embracing proteins: structural themes in aptamer–protein ... · orange coloring on the cartoons...

Embracing proteins: structural themes in aptamer–protein complexesAmy D Gelinas1, Douglas R Davies2 and Nebojsa Janjic1

Available online at www.sciencedirect.com

ScienceDirect

Understanding the structural rules that govern specific, high-

affinity binding characteristic of aptamer–protein interactions is

important in view of the increasing use of aptamers across many

applications. From the modest number of 16 aptamer–protein

structures currently available, trends are emerging. The flexible

phosphodiester backbone allows folding into precise three-

dimensional structures using known nucleic acid motifs as

scaffolds that orient specific functional groups for target

recognition. Still, completely novel motifs essential for structure

and function are found in modified aptamers with diversity-

enhancing side chains. Aptamers and antibodies, two classes of

macromolecules used as affinity reagents with entirely different

backbones and composition, recognize protein epitopes of

similar size and with comparably high shape complementarity.

Addresses1 SomaLogic, Inc., 2945 Wilderness Place, Boulder, CO 80301,

United States2 Beryllium, 7869 NE Day Road West, Bainbridge Island, WA 98110,

United States

Corresponding author: Janjic, Nebojsa ([email protected])

Current Opinion in Structural Biology 2016, 36:122–132

This review comes from a themed issue on Nucleic acids and their

protein complexes

Edited by David MJ Lilley and Anna Marie Pyle

For a complete overview see the Issue and the Editorial

Available online 24th February 2016

http://dx.doi.org/10.1016/j.sbi.2016.01.009

0959-440X/# 2016 The Authors. Published by Elsevier Ltd. This is an

open access article under the CC BY license (http://creativecom-

mons.org/licenses/by/4.0/).

IntroductionA quarter of a century after the invention of SELEX [1,2],

there are now over 10 000 publications describing apta-

mers that bind to proteins, peptides, and small molecules.

With ongoing interest in the use of aptamers as research

reagents, diagnostics and therapeutics, there is consider-

able incentive to understand the structural features that

underlie their high affinity and specificity. Aptamers

bound to small molecules encage their targets and adopt

structures that are only apparent in the presence of the

ligand [3]. In contrast, aptamers bound to proteins utilize

a substantial fraction of their surfaces to engage distinct

epitopes. In this review we will focus on key themes

that have emerged so far from the 16 structures of

aptamer–protein complexes, with special emphasis on


new structural motifs recently observed in modified apta-

mers with diversity-enhancing side chains.

Nucleic acids as ligandsSingle-stranded nucleic acids have a strong tendency to

fold into discrete structures as a consequence of the

propensity of the bases to associate through p–p stacking

and hydrogen bonding interactions. This is further facili-

tated by a highly flexible phosphodiester backbone with

six rotatable bonds that, together with an array of sugar

puckers and a rotatable glycosidic bond, allow the bases to

interact in a wide range of conformations. Success of

SELEX results from the fact that randomized single-

stranded nucleic acid libraries represent a large collection

of shapes that can recognize other molecules through

shape and functional group complementarity.

Our understanding of the structural lexicon of nucleic acids

has increased dramatically since the first structure of tRNA

was reported in 1974 [4,5]. Initial phylogenetic studies and

biochemical probing experiments of ribosomal RNA fol-

lowed by X-ray and NMR structure determinations of

many shorter natural RNAs have established the existence

of numerous structural motifs in RNA [6]. It is not a

coincidence that the first SELEX experiments were done

with randomized RNA libraries in laboratories where

appreciation for the structural complexity of nucleic acids

was already well-established [1,2]. Recurring RNA motifs

such as hairpin loops, internal loops, and pseudoknots as

well as structural elements such as base zippers, ribose

zippers, A-platforms, p-turns, V-turns, a-loops and U-turns

have been extensively reviewed [7–9] and annotated in

databases [10–12]. It is now clear that SELEX with single-

stranded DNA (as well as with other libraries modified at

the 20-position of ribose) is just as feasible [13��,14��].

Together, these motifs provide a rich set of building

blocks for assembling intricate structures and a vocabu-

lary by which newly discovered aptamers can be de-

scribed and classified. As SELEX is increasingly

performed with modified nucleic acid libraries, this vo-

cabulary can clearly be expanded in an entirely new

dimension to include previously unknown motifs.

The original SELEX experiment targeting a protein was

done with bacteriophage T4 DNA polymerase (gp43),

which binds to a stem–loop region of its own mRNA

thereby serving as an autogenous negative regulator of its

expression [2]. Two major sequences with the same

affinity were identified from a randomized loop of

www.sciencedirect.com

http://crossmark.crossref.org/dialog/?doi=10.1016/j.sbi.2016.01.009&domain=pdf

mailto:[email protected]

http://www.sciencedirect.com/science/journal/0959440X/36



http://creativecommons.org/licenses/by/4.0/

http://creativecommons.org/licenses/by/4.0/

http://www.sciencedirect.com/science/journal/0959440X

Structural themes in aptamer–protein complexes Gelinas, Davies and Janjic 123

8 nucleotides: the native sequence and a quadruple

mutant. Once the structures were solved by NMR [15],

it became clear that the second ligand was a structural

mimetic of the native sequence despite being different at

four out of eight positions.

This feature has turned out to be a common theme with

aptamers to nucleic acid binding proteins. A glutamine

tRNA synthetase (GlnRS) aptamer obtained by targeted

randomization of the variable loop of tRNAGln also repre-

sents a quadruple mutant compared to the wild-type

tRNA but maintains the same contacts with the protein

[16]. An RNA aptamer to the MS2 bacteriophage capsid

protein adopts a similar stem–loop motif to the wild-type

RNA translational operator but with three rather than four

nucleotides in the loop and three rather than two base

pairs between the loop and the unpaired adenine in the

stem [17]. Nevertheless, the aptamer and its related

variants make contacts with the protein involving three

key unpaired nucleotides that are strikingly similar to

those of the operator [17,18]. Conversely, an aptamer to

ribosomal S8 protein has a secondary structure completely

different from the native RNA ligand; however, extensive

overlap with the binding interface of the native complex

is maintained after a major conformational change of the

aptamer [19��]. Even the RNA aptamer to NFkB, a

transcription factor known to bind double-stranded

DNA, adopts a stem–loop motif with an asymmetric

internal loop that engages NFkB with an interface that

overlaps essentially completely the key DNA–protein

contacts [20]. Structural plasticity of the RNA scaffold

allows conserved display of functional groups that contact

the protein despite differences in sequence and second-

ary structure.

There are currently twelve structures of aptamer com-

plexes with proteins not known to bind nucleic acids.

Structures of aptamers composed of RNA [21,22], 20-fluoropyrimidine RNA (20-FY RNA) [23,24], L-RNA

[25], L-RNA/DNA hybrid [26], DNA [27–29] and

DNA modified at the 5-position of deoxyuridine

[30,31,32��] have been reported (Figure 1). Originally a

major conceptual leap of faith, it is now clear that these

types of proteins are equally amenable to SELEX.

Structural motifs observed in aptamersAs the basic unit of structure in RNA and DNA, the

double helix is unsurprisingly a common motif in apta-

mer–protein complexes: 14 out of 16 aptamer-protein

complexes have it. Most are capped with hairpin loops

of varying sizes and have non-canonical base pairs, base

mismatches, and/or internal loops (symmetric or asym-

metric), as expected, since such features are known to

increase the structural flexibility and conformational rich-

ness of the helices [33]. Helical regions are also found

in pseudoknots (PDGF-BB and CCL2) and in the context

of a three-way helix junction with unpaired nucleotides at


the strand exchange (vWF) (Figure 1). Base triples and

quadruples have also been observed, and the G-quartet

domain is found in three complexes, either alone (throm-

bin) or as part of the multi-domain motif (C5a and IL-6).

The G-quartet domains have also been observed in

aptamer structures solved by NMR in the absence of

protein [15,34]. All of these motifs can be thought of as

scaffolds that orient specific and often discontiguous

functional groups for precise interactions with comple-

mentary pockets on proteins (Figure 1).

New motifs observed in slow off-rate modifiedaptamers (SOMAmers)Despite considerable success, a sizeable fraction of pro-

teins remain difficult SELEX targets. Compared to pro-

tein-based ligands, aptamers have the advantage of

enormous random libraries that can be efficiently

screened and greater conformational flexibility of the

phosphodiester backbone compared with the peptide

backbone (six rotatable bonds vs. two in proteins), but

a disadvantage in chemical diversity with only four bases

versus twenty amino acids as well as a narrower repertoire

of functional groups. With diversity-enhancing functional

groups at the 5-position of pyrimidines, we have recently

succeeded in identifying high-affinity aptamers against

many previously challenging proteins [35–37], and named

these ligands slow off-rate modified aptamers, or SOMA-

mers (for clarity, SOMAmers are defined as aptamers that

contain both diversity enhancing modifications as well as

slow off-rates). Among functional groups tested, hydro-

phobic, aromatic side chains consistently produce the

best ligands, especially when selection pressure includes

a demand for slow dissociation rates. To antibody experts,

this may not be surprising since amino acids like tyrosine,

tryptophan and phenylalanine are about five times more

common in paratopes (the complement to epitopes on

proteins) than on the rest of antibody surfaces, illustrating

their importance in protein binding [38,39].

With three SOMAmer-protein structures, we are begin-

ning to gain insight into the role such modifications play

in folding and binding. All three co-crystal structures were

obtained with SOMAmers identified from DNA libraries

containing a benzyl moiety at deoxyuridine, although

some of the SOMAmers have additional modifications

introduced as part of post-SELEX optimization. We note

this because the types of new structural motifs are likely

to be dependent on the choice of modifications front-

loaded in the initial random libraries.

Hydrophobic cluster

One of the most apparent and recurring new motifs in

SOMAmers is the hydrophobic cluster. Both target-inter-

face and internal clusters have been observed. In the

PDGF-BB SOMAmer, which is composed of a stem–loop

and a short pseudoknot (Figures 1m and 2a), the eight

modified nucleotides are distributed evenly between the


124 Nucleic acids and their protein complexes

Figure 1

Current Opinion in Structural Biology 2016, 36:122–132 www.sciencedirect.com


Figure 1

Current Opinion in Structural Biology

vWF DNA Aptamer

PfLDH DNA Aptamer NGF SOMAmer

IL-6 SOMAmer

PDGF SOMAmerThrombin DNA Aptamer(j)

(k)

(l)

(m)

(n)

(o)

180º

180º

180º 180º

180º

180º

domains [30]. All eight are in direct contact with the

protein forming an interaction surface with complemen-

tary hydrophobic pockets on the protein. The modified

nucleotides also form an extensive network of face-to-face

and edge-to-face p–p interactions that serve as a bridge

between the two domains (Figure 2a–c). Replacing a

(Figure 1 Legend) Open-book views of protein–SOMAmer and protein–aptam

renderings (�5 kT/e) with the binding interface outlined in orange. Atoms with

surface renderings of the SOMAmers and aptamers. Cartoon representations

interactions. Orange coloring on the cartoons indicates atom(s) in that region

orange hexagon indicates a sugar contact and orange rectangle indicates a b

SOMAmers. RNA aptamers bound to nucleic acid binding proteins are shown

RNA aptamer (PDB ID: 1OOA) and (c) S8/RNA aptamer (PDB ID: 4PDB). RNA

(d) GRK2/RNA aptamer (PDB ID: 3UZS) and (e) minE Lysozyme/RNA aptame

thrombin/20-fluoro RNA aptamer (PDB ID: 3DD2) and (g) IgG/20-fluoro RNA ap

RNA and L-DNA aptamer (DNA nucleotides indicated with an asterisk) (PDB I

aptamer structures are shown in (j–l): (j) thrombin/DNA aptamer (PDB ID: 3QL

(PDB ID: 3ZH2). Protein-SOMAmer structures are shown in (m–o): (m) PDGF/

NGF/SOMAmer (PDB ID: 4ZBN). The electrostatic surface renderings were pr


central aromatic residue (benzyl at position 8) with an

aliphatic isobutyl side chain (equivalent to a Phe to Leu

mutation), disrupts this network and creates a hole in the

structure which reduces binding affinity [30].

er binding interfaces. All proteins are shown as electrostatic surface

in 4 A of their target are colored orange on the stick and transparent

of each SOMAmer and aptamer show base pairing and stacking

within 4 A of the target; orange line indicates a backbone contact,

ase contact and/or modified nucleotide contact, in the case of

in (a–c): (a) MS2 coat protein/RNA aptamer (PDB ID: 6MSF), (b) NFkB/

aptamers bound to non-nucleic acid binding proteins are shown in (d–i):

r (PDB ID: 4M4O) with protein–20-fluoropyrimidine RNA aptamers in (f)

tamer (PDB ID: 3AGV) and protein–Spiegelmer aptamers in (h) C5a/L-

D: 4WB2) and (i) CCL2/L-RNA aptamer (PDB ID: 4R8I). Protein–DNA

P), (k) vWF/DNA aptamer (PDB ID: 3HXO) and (l) PfLDH/DNA aptamer

SOMAmer (PDB ID: 4HQU), (n) IL-6/SOMAmer (PDB ID: 4NI9) and (o)

epared in PyMOL [49] using the APBS plugin [50].



Figure 2

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)


Modified nucleotides of the SOMAmers make distinctive intramolecular contacts and have extraordinary shape complementarity to their targets.

(a) The PDGF-BB SOMAmer binds at the dimer interface of PDGF-BB and utilizes all eight modified nucleotides to create a hydrophobic contact

surface with p-stacking and extensive aromatic edge-to-face interactions as seen in (b,c). (d) The IL-6 SOMAmer wraps around the four-helix

bundle of the IL-6 protein and binds with two distinct domains. (e) The hydrophobic cluster in the G-quartet domain of the IL-6 SOMAmer is

centered on Bn-dU8 which is sandwiched between the uridine bases of Bn-dU7 and Nap-dU12, with Tyr31 on IL-6 completing the quadruple

stack. (f) Benzyl moieties in the stem–loop domain of IL-6 SOMAmer cluster and pack tightly against the protein surface as seen in the

transparent surface representation of Bn-dU22. (g) The triangular prism NGF SOMAmer lacks any helical content and conforms to the NGF

surface, binding at the dimer interface. (h) The benzyl zipper in NGF SOMAmer allows for continuous stacking with bases above and below while

the transparent surface rendering of Bn-dU17 highlights the SOMAmer-NGF surface complementarity. (i) A nonplanar benzyl zipper of Bn-dU10

and Bn-dU27 at the 30 end of the NGF SOMAmer creates a hydrophobic cluster with Bn-dU11 for internal structural stability. Protein structures

are shown in dark and light blue, modified dU nucleotides are colored orange and dA, dC and dG are colored wheat.

Current Opinion in Structural Biology 2016, 36:122–132 www.sciencedirect.com


The IL-6 SOMAmer is also composed of two domains: a

stem-loop domain and a G-quartet domain (Figures 1n

and 2d). Eight out of ten modified nucleotides (four in

each domain) make a direct contact with the protein [31].

The two domains are independent of each other, con-

nected by a flexible hinge with no inter-domain contacts

among the hydrophobic side chains. In the G-quartet

domain, which has the same up-down–up-down topology

as the thrombin G-quartet aptamer but no cross-reactivity

to thrombin (and vice versa), modified nucleotides in the

loops form the interface with the protein. Four hydro-

phobic side chains that contact the protein are organized

in a structured cluster in which the central benzyl group at

position 8 is sandwiched between two uridine bases and

surrounded by two benzyl and one naphthyl group in an

array of edge-to-face p–p interactions. A notable feature

of this cluster is a four-ring stack of p–p interactions

capped with a tyrosyl side chain on the protein

(Figure 2e). The hydrophobic cluster is clearly both an

intramolecular structural motif and a component of the

contact surface. When disconnected from the entire

SOMAmer, the G-quartet domain maintains some affinity

for IL-6 (in contrast to the stem–loop domain), albeit at a

three orders of magnitude lower affinity. A single substi-

tution at position 7 in the G-quartet SOMAmer results in

37-fold improvement in affinity, showing that surfaces

defined by the hydrophobic side chains can be tuned to

optimize interactions with the complementary pockets

[31]. In the stem–loop domain, the four modified nucleo-

tides that contact the protein are oriented toward the

major groove of the helix, with three modified side chains

organized in a distinct non-stacking hydrophobic cluster

(Figure 2f).

In contrast to the PDGF-BB and IL-6 SOMAmers, which

utilize well-known secondary structure motifs, the NGF

SOMAmer adopts an entirely unprecedented structure

resembling an elongated triangular prism in which three

interacting strands are arranged in an S-shaped pattern

held together with a dense network of hydrogen bonds

composed mainly of non-canonical base pairs, base tri-

ples, and base quadruples (Figures 1o and 2g) [32��]. A

remarkable feature of this SOMAmer is a complete

absence of helical character. In contrast to the PDGF-

BB and IL-6 SOMAmers, only four out of nine benzyl

groups contact the protein. The remaining five side

chains appear to serve primarily as internal hydrophobic

clusters that support this highly unusual fold.

Intercalating benzyl zipper

The two hydrophobic clusters of the NGF SOMAmer

exhibit a unique feature in which the benzyl side chains

of two modified nucleotides on adjacent strands are

involved in reciprocal face-to-face p–p stacking interac-

tions with uridine of the opposing base [32��]. The benzyl

groups are separated from the uridine ring by a

distance approximating a base pair and therefore serve

as a pseudo-base in which hydrogen bonding between the


bases is replaced with a covalent bond. In this way, the

intercalating benzyl groups at the 50 end act like the teeth

of a zipper, keeping opposing strands within an appropri-

ate distance for base pairing while maintaining continu-

ous stacking with the bases below and above the zipper

motif (Figure 2h). The other benzyl zipper is found near

the 30 end at the opposite side of this elongated SOMA-

mer as part of a hydrophobic cluster composed of three

modified nucleotides. In this zipper, the interaction be-

tween the benzyl group and the adjacent uridine base is

substantially identical to the 50 end zipper, but with a

twist: the planes defined by the two bases are not parallel,

and the two benzyl groups turn in synchrony to form a

near-perfect face-to-face stacking interaction with the

opposing uridines (Figure 2i). The two benzyl groups

within this twisted zipper interact with a third benzyl

group to form a structured internal cluster.

Far from being a passive means of connecting the side

chain to the base, the amide linkers are capable of

hydrogen bonding with other bases [30] as well as with

residues on protein targets [31]. We have observed an

overwhelming preference for the trans conformation of

the amide bond in the linker, with only 2 out of 27 mod-

ifications across the three SOMAmer structures adopting

the cis conformation. Similarly, only the anti conformation

of the carbonyl group of the C-5 linker with respect to the

C-4 carbonyl group has been observed.

Sugar puckers, glycosidic bonds andbackbone conformationsFull annotation of motifs utilized in the 16 reported co-

crystal structures is beyond the scope of this review.

Nevertheless, we examined the frequency of utilization

of sugar puckers, glycosidic bonds and backbone con-

formations in this limited dataset. Sugar puckers of ten

RNA-based aptamers (broadly defined to include selec-

tions done with T7 RNA polymerase) predominantly

adopt the C30-endo conformation, as expected

(Table 1). Discerning the difference between related

sugar puckers is often difficult, and conformations adja-

cent to C30-endo (such as C10-endo, C20-exo, C40-exo)

may not be within the level of resolution of the reported

structures (1.8–4.5 A) and the geometric restraint pa-

rameters used in structure refinement may be a source of

bias. With these caveats, in seven out of ten aptamers, a

C30-endo-like conformation is observed in more than

80% of nucleotides, comparable to that observed in

tRNA and ribosomal RNA (Table 1). For the six

DNA aptamers, there is a similar preference for the

C20-endo-like conformation (including the adjacent O40-endo, C10-exo and C30-exo conformations). Here, the

outlier is clearly the NGF SOMAmer, which appears to

have a wider spectrum of ribose conformations than any

other aptamer (Table 1), unsurprising given its highly

unusual structure.



Table 1

Sugar conformations, glycosidic bond conformations (darker orange with increasing % in a given conformation), Kd, Sc index and

interface area for rRNA, tRNA and nucleosomal DNA compared to the 16 aptamers. For the Spiegelmers the sugar pucker has been

reversed to account for difference in stereochemistry. The Sc index of 0.75 calculated in this work for the thrombin RNA aptamer is the

same value reported by Long et al. [23]

Sugar Conformation Ref.

endoC2’ exo

C3’ endo

C4’ exo

ΣC3’ endo

O4’ endo

C1’ exo

C2’endo

C3’exo

ΣC2'endo

C4’endo

O4’exo syn anti

RN

A

0 1 85 2 87 0 0 12 0 13 0 0 97 52

0 3 81 0 84 0 0 16 0 16 0 0 97 54

0 0 79 0 79 0 0 14 7 21 0 0 100 17

0 3 97 0 100 0 0 0 0 0 0 0 100 20

0 3 84 0.0 87 0 0 13 0 13 0 0 97 19

160 5 77 11 93 7 0 0 0 7 0 0 99

0 8 76 5 90 2 3 5 0 10 0 0 97 21

5 0 80 5 90 5 0 5 0 10 0 0 90 22

RNA 0 0 80 0 80 0 0 20 0 20 0 0 84 23

RNA 0 8 79 13 100 0 0 0 0 0 0 0 92 24

L-DNA 3 5 43 8 58 0 3 40 0 43 0 0 88 26

0 0 70 3 73 0 3 25 0 28 0 0 88 25

DN

A

0 0 3 5 8 4 22 62 4 92 0 0 87 53

0 0 0 7 7 0 7 73 7 87 7 0 73 29

3 0 13 8 23 8 30 25 15 78 0 0 88 28

0 4 7 4 15 7 7 48 19 81 4 0 78 27

0 0 9 4 13 0 26 57 4 87 0 0 83 30

0 0 6 3 9 6 19 56 9 91 0 0 66 31

Target Comp C1’

50S rRNA rRNAgly-tRNA tRNA

MS2 RNA

NFkB RN A

S8 RNA

GlnRS RNA

Lysozyme RNA

GRK2 RNA

Thrombin 2'-FY

IgG 2’-FY

C5a L-RNA/

CCL2 L-RN A

Nucleosome DNA

Thrombin DNA

vWF DNA

PfLDH DNA

PDGF-BB mDNA

IL-6 mDNA

NGF mDNA 4 7 14 14 39 0 29 29 0 57 4 0 68

3

6

.5 348

.0 870

3 110 0.58 899

1 0.2 .67 2599

.78 403

10 1. .56 614

16 0. .75 1073

.74 477

13 0.02 0.70 913

13 1.

0 2 0.74

0 5 0.64

5 0

2 20 0

2 0

5 0

8 79 0

4 0.71 916

13

27 25 0.75 657

13 0.63 0.75 1011

22 40 0.75 1276

17 0.02 0.80 1225

34 0. .72 1248

32 0.

2 0

2 0.77 1097 32

C3'-endo-like C2'-endo-like

GlycosidicBond

Kd(nM)

SCIndex

InterfaceArea(Å2)

See Refs. [16,17,19��,20–31,32��,52–54].

For glycosidic bonds, the anti conformation is strongly

preferred, as expected (Table 1). The ten RNA-based

aptamers exhibit a range of 0–16% syn conformation

content, with the four aptamers to nucleic acid-binding

proteins exhibiting a lower and narrower range (0–3%)

compared with the ten aptamers to non-nucleic acid-

binding proteins (2–16%). The six DNA-based aptamers,

which include three SOMAmers, overall show a higher synconformation content (13–34%), which is more often

associated with the C20-endo sugar pucker [40]. The

up-down–up-down G-quartet motif (with two syn con-

formations in each tetrad), which comprises the entire

thrombin aptamer and one domain of the IL-6 SOMA-

mer, contributes some of the syn conformation content. In

a recent study of functional RNAs, aptamers were found

to have the largest fraction of syn nucleobases per nucle-

otide [40], indicating the more compact form of the

nucleotide better accommodates the active folded struc-

ture. For RNA aptamers shown in Figure 1, most of the

syn nucleotides are observed in loops and tight turn

regions with no base pairing to other nucleotides

(76%), whereas in DNA aptamers, only 31% are observed

in unpaired nucleotides, with the balance being in stems,

G-quartets or other base-paired motifs. As expected, the

majority of syn nucleotides (83%) are purines [41].


We also examined the backbone conformations using the

method developed by Duarte and Pyle in which two

pseudobonds connecting the phosphorus and C40 carbon

atoms on adjacent nucleotides and their corresponding

pseudoangles (h and u) are used as a shorthand represen-

tation of the six rotamers, which is then amenable to a

two-dimensional representation analogous to the Rama-

chandran plot in proteins [42]. Rather than being an

oversimplification, this method benefits from filtering

out the noise inherent in individual torsion angles (which

tend to make compensating changes within the context of

a canonical A-form helix) and has been successfully used

to identify novel, sequence-independent motifs in RNA

[43]. The h–u plots are particularly useful for detecting

deviations from the standard helical geometry, which is

clearly apparent in all aptamer structures (Figure 1).

For RNA aptamers, 44% of nucleotides have backbone

conformations outside of the helical region (150–1908 for

h and 190–2608 for u) [42] compared with 37% in ribo-

somal RNA and 34% in tRNA and group 1 intron

(Figure 3a,b). Without a precedent in the literature for

h and u bands for DNA helical region, and to make the

analysis of deviations from standard helical geometry

comparable to that used for RNA, we kept the same

h–u ranges (408 for h and 708 for u) but centered them to



Figure 3

0

90

180

270

360DNA Aptamers & SOMAmers

RNA Aptamers(a) (b)

(c) (d)

RNA

DNA

vWF

nucleosome1

nucleosome2

p73

NFY

Est1

FOXM1

Hot1

Pot1

RPA70

TEBP

CCL2

MS2

S8

minE Lysozyme

Group I intron

tRNA

70s ribosome

GRK2NFkB

GlnRS

Thrombin RNA

IgG

C5a

PfLDH

Thrombin DNA

PDGF

IL-6

NGF

90 180

η

θ

90

180

270

360

θ

90

180

270

360

θ

90

180

270

360

θ

270 360 0 90 180

η

270 360

0 90 180

η

270 360 0 90 180

η

270 360


Two-dimensional backbone conformation analysis using pseudoangles h and u. (a) RNA aptamers (b) ribosomal RNA, tRNA and group I intron (c)

DNA aptamers and SOMAmers (d) double stranded DNA and single stranded DNA. The helical range is represented by the intersection of the two

gray bars. Values of pseudoangles are inverted for Speigelmers.

maximize the percentage of nucleotides within the heli-

cal bands (150–1908 for h and 170–2408 for u). With these

assumptions, 60% of nucleotides are outside the helical

region for conventional aptamers and 68% for SOMA-

mers, compared with 27% for dsDNA (Figure 3c,d).

Structures of several ‘single-stranded’ DNAs in the data-

base unsurprisingly are also high in the percent of nucleo-

tides outside the helical region (65%). Aside from being a

measure of the conformational richness of the backbone

of aptamers, this type of analysis may be useful for

identifying specific motifs utilized by RNA and DNA

aptamers that are evolved to recognize proteins with high

affinity, representing an exciting area of future research.

Interaction surfacesProtein epitopes recognized by most aptamers are pre-

dominantly electropositive and, aside from stacking inter-

actions involving the bases, are dominated by hydrogen

bonds and charge–charge interactions. The sum of all

hydrogen bonds and charge–charge interactions increases

approximately linearly with the interface area for the

conventional aptamers ([30,36] and graphical abstract).

The three SOMAmers reported to date fall outside of this


trend line, with fewer polar contacts per interface area,

reflecting their ability to interact with more hydrophobic

surfaces on proteins and resulting in access to a wider

range of accessible epitopes [44]. The sizes of epitopes

engaged by aptamers and SOMAmers combined show

extensive overlap with those observed with antibodies

(560–1300 A2) but cover a wider range (348–2599 A2)

(Table 1). The extreme outlier at the large end of the

range is the GlnRS ‘aptamer’, which is a mutant of

tRNAGln, and therefore clearly a special case. Neverthe-

less, this example illustrates that, since random regions of

any length can be used in SELEX, aptamer interaction

surfaces are not constrained by the more narrowly defined

geometry of antigen combining regions of antibodies.

A high degree of torsional flexibility, a large sequence

space and an array of secondary and tertiary structure

motifs create the possibility for generating ligands with

good shape complementarity with a wide range of con-

tours on protein surfaces. Aptamer structures shown in

Figures 1 and 2 provide visually intuitive evidence that

this is indeed the case. To quantify this observation, we

have used a method described by Lawrence and Colman



Figure 4

Aptamers0.4

0.5

0.6

0.7

0.8

0.9

Sc

Inde

x

Antibodies


Shape complementarity index (Sc) for aptamers compared to

antibodies. The Sc analysis for the 16 aptamer co-crystal structures

has a mean value of 0.71 � 0.05 (SD) (gray line), and a median value

of 0.74 (black line) while the 34 antibodies analyzed have a mean

value of 0.69 � 0.06 (gray line) and a median value of 0.70 (black line).

Symbol coloring: SOMAmers, red; DNA aptamers, orange; RNA

aptamers, blue; antibodies, gray. PDB IDs of antibody/antigen

complexes analyzed are: 4OGY, 5EBE, 4UU9, 4QCI, 4EDW, 4CNI,

2UZI, 2R69, 2BDN, 1OAK, 2R0K, 2Q8B, 2NY7, 2JEL, 2DD8, 2I9L,

1ADQ, 1BJ1, 1BQL, 1BVK, 1C08, 1DQJ, 1E6J, 1EO8, 1FBI, 1FDL,

1G6V. Sc analyses were performed using the program available as

part of the CCP4 suite [51].

[45] to compute the shape complementarity index (SC) for

the 16 aptamer co-crystal structures (Table 1), first ap-

plied to protein–aptamer interaction by Long et al. [23].

SC values for the 16 aptamers have a mean value of

0.71 � 0.05 (SD), and a median value of 0.74. With addi-

tional side chains to follow contours on protein surfaces,

SOMAmers are unsurprisingly at the higher end of the

range with a mean value of 0.76 � 0.03 and a median

value of 0.77 (Table 1, Figure 4). For comparison, SC

values for 34 antibodies in the PDB database exhibit a

mean value of 0.69 � 0.06 and a median value of 0.70

(Figure 4). These two different classes of macromolecules

therefore show a similarly high degree of shape comple-

mentarity with protein surfaces they recognize, a feature

undoubtedly contributing to their high binding affinity

and specificity.

Conformational change on bindingHow much conformational change takes place in proteins

and aptamers as they form complexes? In 15 out of

16 structures, aptamer binding does not change the con-

formation of the protein from that observed in the unbound

state. The observed range of root mean square (RMS)

deviations of the protein backbone atoms of aptamer-


bound and unbound structures is 0.3–1.8 A. For the

remaining structure (NFkB), only the DNA-bound state

is available for comparison, with an RMS value of 5.1 A

overall, but only 1.2 A in the N-terminal domains.

Far less is known about the conformational changes of

aptamers upon binding their protein targets. To date,

there are only two reported solution structures of free

aptamers that have corresponding co-crystal structures:

the S8 aptamer [19��] and the NFkB aptamer [46]. The S8

aptamer study indicates that the RNA undergoes a con-

siderable conformational change upon binding, adopting

a conformation similar to that of the 16S rRNA sequence

that binds to S8. Conversely, the NFkB RNA aptamer

appears to contain a pre-formed stem-loop structure that

resembles the native B-form DNA binding partner. The

RNA undergoes induced fit in an internal loop and the

terminal tetraloop upon binding the protein. Induced fit

has also been observed by NMR studies with NX1838,

the aptamer in pegaptanib [47]. Most aptamers that bind

to small molecules undergo conformational changes upon

binding [3]. These examples, along with those from

ribosomal RNA, tRNA and U1A RNA [48], suggest that

aptamers tend to exhibit greater plasticity in accommo-

dating the conformations that are optimal for complex

formation. The remarkable shape complementarity

exhibited by aptamers (Figure 4) is evidently worth

the thermodynamic cost associated with conformational

change upon binding.

Concluding remarksFrom a small but growing set of structures of aptamers

bound to proteins, some common themes have emerged.

Aptamers take advantage of known nucleic acid structural

motifs to create a scaffold on which nucleotides that

interact with the protein are precisely organized and

displayed. Entirely novel motifs observed with SOMA-

mer structures show that this repertoire can be substan-

tially expanded with modified libraries. The size of the

interaction surface of aptamers is, overall, comparable to

that observed with antibodies, although the range is wider

for aptamers with outliers on both the low and high ends.

Electropositive surfaces are the dominant feature, except

where hydrophobic interactions are made possible with

hydrophobic side chains on modified nucleotides. The

conformational flexibility of the phosphodiester back-

bone allows for extraordinarily high shape complemen-

tarity with epitopes on the protein surface. In most cases,

complex formation takes place without a major change in

the conformation of the protein. These observations,

although obtained from only 16 structures, go a long

way toward providing a solid structural rationale for the

high affinity and specificity observed with aptamers. With

aptamers (including SOMAmers) now identified to more

than 3000 proteins, we clearly have much more to learn

about the structural rules that govern high affinity binding

to proteins.



Conflict of interest statementADG and NJ are employees and shareholders of Soma-

Logic, Inc. DRD is an employee and shareholder of

Beryllium.

Acknowledgements

We thank Dr. Dom Zichi and Thomas Hraha at SomaLogic, Inc. for theirassistance with h–u analysis and Sc index plots. SOMAmer1 is a registeredtrademark of SomaLogic, Inc.

References and recommended readingPapers of particular interest, published within the period of review,have been highlighted as:

� of special interest�� of outstanding interest

1. Ellington AD, Szostak JW: In vitro selection of RNA moleculesthat bind specific ligands. Nature 1990, 346:818-822.

2. Tuerk C, Gold L: Systematic evolution of ligands by exponentialenrichment: RNA ligands to bacteriophage T4 DNApolymerase. Science 1990, 249:505-510.

3. Hermann T, Patel DJ: Adaptive recognition by nucleic acidaptamers. Science 2000, 287:820-825.

4. Kim SH, Sussman JL, Suddath FL, Quigley GJ, McPherson A,Wang AH, Seeman NC, Rich A: The general structure of transferRNA molecules. Proc Natl Acad Sci U S A 1974, 71:4970-4974.

5. Robertus JD, Ladner JE, Finch JT, Rhodes D, Brown RS, Clark BF,Klug A: Structure of yeast phenylalanine tRNA at 3 Aresolution. Nature 1974, 250:546-551.

6. Uhlenbeck OC, Pardi A, Feigon J: RNA structure comes of age.Cell 1997, 90:833-840.

7. Hendrix DK, Brenner SE, Holbrook SR: RNA structural motifs:building blocks of a modular biomolecule. Q Rev Biophys 2005,38:221-243.

8. Holbrook SR: RNA structure: the long and the short of it. CurrOpin Struct Biol 2005, 15:302-308.

9. Leontis NB, Lescoute A, Westhof E: The building blocks andmotifs of RNA architecture. Curr Opin Struct Biol 2006,16:279-287.

10. Klosterman PS, Tamura M, Holbrook SR, Brenner SE: SCOR: astructural classification of RNA database. Nucleic Acids Res2002, 30:392-394.

11. Nagaswamy U, Larios-Sanz M, Hury J, Collins S, Zhang Z, Zhao Q,Fox GE: NCIR: a database of non-canonical interactions inknown RNA structures. Nucleic Acids Res 2002, 30:395-397.

12. Tamura M, Hendrix DK, Klosterman PS, Schimmelman NR,Brenner SE, Holbrook SR: SCOR. Structural classification ofRNA, version 2.0. Nucleic Acids Res 2004, 32:D182-D184.

13.��

McKeague M, McConnell EM, Cruz-Toledo J, Bernard ED, Pach A,Mastronardi E, Zhang X, Beking M, Francis T, Giamberardino Aet al.: Analysis of in vitro aptamer selection parameters. J MolEvol 2015 http://dx.doi.org/10.1007/s00239-015-9708-6.

This paper summarizes two decades of work in the field of SELEX andreports on methods and parameters that most significantly impact thesuccess of a SELEX experiment.

14.��

Ozer A, Pagano JM, Lis JT: New technologies provide quantumchanges in the scale, speed, and success of SELEX methodsand aptamer characterization. Mol Ther Nucleic Acids 2014,3:e183.

This review summarizes recent developments in SELEX methods anddiscusses factors that influence SELEX outcomes.

15. Mirmira SR, Tinoco I Jr: NMR structure of a bacteriophage T4RNA hairpin involved in translational repression. Biochemistry1996, 35:7664-7674.


16. Bullock TL, Sherlin LD, Perona JJ: Tertiary core rearrangementsin a tight binding transfer RNA aptamer. Nat Struct Biol 2000,7:497-504.

17. Convery MA, Rowsell S, Stonehouse NJ, Ellington AD, Hirao I,Murray JB, Peabody DS, Phillips SE, Stockley PG: Crystalstructure of an RNA aptamer-protein complex at 2.8 Aresolution. Nat Struct Biol 1998, 5:133-139.

18. Rowsell S, Stonehouse NJ, Convery MA, Adams CJ, Ellington AD,Hirao I, Peabody DS, Stockley PG, Phillips SE: Crystal structuresof a series of RNA aptamers complexed to the same proteintarget. Nat Struct Biol 1998, 5:970-975.

19.��

Davlieva M, Donarski J, Wang J, Shamoo Y, Nikonowicz EP:Structure analysis of free and bound states of an RNA aptameragainst ribosomal protein S8 from Bacillus anthracis. NucleicAcids Res 2014, 42:10795-10808.

There are only two structures of aptamers in the free and bound state.This is the first example of an RNA aptamer that undergoes a conforma-tional change upon binding its target.

20. Huang DB, Vu D, Cassiday LA, Zimmerman JM, Maher LJ 3rd,Ghosh G: Crystal structure of NF-kappaB (p50)2 complexed toa high-affinity RNA aptamer. Proc Natl Acad Sci U S A 2003,100:9268-9273.

21. Padlan CS, Malashkevich VN, Almo SC, Levy M, Brenowitz M,Girvin ME: An RNA aptamer possessing a novel monovalentcation-mediated fold inhibits lysozyme catalysis by inhibitingthe binding of long natural substrates. RNA 2014, 20:447-461.

22. Tesmer VM, Lennarz S, Mayer G, Tesmer JJ: Molecularmechanism for inhibition of g protein-coupled receptorkinase 2 by a selective RNA aptamer. Structure 2012,20:1300-1309.

23. Long SB, Long MB, White RR, Sullenger BA: Crystal structure ofan RNA aptamer bound to thrombin. RNA 2008, 14:2504-2512.

24. Nomura Y, Sugiyama S, Sakamoto T, Miyakawa S, Adachi H,Takano K, Murakami S, Inoue T, Mori Y, Nakamura Y et al.:Conformational plasticity of RNA for target recognition asrevealed by the 2.15 A crystal structure of a human IgG-aptamer complex. Nucleic Acids Res 2010, 38:7822-7829.

25. Oberthur D, Achenbach J, Gabdulkhakov A, Buchner K,Maasch C, Falke S, Rehders D, Klussmann S, Betzel C: Crystalstructure of a mirror-image L-RNA aptamer (Spiegelmer) incomplex with the natural L-protein target CCL2. Nat Commun2015, 6:6923.

26. Yatime L, Maasch C, Hoehlig K, Klussmann S, Andersen GR,Vater A: Structural basis for the targeting of complementanaphylatoxin C5a using a mixed L-RNA/L-DNA aptamer. NatCommun 2015, 6:6481.

27. Cheung YW, Kwok J, Law AW, Watt RM, Kotaka M, Tanner JA:Structural basis for discriminatory recognition of Plasmodiumlactate dehydrogenase by a DNA aptamer. Proc Natl Acad Sci US A 2013, 110:15967-15972.

28. Huang RH, Fremont DH, Diener JL, Schaub RG, Sadler JE: Astructural explanation for the antithrombotic activity ofARC1172, a DNA aptamer that binds von Willebrand factordomain A1. Structure 2009, 17:1476-1484.

29. Russo Krauss I, Merlino A, Giancola C, Randazzo A, Mazzarella L,Sica F: Thrombin-aptamer recognition: a revealed ambiguity.Nucleic Acids Res 2011, 39:7858-7867.

30. Davies DR, Gelinas AD, Zhang C, Rohloff JC, Carter JD,O’Connell D, Waugh SM, Wolk SK, Mayfield WS, Burgin AB et al.:Unique motifs and hydrophobic interactions shape the bindingof modified DNA ligands to protein targets. Proc Natl Acad SciU S A 2012, 109:19971-19976.

31. Gelinas AD, Davies DR, Edwards TE, Rohloff JC, Carter JD,Zhang C, Gupta S, Ishikawa Y, Hirota M, Nakaishi Y et al.: Crystalstructure of interleukin-6 in complex with a modified nucleicacid ligand. J Biol Chem 2014, 289:8720-8734.

32.��

Jarvis TC, Davies DR, Hisaminato A, Resnicow DI, Gupta S,Waugh SM, Nagabukuro A, Wadatsu T, Hishigaki H, Gawande Bet al.: Non-helical DNA triplex forms a unique aptamer scaffold


http://refhub.elsevier.com/S0959-440X(16)00012-9/sbref0220

































http://dx.doi.org/10.1007/s00239-015-9708-6












































































for high affinity recognition of nerve growth factor. Structure2015, 23:1293-1304.

This paper reports the highly unusual structure of a SOMAmer bound toits target. The SOMAmer lacks any helical content and has an atypicaltriangular prism configuration.

33. Leontis NB, Westhof E: Analysis of RNA motifs. Curr Opin StructBiol 2003, 13:300-308.

34. Phan AT, Kuryavyi V, Ma JB, Faure A, Andreola ML, Patel DJ: Aninterlocked dimeric parallel-stranded DNA quadruplex: apotent inhibitor of HIV-1 integrase. Proc Natl Acad Sci U S A2005, 102:634-639.

35. Gold L, Ayers D, Bertino J, Bock C, Bock A, Brody EN, Carter J,Dalby AB, Eaton BE, Fitzwater T et al.: Aptamer-basedmultiplexed proteomic technology for biomarker discovery.PLoS One 2010, 5:e15004.

36. Rohloff JC, Gelinas AD, Jarvis TC, Ochsner UA, Schneider DJ,Gold L, Janjic N: Nucleic acid ligands with protein-like sidechains: modified aptamers and their use as diagnostic andtherapeutic agents. Mol Ther Nucleic Acids 2014, 3:e201.

37. Vaught JD, Bock C, Carter J, Fitzwater T, Otis M, Schneider D,Rolando J, Waugh S, Wilcox SK, Eaton BE: Expanding thechemistry of DNA for in vitro selection. J Am Chem Soc 2010,132:4141-4151.

38. Mian IS, Bradwell AR, Olson AJ: Structure, function and propertiesof antibody binding sites. J Mol Biol 1991, 217:133-151.

39. Ramaraj T, Angel T, Dratz EA, Jesaitis AJ, Mumey B: Antigen–antibody interface properties: composition, residueinteractions, and features of 53 non-redundant structures.Biochim Biophys Acta 1824, 2012:520-532.

40. Sokoloski JE, Godfrey SA, Dombrowski SE, Bevilacqua PC::Prevalence of syn nucleobases in the active sites of functionalRNAs. RNA 2011, 17:1775-1787.

41. Kligun E, Mandel-Gutfreund Y: The role of RNA conformation inRNA–protein recognition. RNA Biol 2015, 12:720-727.

42. Duarte CM, Pyle AM: Stepping through an RNA structure: anovel approach to conformational analysis. J Mol Biol 1998,284:1465-1478.


43. Wadley LM, Pyle AM: The identification of novel RNAstructural motifs using COMPADRES: an automatedapproach to structural discovery. Nucleic Acids Res 2004,32:6650-6659.

44. Ochsner UA, Green LS, Gold L, Janjic N: Systematic selection ofmodified aptamer pairs for diagnostic sandwich assays.Biotechniques 2014, 56:125-128 130, 132-123.

45. Lawrence MC, Colman PM: Shape complementarity at protein/protein interfaces. J Mol Biol 1993, 234:946-950.

46. Reiter NJ, Maher LJ 3rd, Butcher SE: DNA mimicry by a high-affinity anti-NF-kappaB RNA aptamer. Nucleic Acids Res(36):2008:1227-1236.

47. Lee JH, Jucker F, Pardi A: Imino proton exchange rates imply aninduced-fit binding mechanism for the VEGF165-targetingaptamer, Macugen. FEBS Lett 2008, 582:1835-1839.

48. Williamson JR: Induced fit in RNA–protein recognition. NatStruct Biol 2000, 7:834-837.

49. DeLano WL: The PyMOL Molecular Graphics System. DeLanoScientific; 2002.

50. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA:Electrostatics of nanosystems: application to microtubulesand the ribosome. Proc Natl Acad Sci U S A 2001, 98:10037-10041.

51. Collaborative Computational Project N: The CCP4 suite:programs for protein crystallography. Acta Crystallogr D: BiolCrystallogr 1994, 50:760-763.

52. Ban N, Nissen P, Hansen J, Moore PB, Steitz TA: The completeatomic structure of the large ribosomal subunit at 2.4 Aresolution. Science 2000, 289:905-920.

53. Padavattan S, Shinagawa T, Hasegawa K, Kumasaka T, Ishii S,Kumarevel T: Structural and functional analyses of nucleosomecomplexes with mouse histone variants TH2a and TH2b,involved in reprogramming. Biochem Biophys Res Commun2015, 464:929-935.

54. Qin X, Hao Z, Tian Q, Zhang Z, Zhou C, Xie W: Cocrystalstructures of glycyl-tRNA synthetase in complex with tRNAsuggest multiple conformational states in glycylation. J BiolChem 2014, 289:20359-20369.










































































embracing proteins: structural themes in aptamer–protein ... · orange coloring on the cartoons...

Documents