REVIEW SPECIAL ISSUE: RECONSTITUTING CELL BIOLOGY

Reconstituting the reticular ER network –mechanistic implicationsand open questionsNing Wang and Tom A. Rapoport*

ABSTRACTThe endoplasmic reticulum (ER) is a major membrane-boundorganelle in all eukaryotic cells. This organelle comprisesmorphologically distinct domains, including the nuclear envelopeand peripheral sheets and tubules. The tubules are connected bythree-way junctions into a network. Several membrane proteins havebeen implicated in network formation; curvature-stabilizing proteinsgenerate the tubules themselves, and membrane-anchoredGTPases fuse tubules into a network. Recent experiments haveshown that a tubular network can be formed with reconstitutedproteoliposomes containing the yeast membrane-fusing GTPaseSey1 and a curvature-stabilizing protein of either the reticulon orREEP protein families. The network forms in the presence of GTPand is rapidly disassembled when GTP hydrolysis of Sey1 isinhibited, indicating that continuous membrane fusion is required forits maintenance. Atlastin, the ortholog of Sey1 in metazoans, forms anetwork on its own, serving both as a fusion and curvature-stabilizingprotein. These results show that the reticular ER can be generated bya surprisingly small set of proteins, and represents an energy-dependent steady state between formation and disassembly. Modelsfor the molecular mechanism by which curvature-stabilizing proteinscooperate with fusion GTPases to form a reticular network have beenproposed, but many aspects remain speculative, including thefunction of additional proteins, such as the lunapark protein, andthe mechanism by which the ER interacts with the cytoskeleton. Howthe nuclear envelope and peripheral ER sheets are formed remainmajor unresolved questions in the field. Here, we review reconstitutionexperiments with purified curvature-stabilizing proteins and fusionGTPases, discuss mechanistic implications and point out openquestions.

KEY WORDS: Cell biology, Endoplasmic reticulum, Reconstitution,Reticulon, Atlastin, Lunapark

IntroductionAll organelles have characteristic shapes, but how their morphologyis generated is largely unknown. The endoplasmic reticulum (ER) isa particularly intriguing organelle, as it consists of morphologicallydistinct domains that change during differentiation and the cellcycle. In interphase, the ER consists of the nuclear envelope and aconnected peripheral network of tubules and interspersed sheets(Baumann and Walz, 2001; Shibata et al., 2009; Terasaki et al.,1984; Voeltz et al., 2006). The tubules continuously form, retractand slide along one another, a process facilitated by associatedmolecular motors and dynamic changes of the cytoskeleton. During

mitosis in metazoans, the nuclear envelope disassembles andperipheral ER tubules appear to be transformed into sheets (Luet al., 2009; Wang et al., 2013). How the network is generated andmaintained, and how its morphology changes during the cell cycle,is poorly understood.

The tubules themselves are shaped by two evolutionarilyconserved protein families, the reticulons (Rtns) and REEPs(Yop1 in yeast) (Voeltz et al., 2006). These are ubiquitousmembrane proteins that are both necessary and sufficient togenerate tubules (Hu et al., 2008). They belong to the ten most-abundant membrane proteins in a cell and together occupy ∼10% ofthe membrane surface of the reticular ER in Saccharomycescerevisiae (Hu et al., 2008). The Rtns and REEPS stabilize the highmembrane curvature seen in cross-sections of tubules and sheetedges. How these proteins generate and stabilize membranecurvature is uncertain, as there are no structures available, butthey all contain pairs of closely spaced transmembrane (TM)segments and have an amphipathic helix (Fig. 1A).

Connecting tubules into a network requires membrane fusion,which is mediated by membrane-anchored GTPases, the atlastins(ATLs) in metazoans and Sey1 and related proteins in yeast andplants (Hu et al., 2009; Orso et al., 2009). Mammals have threeclosely related ATLs, with ATL1 being prominently expressed inneurons, and ATL2 and ATL3 more broadly in different cell types(Zhu et al., 2003). These proteins contain a cytoplasmic GTPasedomain, followed by a helical bundle, a membrane-embeddedregion and an amphipathic helix (Fig. 1B). Until recently, themembrane-bound region was thought to comprise two closelyspaced TM segments, but it may actually consist of twointramembrane hairpin loops (Betancourt-Solis et al., 2018),similar to those in the Rtns and REEPs. Crystal structures andbiochemical experiments have led to a model in which ATLmolecules localized to different membranes dimerize through theirGTPase domains, and undergo a conformational change during theGTPase cycle, thereby pulling the two membranes together andfusing them (for a review, see Hu and Rapoport, 2016). Mutations inATLs and REEPs can cause hereditary spastic paraplegia (Hübnerand Kurth, 2014; Salinas et al., 2008; Westrate et al., 2015), aneurodegenerative disease that is characterized by length-dependentaxonal dysfunction or degeneration.

The fusion of membrane tubules generates three-way junctions,which are small, triangular sheets with negatively curved edge lines(Shemesh et al., 2014). It has been suggested that these junctions arestabilized by the lunapark protein (Lnp, also known as LNPK)(Chen et al., 2012, 2015; Shemesh et al., 2014), a conservedmembrane protein containing two closely spaced TM segments andseveral other domains (Fig. 1C).

Peripheral ER sheets might be generated by the Rtns and REEPsstabilizing the highly curved membrane edges. Consistent withthis model, overexpression of these proteins in S. cerevisiae andmammalian cells favors the formation of tubules over sheets (Shibata

HowardHughesMedical Institute and Department of Cell Biology, Harvard MedicalSchool, 240 Longwood Ave, Boston, MA 02115, USA.

*Author for correspondence (tom_rapoport@hms.harvard.edu)

T.A.R., 0000-0001-9911-4216

1

© 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

et al., 2010; Voeltz et al., 2006). Conversely, sheets become moreabundant at the expense of tubules when these proteins are absent(Shibata et al., 2010; Voeltz et al., 2006) or when phosphopholipidsynthesis is increased (Shibata et al., 2010). In higher eukaryotes,sheet formation is stimulated by coiled-coil containing membraneproteins, such as p180 (also known as RRBP1), kinectin (alsoknown as KTN1) and CLIMP-63 (also known as CKAP4), with thelatter forming luminal bridges between the apposing membranes(Shibata et al., 2010).How the nuclear envelope is formed remains unclear.

Experiments in Xenopus egg extracts have suggested that bothATL- and ER SNARE-mediated membrane fusion steps arerequired to form the double membrane around chromatin (Wanget al., 2013), but how exactly membranes bind and fuse on thesurface of chromatin at the end of mitosis remains to be elucidated.Reconstituted systems containing purified proteins have been

instrumental in understanding the mechanism of many biologicalprocesses, including transcription (Ge et al., 1996; Sayre et al.,1992), replication (Waga and Stillman, 1994; Yeeles et al., 2015),protein translocation (Akimaru et al., 1991; Brundage et al., 1990;Görlich and Rapoport, 1993) and ER-associated protein degradation(ERAD) (Baldridge and Rapoport, 2016). This Review focuses onthe ongoing efforts to reconstitute the ER. Our goal is to establishreconstituted systems that recapitulate the formation of different ER

morphologies. As a first step, we concentrated on the formation of atubular ER network. In this Review, we will first discuss thereconstitution of membrane tubules by the curvature-stabilizingproteins of the Rtn and REEP families, then cover the reconstitutionof membrane fusion by the ATL/Sey1 GTPases, and finally theformation of a reticular membrane network by co-reconstitution ofthe curvature-stabilizing proteins and fusion GTPases. At the end ofthis Review, we point out open questions and future directions.

Tubule formation by reconstituted Rtns or REEPsOur initial reconstitution experiments focused on the role of theRtns and REEPs in tubule formation (Hu et al., 2008). Yeast Yop1(the single REEP family member in S. cerevisiae) and Rtn1 wereexpressed in Escherichia coli or S. cerevisiae and purified to nearhomogeneity in detergent. After mixing with phospholipids, thedetergent was removed and the samples were analyzed by negative-stain electron microscopy (EM). At early time points (a few hours)during detergent removal, small particles were observed. Thesemight be small vesicles, but their small size and the lack of anobvious luminal space raises the possibility that they are micellescontaining a mixture of protein, lipids and detergent. Ultimately,after longer times of detergent removal, membrane tubules wereobserved that reached several hundred nanometers in length,indicating that the Rtns and REEPs alone can generate the highmembrane curvature that is characteristic of tubules when viewed incross-section. The tubules had a luminal space, but were muchnarrower than ER tubules observed in yeast cells (17 nm versus30 nm). This can be explained by the significantly higher protein-to-lipid ratio (by a factor of ∼20). Indeed, when Rtn or REEPs wereoverexpressed in yeast, plant or mammalian cells, ER tubulesbecame so narrow that luminal proteins were forced to move to otherareas of the ER (Hu et al., 2008; Tolley et al., 2008). Interestingly,ER tubules with a diameter of ∼20 nm are common in neuronalaxons (Terasaki, 2018), indicating that narrow tubules can bephysiological. However, it is unknown whether these tubulescontain a particularly high concentration of curvature-stabilizingproteins.

The reconstitution experiments are consistent with resultsobtained with more physiological systems. For example,antibodies to an Rtn isoform (Rtn4a) block the formation of atubular ER network in Xenopus egg extracts (Voeltz et al., 2006).Although the exact mode of action of the antibodies remainsunclear, those experiments provided the first evidence that the Rtnsare involved in tubule formation. Further evidence was provided byexperiments with intact cells. Deletion of the Rtns and Yop1 in S.cerevisiae, and depletion of the proteins in mammalian cells,resulted in the conversion of ER tubules into sheets (Anderson andHetzer, 2008; Voeltz et al., 2006; Wang et al., 2016); andoverexpression of these proteins led to long tubules (Voeltz et al.,2006; Wang et al., 2016). Finally, all these proteins localize totubules or sheet edges and avoid the low curvature areas of sheets(Shemesh et al., 2014; Shibata et al., 2010; Voeltz et al., 2006;Wang et al., 2016). Targeting to the high-curvature regions appearsto be mediated by the TM hairpins, which can also be found in allthe other proteins that specifically localize to these areas, such asATLs, Sey1 and Lnp. Taken together with the reconstitutionexperiments, these results indicate that the Rtns and REEPs stabilizehigh membrane curvature.

The mechanism by which these proteins induce and stabilize highmembrane curvature remains unclear. It has been proposed that theyutilize two cooperating mechanisms, hydrophobic insertion(wedging) and scaffolding (Fig. 1D) (reviewed in Shibata et al.,

N C

Amphipathic helix

Lumen

Cytosol

A Reticulons/REEPs

Helix bundle

GTPase

Lumen

CytosolC

N

B ATL/Sey1

C Lunapark

NCoiled-coil

domain

Lumen

Cytosol

P Zn

D

Fig. 1. Schemes for the structures of ER-shaping proteins. (A) Membranetopology of the curvature-stabilizing proteins of the reticulon (Rtn) and REEPfamilies. The double-hairpin trans-membrane (TM) segments (pink), togetherwith the C-terminal amphipathic helix (hydrophobic and hydrophilic surfacesin blue and yellow, respectively) are thought to induce positive curvature ofthe ER membrane. Based on NMR studies, the TM segments are thought tofully span the membrane (Brady et al., 2015). (B) Domains of the fusion ATLGTPase. TheGTPase Sey1 has a similar topology, but the helix bundle is morecomplex (Yan et al., 2015). The proteins are depicted with two fully spanningTM segments, but recent experiments suggest that they have twointramembrane hairpin loops (Betancourt-Solis et al., 2018). (C) Domains ofthe lunapark protein. P is a domain that is phosphorylated during mitosis inhigher eukaryotes. Zn is a Zn2+-finger domain involved in dimerization of theprotein. (D) The Rtns and REEPs are proposed to form arc-shapedoligomers (shown in shades of pink) that bend the membrane (gray).

2

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

2009). The hydrophobic insertion mechanism postulates that thetwo hydrophobic TM hairpins in these proteins, together with theamphipathic helix, form a wedge-like structure that displacesthe lipids preferentially in the outer leaflet of the lipid bilayer, thuscausing local curvature. The scaffolding mechanism assumes thatthe Rtns and REEPs form arc-like oligomers that mold the lipidbilayer into tubules (Fig. 1D). Arc-like scaffolds, rather than rings orspirals that wrap around tubules, are consistent with the observationthat the Rtns also localize to sheet edges and that the diffusion of ERproteins along tubules is not restricted (Shibata et al., 2008, 2010;Shemesh et al., 2014). Structural information on the Rtns andREEPs are required to test these ideas.

Membrane fusion in systems reconstituted with ATL or Sey1The fusion of ER membranes could also be recapitulated withpurified proteins. Drosophila ATL was successfully purified afterexpression in E. coli and reconstituted with phospholipids intoproteoliposomes; the subsequent addition of GTP resulted in thefusion of these vesicles, as initially demonstrated by lipid mixing andsize increase of the vesicles (Orso et al., 2009), and later by contentmixing (Liu et al., 2012). The lipid composition of the membranesdoes not seem to play a major role (Orso et al., 2009). GTPase-competent ATLmolecules need to be present in both vesicles for theirfusion to happen (Orso et al., 2009). Similar results were obtainedwith Sey1, also purified after expression in E. coli (Anwar et al.,2012; Powers et al., 2017). These results indicate that these GTPasesalone can mediate the homotypic fusion of membranes.The reconstitution experiments are again in agreement with

experiments performed in physiological systems, as the depletion ofATLs or the expression of dominant-negative mutants leads to ERtubule fragmentation and the appearance of long, unbranched ERtubules in tissue culture cells (Hu et al., 2009; Wang et al., 2016) orDrosophila melanogaster (Orso et al., 2009). These phenotypes arelikely caused by insufficient fusion between the tubules. Similarly,in Xenopus egg extracts, the addition of a dominant-negative ATLmutant or of non-hydrolysable GTP analogs prevents ER networkformation (Wang et al., 2013). Surprisingly, these reagents alsocause a pre-formed ER network to disintegrate into smallermembrane structures within a few minutes (Wang et al., 2013),indicating that continuous ATL function is required to maintain theintegrity of an ER network.The molecular mechanism of ATL function has been elucidated

from crystal structures of the cytosolic domain (Bian et al., 2011;Byrnes and Sondermann, 2011) and biochemical studies (Bianet al., 2011; Byrnes and Sondermann, 2011; Liu et al., 2012, 2015).These have led to a plausible model for ATL-mediated membranefusion, which begins with the binding of GTP to ATL monomers(Fig. 2; step 1). The monomers then rapidly form a dimer, with thebound GTP molecules buried at the interface. When GTP is

hydrolyzed by the dimer, the two helical bundle domains associatewith one another, generating a tight dimer in the transition state (step2). This interaction likely pulls the two membranes together so thatthey can fuse. The C-terminal amphipathic helix may facilitatemembrane fusion, likely by perturbing the phospholipid bilayer(Liu et al., 2012). Finally, phosphate (Pi) and GDP are released in asequential manner, causing the dissociation of the dimer intomonomers (step 3). All these reactions can occur with ATLmolecules located either in different or the same membranes (trans-and cis-interactions, respectively), but membrane tethering andfusion happen only with trans-interactions (Liu et al., 2015). Theformation and disassembly of cis-dimers is thus a futile cycle.

ATLs also form trans-dimers in a futile manner (Liu et al., 2015;Saini et al., 2014). Experiments with reconstituted proteoliposomeshave shown that multiple GTPase cycles are required for asuccessful fusion event, as the transition state of trans-dimersoften converts them back into monomers without having causedfusion (Liu et al., 2015). Other experiments have indicated that theefficiency of tethering and fusion increases with the density of ATLmolecules in the two membranes (Liu et al., 2015). ATL moleculesappear to associate with one another through their TM segments. Infact, the TM segments cannot be replaced by those from otherproteins, and point mutations in these regions can reduce theefficiency of fusion (Liu et al., 2012). Recent experiments suggestthat the TMs form two intramembrane hairpin loops, rather than twosegments that completely span the membrane (Betancourt-Soliset al., 2018), which may explain their unique role in fusion. Takentogether, it appears that a successful fusion event requires thecooperation of multiple trans-dimerization events: as one dimerreaches the transition state, other dimers forming nearby help tomaintain the tethered state and add to the pulling force exerted onthe opposing membranes. The futile formation of trans-dimerssuggests that a fraction of three-way junctions in cells may actuallyconsist of tethered, rather than fused, tubules, and may explain, atleast in part, why the ER network rapidly disassembles uponinhibition of the GTPase activity of ATL (Wang et al., 2013).

Reconstitution of a reticular ER networkThe next step in reconstituting the ER is to generate a reticular ERnetwork. We therefore tested whether using a curvature-stabilizingprotein, an Rtn or REEP protein, and a fusion GTPase, either ATL orSey1, is sufficient (Powers et al., 2017). To this end, we first purifiedfull-length S. cerevisiae Yop1 and Sey1 after expression in E. coliand reconstituted both proteins together into proteoliposomes.Essentially all molecules were incorporated into the vesicles withtheir cytosolic domains on the outside, as shown by proteaseprotection experiments. The reconstituted vesicles converted into areticular network upon addition of GTP, which could be visualizedwith a hydrophobic fluorescent dye in a light microscope (Fig. 3) or

GDP+PiGTP GDPStep 1 Step 2 Step 3

Fig. 2. Scheme for ATL-mediatedmembrane fusion. In step 1,ATL molecules sitting in differentmembranes bind GTP and associatewith one another. In step 2, GTP ishydrolyzed and the proteins undergoconformational changes that allow thehelical bundles to interact. The twomembranes are pulled together sothat they can fuse. In step 3, Pi andGDP are released, and the ATL dimerdissociates into monomers.

3

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

by EM. Network formation was not observed with the individualproteins. In these experiments, Yop1 could be replaced with anumber of other curvature-stabilizing proteins of the Rtn and REEPfamilies, such as Rtns from S. cerevisiae or Drosophila and REEPsfrom Drosophila and Xenopus. Our titration experiments showedthat the network forms when a physiological molar ratio of Rtn orREEP to Sey1 was used, which is ∼10:1 (Kulak et al., 2014).However, the total protein concentration required was ∼10 timeshigher than in vivo. As was the case in Xenopus extracts (Wanget al., 2013), the network rapidly disassembled when a non-hydrolysable GTP analog was added to inhibit Sey1 activity(Powers et al., 2017). Surprisingly, the entire network was dynamic,with tubules sliding passed each other. As a cytoskeleton, which incells is typically required for ER dynamics, is missing in thesereconstitutions, it is possible that thermal convection on thecoverslip provided the energy for the morphological changes.Taken together, these results indicated that an ER network can begenerated with a small set of proteins, a curvature-stabilizing proteinand a membrane-fusing GTPase.Surprisingly, reconstituted Drosophila ATL was capable of

forming a reticular network on its own (Powers et al., 2017). Thiscan be explained by the fact that ATL has both curvature-stabilizingand fusion activities. Indeed, a GTP-dependent network was formedwhen a GTPase-defective ATLmutant, which retained its curvature-stabilizing activity, was co-reconstituted with Sey1 (Powers et al.,2017). It is currently unclear whether Sey1 also has a curvature-stabilizing activity, as vesicle-flotation experiments showed that itwas less efficiently incorporated into proteoliposomes than ATL.

What is the mechanism of ER network formation?Both the in vivo and in vitro results indicate that the ER networkcorresponds to a steady-state of continuous membrane fusion andfragmentation. Fusion is mediated by the GTPases ATL or Sey1 andfragmentation is caused by the curvature-stabilizing proteins of theRtn and REEP/Yop1 families, which, in the absence of ATLactivity, appear to prefer the higher membrane curvature of smallvesicles to that of tubules. In a steady-state network, fusion mediatedby the GTPases appears to be faster than the breakage of tubules orthe shedding of small vesicles mediated by the curvature-stabilizingproteins, explaining why tubule fission is rarely observed in vivo(Friedman et al., 2010). Under normal conditions, the disassembly

of tubules may preferentially occur at free ends, as the tip of amembrane tubule already corresponds to half a vesicle. However,such free ends are infrequent in mammalian cells and in Xenopusextracts (Friedman et al., 2010; Wang et al., 2013); most tubules areanchored at three-way junctions, and free ends are often associatedwith the tip of a growing microtubule or molecular motor (Friedmanet al., 2010). Fusion of the tips of ER tubules with other tubulescontinuously generates new three-way junctions (Lee and Chen,1988) and thus increases the number of polygons in the tubularnetwork. This process is counteracted by ‘ring closure’ (Lee andChen, 1988), a process in which three-way junctions merge and thusabolish polygons. The dynamics of the ER network is different fromthat of mitochondria, which maintain their structure by a balance offusion and fission reactions (Shaw and Nunnari, 2002).

The Rtns and REEPs have been shown to form oligomers, whichcould contribute to tubule formation (Shibata et al., 2008).However, whether their oligomeric state is different in tubules andvesicles and affected by the fusion GTPases remains to be tested.The roles of the amphipathic helices of the curvature-stabilizingproteins and GTPases in network formation also needs to beclarified.

Reconstitution of Lnp and its implicationsAnother protein that is important for ER network formation is Lnp, aconserved membrane protein containing two closely spaced TMsegments, two coiled-coil regions and a Zn2+-finger domaininvolved in dimerization (Fig. 1C) (Wang et al., 2018). Themammalian protein also has a domain that is phosphorylated duringmitosis. The protein was originally discovered in a genetic screen inS. cerevisiae for ER morphology defects and recognized as acounteracting factor of the fusion GTPase Sey1 (Chen et al., 2012).It was later proposed to stabilize three-way tubular junctions (Chenet al., 2015; Shemesh et al., 2014). However, although Lnp localizespreferentially to three-way junctions of the ER, it is not essential fortheir formation. Indeed, mammalian cells lacking Lnp have moreER sheets, but the tubular network does not completely disappear,particularly remaining at the periphery of the cells (Wang et al.,2016; Zhou et al., 2018).

To better understand the function of Lnp, we purified Xenopus orhuman Lnp after expression in E. coli, before mixing the protein withlipid and removing the detergent (Wang et al., 2018). Surprisingly,negative-stain EM showed that the resulting structures were stackedmembrane discs (bicelles) (Fig. 4A), as confirmed by EM cryo-tomography. Mutagenesis indicated that the Lnp protein isincorporated into the membrane discs in both orientations; thediscs are linked with one another through the cytosolic domains ofLnp (Fig. 4B). The domain phosphorylated during mitosis isimportant for trans-interactions between Lnp molecules, and aphosphomimetic Lnpmutant showed reduced bicelle stacking (Wanget al., 2018). Inactivation of Lnp during mitosis is also suggested byexperiments with Xenopus egg extracts: treatment with dominant-negative Lnp fragments converts the tubular ER into a network ofsmall sheets that are connected by short tubules, which resembles theER morphology seen during mitosis (Wang et al., 2016).

We have proposed that Lnp localizes preferentially to three-wayjunctions because it engages in trans-interactions when junctionscome into close proximity within a 3D network (Wang et al., 2018).Three-way junctions are small, triangular membrane sheets andoffer a better geometry for trans-interactions between multiple Lnpmolecules than tubules. Lnp-containing three-way junctions maynot be able to undergo efficient membrane fusion because Lnpcould prevent ATL from entering the junctional sheets or because

−GTP +GTP

Rhod

amin

e−PE

Yop1 + Sey1

Fig. 3. Generation of a tubular membrane network with purified proteins.Purified Yop1, a curvature-stabilizing protein from S. cerevisiae thatbelongs to the REEP family, was mixed with purified Sey1, the fusionGTPase in S. cerevisiae. Phospholipids, including Rhodamine-labeledphosphoethanolamine (PE) were added, and the detergent was subsequentlyremoved to generate proteoliposomes. The vesicles were incubated withoutor with GTP (left and right panel, respectively). The samples were placed ona coverslip and visualized in a fluorescence microscope. Scale bars: 20 µm.The figure is reproduced from Powers et al., (2017) with permission fromSpringer Nature.

4

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

membranes at these regions become less deformable. In addition,Lnp inhibits ATL-mediated membrane fusion (Zhou et al., 2018). Amechanism in which Lnp counteracts ATL is consistent with theobservations that overexpression of Lnp in mammalian tissueculture cells leads to expanded junctional sheets with ATL located atthe edges and that co-overexpression of ATL restores a normal ERnetwork (Wang et al., 2016). It is also consistent with theobservation that in S. cerevisiae, Lnp and Sey1 have opposingeffects on ER morphology (Chen et al., 2012). The postulatedfunction of Lnp would be important in the center of mammaliancells, where the larger volume allows the ER network to be dense inall three directions. In contrast, at the periphery of the cell, the ERnetwork is essentially two dimensional and Lnp would not berequired to prevent excessive fusion; this also explains why inmammalian cells lacking Lnp, the residual network localizes to theperiphery of the cells (Wang et al., 2016).

Future perspectivesSignificant progress has been made in understanding how themorphology of the ER is achieved, providing a paradigm for howother membrane-bound organelles may be shaped. The recentreconstitution experiments discussed here show that the minimumsystem required to form a reticular ER network consists of acurvature-stabilizing protein, that is, a member of the Rtn or REEPfamilies, and a membrane-fusing GTPase, either ATL or Sey1(Powers et al., 2017). The resulting network represents a dynamicstate that requires continuous GTP hydrolysis by ATL/Sey1.Although in vivo the ER is not formed de novo, but rather frompre-existing membranes by addition of newly synthesized lipids andproteins, the reconstitution experiments allow mechanistic studieson how the final membrane structure is attained.

Further studies are required to fully understand the molecularmechanism of network formation. A major goal is to obtain astructure of the Rtn or REEP proteins, which will shed light on howthese proteins induce high membrane curvature. The possible role ofoligomers formed by these proteins and the significance of their TMhairpins and amphipathic helices also need to be clarified. Furtherstudies are also needed on whether a physical interaction betweenthe curvature-stabilizing proteins and the fusion-mediating GTPasesis required for network formation.

It remains puzzling that there are so many different Rtn andREEP proteins, even within a single cell. Several members of theseprotein families have large cytosolic domains of unknown function.Presumably, these domains bind other proteins and might diversifythe function of the Rtns and REEPs. In addition, several Rtn-likeproteins have been discovered, including Tts1 in S. pombe (Zhangand Oliferenko, 2014; Zhang et al., 2010) and its homologs in otherorganisms, the ER autophagy receptor FAM134 (also known asRETREG1) (Khaminets et al., 2015) and ARL6IP1, which recruitsan inositol 5-phosphatase to the ER (Dong et al., 2018). Whetherthese proteins induce highmembrane curvature and whether such aneffect is required for their functions remains to be investigated.

Another unresolved issue concerns the interaction between theER and the cytoskeleton. In mammalian cells, ER tubules are oftengenerated by the force generated by either microtubule-associatedmolecular motors or by polymerizing microtubule tips (Lee andChen, 1988; Prinz et al., 2000; Terasaki et al., 1986; Waterman-Storer and Salmon, 1998). The proteins that mediate the interactionwith motors or microtubules have not yet been identified. InS. cerevisiae, ER tubules are formed along actin filaments (Estradaet al., 2003). Although the cytoskeleton is required for thegeneration and distribution of the ER network, ultimately thealignment of ER tubules with the cytoskeleton is not perfect. Inaddition, the ER network does not collapse upon depolymerizationof actin filaments in yeast (Prinz et al., 2000), and it retracts onlywith some delay upon depolymerization of microtubules inmammalian cells (Terasaki et al., 1986). In vitro, an ER networkcan be generated in the absence of an intact cytoskeleton (Dreier andRapoport, 2000). Interestingly, p180 and CLIMP-63, the membraneproteins implicated in the generation of peripheral ER sheetsin mammalian cells, contain microtubule-binding domains(Klopfenstein et al., 1998; Ogawa-Goto et al., 2007), but theirprecise function remains unclear.

Peripheral ER sheets seem to be present in all cells, althoughsome may actually be fenestrated (Puhka et al., 2012) or composedof dense clusters of three-way tubular junctions (Nixon-Abell et al.,2016). The abundance of peripheral ER sheets varies among cellsand changes during the cell cycle. Although controversial (Puhkaet al., 2012), in some mammalian tissue culture cells (Lu et al.,2009) and in Xenopus extracts (Wang et al., 2013), ER tubulesconvert into sheets during mitosis. Sheets are generally thought tocorrespond to rough ER (Shibata et al., 2006), that is, regionscontaining membrane-bound ribosomes. The membranes of therough and smooth ER are continuous, and yet the ribosome-studdedarea is kept segregated. How such a distinct rough ER domain isformed remains unclear. The concentration of sheets in one area ofthe cell requires intact polysomes and the presence of the sheet-promoting membrane proteins p180, kinectin and CLIMP-63(Shibata et al., 2010). Perhaps, these membrane proteins linkdifferent polysomes to form a rough ER domain. How peripheral ERsheets are stacked on top of each other in highly secretory cells alsoremains to be investigated. Serial sectioning EM has demonstratedthat the sheets are connected by twisted membrane surfaces,

P Zn

P Zn

P Zn

P Zn

A B

Fig. 4. Bicelle formation by reconstituted lunapark protein. (A) Purifiedlunapark from Xenopus laevis was mixed with lipids (molar ratio 1:5) andthe detergent was removed. The sample was analyzed by negative-stain EM.The figure is reproduced from Wang et al. (2018) where it was published under aCC BY-NC-SA 3.0 licence (https://creativecommons.org/licenses/by-nc-sa/3.0/).(B) Model for lunapark-induced bicelle formation. Lunapark is assumed to sit in themembrane discs in both orientations. The edges of the membrane discs mightbe stabilized by residual detergent. Stacking of the discs is mediated by boththe phosphorylation (P) and Zn2+-finger (Zn) domains of lunapark.

5

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

resembling a parking garage in which the different levels areconnected by helicoidal ramps (Terasaki et al., 2013). However, theproteins required for sheet stacking are currently unknown.Finally, the mechanism of nuclear envelope formation remains

one of the most challenging unresolved issues. Experiments withXenopus extracts have suggested that two membrane-fusion stepsare required, one mediated by ATL to generate an ER network and asecond mediated by ER SNAREs and NSF to form the nuclearenvelope (Wang et al., 2013). However, the role of ER SNARE-mediated fusion is based on dominant-negative reagents and thusrequires further studies in vitro and in vivo. The formation of adouble membrane around chromatin in Xenopus egg extracts doesnot require nuclear pores or lamins (Newport et al., 1990; Sullivanet al., 1993). How membrane fusion is coordinated with chromatindecondensation at the end of mitosis (Ramadan et al., 2007), andhow it is confined to the surface of chromatin are also importantquestions for the future. Ultimately, the reconstitution of nuclearenvelope formation with purified components will be required tounderstand its molecular mechanism.

AcknowledgementsWe acknowledge the important contributions of many former lab members, whichformed the basis for this review.

Competing interestsThe authors declare no competing or financial interests.

FundingOur work in this area is supported by the Howard Hughes Medical Institute.

ReferencesAkimaru, J., Matsuyama, S., Tokuda, H. and Mizushima, S. (1991).Reconstitution of a protein translocation system containing purified SecY, SecE,and SecA from Escherichia coli. Proc. Natl. Acad. Sci. USA 88, 6545-6549.

Anderson, D. J. andHetzer, M.W. (2008). Reshaping of the endoplasmic reticulumlimits the rate for nuclear envelope formation. J. Cell Biol. 182, 911-924.

Anwar, K., Klemm, R. W., Condon, A., Severin, K. N., Zhang, M., Ghirlando, R.,Hu, J., Rapoport, T. A. andPrinz,W. A. (2012). The dynamin-likeGTPase Sey1pmediates homotypic ER fusion in S. cerevisiae. J. Cell Biol. 197, 209-217.

Baldridge, R. D. and Rapoport, T. A. (2016). Autoubiquitination of the Hrd1 ligasetriggers protein retrotranslocation in ERAD. Cell 166, 394-407.

Baumann, O. and Walz, B. (2001). Endoplasmic reticulum of animal cells and itsorganization into structural and functional domains. Int. Rev. Cytol. 205, 149-214.

Betancourt-Solis, M. A., Desai, T. and McNew, J. A. (2018). The atlastinmembrane anchor forms an intramembrane hairpin that does not span thephospholipid bilayer. J. Biol. Chem. 293, 18514-18524.

Bian, X., Klemm, R. W., Liu, T. Y., Zhang, M., Sun, S., Sui, X., Liu, X., Rapoport,T. A. and Hu, J. (2011). Structures of the atlastin GTPase provide insight intohomotypic fusion of endoplasmic reticulum membranes. Proc. Natl. Acad. Sci.USA 108, 3976-3981.

Brady, J. P., Claridge, J. K., Smith, P. G. and Schnell, J. R. (2015). A conservedamphipathic helix is required for membrane tubule formation by Yop1p.Proc. Natl.Acad. Sci. USA 112, E639-E648.

Brundage, L., Hendrick, J. P., Schiebel, E., Driessen, A. J. M. and Wickner, W.(1990). The purified E. coli integral membrane protein SecY E is sufficient forreconstitution of SecA-dependent precursor protein translocation.Cell 62, 649-657.

Byrnes, L. J. and Sondermann, H. (2011). Structural basis for the nucleotide-dependent dimerization of the large G protein atlastin-1/SPG3A. Proc. Natl. Acad.Sci. USA 108, 2216-2221.

Chen, S., Novick, P. and Ferro-Novick, S. (2012). ER network formation requires abalance of the dynamin-like GTPase Sey1p and the Lunapark family memberLnp1p. Nat. Cell Biol. 14, 707-716.

Chen, S., Desai, T., McNew, J. A., Gerard, P., Novick, P. J. and Ferro-Novick, S.(2015). Lunapark stabilizes nascent three-way junctions in the endoplasmicreticulum. Proc. Natl. Acad. Sci. USA 112, 418-423.

Dong, R., Zhu, T., Benedetti, L., Gowrishankar, S., Deng, H., Cai, Y., Wang, X.,Shen, K. and De Camilli, P. (2018). The inositol 5-phosphatase INPP5Kparticipates in the fine control of ER organization. J. Cell Biol. 217, 3577-3592.

Dreier, L. and Rapoport, T. A. (2000). In vitro formation of the endoplasmicreticulum occurs independently of microtubules by a controlled fusion reaction.J. Cell Biol. 148, 883-898.

Estrada, P., Kim, J., Coleman, J., Walker, L., Dunn, B., Takizawa, P., Novick, P.and Ferro-Novick, S. (2003). Myo4p and She3p are required for cortical ERinheritance in Saccharomyces cerevisiae. J. Cell Biol. 163, 1255-1266.

Friedman, J. R., Webster, B. M., Mastronarde, D. N., Verhey, K. J. and Voeltz,G. K. (2010). ER sliding dynamics and ER-mitochondrial contacts occur onacetylated microtubules. J. Cell Biol. 190, 363-375.

Ge, H., Martinez, E., Chiang, C.-M. and Roeder, R. G. (1996). Activator-dependenttranscription bymammalian RNA polymerase II: In vitro reconstitution with generaltranscription factors and cofactors. Methods Enzymol. 274, 57-71.

Gorlich, D. and Rapoport, T. A. (1993). Protein translocation into proteoliposomesreconstituted from purified components of the endoplasmic reticulum membrane.Cell 75, 615-630.

Hu, J. and Rapoport, T. A. (2016). Fusion of the endoplasmic reticulum bymembrane-bound GTPases. Semin. Cell Dev. Biol. 60, 105-111.

Hu, J., Shibata, Y., Voss, C., Shemesh, T., Li, Z., Coughlin, M., Kozlov, M. M.,Rapoport, T. A. and Prinz, W. A. (2008). Membrane proteins of the endoplasmicreticulum induce high-curvature tubules. Science 319, 1247-1250.

Hu, J., Shibata, Y., Zhu, P.-P., Voss, C., Rismanchi, N., Prinz, W. A., Rapoport,T. A. and Blackstone, C. (2009). A class of dynamin-like GTPases involved in thegeneration of the tubular ER network. Cell 138, 549-561.

Hubner, C. A. and Kurth, I. (2014). Membrane-shaping disorders: a commonpathway in axon degeneration. Brain 137, 3109-3121.

Khaminets, A., Heinrich, T., Mari, M., Grumati, P., Huebner, A. K., Akutsu,M., Liebmann, L., Stolz, A., Nietzsche, S., Koch, N. et al. (2015). Regulationof endoplasmic reticulum turnover by selective autophagy. Nature 522,354-358.

Klopfenstein, D. R. C., Kappeler, F. and Hauri, H.-P. (1998). A novel directinteraction of endoplasmic reticulum with microtubules. EMBO J. 17, 6168-6177.

Kulak, N. A., Pichler, G., Paron, I., Nagaraj, N. and Mann, M. (2014). Minimal,encapsulated proteomic-sample processing applied to copy-number estimation ineukaryotic cells. Nat. Methods 11, 319-324.

Lee, C. andChen, L. B. (1988). Dynamic behavior of endoplasmic reticulum in livingcells. Cell 54, 37-46.

Liu, T. Y., Bian, X., Sun, S., Hu, X., Klemm, R. W., Prinz, W. A., Rapoport, T. A.and Hu, J. (2012). Lipid interaction of the C terminus and association of thetransmembrane segments facilitate atlastin-mediated homotypic endoplasmicreticulum fusion. Proc. Natl. Acad. Sci. USA 109, E2146-E2154.

Liu, T. Y., Bian, X., Romano, F. B., Shemesh, T., Rapoport, T. A. and Hu, J.(2015). Cis and trans interactions between atlastin molecules during membranefusion. Proc. Natl. Acad. Sci. USA 112, E1851-E1860.

Lu, L., Ladinsky, M. S. and Kirchhausen, T. (2009). Cisternal organization of theendoplasmic reticulum during mitosis. Mol. Biol. Cell 20, 3471-3480.

Newport, J. W., Wilson, K. L. and Dunphy, W. G. (1990). A lamin-independentpathway for nuclear envelope assembly. J. Cell Biol. 111, 2247-2259.

Nixon-Abell, J., Obara, C. J., Weigel, A. V., Li, D., Legant, W. R., Xu, C. S.,Pasolli, H. A., Harvey, K., Hess, H. F., Betzig, E. et al. (2016). Increasedspatiotemporal resolution reveals highly dynamic dense tubular matrices in theperipheral ER. Science 354, aaf3928.

Ogawa-Goto, K., Tanaka, K., Ueno, T., Tanaka, K., Kurata, T., Sata, T. and Irie, S.(2007). p180 is involved in the interaction between the endoplasmic reticulum andmicrotubules through a novel microtubule-binding and bundling domain.Mol. Biol.Cell 18, 3741-3751.

Orso, G., Pendin, D., Liu, S., Tosetto, J., Moss, T. J., Faust, J. E., Micaroni,M., Egorova, A., Martinuzzi, A., McNew, J. A. et al. (2009). Homotypic fusionof ER membranes requires the dynamin-like GTPase Atlastin. Nature 460,978-983.

Powers, R. E., Wang, S., Liu, T. Y. and Rapoport, T. A. (2017). Reconstitution ofthe tubular endoplasmic reticulum network with purified components. Nature 543,257-260.

Prinz, W. A., Grzyb, L., Veenhuis, M., Kahana, J. A., Silver, P. A. and Rapoport,T. A. (2000). Mutants affecting the structure of the cortical endoplasmic reticulumin Saccharomyces cerevisiae. J. Cell Biol. 150, 461-474.

Puhka, M., Joensuu, M., Vihinen, H., Belevich, I. and Jokitalo, E. (2012).Progressive sheet-to-tubule transformation is a general mechanism forendoplasmic reticulum partitioning in dividing mammalian cells. Mol. Biol. Cell23, 2424-2432.

Ramadan, K., Bruderer, R., Spiga, F. M., Popp, O., Baur, T., Gotta, M. andMeyer,H. H. (2007). Cdc48/p97 promotes reformation of the nucleus by extracting thekinase Aurora B from chromatin. Nature 450, 1258-1262.

Saini, S. G., Liu, C., Zhang, P. and Lee, T. H. (2014). Membrane tethering by theatlastin GTPase depends on GTP hydrolysis but not on forming the cross-overconfiguration. Mol. Biol. Cell 25, 3942-3953.

Salinas, S., Proukakis, C., Crosby, A. andWarner, T. T. (2008). Hereditary spasticparaplegia: clinical features and pathogenetic mechanisms. Lancet Neurol. 7,1127-1138.

Sayre, M. H., Tschochner, H. and Kornberg, R. D. (1992). Reconstitution oftranscription with five purified initiation factors and RNA polymerase II fromSaccharomyces cerevisiae. J. Biol. Chem. 267, 23376-23382.

Shaw, J. M. and Nunnari, J. (2002). Mitochondrial dynamics and division inbudding yeast. Trends Cell Biol. 12, 178-184.

6

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience

Shemesh, T., Klemm, R. W., Romano, F. B., Wang, S., Vaughan, J., Zhuang, X.,Tukachinsky, H., Kozlov, M. M. and Rapoport, T. A. (2014). A model for thegeneration and interconversion of ER morphologies. Proc. Natl. Acad. Sci. USA111, E5243-E5251.

Shibata, Y., Voeltz, G. K. and Rapoport, T. A. (2006). Rough sheets and smoothtubules. Cell 126, 435-439.

Shibata, Y., Voss, C., Rist, J. M., Hu, J., Rapoport, T. A., Prinz, W. A. and Voeltz,G. K. (2008). The reticulon and Dp1/Yop1p proteins form immobile oligomers inthe tubular endoplasmic reticulum. J. Biol. Chem. 283, 18892-18904.

Shibata, Y., Hu, J., Kozlov, M. M. and Rapoport, T. A. (2009). Mechanisms shapingthe membranes of cellular organelles. Annu. Rev. Cell Dev. Biol. 25, 329-354.

Shibata, Y., Shemesh, T., Prinz, W. A., Palazzo, A. F., Kozlov, M. M. andRapoport, T. A. (2010). Mechanisms determining the morphology of theperipheral ER. Cell 143, 774-788.

Sullivan, K. M. C., Busa, W. B. and Wilson, K. L. (1993). Calcium mobilization isrequired for nuclear vesicle fusion in vitro: implications for membrane traffic andIP3 receptor function. Cell 73, 1411-1422.

Terasaki, M. (2018). Axonal endoplasmic reticulum is very narrow. J. Cell Sci. 131,jcs210450.

Terasaki, M., Song, J., Wong, J. R., Weiss, M. J. and Chen, L. B. (1984).Localization of endoplasmic reticulum in living and glutaraldehyde-fixed cells withfluorescent dyes. Cell 38, 101-108.

Terasaki,M., Chen, L.B. andFujiwara,K. (1986).Microtubules and theendoplasmicreticulum are highly interdependent structures. J. Cell Biol. 103, 1557-1568.

Terasaki, M., Shemesh, T., Kasthuri, N., Klemm, R. W., Schalek, R., Hayworth,K. J., Hand, A. R., Yankova, M., Huber, G., Lichtman, J. W. et al. (2013).XStacked endoplasmic reticulum sheets are connected by helicoidal membranemotifs. Cell 154, 285-296.

Tolley, N., Sparkes, I. A., Hunter, P. R., Craddock, C. P., Nuttall, J., Roberts,L. M., Hawes, C., Pedrazzini, E. and Frigerio, L. (2008). Overexpression of aplant reticulon remodels the lumen of the cortical endoplasmic reticulum but doesnot perturb protein transport. Traffic 9, 94-102.

Voeltz, G. K., Prinz, W. A., Shibata, Y., Rist, J. M. and Rapoport, T. A. (2006). Aclass of membrane proteins shaping the tubular endoplasmic reticulum. Cell 124,573-586.

Waga, S. and Stillman, B. (1994). Anatomy of a DNA replication fork revealed byreconstitution of SV40 DNA replication in vitro. Nature 369, 207-212.

Wang, S., Romano, F. B., Field, C. M., Mitchison, T. J. and Rapoport, T. A.(2013). Multiple mechanisms determine ER network morphology during the cellcycle in Xenopus egg extracts. J. Cell Biol. 203, 801-814.

Wang, S., Tukachinsky, H., Romano, F. B. and Rapoport, T. A. (2016).Cooperation of the ER-shaping proteins atlastin, lunapark, and reticulons togenerate a tubular membrane network. eLife 5, e18605.

Wang, S., Powers, R. E., Gold, V. A. M. and Rapoport, T. A. (2018). The ERmorphology-regulating lunapark protein induces the formation of stacked bilayerdiscs. Life Sci. Alliance 1, e201700014.

Waterman-Storer, C. M. and Salmon, E. D. (1998). Endoplasmic reticulummembrane tubules are distributed by microtubules in living cells using threedistinct mechanisms. Curr. Biol. 8, 798-807.

Westrate, L. M., Lee, J. E., Prinz, W. A. and Voeltz, G. K. (2015). Form followsfunction: the importance of endoplasmic reticulum shape. Annu. Rev. Biochem.84, 791-811.

Yan, L., Sun, S., Wang, W., Shi, J., Hu, X. Wang, S. Su, D. Rao, Z. Hu, J. andLou, Z. (2015). Structures of the yeast dynamin-like GTPase Sey1p provideinsight into homotypic ER fusion. J. Cell Biol. 210, 961–972.

Yeeles, J. T. P., Deegan, T. D., Janska, A., Early, A. and Diffley, J. F. X. (2015).Regulated eukaryotic DNA replication origin firing with purified proteins. Nature519, 431-435.

Zhang, D. and Oliferenko, S. (2014). Tts1, the fission yeast homologue of theTMEM33 family, functions in NE remodeling during mitosis. Mol. Biol. Cell 25,2970-2983.

Zhang, D., Vjestica, A. and Oliferenko, S. (2010). The cortical ER network limitsthe permissive zone for actomyosin ring assembly. Curr. Biol. 20, 1029-1034.

Zhou, X., He, Y., Huang, X., Guo, Y., Li, D. and Hu, J. (2018). Reciprocal regulationbetween lunapark and atlastin facilitates ER three-way junction formation. ProteinCell, https://doi.org/10.1007/s13238-018-0595-7.

Zhu, P. P., Patterson, A., Lavoie, B., Stadler, J. Shoeb, M. Patel, R. andBlackstone, C. (2003). Cellular localization, oligomerization, and membraneassociation of the hereditary spastic paraplegia 3A (SPG3A) protein atlastin. J.Biol. Chem. 278, 49063-49071.

7

REVIEW Journal of Cell Science (2019) 132, jcs227611. doi:10.1242/jcs.227611

Journal

ofCe

llScience