Jour

nal o

f Cel

l Sci

ence

RESEARCH ARTICLE

A cascade of ER exit site assembly that is regulated by p125A andlipid signals

David Klinkenberg, Kimberly R. Long, Kuntala Shome, Simon C. Watkins and Meir Aridor*

ABSTRACT

The inner and outer layers of COPII mediate cargo sorting and

vesicle biogenesis. Sec16A and p125A (officially known as

SEC23IP) proteins interact with both layers to control coat activity,

yet the steps directing functional assembly at ER exit sites (ERES)

remain undefined. By using temperature blocks, we find that Sec16A

is spatially segregated from p125A-COPII-coated ERES prior to ER

exit at a step that required p125A. p125A used lipid signals to control

ERES assembly. Within p125A, we defined a C-terminal DDHD

domain found in phospholipases and PI transfer proteins that

recognized PA and phosphatidylinositol phosphates in vitro and

was targeted to PI4P-rich membranes in cells. A conserved central

SAM domain promoted self-assembly and selective lipid recognition

by the DDHD domain. A basic cluster and a hydrophobic interface in

the DDHD and SAM domains, respectively, were required for p125A-

mediated functional ERES assembly. Lipid recognition by the SAM–

DDHD module was used to stabilize membrane association and

regulate the spatial segregation of COPII from Sec16A, nucleating

the coat at ERES for ER exit.

KEY WORDS: COPII, ERES, p125A, Phosphatidylinositol-4-

phosphate, Sec23ip

INTRODUCTIONThe coat protein complex II (COPII) is composed of the small

GTPase Sar1 and the cytosolic Sec23/24 and Sec13/31 proteincomplexes (Antonny and Schekman, 2001). These cytosolicproteins assemble on ER membranes at ER exit sites (ERES) to

select cargo proteins destined for exit and to deform membranesinto buds and vesicles released from the ER. The coat inner layer,Sar1–Sec23/24, binds acidic lipids and presents multiple binding

sites for short peptide sequences, ER exit motifs, thus sortingcargo for incorporation into budded vesicles (Aridor et al., 2001;Aridor et al., 1998; Kuehn et al., 1998; Miller et al., 2003). The

outer layer, which is composed of the Sec13/31 complex, formsan ancestral coat element 1 (ACE1) (Fath et al., 2007) that isrecruited onto the inner layer and polymerizes to form a hedralcage (Stagg et al., 2008). Provision of a catalytic arginine residue

by Sec23 and optimization of the position of catalytic residueswithin the Sar1 GTP-binding site by Sec31 both enhance GTPhydrolysis. A vesicle neck is constricted by the activity of Sar1,

leading to GTPase-dependent vesicle release (Bielli et al., 2005;

Lee et al., 2005; Long et al., 2010). The minimal set of COPII

coat proteins recapitulates basic cargo selection and vesicleformation activities (Antonny et al., 2001; Matsuoka et al., 1998).However, COPII activities measured in minimal reactions are

controlled by interacting proteins that couple sorting and buddingactivities with physiological biosynthetic demands (Zanetti et al.,2011). For example, the adaptation of ERES activities to changesin cargo load is mediated by the activities of mammalian Sec16

(Sec16A) and phosphoinositide 4-kinase (PI4K) IIIa (Farhanet al., 2008). PI4KIIIa generates phosphatidylinositol 4-phosphate (PI4P) on ER membranes where the dynamic

generation of PI4P supports ERES assembly and ER export(Blumental-Perry et al., 2006; Farhan et al., 2008). In yeast,Sec16p interacts with all COPII subunits, enhances their

assembly and potentiates COPII vesicle budding from the ER.The mechanisms by which these two activities, Sec16–COPIIinteractions and PI4P generation, control COPII budding at ERESremain to be defined. A link between Sec16 and lipid signals has

been previously observed. Sec16p substitutes for acidic lipidsin promoting COPII recruitment on synthetic liposomes (Supeket al., 2002). However, Sec16p hinders the functional linkage

between the coat inner and outer layers, leading to inhibitionof Sec31-enhanced GTPase activity of Sar1. The C-terminaldomain of Sec16p, which binds Sar1–Sec23/24, inhibits the

recruitment of the outer layer onto Sec23/24 (Yorimitsu and Sato,2012; Kung et al., 2012; Whittle and Schwartz, 2010). Amechanism that allows for the reversal of Sec16-mediated

inhibition of coat linkage to support ER exit remains to bedefined.

p125A (officially known as SEC23IP) binds Sec31 in cytosoland is recruited to membranes where it also binds Sec23, thus

uniquely linking the two coat layers (Ong et al., 2010).Knockdown and overexpression studies demonstrate that p125Ais required for ERES organization (Iinuma et al., 2007; Shimoiet al., 2005) and cargo export from the ER (Ong et al., 2010).

p125A belongs to a family of PA-preferring phospholipase A1enzymes (Mizoguchi et al., 2000; Shimoi et al., 2005; Tani et al.,1999). We hypothesize that p125A uses lipid signals to reverse

Sec16-inhibited layer coupling and to direct COPII assembly atERES. We now identify a molecular cascade in which Sec16 isspatially segregated from p125A-COPII-coated ERES prior to

exit. p125A functions to decode lipid signals such as PI4P and tosupport the displacement of Sec16A while directing COPIInucleation at ERES, which promotes ER exit.

RESULTSp125A is recruited with COPII to PI4P-enriched liposomesPI4P production at the ER is required for functional ERESorganization (Blumental-Perry et al., 2006; Farhan et al., 2008).We examined whether similar dependency is observed when the

recruitment of mammalian COPII from cytosol is measured on

Department of Cell Biology, University of Pittsburgh School of Medicine, 3500Terrace Street, Pittsburgh, PA 15261, USA.

*Author for correspondence (aridor@pitt.edu)

Received 19 July 2013; Accepted 22 January 2014

� 2014. Published by The Company of Biologists Ltd | Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1765

Jour

nal o

f Cel

l Sci

ence

synthetic defined membranes. PI4P-containing or controlliposomes were incubated with cytosol in the presence of

constitutively active (Sar1H79G, termed Sar1-GTP) or inactive(Sar1T39N, Sar1-GDP) Sar1 proteins, and isolated by floatation insucrose gradients. Effective COPII recruitment from cytosol toliposomes (measured by Sec23) was induced by Sar1 activation

and required PI4P (fractions 1-3; Fig. 1A). The Sar1-dependent

recruitment of the outer COPII cage (Sec31) was directed ontodistinct (albeit not all) Sec23-containing fractions of PI4P-

containing liposomes (Fig. 1B). The results are in agreement withprevious studies monitoring the recruitment of purified yeastCOPII proteins to synthetic membranes (Matsuoka et al., 1998)and support a role for PI4P in coat recruitment, yet they suggest

that the link between inner and outer coat layers might be

Fig. 1. Sar1-dependent COPII and p125A recruitment requires PI4P. (A,B) Sar1-dependent COPII recruitment to floated liposomes is dependent on PI4P.Active (Sar1-GTP) or inactive (Sar1-GDP, both tested at 1 mg, 50 ml final volume) were incubated with rat liver cytosol and synthetic large unilamellarvesicles (LUV, 400 mM) composed of 45% phosphatidylcholine (PC), 35% phosphatidylethanolamine (PE), 10% phosphatidylserine (PS) and 10% cholesterol or35% PC, 35% PE, 10% PS, 10% cholesterol and 10% PI4P at 26˚C for 1 hour and floated onto a sucrose gradient (Bielli et al., 2005). Fractions wereanalyzed by western blot with antibodies against Sec23 (at 1:10,000 dilution) (A) or hSec31a (1:500) (B) (floated fractions as labeled). (C) Recruitment of Sec23to ER membranes is not affected by depletion of p125A. p125A or SNX9 (control) was depleted from rat liver cytosol by immunoprecipitation and p125Adepletion was verified using western blot, as indicated. Sar1-GTP-dependent Sec23 recruitment from control or p125A-depleted cytosol to ER microsomes wasmonitored. Sec23 was recruited by Sar1-GTP (50 ng, 500 ng and 1 mg, final volume 60 mL) in a dose-dependent manner in the absence (lanes 1–4) orpresence (lanes 5–8) of p125A. (D) Floated PI4P-containing liposomes (fractions 1–3) from a COPII recruitment reaction (described in A) were probed for p125A(1:5000), Sec23 and Sec31. (E) Transiently expressed EGFP-tagged p125A colocalizes predominantly with hSec31a (as observed with the endogenous p125Aprotein) at ERES and was juxtaposed to ERGIC (ERGIC53) or cis-Golgi compartments (GPP130). Arrowheads indicate EGFP-p125A localizing with Sec31(antibody dilution 1:100), adjacent to membranes containing ERGIC53 (1:100) and gpp130 (1:500). Scale bars: 5mm.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1766

Jour

nal o

f Cel

l Sci

ence

regulated. p125A regulates ERES assembly and ER export (Onget al., 2010; Shimoi et al., 2005), associates with Sec31 in cytosol

and binds Sec23 using separate segments on its N terminus(Ong et al., 2010). Immunodepletion of p125A from cytosol byimmunoprecipitation, previously shown to co-deplete Sec31 (Onget al., 2010), did not affect the Sar1-dependent recruitment of

COPII inner layer Sec23/24 to ER membranes (Fig. 1C). Therecruitment of p125A to PI4P-containing liposomes mirrored thatof Sec31 (Fig. 1D). Thus, outer layer-associated p125A may link

the two coat layers following initial recruitment of Sar1–Sec23/24 onto ER membranes to control the progression of ERESassembly.

The DDHD and SAM domains cooperate to support lipidrecognition in vitro and binding of PI4P-rich membranesin cellsWe hypothesized that p125A recognizes selective lipid signals toregulate ERES assembly. To test this, we first examined the roleand specificity of selected p125A domains in lipid-signal binding.

p125A binds membranes using its C terminus, which contains aDDHD domain and a sterile a motif (SAM). DDHD domains are,180 residues long (residues 779–989 in p125A) and contain

four conserved residues (DDHD) that can form a putative metalbinding site typically found in phosphoesterase domains. DDHDdomains are found in retinal degeneration B proteins, in the N-

terminal domain-interacting receptor (Nir1-3), where Nir2functions as a phosphatidyl inositol (PI)-transfer protein (Litvaket al., 2005), and in the p125A-containing PLA1 phospholipase

protein family (Sato et al., 2010; Shimoi et al., 2005; Yamashitaet al., 2010). In the context of many multidomain proteins, SAMdomains are common protein-protein interaction motifs thatmodulate function through their ability to homo- or hetero-

associate. Several forms of SAM domains have also been shownto polymerize into larger functional structures (Qiao and Bowie,2005). We analyzed the role of p125A SAM (residues 643–704)

and DDHD domains in vivo and in vitro. In vivo, we examined thecellular localization of the selected EGFP-tagged domains intransiently transfected HeLa cells. In vitro, we analyzed the role

of the domains in lipid recognition and oligomerization usinglipid-blot overlays and sedimentation assays. As previouslyshown for endogenous p125A (Shimoi et al., 2005), EGFP-tagged p125A localized with COPII at ERES (marked by the

outer layer Sec31 subunit) that normally distributed in the cellperiphery or clustered in the perinuclear region (Fig. 1E). Atthese latter sites, ERES were adjacent to, but did not localize

with, the ER to Golgi intermediate compartment (ERGIC53,Fig. 1E), the cis-Golgi (gpp130, Fig. 1E) or the trans-Golginetwork (TGN46, not shown) compartments. At higher

expression levels, EGFP-p125A led to the enlargement ofCOPII-coated ERES membrane structures (Mizoguchi et al.,2000) (Fig. 6 and supplementary material Fig. S3). By contrast,

the isolated EGFP-tagged DDHD domain did not colocalize withSec31 but, strikingly, was distributed in a cytosolic pool and alsoshowed robust association with PI4P-rich Golgi membranes(Fig. 2A). Membrane bound EGFP–DDHD decorated the rims of

ERGIC (ERGIC53), cis-Golgi (gpp130 not shown) and TGN(TGN46, Fig. 2A) compartments, behaving like typical PI4Preporters such as the PH domain of Fapp1 (Weixel et al., 2005).

We prepared a His6-tagged fragment of the C terminusencompassing the DDHD domain (residues 701–989) foranalysis. Shorter fragments were difficult to produce because of

low yields, indicating poor folding. In contrast to its targeting to

PI4P-rich membranes in cells, when measured using lipid blotoverlays the DDHD domain displayed only weak binding

(observed with prolonged exposures) with somewhat broadspecificities toward acidic lipids (Fig. 2B). Given this ineffectivelipid recognition, we produced a larger fragment that includedboth SAM and DDHD domains (643–989) for analysis.

Importantly, the combined domain exerted highly definedspecificity to PI3P, PI5P and, in agreement with our cellularobservations, PI4P (Fig. 2A,C). The domain also recognized PA and

phosphatidylserine (PS), and weakly recognized phosphatidylinositol(3,4) bisphosphate [PI(3,4)P2], yet for the most part did notrecognize more acidic polyphosphate PIs. The results suggest that

inclusion of the SAM domain enhanced selective phospholipidrecognition. It is possible that the SAM domain binds selectivelipids. Alternatively, it may assemble DDHD domains to

optimize lipid-binding avidity. To evaluate these possibilities,we generated a GST-tagged SAM domain. Analysis using lipid-blot overlay showed no lipid binding (not shown). Furthermore,the transiently expressed EGFP-tagged SAM domain remained

cytosolic (Fig. 2F). Structural studies have demonstrated that thehomologous SAM domain of diacyl glycerol kinase (DAGK) d, aprotein that regulates COPII assembly at ERES (Nagaya et al.,

2002), dimerizes and generates oligomeric sheet structures thatfall out of solution upon Zn2+ binding. Zn2+-induced sheetformation is dependent on the ability of the domain to dimerize,

thus providing us with an easy test to monitor SAM oligomerization(Knight et al., 2010). As observed with the SAM domain ofDAGKd, addition of Zn2+ to GST-p125A-SAM domain led to

robust polymerization of the protein, which quantitatively fell outof solution under these conditions (Fig. 2H). Addition of Ca2+ orMn2+ had minimal effects on SAM solubility (not shown). Thesolubility of the control GST protein was also variably affected by

the addition of Zn2+. We thus cleaved the SAM domain from theGST. The non-tagged SAM domain effectively precipitated in thepresence of Zn2+, whereas a dimer mutant remained soluble

(Fig. 2I, see below). The lipid blot overlay analysis suggested thatDDHD-mediated lipid recognition is assisted by the p125A-SAMdomain, which may optimize avidity-based membrane binding.

EGFP–SAM–DDHD also localized to PI4P-enriched Golgi yetoccasionally caused Golgi disassembly (not shown), as observedwith other PI4P-binding domains (Weixel et al., 2005). A GFP-tagged fragment encompassing the linker region between the

SAM and DDHD domains (701–778) remained cytosolic (notshown). Collectively, these results suggest a model in whichcooperative activities of assembled SAM and DDHD domains

promote selective lipid recognition and cellular membranebinding.

Charge and hydrophobic interactions are used by the SAMand DDHD domains to support lipid recognition and assemblyOur in vivo and in vitro analyses (Fig. 2) suggest that lipid

recognition resides within the DDHD domain and is assisted bySAM domain-mediated assembly. To test this hypothesis, wegenerated mutations in these domains and examined theirfunctionality. Within the DDHD domain we focused on a group

of basic residues (851-KGRKR-855) and replaced them withglutamic acid (851-EGEEE-855, termed PI-X). When tested inlipid blot overlay assays, the His6-tagged SAM–DDHDPI-X

domain showed no lipid recognition (Fig. 2D), and transientlyexpressed EGFP-tagged DDHDPI-X lost its Golgi localization andpresented a diffuse cytoplasmic distribution (Fig. 2E). This

localization was also observed when EGFP–SAM–DDHDPI-X

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1767

Jour

nal o

f Cel

l Sci

ence

was analyzed (not shown). The results suggest that the DDHD

domain is required for lipid recognition.We further hypothesized that the SAM domain, which does

not display lipid recognition or cellular targeting in isolation

(Fig. 2F and not shown), supports protein assembly. To testthis hypothesis, we used structural information available for theDAGK d-SAM domain to guide mutagenesis aimed atabolishing protein assembly (Knight et al., 2010; Qiao and

Bowie, 2005). In the DAGK d-SAM domain, hydrophobic

interactions between valine and leucine residues mediate

dimerization required for oligomerization of the domain(Fig. 2G), whereas introduction of a charged residue at thisposition prevented both dimerization and Zn2+-induced high-

order oligomerization. The hydrophobic dimer interface is fullyconserved in p125A, thus we could generate a single residuereplacement in the p125A SAM-domain (L690E) for analysis.Unlike the wild-type protein, GST-SAM(L690E) did not

precipitate in response to Zn2+ (Fig. 2H,I). Thus, in common

Fig. 2. Cooperative lipid recognition by the SAM–DDHD module of p125A. (A) The DDHD domain targets PI4P-rich Golgi membrane in isolation.Transiently expressed EGFP–DDHD domain (779–989) dissociated from Sec31-containing ERES (arrowhead shows lack of colocalization between Sec31-stained ERES and localized EGFP–DDHD). The DDHD domain targets to the periphery of PI4P-enriched membranes and can be seen coating Golgi [rightimage, TGN46) (antibody dilution 1:500)] and ERGIC (ERGIC 53, middle image). Scale bars: 10 mm. (B-F) Selective lipid recognition is dependent on a moduleconsisting of the DDHD and SAM domains. (B) 1 mg/ml of purified His-tagged p125A fragment (701–989) containing the DDHD domain, but not the SAMdomain, was probed on lipid blot overlay using HRP-conjugated antibody against His6 (at 1:500 dilution). The fragment bound weakly to acidic lipids. (C) As in B:extending the fragment to contain the upstream SAM domain (643–989) conferred lipid selectivity to monophosphorylated PIs, PA, PS and PI(3,4)P2. (D) As in C:replacing a basic stretch of residues in the DDHD domain of the SAM–DDHD module with glutamic acid residues (850-KGRKR-854REGEEE, termed PI-X)abolished lipid recognition. (E) The EGFP–DDHD domain is targeted to Golgi membranes (left), whereas EGFP–DDHDPI-X lost Golgi targeting (right). Scale bar:5mm. (F) The EGFP-SAM domain transiently expressed in HeLa cells shows diffuse cytosolic distribution. Scale bar: 5mm. (G) The predicted structure of p125ASAM domain (643–704) (generated in Phyre) with red arrowhead indicating the position of a conserved leucine (690) within a hydrophobic dimer interface.(H) GST-SAM oligomerization is promoted by addition of Zn2+ and is sensitive to the L690E mutation. Buffer with or without the addition of 20 mM Zn(AOc)2was added to GST-SAM or GST-SAML690E (10 mM of each), as indicated. Oligomerization was followed by the precipitation of the proteins from supernatant(S) to pellet (P) fractions (as indicated), using centrifugation and analysis on Coomassie-stained gels. (I) SAM and SAML690E domains (0.455 mM each) wereincubated with 0.455 mM of Zn(AOc)2 and analyzed as in H.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1768

Jour

nal o

f Cel

l Sci

ence

with DAGK d, this single point mutation abolished dimerizationand therefore high order assembly of the domain. Similarly,

while the untagged SAM domain robustly precipitated with Zn2+

addition, SAM(L690E) remained completely soluble. The resultssuggest that the hydrophobic assembly interface within the SAMdomain of p125A is functional.

Traffic inhibition by low temperatures reveals an exclusivelocalization of p125A at ERESp125A may recognize monophosphorylated PIs, including PI4P atERES, or on ERGIC or Golgi membranes adjacent to COPII budsites. To define the site of p125A-membrane binding, we

analyzed the localization of p125A in relation to COPII,ERGIC and Golgi markers in cells incubated at reducedtemperatures that preferentially slow anterograde and retrograde

traffic between the ER and the Golgi. When cells were incubatedat 15 C under conditions that arrest biosynthetic anterogradeand retrograde cargo traffic at ERGIC, ERGIC53 stronglyaccumulated in perinuclear ERGIC compartments and, as

previously reported, ERGIC53 also scattered in defined

puncta throughout the cytoplasm (Saraste and Svensson, 1991)(supplementary material Fig. S2A,C). This distribution was rapidly

reversed when returned to 37 C (not shown). Golgi morphologyremained largely unperturbed under these conditions (supplementarymaterial Fig. S2B). Importantly, at 15 C the number of ERES(marked by Sec31) was reduced, while individual sites were

enlarged and cytosolic COPII was effectively concentrated at thesesites (Fig. 3A). Incubation of cells at 10 C leads to the selectiveaccumulation of folded biosynthetic cargo in COPII-coated ERES

membranes that maintain continuity with the bulk ER membraneand are functional in ER export (Mezzacasa and Helenius, 2002).Under these conditions, ERES further coalesced, collecting COPII

layers in enlarged puncta (Fig. 3A–C). Importantly, p125Aexclusively partitioned with COPII at both 15 C and 10 C[Fig. 3A, and Fig. 7A,B (endogenous p125A)]. p125A-COPII

sites segregated from perinuclear sites where both ERGIC andGolgi compartments largely clustered (supplementary material Fig.S2A–B). To determine whether enlarged COPII-p125A sites coatERES membranes, we analyzed the membrane protein ERGIC53,

which cycles between ER and ERGIC. Similar to endogenous

Fig. 3. p125A associated with COPII-coated ERES duringtemperature-induced traffic inhibition. (A) HeLa cells transientlyexpressing mRFP-p125A (8–12 hours) were maintained at 37˚C, orincubated at 15˚C or 10˚C, as indicated, for 4 hours, then fixed andanalyzed for localization with ERES (hSec31A). (B) Enlarged imagesof boxed areas in A, arrowheads indicate the extensivecolocalization of p125A to ERES. (C) Colocalization of endogenoushSec31a and hSec24c (antibody at 1:100 dilution) in cells incubatedat 10˚C, as in A. Scale bars: 5mm.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1769

Jour

nal o

f Cel

l Sci

ence

protein, overexpressed YFP-p58 (rat ERGIC53) accumulated in theperinuclear region at 15 C or 10 C, and also in distinct scattered

puncta that were particularly evident at 15 C. YFP-p58 also backedup in the ER, most likely due to overexpression (supplementarymaterial Fig. S2C). As a result, YFP-p58 was also arrested atenlarged ERES at 15 C and 10 C to demonstrate that ERES

membranes are coated with p125A and COPII under theseconditions [supplementary material Fig. S2C (Mezzacasa andHelenius, 2002)]. Assembly of ERES at 10 C required functional

p125A (see below, Fig. 7A,B and not shown). Overall, the analysissuggests that p125A exclusively co-segregates with COPII andcoats ERES membranes. Lipid recognition mediated by the

cooperative activity of SAM and DDHD domains supports ERESlocalization.

COPII-p125A-containing ERES segregate from Sec16A duringER exitSec16p substitutes for the requirement of acidic lipids during COPIIassembly on synthetic membranes, whereas Sec16A and PI4KIIIa

are both required during adjustment of ERES assembly with cargoload (Farhan et al., 2008; Supek et al., 2002). We hypothesized that

lipid recognition by p125A may support COPII assembly at stagesthat precede or follow Sec16 regulation. To explore this hypothesis,we localized Sec31, Sec16A and p125A using the temperatureblocks described above. We analyzed the localization of both

endogenous Sec16A (KIAA0310, using specific antibody), as wellas a EGFP-tagged Sec16A with relation to the localization ofSec31 and p125A with similar results (Fig. 4 and Fig. 7A). At

37 C, endogenous Sec16A and transiently expressed EGFP-Sec16A presented ERES localization and diffused cytosolicdistribution. Importantly, slowing traffic at the ER (10 C) or at

ERGIC (15 C), led to robust collection of Sec16A at perinuclearsites adjacent and closely associated with ERGIC53/p58-containing membranes (Fig. 4 and inset in Fig. 4C). The results

support a dynamic distribution of Sec16A between cytosol andmembranes, which is reduced at low temperatures. Surprisingly,Sec31, Sec23, mRFP-p125A and endogenous p125A exhibiteddifferent behavior to the perinuclear clustering of Sec16A under

Fig. 4. ERES coated p125A segregates from Sec16A at lowtemperatures. (A) HeLa cells transiently expressing EGFP-Sec16A (8–12 hours) were maintained at 37˚C, or incubated at15˚C or 10˚C for 4 hours, and the localization of ERES (marked byhSec31a) and EGFP-Sec16A was determined. (B,C) Control HeLacells (B) or cells transiently expressing mRFP-p125A (C) wereincubated at 10˚C and the localization of endogenous Sec16A(antibody at 1:1000 dilution), hSec31a and mRFP-p125A wasdetermined, as indicated. Inset in C shows HeLa cells transientlyexpressing YFP-p58 analyzed for the localization of endogenousSec16A and YFP-p58 during 10˚C incubations. Arrowheadsindicate hSec31 (A,B) or mRFP-p125A (C) sites, segregated fromSec16A. Scale bar: 5 mm.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1770

Jour

nal o

f Cel

l Sci

ence

these conditions, assembling at enlarged scattered ERES andsegregating from Sec16A (Figs 3, 4 and not shown). The enhanced

segregation of Sec16 from COPII-p125A-coated ERES underconditions that either delay (10 C) or enable (15 C) slow cargo exitsuggest that p125A may regulate a late stage that follows the initialnucleation of COPII by Sec16.

Assembly controlled lipid recognition is required to regulateCOPII organization at ERESWe hypothesized that p125A uses lipid recognition to direct COPIIassembly at ERES. Tagged p125A in which specific residuesrequired for SAM-domain assembly and DDHD-supported

lipid recognition (Fig. 2) were mutated (p125API-X, p125AL690E,and SAM and DDHD double mutant p125API-X, L690E)were generated to test this hypothesis. Individual mutations in

the SAM (L690E) or DDHD domain (PI-X) may not besufficient to generate a dominant phenotype because PI-Xmutants may assemble with the endogenous protein and L690Emutants may retain lipid recognition (Fig. 2B-I). Importantly, the

interactions of mutated and wild-type p125A with both COPIIlayers are maintained by the unperturbed N terminus (Ong et al.,

2010). All mutants were expressed at similar levels whenanalyzed by western blots (Fig. 7E and not shown). EGFP-p125A localized to ERES (Fig. 5). EGFP-p125API-X was alsotargeted to ERES. However, a diffused cytosolic component

was clearly evident. For the most part, overexpression ofEGFP-p125API-X did not lead to clustering, as observed with thewild-type protein (not shown and supplementary material Fig.

S3A). Because this mutant retains the ability to interact withendogenous p125A using the dimerization interface, we analyzedboth a SAM oligomerization mutant p125AL690E (Fig. 2H,I) and a

mutant (p125API-X, L690E), in which both the dimerization andlipid recognition abilities were disabled. EGFP-p125AL690E

exhibited diffuse labeling and association with ERES, similar to

the PI-X mutant. Overexpression of EGFP-p125AL690E led toaugmented diffuse cytosolic distribution with occasional targetingto both ERES and the perinuclear Golgi region (Fig. 5).Importantly, in marked contrast to wild type, p125API-X, L690E,

Fig. 5. PI4P binding by the SAM–DDHD module of p125A controlsERES assembly. The localization of transiently expressed (12–14 hours)full-length EGFP-p125A wild type, PI-X, L690E or the double mutant(p125AL690E,PI-X) and ERES (hSec31a) were analyzed in HeLa cells asindicated. The bottom panel shows the expression of a EGFP-p125Achimera where the DDHD domain has been substituted with the PHdomain from Fapp1 in the backbone of an L690E mutant. EGFP-p125API-X

or EGFP-p125AL690E mutants become partly cytosolic, whereas thedouble mutant p125API-X, L690E lost membrane localization and hadcompletely disrupted ERES assembly. Membrane targeting and ERESassembly was partially restored in the Fapp1-PH-containing p125Achimera-expressing cells. Arrowheads indicate a non-transfected cellsurrounded with cells expressing p125API-X, L690E (*). Scale bar: 10 mm.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1771

Jour

nal o

f Cel

l Sci

ence

in which the presumed cooperative lipid-binding module isdisabled, showed no ERES localization and exhibited diffusecytosolic distribution, which was maintained even at very highexpression levels (Fig. 5 and supplementary material Fig. S3A).

These results suggest that selective lipid recognition mediated bythe SAM–DDHD module is required for p125A-membranebinding at ERES. We thus tested whether selective membrane

binding by p125A regulates COPII assembly at ERES.Indeed, p125API-X, L690E became a trans-dominant negativeinhibitor of ERES assembly. Sec31 lost ERES localization in

p125API-X, L690E-expressing cells and remained diffuselylocalized in the cytoplasm (Fig. 5). Moreover, p125API-X, L690E

also inhibited the assembly of ERES when measured at 10 C,

although it did not affect Sec16A localization (not shown). These

results suggest that cooperative lipid recognition derived fromSAM–DDHD activities is required to regulate COPII assembly atERES. The DDHD domain recognized PI4P-rich membranes incells and on lipid blot overlays. Similar to p125API-X, L690E,

deletion of the DDHD domain (D778–989) in the backgroundof SAM domain inactivation (using the L690E mutation,p125AL690E, DDDHD) led to cytosolic dispersion of the protein

and inhibited ERES assembly, as analyzed by Sec31 staining(Fig. 6A). We used this truncation to further test the role ofp125A SAM–DDHD as a selective lipid recognition module

directing ERES assembly, by artificially replacing the domainwith a bona fide PI4P-binding domain, the pleckstrin-homology(PH) domain of Fapp1 (Blumental-Perry et al., 2006). Fapp1-PH

confers selective recognition of PI4P and, as also observed with the

Fig. 6. Lipid recognition by p125A regulates Sec16Adisplacement from ERES. (A) HeLa cells transientlyexpressing (12–14 hours) mRFP-p125AL690E, DDDHD ormRFP-p125AL690E, DDDHD + Fapp1-PH were analyzed forhSec31 localization, as indicated. ERES [hSec31a(antibody dilution 1:200)] are disassembled in cellsexpressing mRFP p125AL690E, DDDHD (asterisks marktransfected cells), and mRFP-p125AL690E, DDDHD + Fapp1-PH

and hSec31a are colocalized (arrows). (B) HeLa cellstransiently expressing mRFP-p125A,mRFP-p125AL690E, DDDHD, EGFP-Sec16A, YFP-Sec23A, asindicated, for 24 hours were fixed and analyzed for thelocalization of transfected proteins. Note the segregation ofEGFP-Sec16A from mRFP-p125A as opposed to the co-assembly of mRFP-p125A with YFP-Sec23a (arrow), andthe collection of EGFP-Sec16A in mRFP-p125AL690E, DDDHD

sites (arrows). Scale bar: 5 mm.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1772

Jour

nal o

f Cel

l Sci

ence

isolated DDHD domain, is targeted in isolation to PI4P-rich Golgi

membranes (Weixel et al., 2005). When expressed in cells, thechimera (p125AL690E, DDDHD, +Fapp1-PH) displayed both cytosolicand punctate distribution, with some preferential targeting to the

Golgi (Fig. 6A and Fig. 5). Importantly, in contrast to expressionof p125API-X, L690E or p125AL690E, DDDHD, which disrupted ERESassembly, the chimera colocalized with and maintained assemblyof endogenous Sec31 at both perinuclear Golgi adjacent sites and

peripheral ERES (Fig. 6A and Fig. 5). The ability to replace the

SAM–DDHD lipid recognition module artificially with a PI4P-binding domain and partially restore ERES assembly supports therole of PI4P in p125A-mediated assembly of ERES.

Lipid recognition controls p125A residency at ERESHow could lipid recognition by p125A control ERES assembly?One possibility is that lipid recognition is required to target

Fig. 7. See next page for legend.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1773

Jour

nal o

f Cel

l Sci

ence

p125A and associated Sec31 proteins to ERES. However, p125A

is targeted to ERES through association of its N terminus withSec23 and Sec31. Alternatively, lipid recognition may regulatethe residency time of p125A and its associated coat proteins on

ERES membranes. To test this possibility, we analyzed thedynamics of p125A proteins and COPII at ERES usingfluorescence recovery after photobleaching (FRAP). HeLa cells

were transiently transfected with constructs expressing YFP-Sec23, EGFP-p125A (supplementary material Fig. S1) or mRFP-p125A. An average of 27–35 measured events for each time-point

collected in three independent experiments is shown. COPIImarked by both YFP-Sec23 (supplementary material Fig. S1B) orCFP-Sec31 (not shown) exhibited typical recovery kinetics(Forster et al., 2006), whereas EGFP-p125A (supplementary

material Fig. S1C) or mRFP-p125A (not shown) both exhibitedsimilar or slightly slower dynamics. Importantly, faster recoverykinetics were observed for EGFP-p125AL690E (supplementary

material Fig. S1D) or EGFP-p125API-X (supplementary materialFig. S1E), providing an explanation for the observed diffusedcytosolic population of these mutants (Fig. 5). The initial

20 seconds of the recovery phases were fitted into singleexponentials (Forster et al., 2006) with overall averaged TK forYFP-Sec23 (3.14 seconds) or EGFP-p125A (3.36 seconds).EGFP-p125AL690E (TK52.98 seconds.) and EGFP-p125API-X

(TK52.56 seconds) showed faster kinetics. Because EGFP-p125API-X, L690E did not assemble at defined sites, it could notbe measured. However, limited association with ERES was

observed when EGFP- p125API-X, L690E was analyzed at reduced

temperatures (not shown), suggesting that enhanced turnover atERES prevented stable localization. The results suggest that

selective lipid recognition by the SAM–DDHD module controlsp125A residency at ERES membranes to regulate COPIIassembly.

p125A functions at a late stage in ERES nucleationp125A-Sec23-Sec31-coated ERES segregated from Sec16Aduring temperature-induced traffic blocks at ERES or ERGIC

(Figs 3,4). We hypothesized that p125A actively displacesSec16A from ERES. We therefore prolonged the expressiontime of p125A from 12 to 16 hours, used in experiments

presented so far, to 24 hours to examine if such overexpressionwould lead to Sec16 displacement from ERES. Overexpression ofp125A led to a marked uniform enlargement of ERES,

as previously demonstrated (mRFP-p125A, Fig. 6B andsupplementary material Fig. S3A and EGFP-p125A, not shown)(Mizoguchi et al., 2000; Shimoi et al., 2005; Tani et al., 1999).The large sites were marked by ER membranes containing

the cargo protein Venus-tsVSV-G (vesicular stomatitis virusglycoprotein). Bud structures containing tsVSV-G were evidenton these structures upon a temperature shift that leads to exit

of tsVSV-G from the ER (super resolution SIM images,supplementary material Fig. S3B and Movies S3-S5). Analysisof tsVSV-G arrival at the cis/medial Golgi (using Endo-H

digestion), under these extreme p125A expression conditions,showed that although traffic was somewhat inhibited it remainedfunctional (supplementary material Fig. S3B). Indeed, the

enlarged ERES collected both layers of COPII, as analyzed bythe colocalization of endogenous Sec31 or co-expressed YFP-Sec23 (Fig. 6B and not shown). We analyzed whether EGFP-Sec16A is displaced from these sites. When expressed at low

levels, mRFP-125A largely colocalized with Sec16A, Sec31 andSec23 at ERES (Figs 3,4). By contrast, p125A-induced large ERESwere clearly lacking EGFP-Sec16A (compare YFP-Sec23-mRFP-

p125A with EGFP-Sec16A-mRFP-p125A; Fig. 6B, middle versustop row). Occasionally, Sec16A was found adjacent to largerounded sites that varied from 500 to 1500 nm in size, as shown by

SIM (supplementary material Fig. S3C). Similar localization wasobserved when endogenous Sec16A was analyzed (not shown).

As Sec16p negates the requirements for acidic lipids inCOPII binding to liposomes (Supek et al., 2002), we

hypothesized that p125A might use PI4P binding to directSec16A displacement, thus stabilizing COPII-membrane binding.We therefore analyzed whether overexpressed p125A proteins in

which the lipid-binding module was inactivated (mRFP-p125API-X, L690E) or deleted (p125AL690E, DDDHD) are alsodefective in Sec16 displacement. In marked contrast to mRFP-

p125A, mRFP- p125API-X, L690E remained dispersed even atvery high expression levels (supplementary material Fig. S3A).Overexpressed p125AL690E, DDDHD also remained largely

cytosolic, but enlarged sites were now occasionally evident(Fig. 6B; supplementary material Fig. S3A). Importantly, thesesites effectively collected EGFP-Sec16A (Fig. 6B). Analysisusing SIM demonstrated that p125AL690E, DDDHD-induced sites

were engulfed within EGFP-Sec16A (supplementary materialFig. S3C; Movies S1,S2). Wild type and p125AL690E, DDDHD

interact similarly with both COPII layers. Therefore, the

presence of a lipid-binding module in p125A controls thedisplacement of Sec16A from COPII-ERES.

The results suggest that PI4P (or lack of) can regulate p125A to

control Sec16A localization. Depletion of PI4KIIIa is known to

Fig. 7. Functional SAM–DDHD module is required for steady-state ER-to-Golgi traffic. (A) Merge images showing the localization of endogenousp125A (green, antibody dilution 1:1000) and hSec31a (red) at ERES arrestedat 10˚C in control and p125A-depleted cells, as indicated (dsRNAi,supplementary material Table S2). Inset shows the localization of endogen-ous Sec16A (green) and hSec31a (red) under similar conditions. (B) Controlor p125A-depleted cells were transfected with RNAi-resistant mRFP-p125Arand traffic was arrested by incubation at 10˚C. The localization of hSec31awas analyzed [merge images are shown, arrowhead indicates depleted cellnext to rescued cells (asterisk)]. (C) HeLa cells stably expressing GFP-tagged N-acetylgalactosaminyltransferase-2 (GalNAcT2-GFP) were treatedwith control or p125A directed dsRNAi (supplementary material Table S2), asindicated, and visualized for GFP. Note the dramatic change in Golgimorphology (arrowheads) with loss of intact Golgi populations and theconcomitant increase in shattered Golgi morphology. Scale bar: 5 mm.(D) Golgi morphology was quantified in HeLa cells by analyzing gpp130localization. Typical observed Golgi morphologies are shown and color-coded, including intact (blue), dispersed (pink) and shattered (vesiculated,green). In some cells, Golgi morphology was not recognizable (labeled asmissing, purple). (E) Analysis of p125A knockdown efficiency by westernblots. Endogenous expression of p125A (middle panel, 1:2500) and actin(lower panel, 1:10,000) are shown. The upper panel shows the comparableexpression of EGFP-p125A-resistant clones (EGFP-p125Ar; GFP antibodydilution 1:10,000) in control and KD cells. EGFP ran below the analyzed areaand is not shown. The expression of EGFP-p125Ar (wild type) and EGFP-p125AL690E, PI-X were also detected by the p125A-specific antibody butrequired prolonged exposures due to partial transfection efficiency of KDcells. (F) The fractional distribution of Golgi morphologies in control, p125A-depleted and rescued cell populations with EGFP-p125Ar and EGFP-p125API-X, L690Er, as indicated. The percentage of cells with either intact Golgi(blue), dispersed Golgi (pink), shattered Golgi (green) or missing Golgi(purple) under each treatment condition is shown. (G) Statistical analysis ofintact Golgi morphology in control and KD cells (mean6s.d.). Threeexperiments were performed for each condition. Ten images were collectedfrom each experiment and Golgi phenotypes were determined for all cellsexpressing EGFP-tagged proteins (93–220 cells per group). UnpairedStudent’s t-test was used to test for significant differences between groups.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1774

Jour

nal o

f Cel

l Sci

ence

inhibit ERES assembly. We asked how such depletion wouldaffect Sec16A localization. siRNA-mediated PI4KIIIa depletion

led to redistribution of hSec31a, which now clustered atperinuclear sites, whereas peripheral ERES numbers werereduced, as previously demonstrated (Farhan et al., 2008).Importantly, PI4KIIIa depletion led to clustering of Sec16A at

perinuclear sites, where colocalization with hSec31a increased(supplementary material Fig. S4). These results support a model(Fig. 8) in which PI4P signals regulate the displacement of

Sec16A from ERES during COPII assembly.

Functional contributions of the SAM–DDHD membrane-binding moduleDepletion of p125A using RNAi (Fig. 7E) leads to disruption ofERES (Shimoi et al., 2005) as further evidenced by the

pronounced loss of ERES in p125A-depleted cells incubated at10 C, under conditions where Sec16A is segregated (Fig. 7A andinset). ERES assembly under these conditions was restored by theexpression of an RNAi-resistant form of mRFP-p125A (Fig. 7B).

p125A depletion causes kinetic inhibition of membrane andsoluble cargo secretion from the ER, leading to the disruption ofGolgi morphology (Ong et al., 2010). To examine the functional

contribution of the SAM–DDHD lipid-binding module to p125Aactivity in ER exit, we replaced endogenous p125A with EGFP-p125API-X, L690E. We analyzed Golgi morphology as a reporter

for overall steady-state traffic activities. Golgi morphology wasanalyzed using GalNAcT2-GFP localization (Fig. 7C) andquantified using gpp130 morphology (Fig. 7D). Morphology

was heterogeneous with four distinct phenotypes (Fig. 7D): (1)intact Golgi localized to the perinuclear region; (2) looselypacked or dispersed Golgi located in the perinuclear region oraround the nuclei; (3) completely shattered (vesiculated) Golgi

dispersed throughout the cell; and (4) cells missing a detectableGolgi compartment, perhaps suggestive of a mitotic cellpopulation. Effective depletion of endogenous p125A was

achieved as previously reported (Ong et al., 2010) (Fig. 7E).Depletion of p125A led to dramatic reduction in the intact Golgi

population and a concomitant increase in shattered morphologywhen compared with control RNAi-treated cells (Fig. 7F,G). The

effect was specific as expression of an RNAi resistant form ofEGFP-p125A (Fig. 7E-G) reversed these effects, namelyeliminating the shattered Golgi morphology and restoring intactGolgi populations (Fig. 7F,G). By contrast, expression of EGFP-

p125API-X, L690E was ineffective, leading to only a partialrestoration of intact Golgi morphology and to a reduction inshattered Golgi compartments (Fig. 7F,G). Overall, these results

suggest that defects in the activity of the SAM–DDHD lipid-binding module of p125A, which led to robust morphologicaldefects in ERES assembly (Fig. 5), to reduced association of

p125A with ERES (supplementary material Fig. S1) and toinhibition of Sec16 segregation from ERES (Figs 4,6 andsupplementary material Fig. S3), also translated into functional

defects in steady-state traffic activities required to maintain Golgimorphology.

DISCUSSIONAlthough COPII core subunits are sufficient in mediating vesiclebiogenesis, COPII-interacting proteins control budding activitiesat ERES. We identified a cascade in which p125A, a protein that

links both COPII layers, uses a lipid recognition module (Figs 1,2) to stabilize membrane binding (supplementary material Fig.S1) and promotes the spatial segregation of COPII from Sec16A

(Figs 3,4,6 and supplementary material Fig. S2–S4), leading tofunctional ERES assembly (Figs 5,7). The results support amodel in which regulatory activities such as lipid signaling

(Blumental-Perry et al., 2006; Farhan et al., 2008; Nagaya et al.,2002; Pathre et al., 2003) control the progression of a molecularcascade that directs COPII nucleation and activity at ERES(Fig. 8).

The SAM–DDHD lipid-binding modulePrevious studies suggested that the C terminus of p125A, which

contains SAM and DDHD domains, is involved in membranebinding. A defined function of SAM domains is protein

Fig. 8. The COPII budding cascade at ERES. (A–C) Following initial association of COPII layers with Sec16 on ER membranes (A), the binding of PI4P byp125A promotes the displacement of Sec16 from COPII inner and outer layers to allow for effective linking between coat layers. This drives coat retention(B) while enhancing GTP hydrolysis to support later steps, including Sar1 exchange by Bet3 on Sec23, and Sar1-induced vesicle neck constriction andfission (C).

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1775

Jour

nal o

f Cel

l Sci

ence

oligomerization (Qiao and Bowie, 2005), which is used toincrease avidity between assembled complexes and substrates.

Our findings suggest the basic assembly activity of SAM isfunctional in p125A. First, we showed that, in common with itsclose homolog, the SAM domain of DAGKd (Knight et al.,2010), the p125A-SAM domain oligomerized when bound to

Zn2+. Second, we demonstrated that the basic assembly interfaceis conserved and introduction of a single point mutationwithin the site abolished oligomerization (Fig. 2). Third, we

demonstrated that the SAM assembly interface within p125A isrequired to support ERES organization and activity (Figs 5-7;supplementary material Figs S1 and S3).

Because p125A-SAM lacked membrane targeting in vivo orlipid-binding activities in vitro, we favor a model in which thisdomain enhances the avidity for lipid recognition by the DDHD

domain, as suggested by lipid blot overlay analysis (Fig. 2). Thisactivity may be shared in other lipid processing enzymes thatfunction in vesicular transport. In DAGKd, a module containingSAM and PH domains is required for inhibition of ERES

assembly although lipid-binding specificities of this PH domainare undefined (Nagaya et al., 2002). The SAM domain of theinositol 5-phosphatase Ship2, which regulates clathrin-mediated

endocytosis, may be similarly required for membrane recognition(Nakatsu et al., 2010). The Ship2-SAM domain further supportsheterodimerization with Arap3 (Raaijmakers et al., 2007). By

analogy, heterodimeric interactions between SAM-DAGKd andSAM-p125A may function to form lipid-recognition andprocessing hubs at ERES.

Lipid recognition most likely resides in the p125A-DDHDdomain. When expressed in cells in isolation, EGFP–DDHD wastargeted to PI4P-rich Golgi membranes (Fig. 2). Mutations in thedomain (DDHDPI-X), which prevented Golgi binding in cells (and

further abolished the targeting of EGFP–SAM–DDHDPI-X

domain, not shown), also abolished lipid recognition by theSAM–DDHD module in vitro (Fig. 2). These mutations probably

did not destabilize DDHDPI-X because the protein did notaggregate in cells and was produced in bacteria at yields higherthan its wild-type version. Moreover, deletion of the DDHD

domain (p125AL690E, DDDHD) abolished membrane binding anddispersed ERES similarly to proteins carrying the DDHDPI-X

mutations (p125API-X, L690E, Figs 5,6 and supplementary materialFig. S3). Future structural studies are needed to define the

contributions of the basic PI-X cluster in lipid recognition.Together, the SAM–DDHD module provides a lipid recognitionunit for p125A that regulates COPII assembly (Fig. 5).

In vitro analysis suggests that the SAM–DDHD modulerecognizes monophosphorylated PIs, PA and PS (Fig. 2).However, several lines of evidence suggest that PI4P might be

a primary target for this module: (1) full-length p125A showsspecificity for PIP recognition (Inoue et al., 2012; Shimoi et al.,2005); (2) p125B, a family member of p125A that localizes to the

Golgi, recognizes PI4P using its SAM and DDHD domains,although individual contributions of these domains were notdefined (Inoue et al., 2012). p125B–SAM–DDHD can replace thep125A lipid-recognition module and the resulting chimera

localizes at ERES (Shimoi et al., 2005); (3) EGFP–DDHDdomain is targeted to PI4P-rich Golgi membranes (Fig. 2); (4)p125A mediated ERES targeting and organizing activity, which is

lost when the SAM dimer interface is disabled (L690E mutant)and the lipid-binding DDHD domain is mutated (PI-X) or deleted(Figs 5,6 and supplementary material Fig. S3), was partially

restored by simple replacement of the DDHD domain with a

well-characterized PI4P-binding domain (Fapp1-PH, Fig. 6); (5)deletion of the DDHD domain in p125A or siRNA-mediated

PI4KIIIa depletion, which reduces PI4P levels, both inhibit thedisplacement of Sec16A from ERES (Fig. 6 and supplementarymaterial Fig. S3,S4); (6) DDHD domains are found in Golgi (PI4P-rich)-targeted proteins, including p125B (Nakajima et al., 2002;

Yamashita et al., 2010) and Nir-2 (Litvak et al., 2005). Collectively,the results suggest that p125A can use PI4P to localize and regulateCOPII activities at ERES.

Role of p125A in ERES regulationA dependency on acidic lipids and in particular PI4P in yeast

COPII assembly, which is abrogated by Sec16p, provided the firstlink between the two activities. However, such dependency isobserved only on synthetic membranes (Supek et al., 2002).

Depletion of the yeast major PI4 kinases PIK1 and Stt4 failed toaffect ER-to-Golgi traffic (Audhya et al., 2000). Subsequentstudies have shown that sequestration of Golgi PI4P in vitro orprolonged PIK1 inactivation in vivo do inhibit ER to Golgi traffic

(Lorente-Rodrıguez and Barlowe, 2011), yet inhibition is exertedon fusion of COPII vesicles with Golgi membranes. Unlike yeast,mammalian COPII vesicles do not fuse with the Golgi but rather

fuse homotypically in the vicinity of ER bud sites and PI4Pformation is initiated at these sites (Blumental-Perry et al., 2006;Farhan et al., 2008).

Sec16A is required to nucleate new ERES, and both Sec16Aand the ER-localized PI4KIIIa are required to maintain these sites(Farhan et al., 2008). Sec16p-COPII interactions hinder

functional linkage between COPII layers, thus inhibiting Sec31-stimulated GAP activity (Kung et al., 2012; Supek et al., 2002;Yorimitsu and Sato, 2012). However, functional linkage isrequired for the completion of vesicle budding (Fromme et al.,

2007). Whereas Sec16p is dispensable for in vitro budding andonly enhances vesicle release, it is a vital component in vivo. Ourresults provide a plausible working model to explain the

seemingly contradictory inhibitory and supportive roles ofSec16. In this model, COPII interactions with Sec16 and withPI4P-p125A each represent sequential steps in a budding cascade

(Fig. 8). Sec16 initiates COPII nucleation on ER membranes yetis physically segregated from the coat as it establishes properlinkage between its layers. Indeed, Sec16A is slightly removedfrom bud sites at steady state (Hughes et al., 2009). At low

temperatures that inhibit ER exit, it is displaced from arrestedp125A-controlled ERES. This enhanced displacement ismaintained when ER exit is active, yet traffic is slowed

subsequently at ERGIC (Fig. 4). Sec16A displacement is alsoenhanced by overexpression of p125A and conversely is inhibitedby inactivation of the lipid-binding module of p125A (Fig. 6;

supplementary material Fig. S3) or by PI4KIIIa depletion(supplementary material Fig. S4). p125A, which interacts withboth of the COPII layers, may provide a mechanism to facilitate

the progression of COPII budding, promoting Sec16 dissociationwhile linking both COPII layers. p125A is required to stabilizeERES following segregation from Sec16A (Fig. 7A,B) and usesits SAM–DDHD lipid binding module in these reactions (Fig. 6

and supplementary material Fig. S3), thus employing lipid signalsto control the assembly cascade. It remains to be determined howp125A-membrane binding regulates the segregation of COPII

from Sec16A. p125A is a late addition in the evolution of COPII,which is lacking in yeast; thus, a mechanism by which Sec16pinhibition of functional coupling between COPII layers is

reversed has yet to be determined.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1776

Jour

nal o

f Cel

l Sci

ence

The physiological role of p125A is unknown. Acute depletion oroverexpression of p125A leads to general traffic defects (Fig. 7 and

supplementary material Fig. S3). p125A depletion interferes withneural crest cell migration in Xenopus and was predicted to be acausative gene in the development of Waardenburg syndrome(McGary et al., 2010). However, p125A-KO mice are relatively

unaffected (Arimitsu et al., 2011), mainly presenting defects inspermiogenesis. Sec16 may provide sufficient support for COPIIactivities in these animals, similar to the documented functionality of

COPII on liposomes that contain Sec16p yet lack acidic lipids (Supeket al., 2002). The identified steps in ERES assembly are probablysubjected to physiological regulation, which can now be defined.

MATERIALS AND METHODSHeLa cells were maintained at sub-confluence in Dulbecco’s modified

Eagle’s media (DMEM) (HyClone Fisher-Scientific) supplemented with

up to 10% fetal bovine serum (FBS) (Serum Source International) and

5% penicillin-streptomycin (Cellgro) under a standard incubation

environment (37 C, 5% CO2). Antibodies against p125A (MSTP053,

Bethyl Lab; AP114511b, Abgent), GFP (24240, Polysciences), ERGIC53

(G1/93, ALX-804-602, Enzo Life Sciences), anti-Sec31A (612350, BD

Transduction Laboratories), Flag (M2, F1804, Sigma-Aldrich), b-Actin

(ab6276, Abcam), GFP (332600, Invitrogen), PI4KIIIa (Cell Signaling),

HRP-conjugated anti-GST antibodies (ab3416-250, Abcam) and HRP-

conjugated anti-His (040905270001, Roche) were used. Mouse

monoclonal anti p125A and rabbit anti KIAA03100 (Sec16L) provided

by Katsuko Tani (Tokyo University of Pharmacy and Life Science,

Hachioji, Tokyo, Japan) were used. All Golgi-specific antibodies were

provided by Adam Linstedt (Carnegie Mellon University, Pittsburgh,

PA). Alexa-conjugated secondary antibodies were from Invitrogen. HRP-

conjugated secondary antibodies were from Pierce.

EGFP-Sec16A was kindly provided by Vivek Malhotra (GRC,

Barcelona, Spain). YFP-p58 was kindly provided by Jennifer Lippincot-

Schwartz (NIH). p125A was excised from pFlag-CMW-6c-p125A, kindly

provided by Katsuko Tani, (Tokyo University of Pharmacy and Life

Science, Tokyo, Japan) using unique HindIII and SmaI restriction sites and

ligated into a modified pEGFP-C1 where a HindIII site was added in-

frame. The construction of EGFP- and mRFP-tagged constructs expressing

p125A, p125A mutants, selected p125A domains and p125A-Fapp1-PH

chimera is detailed in supplementary material Table S1.

Transfection was carried out using Effectene (Qiagen) or Lipofectamine

2000 (Life Sciences) reagents, according to manufacturers’ protocols, with

optimized DNA concentrations. NRK-derived ER microsomes and rat liver

cytosol were prepared as described previously (Rowe et al., 1996). His6-

tagged Sar1 H79G and T39N proteins were purified as described

previously (Aridor et al., 1995; Rowe and Balch, 1995). His6-DDHD

and SAM–DDHD expression in BL 21DE3 (Invitrogen) was induced with

IPTG (0.1 mM) for 4 hours at 37 C and cells were lysed as described

previously (Aridor et al., 1995). 10% N-Lauroyl Sarcosine (MP

Biomedicals) was added to lysates that were further homogenized by

sonication. Cell debris were removed by centrifugation and supernatants

were supplemented with 2% n-Octyl-b-D-glucopyranoside (OG) (Gold

Biotechnology, USA). Solubilized proteins were loaded on Ni-NTA-

Agarose (Qiagen). Protein-bound beads were washed three times with

50 mM Tris-HCl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 1 mM PMSF

and 2% OG, followed by three washes with HNE buffer [50 mM HEPES

(pH57.4), 300 mM NaCl, 1 mM MgCl2, 0.5 mM EGTA, 2% OG] and

three washes with HNE buffer supplemented with 25 mM imidazole (pH

7.4). Bound proteins were eluted with HNE buffer containing 500 mM

imidazole (Frangioni and Neel, 1993). GST-tagged SAM and SAML690E

proteins were purified on GS-Sepharose 4B (GE Healthcare Life Science)

and thrombin cleaved using a standard bulk GST purification protocol.

p125A knockdown-replacement analysisKnockdown of p125A is described in supplementary material Table S2.

For analysis of Golgi morphology, 10 individual fields positive for EGFP

expression were collected from three independent experiments (30

images in total for each condition). All cells in the field were visually

scored for Golgi morphology and p125A expression. Data were analyzed

using Windows Excel 2010 (Microsoft Corporation). Homoscedastic

two-tailed Student’s t-tests on the percentage of intact Golgi were

performed using Microsoft Excel 2010.

Temperature-block analysisHeLa cells expressing FP-proteins for 14–16 hours were incubated in

media supplemented with 20 mM HEPES (pH 7.4) (Fisher Scientific)

and the cells were incubated at 15 C or 10 C for 4 hours. Samples were

fixed and processed for analysis.

Lipid blot-overlayPIP Strips (Echelon) were blocked for 1 hour in TBS-Tween buffer [Tris-

buffered saline (TBS, pH 8.0), 1% Tween-20) supplemented with 3%

BSA (BSA, Fraction V, EMD) and incubated for 2 hours. in the same

buffer supplemented with 1 mg/ml of GST or His6 tagged proteins. Strips

were washed (665 minute incubations) in TBS-Tween and incubated for

1 hour in TBS-Tween supplemented with 3% BSA and HRP-conjugated

murine anti His6 or GST antibodies. Subsequently, strips were washed

(665 min.) in TBS-Tween and visualized using SuperSignal West Dura

Extended Duration Substrate (Thermo Scientific) and HyBlot CL X-Ray

film (Denville Scientific), according to manufacturers’ protocols.

ImmunofluorescenceIndirect immunofluorescence was carried out as described previously

(Aridor et al., 1995). Images were acquired on an Olympus Fluoview

1000 confocal system using an inverted microscope (IX-81 Olympus)

and 606NA 1.42 PLAPON objective. Images were processed using

FV10-ASW V. 02.00.03.10 (Olympus Corporation) and Adobe

Photoshop CS3 (Adobe Photoshop Version: 10.0.1).

Zn2+ based polymerization assayGST-SAM or GST-SAML690E were diluted to a final concentration of

10 mM in 50 mM Tris-HCl (pH57.5), 100 mM NaCl. An equal volume

of 20 mM Zn(OAc)2 in the same buffer was added and polymerization

was estimated by centrifugation at 18,000 g using a cooled microfuge

(Sorvall). Supernatant and pellet fractions were visualized on SDS-PAGE

gels using Coomassie Blue staining. For thrombin-cleaved proteins,

SAM, SAML690E and Zn(OAc)2 (both at 0.455 mM) were incubated in

buffer containing 50 mM Tris-HCl (pH 7.5) and 100 mM NaCl, and

polymerization was determined as above. Proteins were visualized using

SilverQuest Staining Kit (Invitrogen).

Coat recruitment to LUVs or ER microsomesLUV-COPII recruitment assays using rat liver cytosol (RLC) and Sar1

proteins were performed as described previously (Bielli et al., 2005).

p125A or SNX9 were depleted by immunoprecipitation from RLC. The

resulting supernatants were adjusted for protein concentration, verified

for similar Sec23 and HSP70 levels, and used to measure COPII

recruitment to microsomes as described previously (Pathre et al., 2003).

AcknowledgementsWe thank Katsuko Tani and Mitsuo Tagaya (Tokyo University, Japan), William E.Balch (TSRI, La Jolla, CA), Vivek Malhotra (GRC, Barcelona, Spain), AdamLinstedt (CMU, Pittsburgh, PA), and Jennifer Lippincott-Schwartz (NIH, Bethesda,MD) for valuable reagents.

Competing interestsThe authors declare no competing financial interests.

Author contributionsD.K., K.R.L., K.S., S.C.W. and M.A. designed, performed and interpreted variousexperiments. M.A. conceived and directed the project and wrote the manuscriptwith input, editing and comments from all authors.

FundingThe study was supported by the National Institutes of Health (NIH)[R01DK092807 to M.A.]. Deposited in PMC for release after 12 months.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1777

Jour

nal o

f Cel

l Sci

ence

Supplementary materialSupplementary material available online athttp://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.138784/-/DC1

ReferencesAntonny, B. and Schekman, R. (2001). ER export: public transportation by theCOPII coach. Curr. Opin. Cell Biol. 13, 438-443.

Antonny, B., Madden, D., Hamamoto, S., Orci, L. and Schekman, R. (2001).Dynamics of the COPII coat with GTP and stable analogues. Nat. Cell Biol. 3,531-537.

Aridor, M., Bannykh, S. I., Rowe, T. and Balch, W. E. (1995). Sequentialcoupling between COPII and COPI vesicle coats in endoplasmic reticulum toGolgi transport. J. Cell Biol. 131, 875-893.

Aridor, M., Weissman, J., Bannykh, S., Nuoffer, C. and Balch, W. E. (1998).Cargo selection by the COPII budding machinery during export from the ER.J. Cell Biol. 141, 61-70.

Aridor, M., Fish, K. N., Bannykh, S., Weissman, J., Roberts, T. H., Lippincott-Schwartz, J. and Balch, W. E. (2001). The Sar1 GTPase coordinatesbiosynthetic cargo selection with endoplasmic reticulum export site assembly.J. Cell Biol. 152, 213-230.

Arimitsu, N., Kogure, T., Baba, T., Nakao, K., Hamamoto, H., Sekimizu, K.,Yamamoto, A., Nakanishi, H., Taguchi, R., Tagaya, M. et al. (2011). p125/Sec23-interacting protein (Sec23ip) is required for spermiogenesis. FEBS Lett.585, 2171-2176.

Audhya, A., Foti, M. and Emr, S. D. (2000). Distinct roles for the yeastphosphatidylinositol 4-kinases, Stt4p and Pik1p, in secretion, cell growth, andorganelle membrane dynamics. Mol. Biol. Cell 11, 2673-2689.

Bielli, A., Haney, C. J., Gabreski, G., Watkins, S. C., Bannykh, S. I. and Aridor,M. (2005). Regulation of Sar1 NH2 terminus by GTP binding and hydrolysispromotes membrane deformation to control COPII vesicle fission. J. Cell Biol.171, 919-924.

Blumental-Perry, A., Haney, C. J., Weixel, K. M., Watkins, S. C., Weisz, O. A.and Aridor, M. (2006). Phosphatidylinositol 4-phosphate formation at ER exitsites regulates ER export. Dev. Cell 11, 671-682.

Farhan, H., Weiss, M., Tani, K., Kaufman, R. J. and Hauri, H. P. (2008).Adaptation of endoplasmic reticulum exit sites to acute and chronic increases incargo load. EMBO J. 27, 2043-2054.

Fath, S., Mancias, J. D., Bi, X. and Goldberg, J. (2007). Structure andorganization of coat proteins in the COPII cage. Cell 129, 1325-1336.

Forster, R., Weiss, M., Zimmermann, T., Reynaud, E. G., Verissimo, F.,Stephens, D. J. and Pepperkok, R. (2006). Secretory cargo regulates theturnover of COPII subunits at single ER exit sites. Curr. Biol. 16, 173-179.

Frangioni, J. V. and Neel, B. G. (1993). Solubilization and purification ofenzymatically active glutathione S-transferase (pGEX) fusion proteins. Anal.Biochem. 210, 179-187.

Fromme, J. C., Ravazzola, M., Hamamoto, S., Al-Balwi, M., Eyaid, W.,Boyadjiev, S. A., Cosson, P., Schekman, R. and Orci, L. (2007). The geneticbasis of a craniofacial disease provides insight into COPII coat assembly. Dev.Cell 13, 623-634.

Hughes, H., Budnik, A., Schmidt, K., Palmer, K. J., Mantell, J., Noakes, C.,Johnson, A., Carter, D. A., Verkade, P., Watson, P. et al. (2009). Organisationof human ER-exit sites: requirements for the localisation of Sec16 to transitionalER. J. Cell Sci. 122, 2924-2934.

Iinuma, T., Shiga, A., Nakamoto, K., O’Brien, M. B., Aridor, M., Arimitsu, N.,Tagaya, M. and Tani, K. (2007). Mammalian Sec16/p250 plays a role inmembrane traffic from the endoplasmic reticulum. J. Biol. Chem. 282, 17632-17639.

Inoue, H., Baba, T., Sato, S., Ohtsuki, R., Takemori, A., Watanabe, T., Tagaya, M.and Tani, K. (2012). Roles of SAM and DDHD domains in mammalian intracellularphospholipase A1 KIAA0725p. Biochim. Biophys. Acta 1823, 930-939.

Knight, M. J., Joubert, M. K., Plotkowski, M. L., Kropat, J., Gingery, M.,Sakane, F., Merchant, S. S. and Bowie, J. U. (2010). Zinc binding drives sheetformation by the SAM domain of diacylglycerol kinase d. Biochemistry 49, 9667-9676.

Kuehn, M. J., Herrmann, J. M. and Schekman, R. (1998). COPII-cargointeractions direct protein sorting into ER-derived transport vesicles. Nature391, 187-190.

Kung, L. F., Pagant, S., Futai, E., D’Arcangelo, J. G., Buchanan, R., Dittmar,J. C., Reid, R. J., Rothstein, R., Hamamoto, S., Snapp, E. L. et al. (2012).Sec24p and Sec16p cooperate to regulate the GTP cycle of the COPII coat.EMBO J. 31, 1014-1027.

Lee, M. C., Orci, L., Hamamoto, S., Futai, E., Ravazzola, M. and Schekman, R.(2005). Sar1p N-terminal helix initiates membrane curvature and completes thefission of a COPII vesicle. Cell 122, 605-617.

Litvak, V., Dahan, N., Ramachandran, S., Sabanay, H. and Lev, S. (2005).Maintenance of the diacylglycerol level in the Golgi apparatus by the Nir2 proteinis critical for Golgi secretory function. Nat. Cell Biol. 7, 225-234.

Long, K. R., Yamamoto, Y., Baker, A. L., Watkins, S. C., Coyne, C. B., Conway,J. F. and Aridor, M. (2010). Sar1 assembly regulates membrane constrictionand ER export. J. Cell Biol. 190, 115-128.

Lorente-Rodrıguez, A. and Barlowe, C. (2011). Requirement for Golgi-localizedPI(4)P in fusion of COPII vesicles with Golgi compartments. Mol. Biol. Cell 22,216-229.

Matsuoka, K., Orci, L., Amherdt, M., Bednarek, S. Y., Hamamoto, S.,Schekman, R. and Yeung, T. (1998). COPII-coated vesicle formationreconstituted with purified coat proteins and chemically defined liposomes.Cell 93, 263-275.

McGary, K. L., Park, T. J., Woods, J. O., Cha, H. J., Wallingford, J. B. andMarcotte, E. M. (2010). Systematic discovery of nonobvious human diseasemodels through orthologous phenotypes. Proc. Natl. Acad. Sci. USA 107, 6544-6549.

Mezzacasa, A. and Helenius, A. (2002). The transitional ER defines a boundaryfor quality control in the secretion of tsO45 VSV glycoprotein. Traffic 3, 833-849.

Miller, E. A., Beilharz, T. H., Malkus, P. N., Lee, M. C., Hamamoto, S., Orci, L.and Schekman, R. (2003). Multiple cargo binding sites on the COPII subunitSec24p ensure capture of diverse membrane proteins into transport vesicles.Cell 114, 497-509.

Mizoguchi, T., Nakajima, K., Hatsuzawa, K., Nagahama, M., Hauri, H. P.,Tagaya, M. and Tani, K. (2000). Determination of functional regions of p125, anovel mammalian Sec23p-interacting protein. Biochem. Biophys. Res.Commun. 279, 144-149.

Nagaya, H., Wada, I., Jia, Y. J. and Kanoh, H. (2002). Diacylglycerol kinase deltasuppresses ER-to-Golgi traffic via its SAM and PH domains. Mol. Biol. Cell 13,302-316.

Nakajima, K., Sonoda, H., Mizoguchi, T., Aoki, J., Arai, H., Nagahama, M.,Tagaya, M. and Tani, K. (2002). A novel phospholipase A1 with sequencehomology to a mammalian Sec23p-interacting protein, p125. J. Biol. Chem. 277,11329-11335.

Nakatsu, F., Perera, R. M., Lucast, L., Zoncu, R., Domin, J., Gertler, F. B.,Toomre, D. and De Camilli, P. (2010). The inositol 5-phosphatase SHIP2regulates endocytic clathrin-coated pit dynamics. J. Cell Biol. 190, 307-315.

Ong, Y. S., Tang, B. L., Loo, L. S. and Hong, W. (2010). p125A exists as partof the mammalian Sec13/Sec31 COPII subcomplex to facilitate ER-Golgitransport. J. Cell Biol. 190, 331-345.

Pathre, P., Shome, K., Blumental-Perry, A., Bielli, A., Haney, C. J., Alber, S.,Watkins, S. C., Romero, G. and Aridor, M. (2003). Activation of phospholipaseD by the small GTPase Sar1p is required to support COPII assembly and ERexport. EMBO J. 22, 4059-4069.

Qiao, F. and Bowie, J. U. (2005). The many faces of SAM. Sci. STKE 2005, re7.Raaijmakers, J. H., Deneubourg, L., Rehmann, H., de Koning, J., Zhang, Z.,Krugmann, S., Erneux, C. and Bos, J. L. (2007). The PI3K effector Arap3interacts with the PI(3,4,5)P3 phosphatase SHIP2 in a SAM domain-dependentmanner. Cell. Signal. 19, 1249-1257.

Rowe, T. and Balch, W. E. (1995). Expression and purification of mammalian Sarl.Methods Enzymol. 257, 49-53.

Rowe, T., Aridor, M., McCaffery, J. M., Plutner, H., Nuoffer, C. and Balch, W. E.(1996). COPII vesicles derived from mammalian endoplasmic reticulummicrosomes recruit COPI. J. Cell Biol. 135, 895-911.

Saraste, J. and Svensson, K. (1991). Distribution of the intermediate elementsoperating in ER to Golgi transport. J. Cell Sci. 100, 415-430.

Sato, S., Inoue, H., Kogure, T., Tagaya, M. and Tani, K. (2010). Golgi-localizedKIAA0725p regulates membrane trafficking from the Golgi apparatus to theplasma membrane in mammalian cells. FEBS Lett. 584, 4389-4395.

Shimoi, W., Ezawa, I., Nakamoto, K., Uesaki, S., Gabreski, G., Aridor, M.,Yamamoto, A., Nagahama, M., Tagaya, M. and Tani, K. (2005). p125 islocalized in endoplasmic reticulum exit sites and involved in their organization.J. Biol. Chem. 280, 10141-10148.

Stagg, S. M., LaPointe, P., Razvi, A., Gurkan, C., Potter, C. S., Carragher, B.and Balch, W. E. (2008). Structural basis for cargo regulation of COPII coatassembly. Cell 134, 474-484.

Supek, F., Madden, D. T., Hamamoto, S., Orci, L. and Schekman, R. (2002).Sec16p potentiates the action of COPII proteins to bud transport vesicles. J. CellBiol. 158, 1029-1038.

Tani, K., Mizoguchi, T., Iwamatsu, A., Hatsuzawa, K. and Tagaya, M. (1999).p125 is a novel mammalian Sec23p-interacting protein with structural similarityto phospholipid-modifying proteins. J. Biol. Chem. 274, 20505-20512.

Weixel, K. M., Blumental-Perry, A., Watkins, S. C., Aridor, M. and Weisz, O. A.(2005). Distinct Golgi populations of phosphatidylinositol 4-phosphate regulatedby phosphatidylinositol 4-kinases. J. Biol. Chem. 280, 10501-10508.

Whittle, J. R. and Schwartz, T. U. (2010). Structure of the Sec13-Sec16 edgeelement, a template for assembly of the COPII vesicle coat. J. Cell Biol. 190,347-361.

Yamashita, A., Kumazawa, T., Koga, H., Suzuki, N., Oka, S. and Sugiura, T.(2010). Generation of lysophosphatidylinositol by DDHD domain containing 1(DDHD1): Possible involvement of phospholipase D/phosphatidic acid in theactivation of DDHD1. Biochim. Biophys. Acta 1801, 711-720.

Yorimitsu, T. and Sato, K. (2012). Insights into structural and regulatory rolesof Sec16 in COPII vesicle formation at ER exit sites. Mol. Biol. Cell 23, 2930-2942.

Zanetti, G., Pahuja, K. B., Studer, S., Shim, S. and Schekman, R. (2011). COPIIand the regulation of protein sorting in mammals. Nat. Cell Biol. 14, 20-28.

RESEARCH ARTICLE Journal of Cell Science (2014) 127, 1765–1778 doi:10.1242/jcs.138784

1778