukaryotic ER contains an elaborate

network of Ero1-independent OPF pathways that may substitute the Ero1-PDI

pathway, at least in some organisms. The major pathways for disulfide formation

in the ER, that are formed by the PDI family members, the flavin-dependent

sulfhydryl oxidases, peroxidases, and important LMWOs, are reviewed next.

45

Fig. 8. The major pathways for disulfide formation in the human ER. (A) Schematic of

the main Ero1-PDI pathway for disulfide formation, and other major pathways that either

connect to the PDI hub or bypass it. (B) PDI family members and their domain

architecture. White box, active site or active site-like domain with active site (or -like)

motif marked; grey box, structural or substrate-binding domain; black oval,

transmembrane domain; white oval, J domain. The flexible x-linker, that is considered

to modulate the accessibility of the substrate-binding site (Nguyen et al., 2008), is

shown as a small black box between domains b' and a'.

46

OPF hub in the ER: the PDI family

PDI family members are at the center of OPF in the ER by connecting many

important oxidative pathways (Fig. 8A). The PDI family consists of more than

twenty oxidoreductases (Fig. 8B), and it is defined by localization to the ER and

similarity to the canonical PDI (from now on, the word PDI refers to this

archetypical PDI family member, and the family is discussed as PDI family). They

catalyze oxidation, reduction, and isomerization of disulfides utilizing a Cys-X-X-

Cys (where X is non-cysteine and can be the same or different amino acid) active

site motif. However, a few peculiar human PDI family members exist that have

either a Cys-X-X-Ser or Ser-X-X-Cys active site or do not have an active site at all

(Fig. 8B), and they are likely limited to isomerization or chaperone functions, only

(Cai et al., 1994; Appenzeller-Herzog & Ellgaard, 2008b; Hatahet & Ruddock,

2009; Kober et al., 2013).

Active site reduction potentials vary between PDI family members, and some

members may prefer to oxidize substrate dithiols, while others act primarily as

reductases. Although the structural aspects of PDI are reviewed in detail in Chapter

2.4.2, it is worthwhile to discuss already here the underlying determinants that

modulate the active site reduction potential and define the thiol-disulfide exchange

function of the PDI family member. Firstly, a charged residue in between the active

site cysteines has a major impact on defining the role of a PDI family member. For

example, a positively charged histidine lowers the pKa of the N-terminal active site

thiol. This may result in a destabilized active site disulfide, an increase in the

reduction potential, and a preference to oxidize other thiols (Hatahet & Ruddock,

2009). Inversely, if the positively charged histidine of the canonical CGHC motif

is mutated to a non-charged proline, the reduction potential decreases significantly

from ~-170 to ~-230 mV (cf. Fig. 4; Chambers et al., 2010). Secondly, the bulkiness

of the juxtaposed amino acids may influence the reduction potential of the active

site. A bulky proline next to the N-terminal cysteine (CPXC) may strain and

destabilize the active site disulfide, again favoring the oxidative function. Of note,

a proline in the second position does not destabilize the active site disulfide in a

similar manner, as reductases with a stable active site disulfide may have a CXPC

active site motif (Chivers et al., 1997; Hatahet & Ruddock, 2009). To summarize,

stability of the active site disulfide, which is strongly influenced by chemistries of

the juxtaposed amino acids, is the defining factor in deciding whether a PDI family

member primarily acts as an oxidase (unstable active site disulfide, prefers to

47

donate it away) or a reductase (stable active site disulfide, prefers to reduce others

by gaining a disulfide from them).

The exact physiological role of most PDI family members is still unclear. The

select few, that have been characterized, demonstrate a wide variety of functions

and tissue distributions:

1. PDI is a generalist, capable of each of the three described thiol-disulfide

reactions (reduction, oxidation, and isomerization). It exists as a mixture of

different redox states in the ER (circa 50% in fully reduced state and 50% in

semi- or fully oxidized state) (Appenzeller-Herzog & Ellgaard; 2008a).

2. ERp57 is critical for OPF of glycoproteins, and this function is strongly

influenced by interaction with the ER lectins calnexin and calreticulin

(Molinari & Helenius, 1999; Frickel et al., 2002; Pollock et al., 2004).

3. PDIp shows a peculiar tissue enrichment in the pancreas and the brain (Desilva

et al., 1996; Conn et al., 2004).

4. P5 controls the activity of some ER quality control signaling pathways by

reducing disulfides in the signaling components (Eletto et al., 2014).

5. ERp72 is a component of large multiprotein chaperone complexes involved in

biogenesis of immunoglobulins (Meunier et al., 2002).

6. ERdj5 is a dedicated reductase with three CXPC motifs involved in protein

quality control, namely endoplasmic reticulum-associated protein degradation

(ERAD) (for more information, see Chapter 2.5.4.) (Ushioda et al., 2008).

These examples demonstrate that the multitude of PDI family members has evolved

to have differential client repertoire to take part in different quality control features

of the ER - and not just to specialize in certain thiol-disulfide exchange reactions

for OPF.

Flavin-dependent sulfhydryl oxidases

Flavin-dependent sulfhydryl oxidases couple direct reduction of O2 within a FAD

redox center to de novo disulfide formation (Romero et al., 2018). This process

forms equimolar H2O2 (equation 3).

𝑂 2 ˗𝑆𝐻 → ˗𝑆˗𝑆˗ 𝐻 𝑂 (3)

They are net generators of disulfide bonds, whereas many other oxidoreductases

such as PDI family members solely exchange disulfides between thiols. Eukaryotic

flavin-dependent sulfhydryl oxidases have three major subfamilies: Erv1/ALR,

48

quiescin sulfhydryl oxidases (QSOX), and Ero1 (Fass, 2008; Kodali & Thorpe,

2010; Araki & Inaba, 2012). Whereas the Erv1/ALR and QSOX members reside

outside the ER (for example, Erv1p and ALR in the mitochondrial IMS and QSOX

mainly in the post-ER organelles of the secretory pathway or the extracellular

space), all Ero1 isoforms are ER-resident sulfhydryl oxidases. Saccharomyces

cerevisiae, Caenorhabditis elegans, and D. melanogaster express only one isoform

and most mammals, including Homo sapiens and Mus musculus, have two genes

for Ero1 that encode isoforms Ero1α and Ero1β. The main reasons for the

requirement of a second isoform have not been established explicitly. The isoforms

have differential tissue distribution under basal OPF levels: Ero1α is found from a

variety of tissues with high enrichment in esophagus, while Ero1β is enriched

highly in pancreas and other exocrine and endocrine tissues, including stomach and

testis (Pagani et al., 2000; Dias-Gunasekara et al., 2005; Uhlen et al., 2015). Under

stress conditions, both Ero1α and Ero1β have been suggested to be upregulated by

the unfolded protein response (UPR), but by different mechanisms (see Chapter

2.6.2). Importantly, Ero1α has been reported to be upregulated by hypoxia (May et

al., 2005; Gess et al., 2003), but the biochemistry of human Ero1 isoforms at low

oxygen concentrations has not been previously characterized. Despite the

differences, both isoforms seem to have a limited substrate repertoire in vivo, as

mixed disulfides have been reported to form with PDI family members only

(Benham et al., 2000; Mezghrani et al., 2001; Appenzeller-Herzog et al., 2008;

Appenzeller-Herzog et al., 2010). In vitro, PDI is the preferred substrate of all tested

PDI family members (Araki et al., 2013).

QSOX is the other flavin-dependent sulfhydryl oxidase that may exist in the

human ER. The uncertainty of its localization derives from two observations:

Firstly, QSOX contains a clear ER-targeting signal peptide, but it seems to traffic

to various intra- and extracellular places such as secreted fluids, constituents of the

extracellular matrix, plasma and nuclear membrane, and the Golgi (Kodali &

Thorpe, 2010). Secondly, it lacks an ER retention signal, and apparently only one

report by Tury and coworkers (2004) suggests that the membrane-associated form

may be retained in the ER. QSOX is expressed as several poorly characterized

isoforms and splice variants, including a membrane bound form and a cleaved

soluble form, so its localization to the ER is plausible (Kodali & Thorpe, 2010).

Why would the ER express multiple different flavin-dependent sulfhydryl

oxidases? One clear denominator that may eliminate redundancy is substrate

specificity. In contrast to Ero1 isoforms, QSOX seems to have low PDI-oxidase

activity, and it prefers to directly oxidize a variety of reduced unfolded protein

49

substrates, instead (Rancy & Thorpe, 2008). This segregation would allow targeted

oxidation of nascent polypeptides without redundant consumption of reduced PDI

which is required for isomerization – and it would also stress the role of Ero1 as

the main regulator of PDI redox state. Secondly, the structure of QSOX is peculiar:

it contains two thioredoxin-like domains connected to an Erv/ALR domain via a

helix-rich region. The first of the two thioredoxin-like domains contains the

canonical CGHC motif, found also in many PDI family members (Fig. 8B). The

Erv/ALR domain contains two additional CXXC motifs. Mechanistically, it is the

CGHC motif in the thioredoxin-like domain that transfers a disulfide to a folding

substrate. The CGHC motif is then reoxidized by a CXXC motif in the Erv/ALR

domain, and the catalytic site in the Erv/ALR domain is finally reoxidized by FAD

and O2. All in all, the differential substrate specificity, and the structural features of

QSOX would suggest that this enzyme has evolved to fulfill both roles, the central

oxidoreductase and the sulfhydryl oxidase to directly engage in substrate oxidation,

bypassing the PDI hub. This would allow the PDI pathway to remain specifically

under the control of Ero1. Whether QSOX is found in sufficient amounts in the ER

to form an Ero1-independent pathway under physiological conditions is to be

established.

Peroxidases

The sulfhydryl oxidases produce reactive oxygen species (ROS), namely H2O2 as

a side product of their catalytic cycle. In the ER, Ero1 is the main source for H2O2,

and this ROS remains in the ER without leaking to the cytosol (Harding et al., 2003;

Ramming et al., 2014; Konno et al., 2015). Accumulation of H2O2 may lead to

oxidative damage and protein misfolding that can be relieved by externally supplied

antioxidants (Malhotra et al., 2008; Karala et al., 2009). Therefore, it is feasible that

ER must have developed its own antioxidant defense to counter the unspecific

oxidative damage by H2O2.

Recently, three ER-resident peroxidases, peroxiredoxin IV (PrxIV) and

glutathione peroxidases (GPx) 7 and 8, have attracted attention as the main

scavengers of H2O2 in the ER. They very efficiently (Table 1) neutralize H2O2 to

water and form a new disulfide in the process (equation 4) (Utomo et al., 2004;

Tavender et al., 2010; Zito et al., 2010b; Nguyen et al., 2011).

𝐻 𝑂 2 ˗𝑆𝐻 → ˗𝑆˗𝑆˗ 2 𝐻 𝑂 (4)

50

Mechanistically, they generate a disulfide by the actions of a peroxidatic cysteine

(CP) and a resolving cysteine (CR) that are both reduced at the start of the reaction

cycle. CP reacts with H2O2 and is modified to contain sulfenic acid (CP-SOH). CR

then attacks CP-SOH forming a CP-CR disulfide, which can then be transferred to

other proteins (Tavender et al., 2008; Wang et al., 2014). As the peroxidase is in a

reduced state at the end of the catalytic cycle, and no other cellular resources are

consumed during the cycle, the net reaction (equation 5), that combines the partial

reactions by flavin-dependent sulfhydryl oxidases and the peroxidases, is two

disulfides formed from one molecule of O2.

𝑂 2 ˗𝑆𝐻 → ˗𝑆˗𝑆˗ 𝐻 𝑂 (3) 𝐻 𝑂 2 ˗𝑆𝐻 → ˗𝑆˗𝑆˗ 2 𝐻 𝑂 (4) 𝑂 4 ˗𝑆𝐻 → 2 ˗𝑆˗𝑆˗ 2 𝐻 𝑂 (5)

The only known efficient substrates for all the three peroxidases are PDI family

members (Tavender et al., 2010; Zito et al., 2010b; Nguyen et al., 2011; Wang et

al., 2014). Therefore, the peroxidatic disulfide formation is connected to the OPF

hub (Fig. 8A). Of note, the GPx naming convention is outdated and misleading, as

these two GPx isoforms do not have significant glutathione peroxidase activity

(Nguyen et al., 2011) - they are PDI peroxidases, instead.

The three peroxidases catalyze the same reaction in the same cellular

compartment. Do they have redundancy? Likely no, as the current knowledge

seems to place PrxIV in an Ero1-independent pathway, while both GPx are linked

to the Ero1-PDI pathway. Substrate specificity is again one clear denominator, as

GPx7/8 prefer to oxidize PDI, while PrxIV favors ERp46 and P5 over PDI (Nguyen

et al., 2011; Sato et al., 2013; Wang et al., 2014). Konno and colleagues (2015)

have even suggested that an unknown Ero1-independent lumenal H2O2 source

could fuel PrxIV, whereas the GPx isoforms interact directly with Ero1 and utilize

Ero1-derived H2O2 (Nguyen et al., 2011; Wang et al., 2014). The Ero1-

independency of PrxIV is further supported by knockdown studies. Loss of both

Ero1 isoforms in mice cells does not seem to markedly alter cell viability (Zito et

al., 2010a). However, mice cells lacking both Ero1 isoforms and PrxIV lose

viability (Zito et al., 2010b). These experiments are supported by yeast studies that

demonstrated that ER-localized PrxIV can rescue yeast cells with the lethal Ero1

knockout (Zito et al., 2010b). Collectively, these observations suggest that the GPx

isoforms contribute to the main Ero1-PDI pathway and PrvIV may bypass the PDI

51

hub and form an Ero1-PDI independent pathway for OPF in the ER. Speculatively,

this segregation, in combination with the above described PDI-independency of

QSOX, would place Ero1 in even stronger position to regulate the redox state of

PDI and the rates of PDI-driven oxidative protein folding.

Glutathione

Glutathione is a tripeptide (γ-glutamyl‐cysteinyl‐glycine) which contains a thiol

group. It was considered to form the main pathway for disulfide formation in the

ER before the discovery of Ero1 (Hwang et al., 1992). This is not surprising, as

glutathione is an abundant peptide in the ER (estimates range from 4 to 19 mM)

(Dixon et al., 2008; Birk et al., 2013) and capable of carrying out thiol-disulfide

exchange reactions either in the reduced monopeptide form (GSH) or the disulfide-

containing oxidized homodimer (GSSG). Both exchange reactions consist of two

steps that involve formation of a transient protein-glutathione mixed disulfide

intermediate species (Fig. 9). At first glance, the equations look like reverse

reactions of each other. However, the overall rate of reduction depends on GSH

concentration, as both steps of the overall scheme depend on [GSH]. This creates a

kinetic partitioning event for a potential reverse reaction and regeneration of the

starting material at low [GSH]. In contrast, the oxidation series has only one

direction, as the latter steps are independent of [GSSG] (Lappi & Ruddock, 2011).

Therefore, a careful control of redox conditions is a necessity to decide which thiol-

disulfide exchange reactions glutathione is to carry out.

Glutathione forms the main redox buffer in the ER. While the cytosolic GSH

to GSSG ratio has been reported to range from 30:1 to 100:1, values that translate

to a highly reducing environment, the ER is more oxidizing with GSH:GSSG ratios

of 3:1 to 6:1 (Hwang et al., 1992; Birk et al., 2013). This stark difference between

the organelles is affected mainly by two factors. Firstly, ER lacks glutathione

synthesis, and glutathione is transported to the ER in the reduced form from the

cytosol, potentially in a redox-regulated manner (Banhegyi et al., 1999; Ponsero et

al., 2017). Secondly, glutathione redox state dynamically both maintains the

reduced fraction of PDI family members (required for isomerization) and

reoxidizes target proteins (Chakravarthi & Bulleid, 2004; Appenzeller-Herzog et

al., 2010). Therefore, glutathione influences oxidation, reduction, and

isomerization of the ER disulfide pool at least indirectly via PDI family members

(Fig. 8A). To what extent it oxidizes nascent polypeptides directly in vivo is not

well understood. On one hand, glutathione reacts with thiols in unfolded proteins

52

with relatively slow kinetics in vitro (Table 1), arguing that it may not form a major

independent OPF pathway in vivo. On the other hand, the concentration of

glutathione in the ER is extremely high, which may lead to a high cumulative thiol-

disulfide exchange activity. Furthermore, the estimated basal GSH:GSSG ratio in

the ER (see above) was observed to be optimal for RNase A refolding in vitro (Lyles

& Gilbert, 1991). These observations suggest that glutathione has the capability for

efficient oxidation of protein thiols and could bypass the PDI hub to form a major

pathway for OPF (Fig. 8A).

Fig. 9. Reaction mechanisms for PDI reduction (upper reaction) and oxidation (bottom

reaction) by glutathione; adapted from Lappi et al. (2011). Both forward steps of the

reduction series are intermolecular reactions and depend on [GSH], which creates a

potential intramolecular reverse reaction from the mixed disulfide intermediate to the

starting material at low [GSH]. The oxidation series is unidirectional, as the later steps

are intramolecular reactions. The interconversion of the PDI C-terminal thiol to a thiolate

in the oxidation series has been suggested to be controlled by an intramolecular

conformational change that brings the positively charged side chain of Arg120 close to

the active site locale (Lappi et al., 2004; Karala et al., 2010; Lappi et al., 2011).

53

Vitamin-utilizing oxidative systems

Two vitamins and their derivatives potentially influence disulfide formation in the

ER (Fig. 8A). The first pathway utilizes vitamin K and the already described VKOR.

The main difference between the prokaryotic and the eukaryotic VKOR pathways

is that the former is an independent pathway, whereas the latter is coupled to

another metabolic pathway, γ-carboxylation of proteins (Stenflo et al., 1974; Furie

et al., 1999; Schulman et al., 2010). This process is driven by VKOR which reduces

vitamin K derivatives needed for the γ-carboxylation reaction. VKOR is then

recycled back to the reduced state by PDI family members which get oxidized in

the process (Schulman et al., 2010). VKOR pathway is therefore linked to the PDI-

driven OPF network (Fig. 8A). Of note, γ-carboxylation is a modification of a

minor subset of clients, and, as suggested by Ruddock (2012), this pathway is

unlikely to form a major pathway for disulfide formation. A prokaryotic-like

independent VKOR pathway could significantly contribute to the overall disulfide

pool, but such a pathway has not been discovered in the ER, yet.

The other vitamin that could influence disulfide formation is vitamin C

(ascorbate). It is an abundant molecule in the cell found possibly in millimolar

levels (Chaudiere & Ferrari-Iliou, 1999) and best known for its roles as an

antioxidant (Duarte & Lunec, 2005) and in collagen biosynthesis (Kivirikko et al.,

1989). The oxidized redox state of ascorbate, dehydroascorbate (DHA), can

additionally promote refolding of disulfide-containing proteins (Saaranen et al.,

2010; Szarka & Lorincz, 2014). Remarkably, it seems to favor oxidation of

unstructured proteins over the active site of PDI and oxidizes them more efficiently

than glutathione or H2O2 do (Table 1; Karala et al., 2009; Lappi & Ruddock, 2011).

A potential biochemical pathway for DHA formation in the ER has been described

(Zito et al., 2012). It combines reduction of protein cysteinyl-sulfenic acid

modifications to free thiols with concomitant ascorbate oxidation to DHA (Zito et

al., 2012). If the ascorbate to DHA interconversion rates are found to be sufficiently

high in the ER, possibly via the cellular pathway described, this vitamin could form

a major PDI-independent pathway for disulfide formation in the ER (Fig. 8A).

54

2.4 Structural biology of the Ero1-PDI relay

The human Ero1 isoforms have evolved to oxidize specifically PDI, which together

form a regulated key pathway for disulfide bond formation in the ER. This chapter

elucidates the important structural determinants for the specificity of the Ero1-PDI

interaction, and establishes the catalytic and regulatory elements of Ero1, which are

subsequently discussed in Chapters 2.5 and 2.6.

2.4.1 Structure of human Ero1 isoforms

Human Ero1α (445 amino acids, 52 kDa) and Ero1β (434 amino acids, 50 kDa)

share 65.4% amino acid identity and 74.8% similarity with conserved structural,

active site, and regulatory cysteines (Fig. 10A; Pagani et al., 2000). The isoforms

contain two active sites: a CXXC inner active site (Cys394/Cys397 in Ero1α and

Cys393/Cys396 in Eroβ) and a CX4C outer active site (Cys94/Cys99 in Ero1α and

Cys90/Cys95 in Eroβ). The remaining cysteines are either structural or regulatory

in nature. Cys131 in Ero1α forms a key regulatory switch for modulating Ero1

activity (Appenzeller-Herzog et al., 2008; Baker et al., 2008; Appenzeller-Herzog

et al., 2010). This function is discussed more in Chapter 2.6.1. The other proposed

regulatory cysteines, Cys208 and Cys241 in Ero1α, potentially control

conformational change to create a diffusion tunnel for the substrate O2 and the

product H2O2 (see Chapter 2.5.3; Ramming et al., 2015). Three cysteines are

controversial: Cys166 in Ero1α has been regarded non-functional, and many in

vitro studies mutate it to alanine (for example, Inaba et al., 2010; Araki & Nagata,

2011; Masui et al., 2011). However, it is potentially functional, as Appenzeller-

Herzog et al. (2008) suggested that Cys166 may form a disulfide to an as-yet-

unknown partner. Cysteines 85 and 391 form a disulfide that was initially

considered regulatory (Baker et al., 2008) but was later observed to be stable and

therefore likely structural (Araki & Nagata, 2011; Zhang et al., 2014). Whether this

disulfide is reduced under some conditions, like the counterpart in yeast Ero1p

(Sevier et al., 2007), remains to be established. The structural features of Ero1β are

poorly characterized and are typically correlated to those of Ero1α, based on high

conservation of the established residues. The isoform specific Cys262 of Ero1β is

buried and likely does not participate in intra- or intermolecular disulfide bond

(Hansen et al., 2014).

55

Fig. 10. Structural overview of human Ero1 isoforms and PDI. (A) Disulfide patterns of

the inactive states of human Ero1 isoforms. Circles depict cysteines (active site

cysteines are yellow, regulatory cysteines red, and structural cysteines or of unknown

function are grey) and black lines are disulfides. The functions of cysteines 166 and 165

have been unknown before this study, but Appenzeller-Herzog et al. (2008) suggested

that Cys166 may form a disulfide to an unknown partner. The amino acid numberings

are for full-length proteins. (B-C) Crystal structures of inactive Ero1α (PDB code 3AHR)

(B) and PDI (subunit of MTP, PDB code 6I7S) (C). Active site cysteines (and the

regulatory Cys131) and potential disulfides connecting them are colored yellow, FAD

blue, and other elements as per in Fig. 3.

Ero1α is a single-domain enzyme based on two existing crystal structures of mutant

forms that fix Ero1α either in an inactive state or in a state with increased activity,

also called hyperactive state (the inactive state is shown in Fig. 10B) (Inaba et al.,

2010). The FAD cofactor is juxtaposed to the inner active site cysteines and buried

in both structures within a four-helix bundle, a common structural element found

in many other flavin-dependent sulfhydryl oxidases, e.g., Erv1p, Erv2p, ALR, and

56

QSOX (Gross et al., 2002; Wu et al., 2003; Alon et al., 2010; Guo et al., 2012). The

cysteines in the outer active site and the regulatory cysteines 104 and 131 are in a

flexible loop (residues 90-131) which is missing in the structural model of the

hyperactive form and partially missing (residues 95-130) for the inactive state. The

inactive state confirms the existence of the regulatory Cys94-Cys131 disulfide (Fig.

10B). Two other flexible loops are absent in the structures: residues 166-172,

containing the controversial Cys166, and residues 212-238 flanked by the

regulatory cysteines 208 and 241. Other distinguishing features are two

hydrophobic residues mediating PDI interaction: Val101 near the outer active site,

and Trp272 at the tip of a β-hairpin structure (residues Arg265-His279) (Fig. 10B).

Val101 has been shown to be important for efficient PDI oxidation but not for

regulation of Ero1α activity (Zhang et al., 2019). While Trp272 is also crucial for

PDI oxidation (Masui et al., 2011), its roles for any regulatory functions have not

been reported before this work.

2.4.2 Architecture of PDI

Human PDI contains four domains, a, b, b', and a', with each domain having a

thioredoxin-like fold (defined by regular secondary structure elements β-α-β-α-β-

β-α) (Fig. 10C). PDI also contains an acidic C-terminal tail termed c and a linker

region of 19 amino acids called x between domains b' and a'. Together, the domains

and the extensions add up to 491 amino acids and 55 kDa (Freedman et al., 1998;

Alanen et al., 2003). The active sites for thiol-disulfide exchange are in the a and

a' domains. Other domains are non-catalytic: b is likely structural and b' forms the

principle substrate binding site (Fig. 10C). PDI has six cysteines of which four

(Cys53 and Cys56 in a, Cys397 and Cys400 in a’) are in the active site Cys-Gly-

His-Cys (CGHC) motifs, and the remaining two cysteines are in the b' domain and

have no reported functions, excepting S-nitrosylation of Cys343 which has a link

to neurodegenerative diseases (Ogura et al., 2020).

Three crystal structures of (near) full length mature human PDI have been

published: two monomers either in the oxidized or reduced redox states, both

lacking the flexible C-terminal tail (Wang et al., 2013); and a full length PDI as the

beta subunit of the microsomal transfer protein (MTP) (Biterova et al., 2019). The

two monomeric structures show slight conformational differences, mainly how the

a and a' domains are situated around the rigid b-b' axis. The reduced state is ‘closed’

with the a and a’ domains closer to each other, while the oxidized state is more

open and accessible for larger substrates. The open and closed conformations were

57

originally suggested to reflect the catalytic oxidation mechanism, so that in the open

oxidized state PDI binds substrates and releases them in the closed reduced state

(Wang et al., 2013). The real mechanism is likely more complicated though, as full-

length PDI bound to the large alpha subunit of MTP is in the reduced state with an

open conformation (Biterova et al., 2019). The open and closed conformations may

therefore only reflect conformational flexibility of PDI for binding substrates of

different sizes. Regardless, PDI in all the three crystal structures has a U-shaped

structure with the pocket between the domains forming the substrate-binding site

(Fig. 10C).

The PDI active site cysteines are located at the N-termini of the second alpha

helices of the a and a' domains (Fig. 5B, Fig. 10C). While the first cysteine of the

CGHC motif is solvent exposed and located at the N-terminus of the helix, the C-

terminal cysteine is partially buried having limited solvent exposure. The pKa

values of the active site cysteine thiol groups are different and influenced mainly

by two factors: 1) the structure of α-helix, as the N-terminus of the helix has a

permanent positive and the C-terminus permanent negative dipole moment causing

slight decrease in the pKa of the N-terminal cysteine (Kortemme & Creighton,

1995); 2) electrostatic effects from nearby residues (Fig. 5A), for example, from

the histidine residue within the CGHC motif which is partially protonated and has

a net positive charge at physiological pH. The N-terminal cysteine adopts a thiolate

form at physiological pH, as the lowest reported pKa is 4.5 (Kortemme et al., 1996).

In contrast, the pKa value of the C-terminal cysteine is high (up to 12.8; Lappi et

al., 2004), so it will be in the non-charged thiol form for the most part of the

catalytic cycle of PDI. However, the pKa of this residue may be modulated by

conformational changes in the catalytic domain resulting in a decrease of the pKa

value and formation of the thiolate group. This is important during reoxidation of

the active site when the C-terminal cysteine needs to resolve the mixed disulfide

bond between the PDI N-terminal cysteine and a substrate cysteine (discussed in

Chapter 2.5.2).

58

2.5 Structure-function relationship of Ero1-PDI

Efficient substrate oxidation by the Ero1-PDI relay involves intermolecular steps

between the folding substrate and oxidized PDI as well as between Ero1 and PDI

to reoxidize PDI. Intramolecular electron transfer within Ero1 and to O2 will

reoxidize Ero1 to complete the catalytic cycle (Fig. 11). This chapter compiles

reported structure-function details of each step of the catalytic cycle.

2.5.1 Substrate oxidation by PDI

PDI must first bind the substrate for efficient thiol-disulfide exchange. The binding

is mediated by hydrophobic interactions between exposed hydrophobic substrate

residues and hydrophobic residues on the surface of the U-shaped substrate-binding

pocket of PDI (Fig. 10C). b’ forms the principal binding site and is alone sufficient

to bind small substrates such as peptides. However, both a and a’ are additionally

needed for binding larger substrates, for example, non-native proteins (Klappa et

al., 1998) or subunits in protein complexes (Koivunen et al., 2005; Biterova et al.,

2019). The U-shaped binding surface can also form a hydrophobic cavity, when

two PDI monomers form a face-to-face dimer in the oxidized state, potentially

increasing the rate of OPF (Okumura et al., 2019). The binding specificity of PDI

is poorly understood. Whereas PDIp requires a single tyrosine or a tryptophan

residue for efficient binding (Ruddock et al., 2000), PDI has several low affinity

binding sites that accept different hydrophobic residues, and, consequently, it has a

broad binding-capability for non-native proteins. Of note, thiol-disulfide exchange

is not prerequisite for substrate binding, as PDI can also bind various substrates

without any disulfides (Hatahet & Ruddock, 2007).

The thiol-disulfide exchange reaction from an oxidized PDI active site to the

reduced substrate dithiol proceeds essentially with the productive pathway of the

reaction scheme depicted in Fig. 6. This involves mixed disulfide bond formation

between the N-terminal active site cysteine of PDI and a substrate thiol followed

by a subsequent nucleophilic attack of the second substrate thiol to release oxidized

substrate (Fig. 11). To avoid a potential reverse reaction from the mixed disulfide

intermediate, the reactivity of the C-terminal active site cysteine of PDI during this

step is low (high pKa).

59

Fig. 11. Molecular mechanisms of disulfide formation by the Ero1-PDI relay. Substrate

oxidation by PDI and PDI oxidation by Ero1 are carried out by the thiol-disulfide

mechanism described earlier in this study; briefly, oxidized PDI forms mixed disulfide

with a reduced substrate, a substrate thiol resolves the mixed disulfide to form an

oxidized substrate, the reduced PDI is then reoxidized by an oxidized Ero1 outer active

site. For the catalytic cycle to continue, Ero1 outer active site must be reoxidized. This

is carried out by the inner active site that gains a disulfide by de novo oxidation by

FAD/O2. For efficient PDI catalytic cycle and PDI reoxidation, the reactions depicted

must be unidirectional and any reverse reactions (small arrows) must be inhibited. The

pKa of the C-terminal thiol of PDI during substrate oxidation is very high which

eliminates one of the reverse reactions depicted. The pKa values of Ero1 thiols are not

known, but one mechanism that may control the directions of Ero1 reactions is the

mobility of the outer active site which shuttles between the surface of the protein and

the inner active site (Gross et al., 2004).

60

2.5.2 PDI reoxidation by Ero1

Reduced PDI, once dissociated from the oxidized substrate, may bind Ero1 for

reoxidation (Fig. 11). Oxidized PDI will compete for binding oxidized Ero1, but

reduced PDI has approximately 10-fold higher affinity towards the active state of

Ero1 (Masui et al., 2011). The minimal PDI fragment that interacts with Ero1 is

b’xa’ (Wang et al., 2009). The interaction is thought to be mediated by hydrophobic

interactions: Ero1 Trp272 at the tip of the β-hairpin binds to the b’ domain of PDI,

potentially to residues Phe240 or Phe304 (Masui et al., 2011); and Ero1 Val101

recruits the PDI a’ domain to the vicinity of the Ero1 outer active site (Zhang et al.,

2019).

Current knowledge suggests that human Ero1 oxidizes specifically the a'

domain of PDI (Baker et al., 2008; Wang et al., 2009). The a domain is possibly

oxidized by intramolecular electron transfer between the PDI active sites (Araki &

Nagata, 2011) and/or intermolecularly by PDI peroxidases (Wang et al., 2014).

Mechanistically, the disulfide transfer between the Ero1 outer active site and the

PDI a' domain utilizes the described thiol-disulfide exchange mechanism. An Ero1-

PDI mixed disulfide complex can be trapped and coimmunoprecipitated from

mammalian cells or abrogated by Cys94Ser mutation (Appenzeller-Herzog et al.,

2010). It is tempting to speculate that in order to release the oxidized PDI (and

avoid the reverse reaction that reforms the oxidized Ero1), the mixed disulfide

might be resolved by the same principles that were described for the resolution of

the PDI-glutathione mixed disulfide (Fig. 9). In this cycle, the positively charged

side chain of PDI Arg120 is first brought close to the PDI active site locale by a

conformation change. The pKa of the PDI C-terminal active site cysteine is lowered

as a result, and the C-terminal thiol is converted to a thiolate. The more reactive

thiolate state can then resolve the PDI-Ero1 mixed disulfide intermediate, releasing

the oxidized PDI and reduced Ero1.

2.5.3 Ero1 reoxidation

PDI reoxidation results in a reduced Ero1 outer active site. Oxidation of Ero1

combines de novo disulfide formation in the core of Ero1 to thiol-disulfide

exchange reaction to pass the newly generated disulfide from the inner active site

to the outer active site. Therefore, reoxidation of Ero1 consists of a complex series

of chemistries, and some of the specifics are poorly understood. Therefore, for an

61

overall picture of Ero1 reoxidation, we need to review mechanisms from other

characterized flavoenzymes.

The FAD cofactor needs to be oxidized to be able to form a new disulfide to

the inner active site. A typical oxidase flavin cycle proceeds via formation of a

transient radical pair, in which the FAD is in a semi-reduced state and O2 has formed

a superoxide anion (Romero et al., 2018). Typically, the superoxide anion then

reacts to form H2O2 to oxidize FAD, but sometimes superoxide may escape. For

example, ALR releases in vitro one-third of its ROS product as superoxide

(Daithankar et al., 2012). The ratio of the side products for human Ero1 is not

known. The overall reoxidation reaction is spontaneous and irreversible as the

redox potential of the O2/H2O2 pair is significantly higher (+300 mV; see Fig. 4B)

than that of typical protein-bound flavins (+150 to -400 mV) (Romero et al., 2018).

Is O2 the only terminal electron acceptor for human Ero1? Certain sulfhydryl

oxidases are conditional oxidases switching from one electron acceptor to another

based on cellular conditions. For example, the yeast Ero1 homolog, Ero1p, is a

conditional oxidase that alternates between O2 and fumarate under aerobic and

unaerobic conditions, respectively (Kim et al., 2018). The yeast mitochondrial

sulfhydryl oxidase Erv1p consumes oxygen in the test tube (Ang & Lu, 2009; Tang

et al., 2019) but in vivo prefers to pass electrons to cytochrome c (Bihlmaier et al.,

2007; Dabir et al., 2007). Any alternative electron acceptors for human Ero1 have

not been identified, yet.

With the FAD cofactor now oxidized, the inner active site gains a disulfide by

series of chemistries that involve formation and resolution of a covalent adduct

between an inner active site thiol and FAD (Romero et al., 2018). The newly

generated disulfide can then be transferred from the inner active site to the

incoming mobile outer active site by thiol-disulfide exchange (Fig. 11).

Oxygen tunneling

O2 must reach the core of the flavin-dependent oxidase to re-oxidize FAD. Typically,

this is achieved by transient oxygen diffusion tunnels that are formed by dynamic

conformational changes (Romero et al., 2018). A diffusion tunnel has been

proposed for human Ero1α (Ramming et al., 2015). Based on molecular

simulations, the four-helix bundle surrounding the FAD moiety could undergo

conformational change to open a tunnel for substrate O2 and product H2O2 with a

gate (a disulfide between cysteines Cys208 and Cys241) at the distal side of the

enzyme. The gating mechanism could be controlled directly by PDI, which in the

62

reduced form attacks the gating disulfide opening the helix bundle for oxygen entry

(Ramming et al., 2015).

2.6 Quality control of oxidative protein folding

The eukaryotic ER requires flexibility to maintain a balance between the high

folding load and the folding capacity provided by folding factors. If the proteostasis

is destabilized, misfolded proteins may accumulate causing stress and ultimately

even disease states (Mossuto, 2013). The ER can regulate its folding homeostasis

mainly by three mechanisms: i) modulate Ero1 activity, ii) upregulate folding

factors by UPR, and iii) directly remove misfolded protein by ERAD (Fig. 12). The

following chapters describe these main quality control systems in more detail.

2.6.1 Modulation of Ero1 activity

Uncontrolled disulfide formation by Ero1 can lead to accumulation of ROS and

hyperoxidation of PDI which may lead to redox and protein-folding imbalances

and diseases (Cao & Kaufman, 2014). To avoid these, Ero1 activity can be

regulated. Two mechanisms to alter human Ero1 activity have been previously

reported: intramolecular disulfide rearrangement by the inactivation switch

(Cys131 in Ero1α and Cys130 in Ero1β) and post-translational phosphorylation. Of

the two mechanisms, the inactivation switch is more important for downregulation

of ROS and redox state of the ER, as it defines both the on/off states of Ero1.

Mechanistically, to inactivate Ero1α, Cys131 pairs to the active site Cys94 (Fig.

10A) effectively blocking the outer active site of forming a disulfide for PDI

oxidation (Appenzeller-Herzog et al., 2008; Baker et al., 2008; Wang et al., 2011).

This rearrangement process is thought to include also active site Cys99 forming a

disulfide to Cys104 (Fig. 10A; Hansen et al., 2012). The regulatory switch is

controlled directly by PDI in a feedback manner, so that increased levels of reduced

PDI in the ER lead to reduction/rearrangement of the regulatory switch, activation

of Ero1, and generation of new disulfides to the free Cys94-Cys99 outer active site

to oxidize PDI. The formation of the Cys94-Cys131 link, i.e., inactivation of Ero1

may utilize both autonomous and PDI-dependent oxidation of the regulatory

disulfides, when oxidized PDI accumulates, instead (Kim et al., 2012; Shepherd et

al., 2014; Zhang et al., 2014). Kinetic details of either process (activation or

inactivation) have been unknown before this study.

63

Fig. 12. Quality control pathways of protein maturation in the ER, from the unfolded (U)

state via disulfide-containing intermediates (I) to the native (N) state. Efficient

maturation of disulfide-containing proteins to the N state combines oxidation to form

disulfides and isomerization/reduction of (non-native) disulfides by PDI family members.

Reduction is required also for efficient ERAD to avoid accumulation of non-native

proteins and aggregate formation (not depicted). UPR may increase folding capacity of

the ER. The redox states of the ER and PDI modulate activity of Ero1 in a feedback

manner: in reducing conditions, the high levels of reduced PDI (PDIred) activate Ero1,

resulting in increased levels of oxidized PDI (PDIox). In more-oxidizing conditions,

proportion of PDIred decreases, and the levels of inactive Ero1 increase. Only the N state

is efficiently exported from the ER to the organelles of the secretory pathway, including

the Golgi, and ultimately to the extracellular space.

64

The activity of Ero1 was recently suggested to be modulated by a second activity

switch, phosphorylation of Ero1α Ser145. This switch is controlled by a kinase,

FAM20C, which phosphorylates Ero1α in Ero1α-overexpressing HeLa cells in the

Golgi apparatus (with retrograde translocation to the ER by ERp44) and mouse

mammary glands (Zhang et al. 2018). Ser145Glu mutant, which mimics

phosphorylation at this site, enhanced in vitro oxygen consumption activity as well

as J-chain folding in HeLa cells (Zhang et al. 2018). The physiological importance

of this switch for maintaining the redox balance of the ER is still unclear, as the

abundance of the phosphorylated species in vivo as well as any phosphatase for

reversibility of the regulatory switch have not been established, yet.

2.6.2 Unfolded protein response (UPR)

The mammalian ER has evolved an elaborate network of signaling cascades to

battle folding stress, collectively termed the UPR. They ensure that the redox and

protein-folding homeostases are sustained, if the direct modulation of Ero1 activity

is insufficient to maintain the balance. UPR-like mechanisms are crucial also for

other functions such as plasma cell differentiation (Janssens et al., 2014).

The mammalian UPR consists of three signaling cascades mediated by three

different membrane protein families: ATF6, which contains isoforms A and B; IRE1,

with isoforms alpha and beta, and PERK. The goal of each cascade is to transduce

signals from the ER via cytosol to the nucleus to produce various transcription

factors which activate target genes to produce proteins that aim to restore the ER

homeostasis (Fig. 12). The pathways are distinct, they work in parallel, and utilize

unique molecular mechanisms for signal transduction (Walter & Ron, 2011; Hetz

& Papa, 2018).

The first and foremost method to avoid prolonged unfolded protein stress is to

increase the folding capacity to enhance client oxidation and isomerization. This is

carried out mainly by the ATF6 and IRE1 pathways which, as an early response,

upregulate folding factors such as Ero1β (but not Ero1α), PDI family members, and

molecular chaperones as well as components to regulate lipid biosynthesis, ER

expansion and ERAD (Yoshida et al., 1998; Pagani et al., 2000; Yoshida et al., 2003;

Shaffer et al., 2004; Sriburi et al., 2004; Yamamoto et al., 2007; Adachi et al., 2008;

Bommiasamy et al., 2009; Quan et al., 2018). Secondly, IRE1 and PERK reduce

the folding load directly by inhibiting further protein translation (Harding et al.,

2000; Hollien et al., 2009). In a prolonged stress, the PERK pathway may activate

65

pro-apoptotic genes, including the gene encoding for Ero1α, to start programmed

cell death, apoptosis (Marciniak et al., 2004; Li et al., 2009).

Molecular chaperones are prominent targets of the quality control systems of

the ER, specifically, UPR and ERAD (Hetz & Papa, 2018). These targets include

BiP of the HSP70 family which is involved in wide variety functions, from folding

to signaling to protein translocation across the ER membrane (Matlack et al., 1999;

Pobre et al., 2019); GRP94 which is involved in protein folding as well as immune

and antitumor responses (Luo & Lee, 2013); ERp27, a peculiar PDI family member

upregulated by the UPR which contains only the non-catalytic b-type PDI domains

(Fig. 8B) and binds hydrophobic targets as well as ERp57 in a calreticulin/calnexin-

manner (Alanen et al., 2006; Kober et al., 2013); among numerous other ER

chaperones (Fewell et al., 2001). Important for this study, a novel chaperone,

pERp1, is introduced next which is involved in an often-neglected function of

molecular chaperones, assembly of protein quaternary structure (Makhnevych &

Houry, 2012).

pERp1

pERp1 (also known as Mzb1) is a novel 18 kDa soluble ER-resident protein. It

belongs to the canopy (CNPY) protein family which consists of predicted saposin-

like proteins (SAPLIPs). SAPLIPs are defined by an alpha-helical rich fold and six

cysteines that potentially form three disulfide bonds (Shimizu et al., 2009; van

Anken et al., 2009; Schildknegt et al., 2019). pERp1 is expressed in low levels in

B and T lymphocytes, but it is strongly upregulated by a UPR-like mechanism

during plasma cell differentiation. Its link to OPF arises from observations that it

promotes assembly and secretion of immunoglobulins IgA and IgM and influences

activation and function of some integrins (Shimizu et al., 2009; van Anken et al.,

2009; Flach et al., 2010; Andreani et al., 2018; Xiong et al., 2019). The role in IgM

biosynthesis is especially noteworthy. IgM is a challenging ER client, as it can exist

as a pentamer or hexamer of Ig subunits, i.e., it contains at least twenty subunits -

10 light chains and 10 heavy chains which are stabilized by a J chain subunit - and

by more than hundred disulfide bonds (Muller et al., 2013). Knockdown of pERp1

in plasma cells leads to a severe defect in IgM assembly and secretion, suggesting

that pERp1 plays a major role in our immune system (van Anken et al., 2009).

Mechanistic details of how pERp1 promotes OPF of immunoglobulins are still

unclear. Originally, pERp1 was suggested to be an oxidoreductase, as it contains a

CXXC motif, has slight oxidoreductase activity in vitro, and possibly

66

immunoprecipitates in covalent complexes with IgM monomers (Shimizu et al.,

2009; van Anken et al., 2009). However, the covalent nature of the interaction with

IgM was challenged by van Anken and coworkers (2009) who showed that the

presence of a reducing agent DTT did not influence the results of another

immunoprecipitation experiment. A redox state analysis of cysteine mutants

supported this, as cysteines in the CXXC motif seemed to link to other cysteines

far away in the sequence and not form a PDI-like active site. Further cellular studies

suggested that another potential role could be chaperone-assisted prevention of off-

pathway IgM assembly intermediates (heavy chain oligomers) or as a cochaperone

in GRP94 or BiP complexes (Shimizu et al., 2009; Rosenbaum et al., 2014). No

structure or direct biochemical assays for chaperone activity have been reported

before this study to further elucidate its roles as a folding factor.

2.6.3 Endoplasmic reticulum-associated protein degradation (ERAD)

The other quality control mechanisms described herein do not always ensure that

protein molecules fold successfully to the native state. In fact, protein folding is

rather inefficient, and a large portion of ER clients fail to fold correctly (Smith et

al., 2011). The ER has a special pathway called ERAD to monitor and clear out any

terminally misfolded proteins. The same mechanism is also utilized to degrade

some correctly folded and active ER-resident proteins to regulate metabolic

pathways (Foresti et al., 2013; Ruggiano et al., 2014).

ERAD consists of three steps:

1. recognition and processing of ERAD substrates,

2. retrotranslocation across the ER membrane to the cytosol, and

3. cytosolic ubiquitination and proteosomal degradation.

High efficiency of recognition is required to avoid accumulation of misfolded

proteins. On the other hand, the system cannot be overactive, removing productive

folding intermediates for degradation (Ruggiano et al., 2014). One recognition

mechanism of glycosylated proteins is well characterized: a protein complex

consisting of EDEM, ERdj5, and BiP extracts misfolded glycoproteins, but not

folding-competent glycoproteins, from the calnexin quality control cycle, adds a

mannose signal to the glycan to target it for retrotranslocation, and unfolds the

protein for the retrotranslocation and the cytosolic degradation processes

(Hosokawa et al., 2001; Molinari et al., 2003; Oda et al., 2003; Quan et al., 2008;

67

Ushioda et al., 2008). In addition to ERdj5, PDI has been implicated as a potential

reductase for ERAD processing (Molinari et al., 2002; He et al., 2015).

Is disulfide reduction mandatory for ERAD? Several biochemical and

structural studies suggest that it is the reduced unfolded state of the client that is

efficiently retrotranslocated through the ER membrane. For example, disulfide

reduction has been directly observed prior to ERAD for some substrates (Mancini

et al., 2000; Fagioli et al., 2001; Okuda-Shimizu & Hendershot, 2007), redox

conditions of the ER play a major role in controlling ERAD efficiency (Tortorella

et al., 1998; Medrano-Fernandez et al., 2014), and the retrotranslocation channel is

too narrow to fit folded proteins (Schoebel et al., 2017; Ellgaard et al., 2018). As

disulfide reduction seems to be an important prerequisite for ERAD, the main

question under debate is that how are the dedicated reductases such as ERdj5 and

PDI kept in the reduced state? A possible answer is protein thiols, as they have been

suggested to form a larger thiol redox buffer than glutathione itself (Hansen et al.

2009). A likely explanation is also that the reduction of the reductases is connected

to the cytosolic thioredoxin pathways, analogous to the prokaryotic and

mitochondrial reducing pathways (Fig. 7). However, a membrane component that

would transfer electrons between the organelles has not been identified, yet.

Another puzzling problem is a potential futile cycle created by parallel client

oxidation by an active Ero1-PDI relay and the prerequisite of reduction for ERAD

(Fig. 12). Both processes have been characterized separately and are supposedly

active when the concentration of non-native proteins increase in the ER – by the

regulatory feedback and signaling mechanisms described. However, any crosstalk

between the systems, which might help to avoid the futile cycle, has not been

described. Speculatively, such regulation would require either spatial segregation

of the systems, or that they are mutually exclusive systems in the same space by

inverse modulation of their activity.

68

69

3 Aims of the present work

The Ero1 isoforms and PDI form the main pathway for disulfide formation in the

human ER. The previous in vitro studies (for example, Baker et al., 2008; Araki &

Nagata, 2011; Araki et al., 2013) have consistently reported a triphasic activity trace

for Ero1 consisting of an initial redox activation lag phase, a linear maximal rate

period during the catalytic cycle, and a rapid decline of activity at low [O2].

Considering the importance of the Ero1-PDI relay for OPF in the ER, it was

surprising that the previous in vitro studies focused mainly on the kinetics of the

catalytic cycle, and only minor details have been reported for the other two phases

that likely contain information for regulation of Ero1 function. We hypothesized

that one reason for this has been a lack of proper enzyme kinetic tools to study the

complex kinetics of the Ero1 system. The first aims were therefore to generate

proper analytical tools to model the triphasic kinetics, characterize each of the three

phases, and reveal novel aspects of Ero1 function connected to the redox activation

and utilization of substrate O2. Secondly, as the Ero1-PDI relay forms the main

redox controller of the ER (Appenzeller-Herzog et al., 2010), the results obtained

may contain important information for the overall quality control homeostasis of

the ER. In other words, novel regulatory functions of the Ero1-PDI relay may reveal

new insight into how the ER maintains balance between efficient maturation of

disulfide-containing proteins and avoids hyperoxidation of clients. The latter would

have detrimental effects, if uncontrolled, for other quality control features such as

ERAD that utilizes reduced protein substrates (Ellgaard et al., 2018).

The final aim of this work was to gain insight into the non-catalytic aspects of

protein folding. Specifically, we undertook structural studies of a novel protein,

pERp1, which has been suggested to be a chaperone and shown to be a crucial

component of the biogenesis of immunoglobulins as well as other complex

disulfide-rich proteins (see Chapter 2.6.2). The main aim of this project was to

solve the crystal structure of pERp1, which might reveal new insight into the poorly

understood structure-function relationship of how pERp1 promotes

immunoglobulin maturation in the ER.

70

71

4 Methods

The main methods used in this study are listed in Table 2. Further information is

provided in the referred articles.

Table 2. A list of methods and their references.

Method Original publications

Molecular biology techniques

ꞏ Construction of expression vectors I, II, III

ꞏ Generation of expression strains I, II, III

Protein expression and purification I, II, III

Protein analysis

ꞏ Sodium dodecyl sulfate-polyacrylamide gel

electrophoresis (SDS-PAGE) and sample preparation

I

ꞏ Circular dichroism (CD) and analytical reversed phase

high-performance liquid chromatography (rpHPLC)

I

ꞏ Protein concentration and determination of FAD content I

Protein functional studies

ꞏ PDI exchange assay I

ꞏ Ero1 activity assay – oxygen electrode I, II

ꞏ Ero1 activity assay – gel based I

ꞏ Binding assay – isothermal titration calorimetry (ITC) II

ꞏ Binding assay – bilayer interferometry (BLI) II, III

ꞏ Chaperone assay III

Generation and validation of in silico methods for Ero1 kinetics I

Crystallography – crystallization, data collection and

processing, structure determination

III

72

73

5 Results

The main results of I-III are briefly introduced in this section. Chapter 5.1 details

kinetic characterization of the two human Ero1 isoforms, revealing novel

regulatory mechanisms. Chapter 5.2 follows these observations and suggests an

uncharacterized cross-talk between OPF and other quality control features of the

ER, namely ERAD. Chapter 5.3 describes structure-function relationship of pERp1,

a potential chaperone involved in biogenesis of IgA, IgM, and integrins.

5.1 Molecular analysis of human Ero1

5.1.1 Human Ero1 and PDI form a novel Ero1-PDI complex

Human wild type Ero1 isoforms were produced in a folded state in E. coli using the

CyDisCo system. A redox state analysis of purified Ero1 revealed stable mixed

disulfide complex formed between Ero1 and PDI (Fig. 1B in I), but not between

Ero1 and ERp57 (Fig. S3 in I). Both Ero1 isoforms in their respective complexes

were oxidized and contained long-range disulfides, as they migrated faster on non-

reducing gel (Fig. 1B in I), resembling the previously reported inactive OX states

(Benham et al., 2000; Dias-Gunasekara et al., 2005). Other biophysical

experiments (Table S1 in I and S2 in I, Fig. S1 in I, Fig. S2 in I) and FAD:complex

ratio (0.99 ± 0.07 FAD per Ero1α complex; 0.92 ± 0.13 FAD per Ero1β complex)

suggested that the Ero1 isoforms were natively folded.

PDI forms transient mixed disulfides to the Ero1 outer active site cysteines as

well as to the adjacent regulatory cysteines (see Fig. 9 and Chapter 2.6.1). The

stable complexes reported here formed for other reasons than these, as Cys to Ala

mutations in the Ero1 active site and/or the adjacent regulatory cysteines did not

abolish the complexes (Fig. 1C in I). Interestingly, the mixed disulfide formed

between a strongly conserved (see discussion in I) Cys166 in Ero1α or Cys165 in

Ero1β (Fig. 1D in I) and the N-terminal active site cysteine (Cys397) of the a'

domain in PDI (Fig. 1E in I). Cys166 has previously been considered non-

functional (Araki & Nagata, 2011) and is often mutated in in vitro functional studies

(for example, Araki & Nagata, 2011; Masui et al., 2011; Araki et al., 2013).

A redox state analysis revealed that PDI in the complex was a dynamic partner

to Ero1, as exogenous higher-molecular-weight PDI with three FLAG tag

sequences (PDI3FLAG) incorporated to the complex replacing the wild type PDI

74

(Fig. 1F in I). The molar ratio of incorporated PDI to the original PDI could be

modulated by varying the concentration of the supplied PDI variant (Fig. 1F in I).

The results were similar if the redox state of all components were oxidized by

GSSG prior to the analysis. None of other tested PDI family members could replace

the complexed wild-type PDI in similar manner (Fig. S4 in I).

Collectively, these data suggest that Cys166 and 165 are functional, and that

they may interact covalently with PDI. The mixed disulfide complexes were stable

in the absence of exogenous PDI, while supplemented PDI exchanged (relatively

slowly in the tested oxidizing conditions) with the complexed PDI, suggesting that

PDI is a dynamic partner instead of a permanent subunit, when using the

Cys166/165 site.

5.1.2 Generation of a non-linear regression method to analyze Ero1

kinetics

The purified wild type Ero1-PDI complexes, with Ero1 in apparent inactive state,

were expected to undergo redox activation and show catalytic activity in the

presence of exogenous reduced PDI. To test this, we recorded a comprehensive

kinetic data set by measuring oxygen consumption in the presence of the reducing

PDI/GSH natural substrate system (Baker et al., 2008). The resulting oxygen

consumption traces consisted of the three expected phases: the initial non-linear lag

phase for redox activation, a linear maximal catalytic activity phase, and rapid loss

of activity at low [O2] (Fig. 2A in I).

No analytical tools have been reported previously to model the complex

kinetics of the Ero1 systems. We generated a non-linear regression method that

modelled the complete time course of an oxygen consumption trace. The simplest

non-linear equation that fit to the derivative of the oxygen consumption data, i.e.,

the rate of oxygen consumption at any given time point, consisted of Michaelis-

Menten kinetics with additional terms for one- or two-step exponential activation

process and Hill coefficient for potential cooperativity of oxygen binding (Fig. 2B

in I and 2C in I). The regression method could return simultaneously activation rate

constants (which can be transformed to half-time of activation, t1/2), catalytic

turnover number (kcat), half-maximal concentration for oxygen [KM(O2)], and Hill

coefficient (Table 1 in I).

75

The natural substrate assay and the regression methods were validated as

follows:

– Glutathione was kept at physiological concentrations and in high excess over

PDI to observe specifically Ero1 activity, i.e., the rate of disulfide transfer from

PDI to GSH was not limiting the Ero1 to PDI transfer rate (Fig. S8A in I).

– Other PDI-family members showed specificity (Fig. 3B in I and 3C in I)

consistent with previous in vitro studies (Araki et al., 2013): PDI was the most

active substrate and ERp57, ERp46, ERp72, and P5 showed only marginal

activity with both Ero1 isoforms. PDIp showed high activity for Ero1β but low

activity for Ero1α.

– As the main statistical tests for accuracy, residual analyses were carried out for

each fit individually (for example, Fig. 2B in I and 2C in I), and the fits would

be integrated back over the original oxygen consumption data (for example,

Fig. S5 in I).

– The obtained kinetic parameters for the Ero1 systems were cross compared to

those of another sulfhydryl oxidase, yeast Erv1p. Erv1p does not have

regulatory disulfides, and, consistent with this, it showed no lag phase (Fig.

S7A in I and S7B in I). It had stark differences in both KM(O2) and Hill

coefficient compared to those of the Ero1 systems (Table 1 in I), consistent

with Erv1p utilizing cytochrome c instead of O2 as the terminal electron

acceptor in vivo (Bihlmaier et al., 2007; Dabir et al., 2007). These data suggest

that the high affinity for oxygen and high Hill coefficients observed for the

Ero1 systems were genuine features, and not modelling errors.

5.1.3 Differential substrate PDI dependence between the Ero1

isoforms

The activation lag phase and the linear maximal rate period were first characterized

with the validated non-linear regression method. To this end, an oxygen

consumption data set with varying substrate [PDI] was recorded. When kcat values

were plotted versus [PDI] (Fig. 3A in I and S9 in I), similar maximal PDI oxidation

rates but differential midpoints for [PDI] were observed for the wild-type Ero1

isoforms with Ero1α showing clearly higher dependence for [PDI] for efficient

turnover (Table 3). Plots of activation data vs [PDI] (Fig. 4A in I and 4B in I)

suggested that the two Ero1 isoforms had differing PDI dependencies for redox

activation as well (Table 3). Whereas Ero1α showed a two-step activation process

76

with a strong dependence for [PDI] for both steps, the activation of Ero1β collapsed

to a single step [PDI]-independent process already at low [PDI]. These kinetic data

suggest that Ero1α has a high and Ero1β a low overall dependency for reduced

substrate PDI for both efficient redox activation and catalytic turnover. Importantly,

the midpoints for turnover and activation for a given isoform were of similar

magnitude (Table 3), suggesting that both processes, activation and catalytic

turnover likely utilize the same molecular determinants for the Ero1-PDI

interaction, namely the reported β-hairpin of Ero1 interacting with the substrate-

binding pocket of PDI (Masui et al., 2011). This hypothesis was followed up in II

and is described later in this work.

Table 3. Differential PDI dependency of turnover and activation between the Ero1

isoforms. All values are mean ± STD.

Enzyme Maximal turnover1

(s-1)

KM(PDI) for

turnover

(µM)

Activation

process

KM(PDI) for

activation

(µM)

Overall2

[PDI]

dependency

Ero1α-PDI 0.78 ± 0.03 8.5 ± 1.1 Two [PDI]-

dependent

steps

13.3 ± 2.1,

19.0 ± 3.3

high

Ero1β-PDI 1.06 ± 0.02 2.8 ± 0.3 One [PDI]-

dependent,

one [PDI]-

independent

step

4.6 ± 0.8 low

1at saturated [PDI] and [O2], 2accounting both redox activation and turnover.

Cys to Ala mutagenesis was carried out to dissect which disulfides are reduced

during activation, potentially explaining the differences in the activation processes

between the isoforms. Two regulatory Ero1α mutants, Cys104/131Ala and

Cys208/241Ala, that have the regulatory switch and the gating disulfide (see

Chapters 2.5.3 and 2.6.1) mutated, respectively, fit to the two-step activation

process with no major differences in activation rates compared to the wild type (Fig.

4E in I). In contrast, the Ero1α Cys166Ala mutant, which purified as a monomer,

fit to a single step activation scheme, and significantly increased the activation rate

(Fig. 4E in I). A similar study for Ero1β could not be conducted, as only the

Cys100/130Ala (a hyperactive mutant based on residue homology to Ero1α) could

be efficiently produced and analyzed. This mutant, in contrast to the hyperactive

mutant of Ero1α, showed kinetics which collapsed to the PDI-independent

77

activation process and significantly increased activation rate - further highlighting

the differences in the activation mechanisms between the two isoforms.

The activation kinetics of the Ero1α regulatory mutants were expected to

deviate more strongly from those of the wild type Ero1, as the mutants require

fewer regulatory disulfide reduction steps. As this was not the case, we next tested

the possibility that the rate-limiting steps for the two PDI-dependent activation

steps are not defined by reduction of regulatory disulfides, but by the PDI exchange

mechanism that was described earlier (Chapter 5.1.1). In such a scenario, the

regulatory disulfides may get reduced, but the reduction events are not observable

in our assays, as they get masked by the slow PDI exchange reactions. We studied

this hypothesis with a gel-based redox state analysis. We mixed Ero1α complex

with exogenous PDI3FLAG in the reducing activity assay conditions and quenched

an aliquot at constant intervals with NEM to stop thiol-disulfide exchange reactions

(Fig. S12A in I). As a result, the wild type complex disappeared with two-step

kinetics (Fig. S12B in I), resulting in a halftime of disappearance (1.5 min) similar

to the halftime of activation observed in the activity assays (cf. Table 1 in I).

Concomitant, a higher-molecular-weight complex with incorporated PDI3FLAG

appeared with a consistent halftime (1.7 min). These results suggest that

exchanging substrate PDI molecules, instead of reduction of regulatory disulfides,

forms the rate-limiting steps during redox activation of Ero1α.

5.1.4 A novel regulatory mechanism allows efficient operation in

hypoxic conditions

Another poorly characterized kinetic phase for human Ero1 is the rapid loss of

activity at low [O2]. The initial experiment (Table 1 in I) showed that both Ero1

isoforms had low KM(O2) and high Hill coefficients of >3. The high Hill

coefficients were consistent in the PDI titration assays (4.0 ± 1.2 for Ero1α and 3.2

± 0.7 for Ero1β; mean ± STD; Chapter 5.1.3), suggesting that the Ero1 systems

potentially exhibit cooperative binding for oxygen. In contrast, Erv1p had a high

KM(O2) and a Hill coefficient of 1 (Table 1 in I), implying no cooperativity of

oxygen binding.

The surprising apparent cooperativity was further characterized by modeling

[O2] thresholds of inactivation in silico. We compared the kinetics of wild type Ero1

using the experimentally derived kinetic values (Table 1 in I) to those of a

theoretical enzyme that utilized the same KM(O2) but a Hill coefficient of 1.

Whereas the theoretical enzyme (without cooperativity) had a broad [O2] range of

78

inactivation (90% to 10% activity) of 44.4 µM and created hyperhypoxic conditions

by effectively consuming all oxygen from the system (Fig. 5A in I), both Ero1

isoforms retained full activity till very low [O2] and inactivated before consuming

all oxygen from the system (Fig. 5A in I). This effect was markedly apparent for

Ero1α which showed an inactivation range of only 4.3 µM (Table 4). These data

suggest that the Ero1 isoforms have been fine-tuned to operate efficiently at low

oxygen concentrations and this feature is achieved by the sigmoidal kinetics

permitted by the high Hill coefficients.

Table 4. Experimental kinetic data at low [O2] for Ero1α vs a theoretical enzyme without

a Hill coefficient.

Enzyme KM(O2)

(µM)

Hill

coefficient

[O2]90%

(µM)

[O2]10%

(µM)

inactivation

range1 (µM)

Ero1α-PDI 5.0 5.2 7.6 3.3 4.3

Theoretical enzyme 5.0 1.0 45.0 0.6 44.4

1[O2]90% - [O2]10%; [O2]90%, oxygen concentration at 90% activity; [O2]10%, oxygen concentration at 10%

activity.

5.2 Feedback inhibition of the Ero1α-PDI system

This work demonstrated that human Ero1α requires high concentrations of

exogenous reduced PDI for efficient redox activation and catalytic turnover. The

kinetic data suggested also that the activation and turnover events are likely

mediated by the same molecular determinants, specifically, the protruding β-

hairpin of Ero1α interacting with the hydrophobic substrate-binding site of PDI

(Fig. 10). These observations proposed a paradoxical hypothesis: Are the redox

activation steps and the catalytic activity of Ero1α inhibited during unfolded protein

stress, if the PDI substrate-binding site is saturated by non-native proteins? Non-

native proteins with exposed hydrophobic residues bind the b’ domain of PDI

(Klappa et al., 1998) which is also needed to bind Ero1α. We further hypothesized

that the inhibitory effect would not be paradoxical if some other quality control

feature of the ER benefited from the inhibition. One candidate is ERAD which

utilizes reduced protein substrates (Ellgaard et al., 2018). Inhibition of Ero1α

activity could therefore form a feedback inhibition loop that prevents futile redox

cycles, retaining non-native proteins in more reduced state for ERAD (Fig. 1 in II).

79

5.2.1 The Ero1α-PDI interaction is inhibited by non-native proteins

To test the above hypotheses, we carried out comprehensive activity assays for

Ero1α complex using different variants of BPTI as the potential inhibitors. We

performed the assays in the presence or absence of native BPTI (N) or BPTI mutant

variants Cys30/51Ala and Cys14/38Ala that mimic the late folding intermediates

N* and NSS

HH (Fig. 3) and are incapable of forming N. When N was in clear excess

over PDI, no inhibition of Ero1α activation or catalytic cycle was observed,

irrespective of whether N was added at the start (denoted as pre-activation assay

from now on; Fig. 2B in II) or with circa 50% oxygen left in the system, when

Ero1α had completely activated (post-activation assay, Fig. 2C in II). In contrast,

N* and NSS

HH severely compromised Ero1α activity, with both intermediates causing

clearly extended lag phases (Fig. 2B in II). Whereas N* inhibited Ero1α for the

duration of the measurement, NSS

HH lost its inhibitory effect at later time points,

suggesting that late folding intermediates may not inhibit the catalytic cycle. These

results were supported by post-activation assays (Fig. 2C in II), resulting in

uninhibited activity in the presence of NSS

HH and an exponential loss of activity in the

presence of N*. The exponential activity loss could be modelled in silico (Fig. 2D

in II), and the returned halftime of inactivation of circa 128 s resembled the halftime

of activation of the uninhibited reaction (126 ± 7 s), suggesting that the exponential

loss was caused by redox inactivation instead of compromised catalytic activity.

All in all, these results suggest that natively folded proteins do not inhibit Ero1

activity, as they interact with neither Ero1α nor PDI; late folding intermediates

inhibit severely the redox activation steps but not the catalytic cycle.

5.2.2 Quantification of the inhibition

We quantified the inhibition with small redox insensitive hydrophobic molecules

that contain no cysteines. First, we tested a hydrophobic peptide, KFWWFS, which

has been reported to be bound by PDI (Byrne et al., 2009). Potent inhibition by the

peptide was observed in both the pre- and post-activation assays (Fig. 3A in II). To

quantify the inhibition, we kept [PDI] constant and varied [KFWWFS]. Both dose-

dependent extension of the lag phase and inhibition of the catalytic cycle were

observed (Fig. 3B in II). A plot of kcat vs [KFWWFS] (Fig. 3C in II) revealed

midpoint for inhibition of 4.9 ± 0.6 μM. Binding studies by ITC (Fig. 4A in II) and

BLI (Fig. 4B in II) demonstrated that PDI binds KFWWFS with affinity (Kd of 6.8

μM and 12.7 μM, respectively) of the same magnitude as the midpoint of inhibition,

80

suggesting that the inhibition is directly linked to competitive binding between the

peptide and Ero1α to the substrate binding site of PDI. We confirmed this

hypothesis by measuring the affinity of PDI to immobilized protruding β-hairpin

of Ero1α in the presence or absence of excess KFWWFS. In the absence of

KFWWFS, PDI bound to the hairpin with a Kd of 4.0 μM (Fig. 4C in II). When

KFWWFS was added in excess over PDI/hairpin, the PDI to hairpin binding signal

was decreased considerably (Fig. 4C in II).

Two other potential small-molecule inhibitors, the endogenous hormone 17β-

estradiol (17β-E2) (Fu et al., 2011) and the exogenous endocrine disruptor

bisphenol A (BPA; Hiroi et al., 2006) were tested. Both potential inhibitors, but not

a control compound 17α-estradiol (17α-E2), compromised Ero1α activity severely

in both the pre- and post-activation assays (Fig. 5A in II and 5B in II). The activity

of Erv1p was not inhibited to such degree, as it lacks the β-hairpin structure and

likely utilizes different binding mechanism (Guo et al., 2012; Fig. 5D in II). The

potent inhibition by 17β-E2 and BPA was confirmed not to be caused by the

inhibitors binding directly to Ero1α, as activity was regained in the presence of the

inhibitors and DTT (Fig. 5C in II, cf. Fig. 5A in II). To conclude, the quantification

data from this assay and the peptide assay confirm the observations from the protein

folding intermediate assays that the inhibition of Ero1 activity resulted from loss

of Ero1-PDI interaction due to saturation of the PDI substrate-binding site by the

inhibitors.

5.2.3 The inhibition is reversible

The feedback inhibition mechanism suggested (Fig. 1 in II) would have to be

reversible to have physiological significance. We tested the reversibility in vitro by

two methods:

1. the inhibition caused by non-native proteins should subside when the non-

native proteins fold to the native state;

2. increased concentration of folding factors should decrease the inhibition, as the

added factors sequester the inhibitors.

In both cases, loss of inhibition would restore the Ero1α-PDI interaction and

reinitiate oxygen consumption. To test if the inhibition subsides by folding, we

carried out oxygen consumption assays in the presence of denatured and

completely reduced wild-type BPTI (rBPTI). rBPTI behaves differently from the

late-stage folding intermediates tested earlier, as its folding pathway (Fig. 3B) is

81

not obstructed at any point, allowing it to form the early- and the late-stage folding

intermediates as well as the natively folded state. In the pre-activation assay, rBPTI

severely inhibited Ero1α activity at early time points, when no substantial folding

of rBPTI had occurred (Fig. 6A in II). At later time points, Ero1α slowly regained

activity (Fig. 6A in II), consistent with inhibition subsiding with BPTI folding and

N not inhibiting Ero1α (Fig. 2 in II). Interestingly, when rBPTI was tested in the

post-activation assay (Fig. 6B in II), severe inhibition of the catalytic activity was

observed. This suggests that early-stage folding intermediates inhibit both

activation and catalytic cycle, while the late-stage folding intermediates inhibit

mainly redox activation (cf. Fig. 2C in II).

To test the second hypothesis, we repeated the assay in the presence of rBPTI,

but, this time, before substantial amount of folded BPTI had accumulated, we

injected additional PDI to the assay. When [PDI] increased over [BPTI], Ero1α

regained activity immediately (Fig. 6A in II), supporting the second hypothesis that

folding factors sequester the non-native proteins and allow Ero1α to regain activity.

Together, these two assays demonstrate that the witnessed inhibition is reversible

and may therefore form a physiologically relevant feedback inhibition mechanism.

5.3 Structure-function relationship of human pERp1

One aim of this work was to expand the understanding of protein folding from

catalyzed folding to the non-catalytic aspect of protein maturation in the ER,

specifically, the action of molecular chaperones that are important for OPF but may

not directly promote disulfide formation. We chose pERp1 as the model protein, as

it is currently classified as either a chaperone or a co-chaperone. pERp1 promotes

the biogenesis of some complex disulfide-containing proteins, for example, IgA

and IgM, but otherwise it is poorly characterized (see Chapter 2.6.2). Cell-based

studies have demonstrated the physiological importance of pERp1 for the humoral

immune response, but the exact molecular mechanisms of its action are still poorly

understood. Here, high-resolution crystal structure of human pERp1 is presented to

address this issue. Structural and bioinformatic studies reveal relationship and

differences to other SAPLIPs. Functional studies elucidate its role as a chaperone

involved in biogenesis of immunoglobulins.

82

5.3.1 High-resolution crystal structure of pERp1

We determined the crystal structure of human pERp1 to 1.4 Å resolution (Table 1

in III). Two chains were in the asymmetric unit, and all residues except for

disordered N- and C-terminal regions could be modelled. The structure consisted

of the predicted α-helical saposin-like fold with five helices linked by three

conserved disulfides (Fig. 2 in III). Four inserts were identified that are missing in

the saposin C structure (Fig. 3A in III): The disordered N- and C-terminal regions

and two disordered regions linking helices 1 and 2 (an 11 amino acid loop) and

helices 2 and 3 (a 37 amino acid region with a three stranded β-sheet).

Two potential dimers of pERp1 were identified:

1. The two chains in the asymmetric unit formed an intermolecular antiparallel β-

sheet between the 37-residue inserts. This interaction was mediated mainly by

hydrophobic interactions (Fig. 2B in III).

2. PISA analysis identified another dimerization site formed from helices 2 and 3

and N-terminus (Fig. 2C in III). The chains in this dimer would be connected

by salt bridges, with a negatively charged patch in one chain interacting with a

positively charged patch in the other.

While either of the homodimers could represent a biologically relevant form of

pERp1, analytical techniques (Fig. S3 in III) suggested that the pERp1 used in this

study was a monomer in solution. As the residues involved in dimerization are not

conserved in mammalia (see below), and saposins and other SAPLIPs (Ahn et al.,

2003; Rossman et al., 2008; Bryksa et al., 2011; Garrido-Arandia et al., 2018)

utilize different dimerization mechanisms, the observed homodimers may not be

relevant for the physiological function of pERp1.

5.3.2 Mapping function onto the structure

pERp1 is important for many physiological processes, including integrin

maturation and immune response (Shimizu et al., 2009; van Anken et al., 2009;

Flach et al., 2010; Xiong et al., 2019). However, molecular mechanisms of action

are poorly understood. We were interested if the structure reported here would

reveal new insight for the reported functions. Saposin C is known to undergo

conformational changes and possess “open” and “closed” conformations that are

related to function (Fig. 4A in III). To form the open conformation in pERp1,

helices 2, 3, and 4 as well as the 37-residue insert would be required to undergo

83

major structural rearrangements. The insert packs via hydrophobic interactions

against helices 1 and 3 (Fig. 4B in III and 4C in III), which is not present in the

saposin C fold (Ahn et al., 2006). Therefore, the structural features presented here

seem to disallow formation of a similar open conformation that has been described

for saposin C.

pERp1 contains the CXXC motif found in many oxidoreductases, and it has

been shown to have slight oxidoreductase activity (van Anken et al., 2009).

However, the structural analysis in this work suggests that pERp1 is unlikely to be

an oxidoreductase: each disulfide is buried, and the structure confirms the earlier

observations (van Anken et al., 2009) that the cysteines in the CXXC motif link to

other cysteines far away in the sequence. As pERp1 is unlikely to have the reported

dynamic open/closed states of the saposin, the three disulfides in pERp1 are likely

structural instead of functional.

As the structural features could not be directly linked to any functions, we

carried out bioinformatics analyses next. A structure-based conservation analysis

of 167 pERp1 sequences of different eukaryotic organisms or of only mammalian

species revealed 18 or 36 fully conserved amino acids, respectively (Fig. 5A in III).

We focused on the mammalian data set, as the former data set could be affected by

parallel divergent evolution of interacting proteins. A map of the 36 fully conserved

residues on tertiary structure (Fig. 5B in III) revealed that most of the conserved

residues are buried and stabilize native structure. Five or eight residues formed a

highly acidic surface exposed patch (Fig. 5B in III). The mammalian data set also

revealed that pERp1 has lower overall sequence conservation compared to other

CNPY family members (Table 2 in III).

5.3.3 Biochemical assays to elucidate function

The above structural and bioinformatics analyses revealed mainly structure-

stabilizing features, while the function of the protein remained unclear. We decided

to elucidate the function of pERp1 next by biochemical assays. pERp1 has been

suggested to be either a chaperone or a cochaperone (Shimizu et al., 2009; van

Anken et al., 2009; Rosenbaum et al., 2014). We tested if pERp1 is a general

chaperone by a classical chaperone assay based on inhibition of aggregation of

thermally denatured citrate synthase (CS). Whereas molecular chaperones BiP and

ERp27 strongly inhibited aggregation of denatured CS, pERp1, even at high 10x

molar excess over CS, showed no inhibition of aggregation (Fig. 6A in III). The

84

lack of chaperone activity was not caused by thermal denaturation of pERp1 (Fig.

6B in III).

pERp1 is clearly not a general chaperone, so we next tested its substrate

specificity. pERp1 has been reported to interact with only a relatively small set of

proteins, with IgA and IgM representing the most characterized substrates (Shimizu

et al., 2009; van Anken et al., 2009; Xiong et al., 2019). Furthermore, the C-

terminal extension of IgA has been suggested to interact with pERp1 in vivo (Xiong

et al., 2019). As the extension is predicted to contain a β-strand and is highly

conserved (Fig. 6C in III), we hypothesized that the interaction between IgA and

pERp1 is potentially mediated by formation of an intermolecular β-sheet between

the IgA extension and the 37-residue insert in pERp1. We tested binding of IgA

with and without the C-terminal extension using BLI. Neither IgA form interacted

with pERp1 (Fig. 6D in III).

85

6 Discussion

Efficient maturation of disulfide-containing proteins in the mammalian ER is a

complex process that requires the concerted effort of folding catalysts and

chaperones as well as other quality control features such as UPR and ERAD.

Although the Ero1-PDI pathway became a focus of interest as the main pathway

for disulfide bond formation in the ER already two decades ago (Frand & Kaiser,

1998; Cabibbo et al., 2000; Pagani et al., 2000), no comprehensive enzyme kinetic

studies, that characterized each of the three major phases of Ero1 kinetics (redox

activation, catalytic cycle, rapid decline of activity at low [O2]), have been reported

for the pathway. In the present study, we reported comprehensive enzyme kinetic

measurements for various forms of human Ero1 isoforms, kinetic tools that could

analyze the complex triphasic kinetics, and revealed novel structural and regulatory

features for OPF and for other quality control mechanisms of the ER. These

features included:

1. a stable mixed disulfide complex between PDI and inactive Ero1, utilizing a

cysteine that has been previously considered non-functional;

2. a mechanism that allows efficient disulfide formation in hypoxic conditions

and inactivation before hyperhypoxia;

3. differential PDI dependency for activation and catalytic cycle between the

human Ero1 isoforms;

4. feedback inhibition of Ero1α activity by non-native proteins that hints at a

previously uncharacterized crosstalk between oxidative protein folding and

ERAD.

We identified a stable heterodimeric mixed disulfide complex formed between

human PDI and Ero1α or Ero1β in the inactive state. The mixed disulfide formed

between conserved Cys166 of Ero1α or Cys165 of Ero1β and Cys397 of the a'

domain of PDI. Two technical reasons stand out why these complexes have

remained uncharacterized to date: i) Cys166 has previously been considered non-

functional, and several studies mutated it to alanine (for example, Inaba et al.,

2010; Araki & Nagata, 2011; Masui et al., 2011; Araki et al., 2013); ii) Ero1 and

PDI may form at least three other transient mixed disulfide complexes [the catalytic

complex and two regulatory interactions (Appenzeller-Herzog et al., 2008;

Appenzeller-Herzog et al., 2010; Ramming et al., 2015)], and no efficient methods

have been reported to trap and isolate a specific mixed disulfide state in vivo.

Interestingly, the complex reported in this study may have been observed but not

86

followed up by others: Benham and coworkers (2000) reported in an elegant in

vitro translation study that the levels of Ero1-PDI mixed disulfide complex

increased concomitant with Ero1α shifting from the active OX1 to the inactive OX2

state. An endogenous Ero1-PDI mixed disulfide complex was later observed at

steady state from HT and HeLa cell lysates with Ero1α primarily in the OX2 state

(Benham et al., 2013). Furthermore, an immunoprecipitation experiment of Bertoli

and coworkers (2004) demonstrated that Ero1α Cys166Ala mutant bound

significantly less PDI than an unmutated Ero1α and Appenzeller-Herzog et al.

(2008) suggested that Cys166 forms a disulfide, but the partner cysteine could not

be identified. Hence, the inactivation pathway of human Ero1 consists of

intermolecular interactions with PDI involving Cys166/165, with the reported

stable mixed disulfide complex potentially representing the final state of Ero1

inactivation (see Fig. 5B in I and the discussion below).

The analysis of the triphasic kinetic profiles of human Ero1 systems

demonstrated that the wild type complexes were active with catalytic efficiencies

(second-order rate constant Kcat / KM) at saturating [PDI] and [O2] of 9.1 x 104 M-

1s-1 (Ero1α) and 3.8 x 105 M-1s-1 (Ero1β), values that are consistent with typical

enzyme-catalyzed disulfide formation rates (cf. DsbA/DsbB couple in Table 1).

Importantly, at low [O2], we observed a novel regulatory mechanism of Ero1

activity, which was a combination of high affinity for substrate oxygen and a high

Hill coefficient (>3). The high Hill coefficient was present for all tested wild type

and redox mutants of Ero1α and Ero1β as well as at all tested substrate PDI

concentrations, suggesting that cooperative binding of substrate oxygen may

modulate the activity of both Ero1 isoforms, and that redox states of the enzyme

are unlikely to play part in the apparent cooperativity. Possible explanations for the

cooperativity include classical cooperativity (multiple oxygen-binding sites),

although we found no evidence of Ero1 forming homo-oligomeric states during the

catalytic cycle, or the monomeric glucokinase-type cooperativity which arises from

order-disorder transitions in a single substrate-binding site (Larion et al., 2012;

Whittington et al., 2015).

Physiologically, the low KM(O2) and the apparent cooperativity have important

implications for OPF. Both Ero1 isoforms retain high activity to low [O2] with a

drastic reduction of activity within an extremely narrow oxygen concentration

range (Fig. 5A in I). This effect was very strong for Ero1α which was still fully

active with ca. 5% oxygen left in the system, whereas a theoretical enzyme without

cooperativity had lost much of its activity by this point (Fig. 5A in I). Another

important observation was that the theoretical enzyme was capable of consuming

87

all oxygen from the system, while Ero1α effectively inactivated with circa 2 µM

(0.8%) oxygen left in the system (Fig. 5A in I). In other words, Ero1α has been

fine-tuned to function efficiently in hypoxic conditions and avoids formation of

hyperhypoxic conditions that might be lethal for the cell. Cell-based studies have

demonstrated that hypoxia upregulates Ero1α in vivo (May et al., 2005; Gess et al.,

2003). Our data suggests that the reported upregulation is a feasible means to

increase the folding capacity and disulfide formation rates of the ER in hypoxic

conditions when availability of substrate O2 is scarce.

We characterized for the first time the early lag phase of the Ero1 kinetics. The

data revealed differential PDI dependency for activation of the human Ero1

isoforms. Whereas Ero1α activated with two [PDI]-dependent steps, the kinetics of

Ero1β collapsed to a [PDI]-independent activation process already at low [PDI].

These differences have important consequences for in vivo function: Ero1α has a

strong dependence for exogenous reduced PDI to both activate and turn over

efficiently. Ero1β, on the other hand, will activate and show higher disulfide

formation rates even under conditions with low availability of reduced PDI. Such

a situation may occur in cells with extremely high protein folding load, for example,

professional secretory cells such as insulin-secreting pancreatic cells, where Ero1β

is enriched (Pagani et al., 2000; Dias-Gunasekara et al., 2005).

It is important to note that the differences in the activation kinetics do not

necessarily mean that the Ero1 isoforms have different activation pathways. The

simplest explanation is that the isoforms utilize similar activation pathways, but

they have different rate-limiting reactions in the individual steps. We developed a

scheme for the activation and inactivation pathways based on this view (Fig. 5B in

I) that is consistent with the activation assays utilizing redox mutants (Fig 4 in I)

and with the PDI exchange assays (Fig. 1F in I and Fig. S12 in I). In the scheme,

the oxidized Ero1-PDI mixed disulfide complex is the ultimate inactivated state of

human Ero1. The complex contains three regulatory disulfides - the mixed disulfide

and two intramolecular regulatory disulfides in Ero1 - each of which need to be

reduced for activation. The complex is relatively stable in the absence of exogenous

reduced PDI, as no net change takes place in the complex in the presence of

exogenous oxidized PDI (Fig. S13 in I). When exogenous reduced PDI is available,

the oxidized PDI in the complex is exchanged to reduced PDI, which initiates

activation. Two rounds of PDI exchange is required for full activation: the first one

leads to the reduction of the mixed disulfide and one of the intramolecular

regulatory disulfides, the second step to the reduction of the final regulatory

disulfide. For Ero1α, reductions of the intramolecular regulatory disulfides are

88

relatively fast compared to the slow PDI exchange reactions; the latter were

observed as the two [PDI]-dependent kinetic steps. Consistent with this, the Ero1α

Cys166Ala mutant activated with a single step and with significantly faster kinetics

(Fig. 4E in I), suggesting that the first slow PDI exchange step was omitted with

this monomeric mutant. In contrast, Ero1β exchanges PDI extremely fast, to such

degree that the gel-based activation assay (cf. Fig. S12 in I) could not be carried

out for this isoform. This would suggest that reduction of regulatory disulfides, of

which one was observed at low [PDI] only, form the rate limiting steps for Ero1β.

Unfortunately, we could not explicitly identify which two intramolecular disulfides

were reduced in the activation pathway and the specifics of the [PDI]-independent

activation event of Ero1β. One of the reduced disulfides is likely the well-

characterized regulatory switch (Cys94-Cys131 in Ero1α, Cys90-Cys130 in Ero1β),

supported by similar kinetics between wild-type Ero1α and the regulatory switch

mutant (activation rate limited by PDI exchange) and increased activation rate of

the Ero1β regulatory switch mutant (reduction rate-limiting). The second step is

unlikely to be reduction of the gating disulfide as the Ero1α gating mutant activated

with two [PDI]-dependent steps and without major differences in activation rates.

One possible explanation is that the second reduced intramolecular disulfide is

Cys85-Cys391 (in Ero1α) and Cys81-Cys390 (in Ero1β). This disulfide was

initially thought regulatory (Baker et al., 2008) but later reconsidered structural

(Araki & Nagata, 2011; Zhang et al., 2014). We could not produce a mutant without

this controversial disulfide in sufficient amounts to confirm this hypothesis.

Based on the strong PDI-dependency of Ero1α, we hypothesized that the

activity of Ero1α might be inhibited by molecules that compete for binding PDI.

This could potentially create a feedback inhibition mechanism to avoid futile

oxidative cycles in conditions when proteins need to be in a more-reduced state, for

example, for efficient ERAD. We observed this to be true in vitro for redox

insensitive low-molecular-weight hydrophobic compounds such as natural 17β-E2

and a peptide containing no cysteines as well as for larger non-native proteins,

including a completely unfolded protein and late-stage folding intermediates.

Whereas the activated Ero1α-PDI catalytic cycle was inhibited only by a subset of

inhibitors, the activation steps were inhibited by all the tested inhibitors. As we also

observed a kinetic partitioning in the catalytic cycle that allowed a small portion of

Ero1α (circa 1.4% at the tested conditions) to become inactivated each catalytic

cycle, with kinetics that were proportional to the activation kinetics, we propose

that the inhibition of activation may present the primary mechanism for the

feedback inhibition mechanism. We developed a scheme based on these results (Fig.

89

7 in II). During basal protein synthesis, PDI, which may reach millimolar levels in

the ER (Hillson et al., 1984; Lyles & Gilbert, 1991), is likely in excess over non-

native proteins, retaining balance with an active Ero1α between the folding load

and the folding capacity. If the concentration of non-native proteins exceeds that of

the folding capacity – that consists of PDI and other folding catalysts as well as

molecular chaperones – activation of Ero1α is inhibited. This effectively shuts

down the whole Ero1α-PDI relay, as any currently active Ero1α molecules will

inactivate, per the kinetic partitioning described above, and remain inactive due to

the inhibitors sequestering reduced PDI. This inhibition happens independent of

cysteines, suggesting that all non-native proteins that contain exposed hydrophobic

residues and bind PDI may inhibit the Ero1α-PDI pathway.

The inhibition by non-native proteins was found to be reversible by removal of

the inhibiting non-native proteins by folding as well as by increased levels of

folding factors that sequestered the inhibitors, mimicking ERAD and UPR,

respectively. In each case, the physiological impact is an increase in the level of

free PDI with an unoccupied substrate-binding site, that allows reactivation of

Ero1α and restoration of OPF. Therefore, the inhibition of the Ero1α activation

pathway by non-native proteins may present a reversible regulatory mechanism to

avoid futile oxidative cycles and to decrease the dependence on the reductive

component of the ERAD, saving cellular resources.

pERp1

This work presented a novel structure for a member of the CNPY family.

Specifically, we solved the high-resolution crystal structure of pERp1, a folding

factor that has been shown by cell-based studies to be important for efficient

maturation of some complex disulfide-containing proteins in the ER, for example,

integrins and oligomeric assembly of IgA and IgM (Shimizu et al., 2009; van Anken

et al., 2009; Flach et al., 2010; Andreani et al., 2018; Xiong et al., 2019). The

structure confirms that pERp1 is a SAPLIP, as is also predicted for other CNPY

family members (Schildknegt et al., 2019). It has large and flexible inserts at both

termini and inserted between alpha helices that are missing in the canonical saposin

fold (Ahn et al., 2006). While previous studies have clearly demonstrated the

importance of pERp1 for the humoral immune response (Shimizu et al., 2009; van

Anken et al., 2009; Flach et al., 2010; Andreani et al., 2018; Xiong et al., 2019),

the molecular mechanisms of action of pERp1 have remained elusive.

Unfortunately, the structural analyses or the bioinformatics analyses presented here

90

could not explicitly answer the question either. Overall sequence conservation of

pERP1 was found to be low compared to other CNPY family members (Table 2 in

III), which may reflect higher selective pressure of genes related to immune

function (Shultz & Sackton, 2019). We could eliminate potential functions of

pERp1 by comparing the novel structure to the structure of saposin C and by

biochemical assays:

1. The structure of pERp1 is in the ‘closed’ state of saposin and likely cannot form

the ‘open’ conformation as readily as saposin due to additional interactions

between the inserts and the SAPLIP scaffold. As saposins are involved in lipid

binding (Vaccaro et al., 1999; Bruhn, 2005), and there are no reports of CNPY

family members interacting with lipids, it is possible that the open/closed

dynamics are specific for saposins and lipid binding.

2. Instead of forming a redox active CXXC motif, the six cysteines of pERp1

likely form the three buried structural disulfides like in the saposin fold (Ahn

et al., 2006). This also suggests that previous cysteine mutations that inhibited

in vivo function (Shimizu et al., 2009) disrupted structure.

3. pERp1 does not interact with unfolded proteins, so it is not a general chaperone.

4. Binding studies showed that pERp1 does not interact directly with the C-

terminal extension of IgA.

The final result presented above seem to be at odds with the recent report by Xiong

and colleagues (2019) whose in vivo data suggested an interaction between pERp1

and the C-terminal extension of IgA. We could not reproduce this interaction in

vitro. The recently characterized function of MANF, an ER-resident protein

involved in protein folding (Yan et al., 2019) may resolve this contradiction. MANF

contains a SAPLIP domain (Fig. 7A in III) which, like pERp1, is poorly conserved

(Fig. 7D in III) but has significant structural homology to pERp1 (Fig. 7C in III).

Its action in protein folding is modulated by the molecular chaperone BiP, and this

interaction involves the C-terminal SAF-A/B, Acinus, and PIAS domain, but not

the SAPLIP domain (Fig. 7B in III). Based on these functional and structural

similarities, we propose that the CNPY family should be grouped with MANF. Two

other proteins were identified that may belong to this group: CRELD1 and

CRELD2. Both proteins are ER-resident proteins that are involved in protein

folding (D'Alessandro et al., 2018; Hartley et al., 2013). They contain six cysteines

and a CXXC motif in a domain that InterPro predicts to be similar to the SAPLIP

domains of the CNPY family members. The six cysteines of CRELD are conserved

(Fig. 7E in III) and in analogous positions to those of pERp1 (Fig. 7A in III),

91

suggesting that they form the three structural disulfides and that both CRELD

indeed contain a SAPLIP domain. All in all, the CNPY family, MANF, and CRELD

may form a functional superfamily, and the members of this superfamily would be

regulators of protein folding with their SAPLIP domains acting as scaffolds to add

extensions that regulate the action of other folding factors. This would be consistent

with pERp1 interacting indirectly with the C-terminal extension of IgA, i.e., pERp1

regulates another folding factor which interacts with IgA.

Conclusions

The activity of human Ero1 isoforms is controlled by the redox state of the ER,

turning it on and off by a regulatory disulfide in reducing and oxidizing conditions,

respectively (Appenzeller-Herzog et al., 2008; Baker et al., 2008; Appenzeller-

Herzog et al., 2010). The results of the enzyme kinetic studies in the present work

have demonstrated that the activity of Ero1 and oxidative protein folding in the ER

are modulated on multiple other previously uncharacterized levels. These layers

include another “on/off switch” at low substrate oxygen concentrations that

potentially allows Ero1 to be shut down before consuming all precious oxygen from

the cell, and, inversely, allows OPF to rapidly initiate when the oxygen levels start

to rise; as well as by the controversial-sounding inhibition of OPF by unfolded

proteins that becomes actual in the larger concept of quality control network with

one reversibly downregulated quality control pathway allowing another pathway to

act more efficiently. These quality control pathways, e.g., UPR and ERAD, have

been characterized well individually, and, to the best of our knowledge, this is the

first time that a potential direct cross-talk between OPF and ERAD have been

reported. These novel layers and interconnections of the quality control features as

well as the novel SAPLIP structure and superfamily reported demonstrate that OPF

is an intriguing phenomenon that is still poorly understood, and the results of the

present work open up new fascinating possibilities for future protein-folding work.

92

93

7 Future prospects

We carried out comprehensive in vitro enzyme kinetic studies for both human Ero1

isoforms, revealing several new regulatory mechanisms for OPF. These studies

were consistent with previous works by others, and they detailed molecular

mechanisms of actions for previously observed physiological phenomena such as

the hypoxia link. However, the background that the present work created for the

novel regulatory mechanisms would benefit from further validation by cell-based

studies. Especially, the differences in the [PDI]-dependency between the Ero1

isoforms would greatly benefit from cellular studies and potentially explicitly

answer the fundamental question of why many mammalian species have two Ero1

isoforms. Additionally, establishing the OPF-ERAD link in cells would perhaps

allow ERAD to be studied in a more efficient way.

Another interesting future prospect is elucidation of the apparent cooperativity

for oxygen binding of human Ero1 isoforms. Apparently, substrate cooperativity

has not yet been described for a sulfhydryl oxidase. Molecular details of this feature

could therefore allow us to understand sulfhydryl oxidases in a new way but would

also potentially benefit biotechnology and industry. Many important protein-based

drugs contain disulfides and one major problem in their production – in addition to

formation of disulfides – is oxygen availability during protein expression. The

present work demonstrated that Erv1p, which is an important component together

with human PDI of the CyDisCo system, utilizes molecular oxygen relatively

poorly as the terminal electron acceptor. Preliminary unpublished data has

suggested that production of disulfide-rich proteins with CyDisCo is often oxygen-

limited during shake flask production, and that Ero1 can replace Erv1p in the

CyDisCo system to some extent. Establishing the molecular details of how Ero1

utilizes oxygen and how the apparent cooperativity works could potentially benefit

production of complex disulfide-rich proteins.

Finally, as both Ero1 and pERp1 have implications in the medical field,

knowledge about their molecular functions may contribute to development of new

therapeutic and diagnostic tools. Ero1 has been demonstrated to be upregulated in

cancers, is a predictor of poor prognosis of breast cancer, and the first selective

inhibitor was published already a decade ago (Blais et al., 2010; Kutomi et al., 2013,

Shergalis et al., 2020). These works have likely relied on the assumption that human

Ero1 is a monomeric protein (when not involved in catalytic disulfide transfer to

PDI). If Ero1 is found in significant amounts in vivo in the stable heterodimeric

inactive state with PDI, this state might serve as a novel therapeutic target. pERp1

94

is crucial for immune response and, thus, it has been reported to be involved in

many immune system-related diseases and cancers (for example, Matsumura et al.,

2012; Wong et al., 2013; Kanda et al., 2016; Miyagawa-Hayashino et al., 2018;

Guzeldemir-Akcakanat et al., 2019). The structure presented in this work will be

important for establishing the potential of pERp1 as a therapeutic target.

95

List of references

Abkevich, V. I., Gutin, A. M., & Shakhnovich, E. I. (1994). Specific nucleus as the transition state for protein folding: Evidence from the lattice model. Biochemistry, 33(33), 10026-10036. doi:10.1021/bi00199a029

Adachi, Y., Yamamoto, K., Okada, T., Yoshida, H., Harada, A., & Mori, K. (2008). ATF6 is a transcription factor specializing in the regulation of quality control proteins in the endoplasmic reticulum. Cell Structure and Function, 33(1), 75-89. doi:JST.JSTAGE/csf/07044

Ahn, V. E., Faull, K. F., Whitelegge, J. P., Fluharty, A. L., & Privé, G. G. (2003). Crystal structure of saposin B reveals a dimeric shell for lipid binding. Proceedings of the National Academy of Sciences of the United States of America, 100(1), 38-43. doi:0136947100

Ahn, V. E., Leyko, P., Alattia, J. R., Chen, L., & Privé, G. G. (2006). Crystal structures of saposins A and C. Protein Science : A Publication of the Protein Society, 15(8), 1849-1857. doi:ps.062256606

Alanen, H. I., Salo, K. E., Pekkala, M., Siekkinen, H. M., Pirneskoski, A., & Ruddock, L. W. (2003). Defining the domain boundaries of the human protein disulfide isomerases. Antioxidants & Redox Signaling, 5(4), 367-374. doi:10.1089/152308603768295096

Alanen, H. I., Williamson, R. A., Howard, M. J., Hatahet, F. S., Salo, K. E., Kauppila, A., . . . Ruddock, L. W. (2006). ERp27, a new non-catalytic endoplasmic reticulum-located human protein disulfide isomerase family member, interacts with ERp57. The Journal of Biological Chemistry, 281(44), 33727-33738. doi:M604314200

Alon, A., Heckler, E. J., Thorpe, C., & Fass, D. (2010). QSOX contains a pseudo-dimer of functional and degenerate sulfhydryl oxidase domains. FEBS Letters, 584(8), 1521-1525. doi:10.1016/j.febslet.2010.03.001

Andreani, V., Ramamoorthy, S., Pandey, A., Lupar, E., Nutt, S. L., Lammermann, T., & Grosschedl, R. (2018). Cochaperone Mzb1 is a key effector of Blimp1 in plasma cell differentiation and beta1-integrin function. Proceedings of the National Academy of Sciences of the United States of America, 115(41), E9630-E9639. doi:10.1073/pnas.1809739115

Anfinsen, C. B. (1973). Principles that govern the folding of protein chains. Science (New York, N.Y.), 181(4096), 223-230. doi:10.1126/science.181.4096.223

ANFINSEN, C. B., HABER, E., SELA, M., & WHITE, F. H.,Jr. (1961). The kinetics of formation of native ribonuclease during oxidation of the reduced polypeptide chain. Proceedings of the National Academy of Sciences of the United States of America, 47, 1309-1314. doi:10.1073/pnas.47.9.1309

Ang, S. K., & Lu, H. (2009). Deciphering structural and functional roles of individual disulfide bonds of the mitochondrial sulfhydryl oxidase Erv1p. The Journal of Biological Chemistry, 284(42), 28754-28761. doi:10.1074/jbc.M109.021113

96

Apetri, A. C., & Horwich, A. L. (2008). Chaperonin chamber accelerates protein folding through passive action of preventing aggregation. Proceedings of the National Academy of Sciences of the United States of America, 105(45), 17351-17355. doi:10.1073/pnas.0809794105

Appenzeller-Herzog, C., & Ellgaard, L. (2008). The human PDI family: Versatility packed into a single fold. Biochimica Et Biophysica Acta, 1783(4), 535-548. doi:S0167-4889(07)00281-9

Appenzeller-Herzog, C., & Ellgaard, L. (2008). In vivo reduction-oxidation state of protein disulfide isomerase: The two active sites independently occur in the reduced and oxidized forms. Antioxidants & Redox Signaling, 10(1), 55-64. doi:10.1089/ars.2007.1837

Appenzeller-Herzog, C., Riemer, J., Christensen, B., Sorensen, E. S., & Ellgaard, L. (2008). A novel disulphide switch mechanism in Ero1alpha balances ER oxidation in human cells. The EMBO Journal, 27(22), 2977-2987. doi:10.1038/emboj.2008.202

Appenzeller-Herzog, C., Riemer, J., Zito, E., Chin, K. T., Ron, D., Spiess, M., & Ellgaard, L. (2010). Disulphide production by Ero1alpha-PDI relay is rapid and effectively regulated. The EMBO Journal, 29(19), 3318-3329. doi:10.1038/emboj.2010.203

Araki, K., Iemura, S., Kamiya, Y., Ron, D., Kato, K., Natsume, T., & Nagata, K. (2013). Ero1-alpha and PDIs constitute a hierarchical electron transfer network of endoplasmic reticulum oxidoreductases. The Journal of Cell Biology, 202(6), 861-874. doi:10.1083/jcb.201303027

Araki, K., & Inaba, K. (2012). Structure, mechanism, and evolution of Ero1 family enzymes. Antioxidants & Redox Signaling, 16(8), 790-799. doi:10.1089/ars.2011.4418 [doi]

Araki, K., & Nagata, K. (2011). Functional in vitro analysis of the ERO1 protein and protein-disulfide isomerase pathway. The Journal of Biological Chemistry, 286(37), 32705-32712. doi:10.1074/jbc.M111.227181

Baker, K. M., Chakravarthi, S., Langton, K. P., Sheppard, A. M., Lu, H., & Bulleid, N. J. (2008). Low reduction potential of Ero1alpha regulatory disulphides ensures tight control of substrate oxidation. The EMBO Journal, 27(22), 2988-2997. doi:10.1038/emboj.2008.230

Banhegyi, G., Lusini, L., Puskas, F., Rossi, R., Fulceri, R., Braun, L., . . . Benedetti, A. (1999). Preferential transport of glutathione versus glutathione disulfide in rat liver microsomal vesicles. The Journal of Biological Chemistry, 274(18), 12213-12216. doi:10.1074/jbc.274.18.12213

Bass, R., Ruddock, L. W., Klappa, P., & Freedman, R. B. (2004). A major fraction of endoplasmic reticulum-located glutathione is present as mixed disulfides with protein. The Journal of Biological Chemistry, 279(7), 5257-5262. doi:10.1074/jbc.M304951200

Benham, A. M., Cabibbo, A., Fassio, A., Bulleid, N., Sitia, R., & Braakman, I. (2000). The CXXCXXC motif determines the folding, structure and stability of human Ero1-lalpha. The EMBO Journal, 19(17), 4493-4502. doi:10.1093/emboj/19.17.4493

97

Benham, A. M., van Lith, M., Sitia, R., & Braakman, I. (2013). Ero1-PDI interactions, the response to redox flux and the implications for disulfide bond formation in the mammalian endoplasmic reticulum. Philosophical Transactions of the Royal Society of London.Series B, Biological Sciences, 368(1617), 20110403. doi:10.1098/rstb.2011.0403

Bertoli, G., Simmen, T., Anelli, T., Molteni, S. N., Fesce, R., & Sitia, R. (2004). Two conserved cysteine triads in human Ero1alpha cooperate for efficient disulfide bond formation in the endoplasmic reticulum. The Journal of Biological Chemistry, 279(29), 30047-30052. doi:10.1074/jbc.M403192200

Bien, M., Longen, S., Wagener, N., Chwalla, I., Herrmann, J. M., & Riemer, J. (2010). Mitochondrial disulfide bond formation is driven by intersubunit electron transfer in Erv1 and proofread by glutathione. Molecular Cell, 37(4), 516-528. doi:10.1016/j.molcel.2010.01.017

Bihlmaier, K., Mesecke, N., Terziyska, N., Bien, M., Hell, K., & Herrmann, J. M. (2007). The disulfide relay system of mitochondria is connected to the respiratory chain. The Journal of Cell Biology, 179(3), 389-395. doi:jcb.200707123

Birk, J., Meyer, M., Aller, I., Hansen, H. G., Odermatt, A., Dick, T. P., . . . Appenzeller-Herzog, C. (2013). Endoplasmic reticulum: Reduced and oxidized glutathione revisited. Journal of Cell Science, 126(Pt 7), 1604-1617. doi:10.1242/jcs.117218

Biterova, E. I., Isupov, M. N., Keegan, R. M., Lebedev, A. A., Sohail, A. A., Liaqat, I., . . . Ruddock, L. W. (2019). The crystal structure of human microsomal triglyceride transfer protein. Proceedings of the National Academy of Sciences of the United States of America, 116(35), 17251-17260. doi:10.1073/pnas.1903029116

Blais, J. D., Chin, K. T., Zito, E., Zhang, Y., Heldman, N., Harding, H. P., . . . Ron, D. (2010). A small molecule inhibitor of endoplasmic reticulum oxidation 1 (ERO1) with selectively reversible thiol reactivity. The Journal of Biological Chemistry, 285(27), 20993-21003. doi:10.1074/jbc.M110.126599

Blobel, G. (1980). Intracellular protein topogenesis. Proceedings of the National Academy of Sciences of the United States of America, 77(3), 1496-1500. doi:10.1073/pnas.77.3.1496

Bommiasamy, H., Back, S. H., Fagone, P., Lee, K., Meshinchi, S., Vink, E., . . . Brewer, J. W. (2009). ATF6alpha induces XBP1-independent expansion of the endoplasmic reticulum. Journal of Cell Science, 122(Pt 10), 1626-1636. doi:10.1242/jcs.045625 [doi]

Braakman, I., Hoover-Litty, H., Wagner, K. R., & Helenius, A. (1991). Folding of influenza hemagglutinin in the endoplasmic reticulum. The Journal of Cell Biology, 114(3), 401-411. doi:10.1083/jcb.114.3.401

Brinker, A., Pfeifer, G., Kerner, M. J., Naylor, D. J., Hartl, F. U., & Hayer-Hartl, M. (2001). Dual function of protein confinement in chaperonin-assisted protein folding. Cell, 107(2), 223-233. doi:S0092-8674(01)00517-7

Bruhn, H. (2005). A short guided tour through functional and structural features of saposin-like proteins. The Biochemical Journal, 389(Pt 2), 249-257. doi:BJ20050051

98

Bryksa, B. C., Bhaumik, P., Magracheva, E., De Moura, D. C., Kurylowicz, M., Zdanov, A., . . . Yada, R. Y. (2011). Structure and mechanism of the saposin-like domain of a plant aspartic protease. The Journal of Biological Chemistry, 286(32), 28265-28275. doi:10.1074/jbc.M111.252619

Bulleid, N. J., & Freedman, R. B. (1988). Defective co-translational formation of disulphide bonds in protein disulphide-isomerase-deficient microsomes. Nature, 335(6191), 649-651. doi:10.1038/335649a0

Byrne, L. J., Sidhu, A., Wallis, A. K., Ruddock, L. W., Freedman, R. B., Howard, M. J., & Williamson, R. A. (2009). Mapping of the ligand-binding site on the b' domain of human PDI: Interaction with peptide ligands and the x-linker region. The Biochemical Journal, 423(2), 209-217. doi:10.1042/BJ20090565

Cabibbo, A., Pagani, M., Fabbri, M., Rocchi, M., Farmery, M. R., Bulleid, N. J., & Sitia, R. (2000). ERO1-L, a human protein that favors disulfide bond formation in the endoplasmic reticulum. The Journal of Biological Chemistry, 275(7), 4827-4833.

Cai, H., Wang, C. C., & Tsou, C. L. (1994). Chaperone-like activity of protein disulfide isomerase in the refolding of a protein with no disulfide bonds. The Journal of Biological Chemistry, 269(40), 24550-24552.

Cao, S. S., & Kaufman, R. J. (2014). Endoplasmic reticulum stress and oxidative stress in cell fate decision and human disease. Antioxidants & Redox Signaling, 21(3), 396-413. doi:10.1089/ars.2014.5851

Cao, X., Lilla, S., Cao, Z., Pringle, M. A., Oka, O. B. V., Robinson, P. J., . . . Bulleid, N. J. (2020). The mammalian cytosolic thioredoxin reductase pathway acts via a membrane protein to reduce ER-localised proteins. Journal of Cell Science, 133(8), jcs241976. doi: 10.1242/jcs.241976. doi:10.1242/jcs.241976

Chakravarthi, S., & Bulleid, N. J. (2004). Glutathione is required to regulate the formation of native disulfide bonds within proteins entering the secretory pathway. The Journal of Biological Chemistry, 279(38), 39872-39879. doi:10.1074/jbc.M406912200

Chambers, J. E., Tavender, T. J., Oka, O. B., Warwood, S., Knight, D., & Bulleid, N. J. (2010). The reduction potential of the active site disulfides of human protein disulfide isomerase limits oxidation of the enzyme by Ero1alpha. The Journal of Biological Chemistry, 285(38), 29200-29207. doi:10.1074/jbc.M110.156596

Chang, J. Y. (1997). A two-stage mechanism for the reductive unfolding of disulfide-containing proteins. The Journal of Biological Chemistry, 272(1), 69-75.

Chang, J. Y. (2011). Diverse pathways of oxidative folding of disulfide proteins: Underlying causes and folding models. Biochemistry, 50(17), 3414-3431. doi:10.1021/bi200131j

Chatrenet, B., & Chang, J. Y. (1992). The folding of hirudin adopts a mechanism of trial and error. The Journal of Biological Chemistry, 267(5), 3038-3043.

Chatrenet, B., & Chang, J. Y. (1993). The disulfide folding pathway of hirudin elucidated by stop/go folding experiments. The Journal of Biological Chemistry, 268(28), 20988-20996.

99

Chaudiere, J., & Ferrari-Iliou, R. (1999). Intracellular antioxidants: From chemical to biochemical mechanisms. Food and Chemical Toxicology : An International Journal Published for the British Industrial Biological Research Association, 37(9-10), 949-962. doi:S0278-6915(99)00090-3

Chivers, P. T., Prehoda, K. E., & Raines, R. T. (1997). The CXXC motif: A rheostat in the active site. Biochemistry, 36(14), 4061-4066. doi:10.1021/bi9628580 [doi]

CLELAND, W. W. (1964). Dithiothreitol, a new protective reagent for sh groups. Biochemistry, 3, 480-482. doi:10.1021/bi00892a002

Conn, K. J., Gao, W., McKee, A., Lan, M. S., Ullman, M. D., Eisenhauer, P. B., . . . Wells, J. M. (2004). Identification of the protein disulfide isomerase family member PDIp in experimental parkinson's disease and lewy body pathology. Brain Research, 1022(1-2), 164-172. doi:10.1016/j.brainres.2004.07.026

Dabir, D. V., Leverich, E. P., Kim, S. K., Tsai, F. D., Hirasawa, M., Knaff, D. B., & Koehler, C. M. (2007). A role for cytochrome c and cytochrome c peroxidase in electron shuttling from Erv1. The EMBO Journal, 26(23), 4801-4811. doi:7601909

Daithankar, V. N., Wang, W., Trujillo, J. R., & Thorpe, C. (2012). Flavin-linked erv-family sulfhydryl oxidases release superoxide anion during catalytic turnover. Biochemistry, 51(1), 265-272. doi:10.1021/bi201672h

D'Alessandro, M., Richard, M., Stigloher, C., Gache, V., Boulin, T., Richmond, J. E., & Bessereau, J. L. (2018). CRELD1 is an evolutionarily-conserved maturational enhancer of ionotropic acetylcholine receptors. Elife, 7, 10.7554/eLife.39649. doi:10.7554/eLife.39649

Desilva, M. G., Lu, J., Donadel, G., Modi, W. S., Xie, H., Notkins, A. L., & Lan, M. S. (1996). Characterization and chromosomal localization of a new protein disulfide isomerase, PDIp, highly expressed in human pancreas. DNA and Cell Biology, 15(1), 9-16. doi:10.1089/dna.1996.15.9

Dias-Gunasekara, S., Gubbens, J., van Lith, M., Dunne, C., Williams, J. A., Kataky, R., . . . Benham, A. M. (2005). Tissue-specific expression and dimerization of the endoplasmic reticulum oxidoreductase Ero1beta. The Journal of Biological Chemistry, 280(38), 33066-33075. doi:M505023200

Dill, K. A. (1990). Dominant forces in protein folding. Biochemistry, 29(31), 7133-7155. doi:10.1021/bi00483a001

Dinner, A. R., Sali, A., Smith, L. J., Dobson, C. M., & Karplus, M. (2000). Understanding protein folding via free-energy surfaces from theory and experiment. Trends in Biochemical Sciences, 25(7), 331-339. doi:S0968-0004(00)01610-8

DINTZIS, H. M. (1961). Assembly of the peptide chains of hemoglobin. Proceedings of the National Academy of Sciences of the United States of America, 47, 247-261. doi:10.1073/pnas.47.3.247

Dixon, B. M., Heath, S. H., Kim, R., Suh, J. H., & Hagen, T. M. (2008). Assessment of endoplasmic reticulum glutathione redox status is confounded by extensive ex vivo oxidation. Antioxidants & Redox Signaling, 10(5), 963-972. doi:10.1089/ars.2007.1869

Dobson, C. M. (2003). Protein folding and misfolding. Nature, 426(6968), 884-890. doi:10.1038/nature02261

100

Duarte, T. L., & Lunec, J. (2005). Review: When is an antioxidant not an antioxidant? A review of novel actions and reactions of vitamin C. Free Radical Research, 39(7), 671-686. doi:K5WNV67523R585N3

Dutton, R. J., Boyd, D., Berkmen, M., & Beckwith, J. (2008). Bacterial species exhibit diversity in their mechanisms and capacity for protein disulfide bond formation. Proceedings of the National Academy of Sciences of the United States of America, 105(33), 11933-11938. doi:10.1073/pnas.0804621105

Eletto, D., Eletto, D., Dersh, D., Gidalevitz, T., & Argon, Y. (2014). Protein disulfide isomerase A6 controls the decay of IRE1alpha signaling via disulfide-dependent association. Molecular Cell, 53(4), 562-576. doi:10.1016/j.molcel.2014.01.004

Ellgaard, L., McCaul, N., Chatsisvili, A., & Braakman, I. (2016). Co- and post-translational protein folding in the ER. Traffic (Copenhagen, Denmark), 17(6), 615-638. doi:10.1111/tra.12392

Ellgaard, L., Sevier, C. S., & Bulleid, N. J. (2018). How are proteins reduced in the endoplasmic reticulum? Trends in Biochemical Sciences, 43(1), 32-43. doi:S0968-0004(17)30200-1

Ellis, R. J. (2001). Macromolecular crowding: An important but neglected aspect of the intracellular environment. Current Opinion in Structural Biology, 11(1), 114-119. doi:S0959-440X(00)00172-X

Fagioli, C., Mezghrani, A., & Sitia, R. (2001). Reduction of interchain disulfide bonds precedes the dislocation of ig-mu chains from the endoplasmic reticulum to the cytosol for proteasomal degradation. The Journal of Biological Chemistry, 276(44), 40962-40967. doi:10.1074/jbc.M107456200

Fass, D. (2008). The erv family of sulfhydryl oxidases. Biochimica Et Biophysica Acta, 1783(4), 557-566. doi:S0167-4889(07)00280-7

Fass, D., & Thorpe, C. (2018). Chemistry and enzymology of disulfide cross-linking in proteins. Chemical Reviews, 118(3), 1169-1198. doi:10.1021/acs.chemrev.7b00123

Ferbitz, L., Maier, T., Patzelt, H., Bukau, B., Deuerling, E., & Ban, N. (2004). Trigger factor in complex with the ribosome forms a molecular cradle for nascent proteins. Nature, 431(7008), 590-596. doi:10.1038/nature02899

Fewell, S. W., Travers, K. J., Weissman, J. S., & Brodsky, J. L. (2001). The action of molecular chaperones in the early secretory pathway. Annual Review of Genetics, 35, 149-191. doi:35/1/149

Finkelstein, A. V. (2018). 50+ years of protein folding. Biochemistry.Biokhimiia, 83(Suppl 1), S3-S18. doi:10.1134/S000629791814002X

Finkelstein, A. V., & Ptitsyn, O. (2016). Protein physics: A course of lectures Elsevier Science.

Fischer, M., Horn, S., Belkacemi, A., Kojer, K., Petrungaro, C., Habich, M., . . . Riemer, J. (2013). Protein import and oxidative folding in the mitochondrial intermembrane space of intact mammalian cells. Molecular Biology of the Cell, 24(14), 2160-2170. doi:10.1091/mbc.E12-12-0862

101

Flach, H., Rosenbaum, M., Duchniewicz, M., Kim, S., Zhang, S. L., Cahalan, M. D., . . . Grosschedl, R. (2010). Mzb1 protein regulates calcium homeostasis, antibody secretion, and integrin activation in innate-like B cells. Immunity, 33(5), 723-735. doi:10.1016/j.immuni.2010.11.013

Fonin, A. V., Darling, A. L., Kuznetsova, I. M., Turoverov, K. K., & Uversky, V. N. (2018). Intrinsically disordered proteins in crowded milieu: When chaos prevails within the cellular gumbo. Cellular and Molecular Life Sciences : CMLS, 75(21), 3907-3929. doi:10.1007/s00018-018-2894-9

Foresti, O., Ruggiano, A., Hannibal-Bach, H. K., Ejsing, C. S., & Carvalho, P. (2013). Sterol homeostasis requires regulated degradation of squalene monooxygenase by the ubiquitin ligase Doa10/Teb4. Elife, 2, e00953. doi:10.7554/eLife.00953

Frand, A. R., & Kaiser, C. A. (1998). The ERO1 gene of yeast is required for oxidation of protein dithiols in the endoplasmic reticulum. Molecular Cell, 1(2), 161-170. doi:S1097-2765(00)80017-9

Freedman, R. B., Gane, P. J., Hawkins, H. C., Hlodan, R., McLaughlin, S. H., & Parry, J. W. (1998). Experimental and theoretical analyses of the domain architecture of mammalian protein disulphide-isomerase. Biological Chemistry, 379(3), 321-328. doi:10.1515/bchm.1998.379.3.321

Frickel, E. M., Riek, R., Jelesarov, I., Helenius, A., Wuthrich, K., & Ellgaard, L. (2002). TROSY-NMR reveals interaction between ERp57 and the tip of the calreticulin P-domain. Proceedings of the National Academy of Sciences of the United States of America, 99(4), 1954-1959. doi:10.1073/pnas.042699099

Fu, X. M., Wang, P., & Zhu, B. T. (2011). Characterization of the estradiol-binding site structure of human protein disulfide isomerase (PDI). PloS One, 6(11), e27185. doi:10.1371/journal.pone.0027185

Furie, B., Bouchard, B. A., & Furie, B. C. (1999). Vitamin K-dependent biosynthesis of gamma-carboxyglutamic acid. Blood, 93(6), 1798-1808.

Gamerdinger, M., Kobayashi, K., Wallisch, A., Kreft, S. G., Sailer, C., Schlomer, R., . . . Deuerling, E. (2019). Early scanning of nascent polypeptides inside the ribosomal tunnel by NAC. Molecular Cell, 75(5), 996-1006.e8. doi:S1097-2765(19)30487-3

Ganderton, T., Berndt, M. C., Chesterman, C. N., & Hogg, P. J. (2007). Hypothesis for control of von willebrand factor multimer size by intra-molecular thiol-disulphide exchange. Journal of Thrombosis and Haemostasis : JTH, 5(1), 204-206. doi:JTH2292

Garrido-Arandia, M., Cuevas-Zuviría, B., Díaz-Perales, A., & Pacios, L. F. (2018). A comparative study of human saposins. Molecules (Basel, Switzerland), 23(2), 422. doi: 10.3390/molecules23020422. doi:10.3390/molecules23020422

Gess, B., Hofbauer, K. H., Wenger, R. H., Lohaus, C., Meyer, H. E., & Kurtz, A. (2003). The cellular oxygen tension regulates expression of the endoplasmic oxidoreductase ERO1-lalpha. European Journal of Biochemistry, 270(10), 2228-2235. doi:3590

Grauschopf, U., Fritz, A., & Glockshuber, R. (2003). Mechanism of the electron transfer catalyst DsbB from escherichia coli. The EMBO Journal, 22(14), 3503-3513. doi:10.1093/emboj/cdg356

102

Gross, E., Kastner, D. B., Kaiser, C. A., & Fass, D. (2004). Structure of Ero1p, source of disulfide bonds for oxidative protein folding in the cell. Cell, 117(5), 601-610. doi:S0092867404004180

Gross, E., Sevier, C. S., Vala, A., Kaiser, C. A., & Fass, D. (2002). A new FAD-binding fold and intersubunit disulfide shuttle in the thiol oxidase Erv2p. Nature Structural Biology, 9(1), 61-67. doi:10.1038/nsb740

Guo, P. C., Ma, J. D., Jiang, Y. L., Wang, S. J., Bao, Z. Z., Yu, X. J., . . . Zhou, C. Z. (2012). Structure of yeast sulfhydryl oxidase erv1 reveals electron transfer of the disulfide relay system in the mitochondrial intermembrane space. The Journal of Biological Chemistry, 287(42), 34961-34969. doi:10.1074/jbc.M112.394759

Guzeldemir-Akcakanat, E., Alkan, B., Sunnetci-Akkoyunlu, D., Gurel, B., Balta, V. M., Kan, B., . . . Savli, H. (2019). Molecular signatures of chronic periodontitis in gingiva: A genomic and proteomic analysis. Journal of Periodontology, 90(6), 663-673. doi:10.1002/JPER.18-0477

Hansen, H. G., Soltoft, C. L., Schmidt, J. D., Birk, J., Appenzeller-Herzog, C., & Ellgaard, L. (2014). Biochemical evidence that regulation of Ero1beta activity in human cells does not involve the isoform-specific cysteine 262. Bioscience Reports, 34(2), 10.1042/BSR20130124. Print 2014 Apr 1. doi:e00103

Hansen, R. E., Roth, D., & Winther, J. R. (2009). Quantifying the global cellular thiol-disulfide status. Proceedings of the National Academy of Sciences of the United States of America, 106(2), 422-427. doi:10.1073/pnas.0812149106

Harding, H. P., Novoa, I., Zhang, Y., Zeng, H., Wek, R., Schapira, M., & Ron, D. (2000). Regulated translation initiation controls stress-induced gene expression in mammalian cells. Molecular Cell, 6(5), 1099-1108. doi:S1097-2765(00)00108-8

Harding, H. P., Zhang, Y., Zeng, H., Novoa, I., Lu, P. D., Calfon, M., . . . Ron, D. (2003). An integrated stress response regulates amino acid metabolism and resistance to oxidative stress. Molecular Cell, 11(3), 619-633. doi:S1097-2765(03)00105-9

Hartl, F. U., Bracher, A., & Hayer-Hartl, M. (2011). Molecular chaperones in protein folding and proteostasis. Nature, 475(7356), 324-332. doi:10.1038/nature10317

Hartl, F. U., & Hayer-Hartl, M. (2009). Converging concepts of protein folding in vitro and in vivo. Nature Structural & Molecular Biology, 16(6), 574-581. doi:10.1038/nsmb.1591

Hartley, C. L., Edwards, S., Mullan, L., Bell, P. A., Fresquet, M., Boot-Handford, R. P., & Briggs, M. D. (2013). Armet/Manf and Creld2 are components of a specialized ER stress response provoked by inappropriate formation of disulphide bonds: Implications for genetic skeletal diseases. Human Molecular Genetics, 22(25), 5262-5275. doi:10.1093/hmg/ddt383

Hatahet, F., & Ruddock, L. W. (2007). Substrate recognition by the protein disulfide isomerases. The FEBS Journal, 274(20), 5223-5234. doi:EJB6058

Hatahet, F., & Ruddock, L. W. (2009). Protein disulfide isomerase: A critical evaluation of its function in disulfide bond formation. Antioxidants & Redox Signaling, 11(11), 2807-2850. doi:10.1089/ARS.2009.2466

103

Hatahet, F., & Ruddock, L. W. (2013). Topological plasticity of enzymes involved in disulfide bond formation allows catalysis in either the periplasm or the cytoplasm. Journal of Molecular Biology, 425(18), 3268-3276. doi:10.1016/j.jmb.2013.04.034

He, K., Cunningham, C. N., Manickam, N., Liu, M., Arvan, P., & Tsai, B. (2015). PDI reductase acts on akita mutant proinsulin to initiate retrotranslocation along the Hrd1/Sel1L-p97 axis. Molecular Biology of the Cell, 26(19), 3413-3423. doi:10.1091/mbc.E15-01-0034

Hetz, C., & Papa, F. R. (2018). The unfolded protein response and cell fate control. Molecular Cell, 69(2), 169-181. doi:S1097-2765(17)30444-6

Hillson, D. A., Lambert, N., & Freedman, R. B. (1984). Formation and isomerization of disulfide bonds in proteins: Protein disulfide-isomerase. Methods in Enzymology, 107, 281-294. doi:10.1016/0076-6879(84)07018-x

Hiroi, T., Okada, K., Imaoka, S., Osada, M., & Funae, Y. (2006). Bisphenol A binds to protein disulfide isomerase and inhibits its enzymatic and hormone-binding activities. Endocrinology, 147(6), 2773-2780. doi:en.2005-1235

Hollien, J., Lin, J. H., Li, H., Stevens, N., Walter, P., & Weissman, J. S. (2009). Regulated Ire1-dependent decay of messenger RNAs in mammalian cells. The Journal of Cell Biology, 186(3), 323-331. doi:10.1083/jcb.200903014

Holtkamp, W., Kokic, G., Jager, M., Mittelstaet, J., Komar, A. A., & Rodnina, M. V. (2015). Cotranslational protein folding on the ribosome monitored in real time. Science (New York, N.Y.), 350(6264), 1104-1107. doi:10.1126/science.aad0344

Hosokawa, N., Wada, I., Hasegawa, K., Yorihuzi, T., Tremblay, L. O., Herscovics, A., & Nagata, K. (2001). A novel ER alpha-mannosidase-like protein accelerates ER-associated degradation. EMBO Reports, 2(5), 415-422. doi:10.1093/embo-reports/kve084

Huber-Wunderlich, M., & Glockshuber, R. (1998). A single dipeptide sequence modulates the redox properties of a whole enzyme family. Folding & Design, 3(3), 161-171. doi:S1359-0278(98)00024-8

Hwang, C., Sinskey, A. J., & Lodish, H. F. (1992). Oxidized redox state of glutathione in the endoplasmic reticulum. Science (New York, N.Y.), 257(5076), 1496-1502.

Inaba, K., Masui, S., Iida, H., Vavassori, S., Sitia, R., & Suzuki, M. (2010). Crystal structures of human Ero1alpha reveal the mechanisms of regulated and targeted oxidation of PDI. The EMBO Journal, 29(19), 3330-3343. doi:10.1038/emboj.2010.222

International Human Genome Sequencing Consortium. (2004). Finishing the euchromatic sequence of the human genome. Nature, 431(7011), 931-945. doi:nature03001

Jahn, T. R., & Radford, S. E. (2005). The yin and yang of protein folding. The FEBS Journal, 272(23), 5962-5970. doi:EJB5021

Janssens, S., Pulendran, B., & Lambrecht, B. N. (2014). Emerging functions of the unfolded protein response in immunity. Nature Immunology, 15(10), 910-919. doi:10.1038/ni.2991

Kadokura, H., & Beckwith, J. (2010). Mechanisms of oxidative protein folding in the bacterial cell envelope. Antioxidants & Redox Signaling, 13(8), 1231-1246. doi:10.1089/ars.2010.3187

104

Kaiser, C. M., Chang, H. C., Agashe, V. R., Lakshmipathy, S. K., Etchells, S. A., Hayer-Hartl, M., . . . Barral, J. M. (2006). Real-time observation of trigger factor function on translating ribosomes. Nature, 444(7118), 455-460. doi:nature05225

Kaiser, C. M., Goldman, D. H., Chodera, J. D., Tinoco, I.,Jr, & Bustamante, C. (2011). The ribosome modulates nascent protein folding. Science (New York, N.Y.), 334(6063), 1723-1727. doi:10.1126/science.1209740

Kanda, M., Tanaka, C., Kobayashi, D., Tanaka, H., Shimizu, D., Shibata, M., . . . Kodera, Y. (2016). Epigenetic suppression of the immunoregulator MZB1 is associated with the malignant phenotype of gastric cancer. International Journal of Cancer, 139(10), 2290-2298. doi:10.1002/ijc.30286

Karala, A. R., Lappi, A. K., & Ruddock, L. W. (2010). Modulation of an active-site cysteine pKa allows PDI to act as a catalyst of both disulfide bond formation and isomerization. Journal of Molecular Biology, 396(4), 883-892. doi:10.1016/j.jmb.2009.12.014

Karala, A. R., Lappi, A. K., Saaranen, M. J., & Ruddock, L. W. (2009). Efficient peroxide-mediated oxidative refolding of a protein at physiological pH and implications for oxidative folding in the endoplasmic reticulum. Antioxidants & Redox Signaling, 11(5), 963-970. doi:10.1089/ARS.2008.2326

Kim, S., Kim, C. M., Son, Y. J., Choi, J. Y., Siegenthaler, R. K., Lee, Y., . . . Park, H. H. (2018). Molecular basis of maintaining an oxidizing environment under anaerobiosis by soluble fumarate reductase. Nature Communications, 9(1), 4867-018-07285-9. doi:10.1038/s41467-018-07285-9

Kim, S., Sideris, D. P., Sevier, C. S., & Kaiser, C. A. (2012). Balanced Ero1 activation and inactivation establishes ER redox homeostasis. The Journal of Cell Biology, 196(6), 713-725. doi:10.1083/jcb.201110090

Kim, Y. E., Hipp, M. S., Bracher, A., Hayer-Hartl, M., & Hartl, F. U. (2013). Molecular chaperone functions in protein folding and proteostasis. Annual Review of Biochemistry, 82, 323-355. doi:10.1146/annurev-biochem-060208-092442

Kivirikko, K. I., Myllyla, R., & Pihlajaniemi, T. (1989). Protein hydroxylation: Prolyl 4-hydroxylase, an enzyme with four cosubstrates and a multifunctional subunit. FASEB Journal : Official Publication of the Federation of American Societies for Experimental Biology, 3(5), 1609-1617.

Klappa, P., Ruddock, L. W., Darby, N. J., & Freedman, R. B. (1998). The b' domain provides the principal peptide-binding site of protein disulfide isomerase but all domains contribute to binding of misfolded proteins. The EMBO Journal, 17(4), 927-935. doi:10.1093/emboj/17.4.927

Kober, F. X., Koelmel, W., Kuper, J., Drechsler, J., Mais, C., Hermanns, H. M., & Schindelin, H. (2013). The crystal structure of the protein-disulfide isomerase family member ERp27 provides insights into its substrate binding capabilities. The Journal of Biological Chemistry, 288(3), 2029-2039. doi:10.1074/jbc.M112.410522

Koch, J. R., & Schmid, F. X. (2014). Mia40 combines thiol oxidase and disulfide isomerase activity to efficiently catalyze oxidative folding in mitochondria. Journal of Molecular Biology, 426(24), 4087-4098. doi:S0022-2836(14)00573-7

105

Kodali, V. K., & Thorpe, C. (2010). Oxidative protein folding and the quiescin-sulfhydryl oxidase family of flavoproteins. Antioxidants & Redox Signaling, 13(8), 1217-1230. doi:10.1089/ars.2010.3098

Koivunen, P., Salo, K. E., Myllyharju, J., & Ruddock, L. W. (2005). Three binding sites in protein-disulfide isomerase cooperate in collagen prolyl 4-hydroxylase tetramer assembly. The Journal of Biological Chemistry, 280(7), 5227-5235. doi:M412480200

Kojer, K., Bien, M., Gangel, H., Morgan, B., Dick, T. P., & Riemer, J. (2012). Glutathione redox potential in the mitochondrial intermembrane space is linked to the cytosol and impacts the Mia40 redox state. The EMBO Journal, 31(14), 3169-3182. doi:10.1038/emboj.2012.165

Kolb, V. A., Makeyev, E. V., & Spirin, A. S. (1994). Folding of firefly luciferase during translation in a cell-free system. The EMBO Journal, 13(15), 3631-3637.

Komar, A. A., Kommer, A., Krasheninnikov, I. A., & Spirin, A. S. (1997). Cotranslational folding of globin. The Journal of Biological Chemistry, 272(16), 10646-10651. doi:10.1074/jbc.272.16.10646

Konno, T., Pinho Melo, E., Lopes, C., Mehmeti, I., Lenzen, S., Ron, D., & Avezov, E. (2015). ERO1-independent production of H2O2 within the endoplasmic reticulum fuels Prdx4-mediated oxidative protein folding. The Journal of Cell Biology, 211(2), 253-259. doi:10.1083/jcb.201506123

Kortemme, T., & Creighton, T. E. (1995). Ionisation of cysteine residues at the termini of model alpha-helical peptides. relevance to unusual thiol pKa values in proteins of the thioredoxin family. Journal of Molecular Biology, 253(5), 799-812. doi:S0022-2836(85)70592-X

Kortemme, T., Darby, N. J., & Creighton, T. E. (1996). Electrostatic interactions in the active site of the N-terminal thioredoxin-like domain of protein disulfide isomerase. Biochemistry, 35(46), 14503-14511. doi:10.1021/bi9617724 [doi]

Kutomi, G., Tamura, Y., Tanaka, T., Kajiwara, T., Kukita, K., Ohmura, T., . . . Hirata, K. (2013). Human endoplasmic reticulum oxidoreductin 1-α is a novel predictor for poor prognosis of breast cancer. Cancer Science, 104(8), 1091-1096. doi:10.1111/cas.12177

Langer, T., Lu, C., Echols, H., Flanagan, J., Hayer, M. K., & Hartl, F. U. (1992). Successive action of DnaK, DnaJ and GroEL along the pathway of chaperone-mediated protein folding. Nature, 356(6371), 683-689. doi:10.1038/356683a0

Lappi, A. K., Lensink, M. F., Alanen, H. I., Salo, K. E., Lobell, M., Juffer, A. H., & Ruddock, L. W. (2004). A conserved arginine plays a role in the catalytic cycle of the protein disulphide isomerases. Journal of Molecular Biology, 335(1), 283-295. doi:S0022283603013445

Lappi, A. K., & Ruddock, L. W. (2011). Reexamination of the role of interplay between glutathione and protein disulfide isomerase. Journal of Molecular Biology, 409(2), 238-249. doi:10.1016/j.jmb.2011.03.024

Larion, M., Salinas, R. K., Bruschweiler-Li, L., Miller, B. G., & Bruschweiler, R. (2012). Order-disorder transitions govern kinetic cooperativity and allostery of monomeric human glucokinase. PLoS Biology, 10(12), e1001452. doi:10.1371/journal.pbio.1001452

106

Leopold, P. E., Montal, M., & Onuchic, J. N. (1992). Protein folding funnels: A kinetic approach to the sequence-structure relationship. Proceedings of the National Academy of Sciences of the United States of America, 89(18), 8721-8725. doi:10.1073/pnas.89.18.8721

Li, G., Mongillo, M., Chin, K. T., Harding, H., Ron, D., Marks, A. R., & Tabas, I. (2009). Role of ERO1-alpha-mediated stimulation of inositol 1,4,5-triphosphate receptor activity in endoplasmic reticulum stress-induced apoptosis. The Journal of Cell Biology, 186(6), 783-792. doi:10.1083/jcb.200904060

Li, Y., & Camacho, P. (2004). Ca2+-dependent redox modulation of SERCA 2b by ERp57. The Journal of Cell Biology, 164(1), 35-46. doi:10.1083/jcb.200307010

Libich, D. S., Tugarinov, V., & Clore, G. M. (2015). Intrinsic unfoldase/foldase activity of the chaperonin GroEL directly demonstrated using multinuclear relaxation-based NMR. Proceedings of the National Academy of Sciences of the United States of America, 112(29), 8817-8823. doi:10.1073/pnas.1510083112

Libich, D. S., Tugarinov, V., & Clore, G. M. (2015). Reply to marchenko et al.: Flux analysis of GroEL-assisted protein folding/unfolding. Proceedings of the National Academy of Sciences of the United States of America, 112(50), E6833-4. doi:10.1073/pnas.1520474112

Lundstrom, J., & Holmgren, A. (1993). Determination of the reduction-oxidation potential of the thioredoxin-like domains of protein disulfide-isomerase from the equilibrium with glutathione and thioredoxin. Biochemistry, 32(26), 6649-6655. doi:10.1021/bi00077a018

Luo, B., & Lee, A. S. (2013). The critical roles of endoplasmic reticulum chaperones and unfolded protein response in tumorigenesis and anticancer therapies. Oncogene, 32(7), 805-818. doi:10.1038/onc.2012.130

Lyles, M. M., & Gilbert, H. F. (1991). Catalysis of the oxidative folding of ribonuclease A by protein disulfide isomerase: Dependence of the rate on the composition of the redox buffer. Biochemistry, 30(3), 613-619. doi:10.1021/bi00217a004

Makhnevych, T., & Houry, W. A. (2012). The role of Hsp90 in protein complex assembly. Biochimica et Biophysica Acta, 1823(3), 674-682. doi:10.1016/j.bbamcr.2011.09.001

Malhotra, J. D., Miao, H., Zhang, K., Wolfson, A., Pennathur, S., Pipe, S. W., & Kaufman, R. J. (2008). Antioxidants reduce endoplasmic reticulum stress and improve protein secretion. Proceedings of the National Academy of Sciences of the United States of America, 105(47), 18525-18530. doi:10.1073/pnas.0809677105

Mancini, R., Fagioli, C., Fra, A. M., Maggioni, C., & Sitia, R. (2000). Degradation of unassembled soluble ig subunits by cytosolic proteasomes: Evidence that retrotranslocation and degradation are coupled events. The FASEB Journal, 14(5), 769-778. doi:10.1096/fasebj.14.5.769

Marciniak, S. J., Yun, C. Y., Oyadomari, S., Novoa, I., Zhang, Y., Jungreis, R., . . . Ron, D. (2004). CHOP induces death by promoting protein synthesis and oxidation in the stressed endoplasmic reticulum. Genes & Development, 18(24), 3066-3077. doi:18/24/3066

107

Martin, J., & Hartl, F. U. (1997). The effect of macromolecular crowding on chaperonin-mediated protein folding. Proceedings of the National Academy of Sciences of the United States of America, 94(4), 1107-1112. doi:10.1073/pnas.94.4.1107

Massey, V. (1994). Activation of molecular oxygen by flavins and flavoproteins. The Journal of Biological Chemistry, 269(36), 22459-22462.

Masui, S., Vavassori, S., Fagioli, C., Sitia, R., & Inaba, K. (2011). Molecular bases of cyclic and specific disulfide interchange between human ERO1alpha protein and protein-disulfide isomerase (PDI). The Journal of Biological Chemistry, 286(18), 16261-16271. doi:10.1074/jbc.M111.231357

Matlack, K. E., Misselwitz, B., Plath, K., & Rapoport, T. A. (1999). BiP acts as a molecular ratchet during posttranslational transport of prepro-alpha factor across the ER membrane. Cell, 97(5), 553-564. doi:S0092-8674(00)80767-9

Matsumura, S., Imoto, I., Kozaki, K., Matsui, T., Muramatsu, T., Furuta, M., . . . Inazawa, J. (2012). Integrative array-based approach identifies MZB1 as a frequently methylated putative tumor suppressor in hepatocellular carcinoma. Clinical Cancer Research : An Official Journal of the American Association for Cancer Research, 18(13), 3541-3551. doi:10.1158/1078-0432.CCR-11-1007

Mattevi, A. (2006). To be or not to be an oxidase: Challenging the oxygen reactivity of flavoenzymes. Trends in Biochemical Sciences, 31(5), 276-283. doi:S0968-0004(06)00080-6

May, D., Itin, A., Gal, O., Kalinski, H., Feinstein, E., & Keshet, E. (2005). Ero1-L alpha plays a key role in a HIF-1-mediated pathway to improve disulfide bond formation and VEGF secretion under hypoxia: Implication for cancer. Oncogene, 24(6), 1011-1020. doi:1208325

Medrano-Fernandez, I., Fagioli, C., Mezghrani, A., Otsu, M., & Sitia, R. (2014). Different redox sensitivity of endoplasmic reticulum associated degradation clients suggests a novel role for disulphide bonds in secretory proteins. Biochemistry and Cell Biology, 92(2), 113-118. doi:10.1139/bcb-2013-0090

Mendoza, J. A., Jarstfer, M. B., & Goldenberg, D. P. (1994). Effects of amino acid replacements on the reductive unfolding kinetics of pancreatic trypsin inhibitor. Biochemistry, 33(5), 1143-1148. doi:10.1021/bi00171a013

Meunier, L., Usherwood, Y. K., Chung, K. T., & Hendershot, L. M. (2002). A subset of chaperones and folding enzymes form multiprotein complexes in endoplasmic reticulum to bind nascent proteins. Molecular Biology of the Cell, 13(12), 4456-4469. doi:10.1091/mbc.e02-05-0311

Mezghrani, A., Fassio, A., Benham, A., Simmen, T., Braakman, I., & Sitia, R. (2001). Manipulation of oxidative protein folding and PDI redox state in mammalian cells. The EMBO Journal, 20(22), 6288-6296. doi:10.1093/emboj/20.22.6288

Miyagawa-Hayashino, A., Yoshifuji, H., Kitagori, K., Ito, S., Oku, T., Hirayama, Y., . . . Tsuruyama, T. (2018). Increase of MZB1 in B cells in systemic lupus erythematosus: Proteomic analysis of biopsied lymph nodes. Arthritis Research & Therapy, 20(1), 13-018-1511-5. doi:10.1186/s13075-018-1511-5

108

Molinari, M., Calanca, V., Galli, C., Lucca, P., & Paganetti, P. (2003). Role of EDEM in the release of misfolded glycoproteins from the calnexin cycle. Science (New York, N.Y.), 299(5611), 1397-1400. doi:10.1126/science.1079474

Molinari, M., Galli, C., Piccaluga, V., Pieren, M., & Paganetti, P. (2002). Sequential assistance of molecular chaperones and transient formation of covalent complexes during protein degradation from the ER. The Journal of Cell Biology, 158(2), 247-257. doi:10.1083/jcb.200204122

Molinari, M., & Helenius, A. (1999). Glycoproteins form mixed disulphides with oxidoreductases during folding in living cells. Nature, 402(6757), 90-93. doi:10.1038/47062

Mossuto, M. F. (2013). Disulfide bonding in neurodegenerative misfolding diseases. International Journal of Cell Biology, 2013, 318319. doi:10.1155/2013/318319

Muller, R., Grawert, M. A., Kern, T., Madl, T., Peschek, J., Sattler, M., . . . Buchner, J. (2013). High-resolution structures of the IgM fc domains reveal principles of its hexamer formation. Proceedings of the National Academy of Sciences of the United States of America, 110(25), 10183-10188. doi:10.1073/pnas.1300547110

Nagy, P. (2013). Kinetics and mechanisms of thiol-disulfide exchange covering direct substitution and thiol oxidation-mediated pathways. Antioxidants & Redox Signaling, 18(13), 1623-1641. doi:10.1089/ars.2012.4973

Nguyen, V. D., Saaranen, M. J., Karala, A. R., Lappi, A. K., Wang, L., Raykhel, I. B., . . . Ruddock, L. W. (2011). Two endoplasmic reticulum PDI peroxidases increase the efficiency of the use of peroxide during disulfide bond formation. Journal of Molecular Biology, 406(3), 503-515. doi:10.1016/j.jmb.2010.12.039

Nguyen, V. D., Wallis, K., Howard, M. J., Haapalainen, A. M., Salo, K. E., Saaranen, M. J., . . . Williamson, R. A. (2008). Alternative conformations of the x region of human protein disulphide-isomerase modulate exposure of the substrate binding b' domain. Journal of Molecular Biology, 383(5), 1144-1155. doi:10.1016/j.jmb.2008.08.085

Oda, Y., Hosokawa, N., Wada, I., & Nagata, K. (2003). EDEM as an acceptor of terminally misfolded glycoproteins released from calnexin. Science (New York, N.Y.), 299(5611), 1394-1397. doi:10.1126/science.1079181

Ogura, J., Ruddock, L. W., & Mano, N. (2020). Cysteine 343 in the substrate binding domain is the primary S-nitrosylated site in protein disulfide isomerase. Free Radical Biology & Medicine, doi:S0891-5849(20)31165-5

Okuda-Shimizu, Y., & Hendershot, L. M. (2007). Characterization of an ERAD pathway for nonglycosylated BiP substrates, which require herp. Molecular Cell, 28(4), 544-554. doi:S1097-2765(07)00624-7

Okumura, M., Noi, K., Kanemura, S., Kinoshita, M., Saio, T., Inoue, Y., . . . Inaba, K. (2019). Dynamic assembly of protein disulfide isomerase in catalysis of oxidative folding. Nature Chemical Biology, 15(5), 499-509. doi:10.1038/s41589-019-0268-8

Pagani, M., Fabbri, M., Benedetti, C., Fassio, A., Pilati, S., Bulleid, N. J., . . . Sitia, R. (2000). Endoplasmic reticulum oxidoreductin 1-lbeta (ERO1-lbeta), a human gene induced in the course of the unfolded protein response. The Journal of Biological Chemistry, 275(31), 23685-23692. doi:10.1074/jbc.M003061200

109

Pan, Q., Shai, O., Lee, L. J., Frey, B. J., & Blencowe, B. J. (2008). Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nature Genetics, 40(12), 1413-1415. doi:10.1038/ng.259

Pobre, K. F. R., Poet, G. J., & Hendershot, L. M. (2019). The endoplasmic reticulum (ER) chaperone BiP is a master regulator of ER functions: Getting by with a little help from ERdj friends. The Journal of Biological Chemistry, 294(6), 2098-2108. doi:10.1074/jbc.REV118.002804

Poet, G. J., Oka, O. B., van Lith, M., Cao, Z., Robinson, P. J., Pringle, M. A., . . . Bulleid, N. J. (2017). Cytosolic thioredoxin reductase 1 is required for correct disulfide formation in the ER. The EMBO Journal, 36(5), 693-702. doi:10.15252/embj.201695336

Pollock, S., Kozlov, G., Pelletier, M. F., Trempe, J. F., Jansen, G., Sitnikov, D., . . . Thomas, D. Y. (2004). Specific interaction of ERp57 and calnexin determined by NMR spectroscopy and an ER two-hybrid system. The EMBO Journal, 23(5), 1020-1029. doi:10.1038/sj.emboj.7600119

Ponsero, A. J., Igbaria, A., Darch, M. A., Miled, S., Outten, C. E., Winther, J. R., . . . Toledano, M. B. (2017). Endoplasmic reticulum transport of glutathione by Sec61 is regulated by Ero1 and bip. Molecular Cell, 67(6), 962-973.e5. doi:S1097-2765(17)30609-3

Puig, A., & Gilbert, H. F. (1994). Protein disulfide isomerase exhibits chaperone and anti-chaperone activity in the oxidative refolding of lysozyme. The Journal of Biological Chemistry, 269(10), 7764-7771.

Qin, M., Wang, W., & Thirumalai, D. (2015). Protein folding guides disulfide bond formation. Proceedings of the National Academy of Sciences of the United States of America, 112(36), 11241-11246. doi:10.1073/pnas.1503909112

Quan, E. M., Kamiya, Y., Kamiya, D., Denic, V., Weibezahn, J., Kato, K., & Weissman, J. S. (2008). Defining the glycan destruction signal for endoplasmic reticulum-associated degradation. Molecular Cell, 32(6), 870-877. doi:10.1016/j.molcel.2008.11.017

Quan, H., Fan, Q., Li, C., Wang, Y. Y., & Wang, L. (2018). The transcriptional profiles and functional implications of long non-coding RNAs in the unfolded protein response. Scientific Reports, 8(1), 4981-018-23289-3. doi:10.1038/s41598-018-23289-3

Rakauskiene, G. A., Cenas, N. K., & Kulys, J. J. (1989). A 'branched' mechanism of the reverse reaction of yeast glutathione reductase. an estimation of the enzyme standard potential values from the steady-state kinetics data. FEBS Letters, 243(1), 33-36. doi:10.1016/0014-5793(89)81212-8

Ramming, T., Hansen, H. G., Nagata, K., Ellgaard, L., & Appenzeller-Herzog, C. (2014). GPx8 peroxidase prevents leakage of H2O2 from the endoplasmic reticulum. Free Radical Biology & Medicine, 70, 106-116. doi:10.1016/j.freeradbiomed.2014.01.018

Ramming, T., Okumura, M., Kanemura, S., Baday, S., Birk, J., Moes, S., . . . Appenzeller-Herzog, C. (2015). A PDI-catalyzed thiol-disulfide switch regulates the production of hydrogen peroxide by human Ero1. Free Radical Biology & Medicine, 83, 361-372. doi:10.1016/j.freeradbiomed.2015.02.011

110

Rancy, P. C., & Thorpe, C. (2008). Oxidative protein folding in vitro: A study of the cooperation between quiescin-sulfhydryl oxidase and protein disulfide isomerase. Biochemistry, 47(46), 12047-12056. doi:10.1021/bi801604x

Rapoport, T. A. (2007). Protein translocation across the eukaryotic endoplasmic reticulum and bacterial plasma membranes. Nature, 450(7170), 663-669. doi:nature06384

Raykhel, I., Alanen, H., Salo, K., Jurvansuu, J., Nguyen, V. D., Latva-Ranta, M., & Ruddock, L. (2007). A molecular specificity code for the three mammalian KDEL receptors. The Journal of Cell Biology, 179(6), 1193-1204. doi:jcb.200705180

Rietsch, A., Bessette, P., Georgiou, G., & Beckwith, J. (1997). Reduction of the periplasmic disulfide bond isomerase, DsbC, occurs by passage of electrons from cytoplasmic thioredoxin. Journal of Bacteriology, 179(21), 6602-6608. doi:10.1128/jb.179.21.6602-6608.1997

Rivas, G., Fernandez, J. A., & Minton, A. P. (1999). Direct observation of the self-association of dilute proteins in the presence of inert macromolecules at high concentration via tracer sedimentation equilibrium: Theory, experiment, and biological significance. Biochemistry, 38(29), 9379-9388. doi:10.1021/bi990355z

Romero, E., Gomez Castellanos, J. R., Gadda, G., Fraaije, M. W., & Mattevi, A. (2018). Same substrate, many reactions: Oxygen activation in flavoenzymes. Chemical Reviews, 118(4), 1742-1769. doi:10.1021/acs.chemrev.7b00650

Rosenbaum, M., Andreani, V., Kapoor, T., Herp, S., Flach, H., Duchniewicz, M., & Grosschedl, R. (2014). MZB1 is a GRP94 cochaperone that enables proper immunoglobulin heavy chain biosynthesis upon ER stress. Genes & Development, 28(11), 1165-1178. doi:10.1101/gad.240762.114

Rossmann, M., Schultz-Heienbrok, R., Behlke, J., Remmel, N., Alings, C., Sandhoff, K., . . . Maier, T. (2008). Crystal structures of human saposins C andD: Implications for lipid recognition and membrane interactions. Structure (London, England : 1993), 16(5), 809-817. doi:10.1016/j.str.2008.02.016

Rost, J., & Rapoport, S. (1964). Reduction-potential of glutathione. Nature, 201, 185. doi:10.1038/201185a0

Ruddock, L. W. (2012). Low-molecular-weight oxidants involved in disulfide bond formation. Antioxidants & Redox Signaling, 16(10), 1129-1138. doi:10.1089/ars.2011.4481

Ruddock, L. W., Freedman, R. B., & Klappa, P. (2000). Specificity in substrate binding by protein folding catalysts: Tyrosine and tryptophan residues are the recognition motifs for the binding of peptides to the pancreas-specific protein disulfide isomerase PDIp. Protein Science : A Publication of the Protein Society, 9(4), 758-764. doi:10.1110/ps.9.4.758

Ruggiano, A., Foresti, O., & Carvalho, P. (2014). Quality control: ER-associated degradation: Protein quality control and beyond. The Journal of Cell Biology, 204(6), 869-879. doi:10.1083/jcb.201312042

Saaranen, M. J., Karala, A. R., Lappi, A. K., & Ruddock, L. W. (2010). The role of dehydroascorbate in disulfide bond formation. Antioxidants & Redox Signaling, 12(1), 15-25. doi:10.1089/ars.2009.2674

111

Saaranen, M. J., & Ruddock, L. W. (2013). Disulfide bond formation in the cytoplasm. Antioxidants & Redox Signaling, 19(1), 46-53. doi:10.1089/ars.2012.4868

Sali, A., Shakhnovich, E., & Karplus, M. (1994). Kinetics of protein folding. A lattice model study of the requirements for folding to the native state. Journal of Molecular Biology, 235(5), 1614-1636. doi:S0022-2836(84)71110-7

Samelson, A. J., Bolin, E., Costello, S. M., Sharma, A. K., O'Brien, E. P., & Marqusee, S. (2018). Kinetic and structural comparison of a protein's cotranslational folding and refolding pathways. Science Advances, 4(5), eaas9098. doi:10.1126/sciadv.aas9098

Sarvas, M., Harwood, C. R., Bron, S., & van Dijl, J. M. (2004). Post-translocational folding of secretory proteins in gram-positive bacteria. Biochimica Et Biophysica Acta, 1694(1-3), 311-327. doi:S0167488904000795

Sato, Y., Kojima, R., Okumura, M., Hagiwara, M., Masui, S., Maegawa, K., . . . Inaba, K. (2013). Synergistic cooperation of PDI family members in peroxiredoxin 4-driven oxidative protein folding. Scientific Reports, 3, 2456. doi:10.1038/srep02456

Schildknegt, D., Lodder, N., Pandey, A., Egmond, M., Pena, F., Braakman, I., & van der Sluijs, P. (2019). Characterization of CNPY5 and its family members. Protein Science : A Publication of the Protein Society, 28(7), 1276-1289. doi:10.1002/pro.3635

Schoebel, S., Mi, W., Stein, A., Ovchinnikov, S., Pavlovicz, R., DiMaio, F., . . . Liao, M. (2017). Cryo-EM structure of the protein-conducting ERAD channel Hrd1 in complex with Hrd3. Nature, 548(7667), 352-355. doi:10.1038/nature23314

Schuler, B., & Hofmann, H. (2013). Single-molecule spectroscopy of protein folding dynamics--expanding scope and timescales. Current Opinion in Structural Biology, 23(1), 36-47. doi:10.1016/j.sbi.2012.10.008

Schulman, S., Wang, B., Li, W., & Rapoport, T. A. (2010). Vitamin K epoxide reductase prefers ER membrane-anchored thioredoxin-like redox partners. Proceedings of the National Academy of Sciences of the United States of America, 107(34), 15027-15032. doi:10.1073/pnas.1009972107

Sevier, C. S., Qu, H., Heldman, N., Gross, E., Fass, D., & Kaiser, C. A. (2007). Modulation of cellular disulfide-bond formation and the ER redox environment by feedback regulation of Ero1. Cell, 129(2), 333-344. doi:S0092-8674(07)00325-X

Shaffer, A. L., Shapiro-Shelef, M., Iwakoshi, N. N., Lee, A. H., Qian, S. B., Zhao, H., . . . Staudt, L. M. (2004). XBP1, downstream of blimp-1, expands the secretory apparatus and other organelles, and increases protein synthesis in plasma cell differentiation. Immunity, 21(1), 81-93. doi:10.1016/j.immuni.2004.06.010

Shepherd, C., Oka, O. B., & Bulleid, N. J. (2014). Inactivation of mammalian Ero1alpha is catalysed by specific protein disulfide-isomerases. The Biochemical Journal, 461(1), 107-113. doi:10.1042/BJ20140234

Shergalis, A. G., Hu, S., Bankhead, A.,3rd, & Neamati, N. (2020). Role of the ERO1-PDI interaction in oxidative protein folding and disease. Pharmacology & Therapeutics, 210, 107525. doi:S0163-7258(20)30053-X

112

Shimizu, Y., Meunier, L., & Hendershot, L. M. (2009). pERp1 is significantly up-regulated during plasma cell differentiation and contributes to the oxidative folding of immunoglobulin. Proceedings of the National Academy of Sciences of the United States of America, 106(40), 17013-17018. doi:10.1073/pnas.0811591106

Singh, A. K., Bhattacharyya-Pakrasi, M., & Pakrasi, H. B. (2008). Identification of an atypical membrane protein involved in the formation of protein disulfide bonds in oxygenic photosynthetic organisms. The Journal of Biological Chemistry, 283(23), 15762-15770. doi:10.1074/jbc.M800982200

Shultz, A. J., & Sackton, T. B. (2019). Immune genes are hotspots of shared positive selection across birds and mammals. Elife, 8, 10.7554/eLife.41815. doi:10.7554/eLife.41815

Smith, M. H., Ploegh, H. L., & Weissman, J. S. (2011). Road to ruin: Targeting proteins for degradation in the endoplasmic reticulum. Science (New York, N.Y.), 334(6059), 1086-1090. doi:10.1126/science.1209235

Sriburi, R., Jackowski, S., Mori, K., & Brewer, J. W. (2004). XBP1: A link between the unfolded protein response, lipid biosynthesis, and biogenesis of the endoplasmic reticulum. The Journal of Cell Biology, 167(1), 35-41. doi:jcb.200406136

Staley, J. P., & Kim, P. S. (1990). Role of a subdomain in the folding of bovine pancreatic trypsin inhibitor. Nature, 344(6267), 685-688. doi:10.1038/344685a0

Stenflo, J., Fernlund, P., Egan, W., & Roepstorff, P. (1974). Vitamin K dependent modifications of glutamic acid residues in prothrombin. Proceedings of the National Academy of Sciences of the United States of America, 71(7), 2730-2733. doi:10.1073/pnas.71.7.2730

Stigler, J., Ziegler, F., Gieseke, A., Gebhardt, J. C., & Rief, M. (2011). The complex folding network of single calmodulin molecules. Science (New York, N.Y.), 334(6055), 512-516. doi:10.1126/science.1207598

Szarka, A., & Lorincz, T. (2014). The role of ascorbate in protein folding. Protoplasma, 251(3), 489-497. doi:10.1007/s00709-013-0560-5

Tanford, C. (1962). Contribution of hydrophobic interactions to the stability of the globular conformation of proteins. Journal of the American Chemical Society, 84(22), 4240-4247. doi:10.1021/ja00881a009

Tang, X., Ang, S. K., Ceh-Pavia, E., Heyes, D. J., & Lu, H. (2019). Kinetic characterisation of Erv1, a key component for protein import and folding in yeast mitochondria. The FEBS Journal, doi:10.1111/febs.15077

Tang, Y. C., Chang, H. C., Roeben, A., Wischnewski, D., Wischnewski, N., Kerner, M. J., . . . Hayer-Hartl, M. (2006). Structural features of the GroEL-GroES nano-cage required for rapid folding of encapsulated protein. Cell, 125(5), 903-914. doi:S0092-8674(06)00560-5

Tavender, T. J., Sheppard, A. M., & Bulleid, N. J. (2008). Peroxiredoxin IV is an endoplasmic reticulum-localized enzyme forming oligomeric complexes in human cells. The Biochemical Journal, 411(1), 191-199. doi:BJ20071428

113

Tavender, T. J., Springate, J. J., & Bulleid, N. J. (2010). Recycling of peroxiredoxin IV provides a novel pathway for disulphide formation in the endoplasmic reticulum. The EMBO Journal, 29(24), 4185-4197. doi:10.1038/emboj.2010.273

Thommen, M., Holtkamp, W., & Rodnina, M. V. (2017). Co-translational protein folding: Progress and methods. Current Opinion in Structural Biology, 42, 83-89. doi:S0959-440X(16)30207-X

Tien, A. C., Rajan, A., Schulze, K. L., Ryoo, H. D., Acar, M., Steller, H., & Bellen, H. J. (2008). Ero1L, a thiol oxidase, is required for notch signaling through cysteine bridge formation of the Lin12-notch repeats in drosophila melanogaster. The Journal of Cell Biology, 182(6), 1113-1125. doi:10.1083/jcb.200805001

Tortorella, D., Story, C. M., Huppa, J. B., Wiertz, E. J., Jones, T. R., Bacik, I., . . . Ploegh, H. L. (1998). Dislocation of type I membrane proteins from the ER to the cytosol is sensitive to changes in redox potential. The Journal of Cell Biology, 142(2), 365-376. doi:10.1083/jcb.142.2.365

Tsai, A., Kornberg, G., Johansson, M., Chen, J., & Puglisi, J. D. (2014). The dynamics of SecM-induced translational stalling. Cell Reports, 7(5), 1521-1533. doi:S2211-1247(14)00335-0

Tury, A., Mairet-Coello, G., Poncet, F., Jacquemard, C., Risold, P. Y., Fellmann, D., & Griffond, B. (2004). QSOX sulfhydryl oxidase in rat adenohypophysis: Localization and regulation by estrogens. The Journal of Endocrinology, 183(2), 353-363. doi:183/2/353

Udgaonkar, J. B., & Baldwin, R. L. (1988). NMR evidence for an early framework intermediate on the folding pathway of ribonuclease A. Nature, 335(6192), 694-699. doi:10.1038/335694a0

Uhlen, M., Fagerberg, L., Hallstrom, B. M., Lindskog, C., Oksvold, P., Mardinoglu, A., . . . Ponten, F. (2015). Proteomics. tissue-based map of the human proteome. Science (New York, N.Y.), 347(6220), 1260419. doi:10.1126/science.1260419

Ushioda, R., Hoseki, J., Araki, K., Jansen, G., Thomas, D. Y., & Nagata, K. (2008). ERdj5 is required as a disulfide reductase for degradation of misfolded proteins in the ER. Science (New York, N.Y.), 321(5888), 569-572. doi:10.1126/science.1159293

Ushioda, R., Miyamoto, A., Inoue, M., Watanabe, S., Okumura, M., Maegawa, K. I., . . . Nagata, K. (2016). Redox-assisted regulation of Ca2+ homeostasis in the endoplasmic reticulum by disulfide reductase ERdj5. Proceedings of the National Academy of Sciences of the United States of America, 113(41), E6055-E6063. doi:1605818113

Utomo, A., Jiang, X., Furuta, S., Yun, J., Levin, D. S., Wang, Y. C., . . . Lee, W. H. (2004). Identification of a novel putative non-selenocysteine containing phospholipid hydroperoxide glutathione peroxidase (NPGPx) essential for alleviating oxidative stress generated from polyunsaturated fatty acids in breast cancer cells. The Journal of Biological Chemistry, 279(42), 43522-43529. doi:10.1074/jbc.M407141200

Vaccaro, A. M., Salvioli, R., Tatti, M., & Ciaffoni, F. (1999). Saposins and their interaction with lipids. Neurochemical Research, 24(2), 307-314. doi:10.1023/a:1022530508763

114

van Anken, E., Pena, F., Hafkemeijer, N., Christis, C., Romijn, E. P., Grauschopf, U., . . . Braakman, I. (2009). Efficient IgM assembly and secretion require the plasma cell induced endoplasmic reticulum protein pERp1. Proceedings of the National Academy of Sciences of the United States of America, 106(40), 17019-17024. doi:10.1073/pnas.0903036106

Walter, P., & Ron, D. (2011). The unfolded protein response: From stress pathway to homeostatic regulation. Science (New York, N.Y.), 334(6059), 1081-1086. doi:10.1126/science.1209038

Wang, C., Li, W., Ren, J., Fang, J., Ke, H., Gong, W., . . . Wang, C. C. (2013). Structural insights into the redox-regulated dynamic conformations of human protein disulfide isomerase. Antioxidants & Redox Signaling, 19(1), 36-45. doi:10.1089/ars.2012.4630

Wang, L., Li, S. J., Sidhu, A., Zhu, L., Liang, Y., Freedman, R. B., & Wang, C. C. (2009). Reconstitution of human Ero1-Lalpha/protein-disulfide isomerase oxidative folding pathway in vitro. position-dependent differences in role between the a and a' domains of protein-disulfide isomerase. The Journal of Biological Chemistry, 284(1), 199-206. doi:10.1074/jbc.M806645200

Wang, L., Zhang, L., Niu, Y., Sitia, R., & Wang, C. C. (2014). Glutathione peroxidase 7 utilizes hydrogen peroxide generated by Ero1alpha to promote oxidative protein folding. Antioxidants & Redox Signaling, 20(4), 545-556. doi:10.1089/ars.2013.5236

Wang, L., Zhu, L., & Wang, C. C. (2011). The endoplasmic reticulum sulfhydryl oxidase Ero1beta drives efficient oxidative protein folding with loose regulation. The Biochemical Journal, 434(1), 113-121. doi:10.1042/BJ20101357

Wang, X., Wang, L., Wang, X., Sun, F., & Wang, C. C. (2012). Structural insights into the peroxidase activity and inactivation of human peroxiredoxin 4. The Biochemical Journal, 441(1), 113-118. doi:10.1042/BJ20110380

Waudby, C. A., Dobson, C. M., & Christodoulou, J. (2019). Nature and regulation of protein folding on the ribosome. Trends in Biochemical Sciences, 44(11), 914-926. doi:S0968-0004(19)30141-0

Wedemeyer, W. J., Welker, E., Narayan, M., & Scheraga, H. A. (2000). Disulfide bonds and protein folding. Biochemistry, 39(15), 4207-4216. doi:bi992922o

Weissman, J. S., & Kim, P. S. (1991). Reexamination of the folding of BPTI: Predominance of native intermediates. Science (New York, N.Y.), 253(5026), 1386-1393. doi:10.1126/science.1716783

Weitzmann, A., Baldes, C., Dudek, J., & Zimmermann, R. (2007). The heat shock protein 70 molecular chaperone network in the pancreatic endoplasmic reticulum - a quantitative approach. The FEBS Journal, 274(19), 5175-5187. doi:EJB6039

Whittington, A. C., Larion, M., Bowler, J. M., Ramsey, K. M., Bruschweiler, R., & Miller, B. G. (2015). Dual allosteric activation mechanisms in monomeric human glucokinase. Proceedings of the National Academy of Sciences of the United States of America, 112(37), 11553-11558. doi:10.1073/pnas.1506664112

Wolynes, P. G., Onuchic, J. N., & Thirumalai, D. (1995). Navigating the folding routes. Science (New York, N.Y.), 267(5204), 1619-1620. doi:10.1126/science.7886447

115

Wong, L. Y., Brulois, K., Toth, Z., Inn, K. S., Lee, S. H., O'Brien, K., . . . Jung, J. U. (2013). The product of kaposi's sarcoma-associated herpesvirus immediate early gene K4.2 regulates immunoglobulin secretion and calcium homeostasis by interacting with and inhibiting pERP1. Journal of Virology, 87(22), 12069-12079. doi:10.1128/JVI.01900-13

Wu, C. K., Dailey, T. A., Dailey, H. A., Wang, B. C., & Rose, J. P. (2003). The crystal structure of augmenter of liver regeneration: A mammalian FAD-dependent sulfhydryl oxidase. Protein Science, 12(5), 1109-1118. doi:10.1110/ps.0238103

Xiong, E., Li, Y., Min, Q., Cui, C., Liu, J., Hong, R., . . . Wang, J. Y. (2019). MZB1 promotes the secretion of J-chain-containing dimeric IgA and is critical for the suppression of gut inflammation. Proceedings of the National Academy of Sciences of the United States of America, 116(27), 13480-13489. doi:10.1073/pnas.1904204116

Yamamoto, K., Sato, T., Matsui, T., Sato, M., Okada, T., Yoshida, H., . . . Mori, K. (2007). Transcriptional induction of mammalian ER quality control proteins is mediated by single or combined action of ATF6alpha and XBP1. Developmental Cell, 13(3), 365-376. doi:S1534-5807(07)00300-0

Yan, X., Hoek, T. A., Vale, R. D., & Tanenbaum, M. E. (2016). Dynamics of translation of single mRNA molecules in vivo. Cell, 165(4), 976-989. doi:10.1016/j.cell.2016.04.034

Yan, Y., Rato, C., Rohland, L., Preissler, S., & Ron, D. (2019). MANF antagonizes nucleotide exchange by the endoplasmic reticulum chaperone BiP. Nature Communications, 10(1), 541-019-08450-4. doi:10.1038/s41467-019-08450-4

Yoshida, H., Haze, K., Yanagi, H., Yura, T., & Mori, K. (1998). Identification of the cis-acting endoplasmic reticulum stress response element responsible for transcriptional induction of mammalian glucose-regulated proteins. involvement of basic leucine zipper transcription factors. The Journal of Biological Chemistry, 273(50), 33741-33749. doi:10.1074/jbc.273.50.33741

Yoshida, H., Matsui, T., Hosokawa, N., Kaufman, R. J., Nagata, K., & Mori, K. (2003). A time-dependent phase shift in the mammalian unfolded protein response. Developmental Cell, 4(2), 265-271. doi:S1534-5807(03)00022-4

Zeida, A., Babbush, R., Lebrero, M. C., Trujillo, M., Radi, R., & Estrin, D. A. (2012). Molecular basis of the mechanism of thiol oxidation by hydrogen peroxide in aqueous solution: Challenging the SN2 paradigm. Chemical Research in Toxicology, 25(3), 741-746. doi:10.1021/tx200540z

Zhang, G., Hubalewska, M., & Ignatova, Z. (2009). Transient ribosomal attenuation coordinates protein synthesis and co-translational folding. Nature Structural & Molecular Biology, 16(3), 274-280. doi:10.1038/nsmb.1554

Zhang, J., Zhu, Q., Wang, X., Yu, J., Chen, X., Wang, J., . . . Wang, L. (2018). Secretory kinase Fam20C tunes endoplasmic reticulum redox state via phosphorylation of Ero1alpha. The EMBO Journal, 37(14), 10.15252/embj.201798699. Epub 2018 Jun 1. doi:e98699

116

Zhang, L., Niu, Y., Zhu, L., Fang, J., Wang, X., Wang, L., & Wang, C. C. (2014). Different interaction modes for protein-disulfide isomerase (PDI) as an efficient regulator and a specific substrate of endoplasmic reticulum oxidoreductin-1alpha (Ero1alpha). The Journal of Biological Chemistry, 289(45), 31188-31199. doi:10.1074/jbc.M114.602961

Zhang, Y., Li, T., Zhang, L., Shangguan, F., Shi, G., Wu, X., . . . Wang, L. (2019). Targeting the functional interplay between endoplasmic reticulum oxidoreductin-1alpha and protein disulfide isomerase suppresses the progression of cervical cancer. Ebiomedicine, 41, 408-419. doi:S2352-3964(19)30119-7

Zimmerman, S. B., & Trach, S. O. (1991). Estimation of macromolecule concentrations and excluded volume effects for the cytoplasm of escherichia coli. Journal of Molecular Biology, 222(3), 599-620. doi:0022-2836(91)90499-V

Zito, E., Chin, K. T., Blais, J., Harding, H. P., & Ron, D. (2010). ERO1-beta, a pancreas-specific disulfide oxidase, promotes insulin biogenesis and glucose homeostasis. The Journal of Cell Biology, 188(6), 821-832. doi:10.1083/jcb.200911086

Zito, E., Hansen, H. G., Yeo, G. S., Fujii, J., & Ron, D. (2012). Endoplasmic reticulum thiol oxidase deficiency leads to ascorbic acid depletion and noncanonical scurvy in mice. Molecular Cell, 48(1), 39-51. doi:10.1016/j.molcel.2012.08.010

Zito, E., Melo, E. P., Yang, Y., Wahlander, A., Neubert, T. A., & Ron, D. (2010). Oxidative protein folding by an endoplasmic reticulum-localized peroxiredoxin. Molecular Cell, 40(5), 787-797. doi:10.1016/j.molcel.2010.11.010

117

Original publications

I Moilanen, A., Korhonen, K., Saaranen, M. J., & Ruddock, L. W. (2018). Molecular analysis of human Ero1 reveals novel regulatory mechanisms for oxidative protein folding. Life Science Alliance, 1(3), e201800090. doi:10.26508/lsa.201800090

II Moilanen, A., & Ruddock, L. W. (2020). Non-native proteins inhibit the ER oxidoreductin 1 (Ero1)-protein disulfide-isomerase relay when protein folding capacity is exceeded. The Journal of Biological Chemistry, 295(26), 8647-8655. doi:10.1074/jbc.RA119.011766

III Sowa, S. T., Moilanen, A., Biterova, E., Saaranen, M. J., Lehtiö, L., & Ruddock L. W. High-resolution crystal structure of human pERp1, a saposin-like protein crucial for IgA, IgM and integrin maturation in the endoplasmic reticulum. In press.

Reprinted with permission from Life Science Alliance (I) and American Society for

Biochemistry and Molecular Biology (II).

Original publications are not included in the electronic version of the dissertation.

118

A C T A U N I V E R S I T A T I S O U L U E N S I S

Book orders:Virtual book store

http://verkkokauppa.juvenesprint.fi

S E R I E S A S C I E N T I A E R E R U M N A T U R A L I U M

738. Rohunen, Anna (2019) Advancing information privacy concerns evaluation inpersonal data intensive services

739. Haghighatkhah, Alireza (2020) Test case prioritization using build history and testdistances : an approach for improving automotive regression testing incontinuous integration environments

740. Santos Parrilla, Adrian (2020) Analyzing families of experiments in softwareengineering

741. Kynkäänniemi, Sanna-Mari (2020) The relationship between the reindeer (Rangifertarandus tarandus) and the ectoparasitic deer ked (Lipoptena cervi) : reindeerwelfare aspects

742. Grundstrom, Casandra (2020) Health data as an enabler of digital transformation: a single holistic case study of connected insurance

743. Honka, Johanna (2020) Evolutionary and conservation genetics of Europeandomestic and wild geese

744. Alasaarela, Mervi (2020) Tietojärjestelmän käytön vaikutus laatuun jatuottavuuteen sairaalaorganisaatiossa palveluhenkilöstön kokemana

745. Korhonen, Tanja (2020) Tools and methods to support the key phases of seriousgame development in the health sector

746. Alahuhta, Kirsi (2020) Consequences of incomplete demographic information onecological modelling of plant populations with hidden life-history stages

747. Tyrmi, Jaakko (2020) Genomics and bioinformatics of local adaptation : studies ontwo non-model plants and a software for bioinformatics workflow management

748. Lavrinienko, Anton (2020) The effects of exposure to radionuclide contaminationon microbiota of wild mammals

749. Kainulainen, Tuomo (2020) Furfural-based 2,2?-bifurans : synthesis andapplications in polymers

750. Varila, Toni (2020) New, biobased carbon foams

751. Tikka, Piiastiina (2020) Persuasive user experiences in behaviour change supportsystems : avoiding bottlenecks along the way to full potential of persuasivetechnology

752. Raulamo-Jurvanen, Päivi (2020) Evaluating and selecting software test automationtools : synthesizing empirical evidence from practitioners

753. Korhonen, Olli (2020) Service pathway personalization in digital health services

UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Postdoctoral researcher Jani Peräntie

University Lecturer Anne Tuomisto

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

University Lecturer Anu Soikkeli

University Lecturer Santeri Palviainen

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-2846-4 (Paperback)ISBN 978-952-62-2847-1 (PDF)ISSN 0355-3191 (Print)ISSN 1796-220X (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAA

SCIENTIAE RERUM NATURALIUM

U N I V E R S I TAT I S O U L U E N S I SACTAA

SCIENTIAE RERUM NATURALIUM

OULU 2021

A 754

Antti Moilanen

NOVEL REGULATORY MECHANISMS AND STRUCTURAL ASPECTS OF OXIDATIVE PROTEIN FOLDING

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF BIOCHEMISTRY AND MOLECULAR MEDICINE

A 754

AC

TAA

ntti Moilanen