In vitro reconstitution of ER-stress induced ATF6transport in COPII vesiclesAdam J. Schindler and Randy Schekman1

Department of Molecular and Cell Biology and Howard Hughes Medical Institute, Barker Hall, University of California, Berkeley, CA 94720-3202

Contributed by Randy Schekman, September 9, 2009 (sent for review July 27, 2009)

The transcription factor ATF6 is held as a membrane precursor inthe endoplasmic reticulum (ER), and is transported and proteolyti-cally processed in the Golgi apparatus under conditions of un-folded protein response stress. We show that during stress, ATF6forms an interaction with COPII, the protein complex required forvesicular traffic of cargo proteins from the ER. Using an in vitrobudding reaction that recapitulates the ER-stress induced transportof ATF6, we show that no cytoplasmic proteins other than COPII arenecessary for transport. ATF6 is retained in the ER by associationwith the chaperone BiP (GRP78). In the in vitro reaction, theATF6-BiP complex disassembles when membranes are treated withreducing agent and ATP. A hybrid protein with the ATF6 cytoplas-mic domain replaced by a constitutive sorting signal (Sec22bSNARE) retains stress-responsive transport in vivo and in vitro.These results suggest that unfolded proteins or an ER luminal �SHreactive bond controls BiP-ATF6 stability and access of ATF6 to theCOPII budding machinery.

regulated transport � unfolded protein response � BiP �prebudding complex

Endoplasmic reticulum (ER) homeostasis can be perturbed byup-regulation of secretory proteins or by disruption of

protein processing, a condition termed ER stress (1). ER stresscauses increased protein misfolding and leads to activation of theunfolded protein response (UPR), an evolutionarily conservedpathway to alleviate ER stress (2). The UPR leads to a slowdownin protein synthesis, an upregulation of ER chaperones, and anupregulation of ER-associated protein degradation.

The three effector proteins for the UPR are IRE1, PERK, andATF6. During the UPR, IRE and PERK remain in the ER andact via cytoplasmic effectors, whereas ATF6 is transported to theGolgi complex. In the Golgi, ATF6 is cleaved by Site-1 and Site-2proteases (3). These cleavages release the N-terminal cytoplas-mic domain, which contains a bZIP motif that binds DNA at ERstress response elements that control expression of the ERchaperones BiP (GRP78) and GRP94 (4).

Transport of ATF6 and other cargo proteins likely involves theCOPII coat, a complex of five cytoplasmic proteins that selectscargo at the ER membrane and pinches off membranes to formvesicles. The five proteins are the GTPase Sar1, which isrecruited to the ER membrane and initiates coat formation byGDP to GTP exchange; the heterodimeric complex Sec23/Sec24,which binds to Sar1�GTP; and a second dimeric complex,Sec13/Sec31, which binds to Sec23/24 and provides the curvaturenecessary for vesicle fission (5). Cargo is selected by interactionsbetween domains on the Sec24 subunit and cytoplasmic motifson cargo proteins (6). Inhibition of COPII by overexpression ofdominant negative Sar1 has been shown to block transport ofATF6 (7).

A central question in ATF6 function is how luminal stress isconverted into recognition by the cytoplasmic COPII complex.The luminal chaperone BiP binds ATF6 stably in unstressed cellsand dissociates specifically during stress (8). The triggers for thisrelease are unclear. Structural studies of IRE1, which is alsocontrolled by BiP, have posited direct recognition of misfoldedproteins that induces a conformation change to the active state

(9, 10). Whether ATF6 is triggered by a similar mechanism ofmisfolded protein binding is unclear.

Evidence supports a model in which accessory proteins arenecessary for ATF6 transport. When the luminal domain ofATF6 was fused to the cytoplasmic domain of the ER-localizedprotein LZIP, the chimera still trafficked to the Golgi during ERstress (11), demonstrating that the ATF6 cytoplasmic domain isnot required for transport. A sorting signal may be provided bya cargo receptor that engages ATF6 in the ER lumen and COPIIin the cytoplasm, an interaction occurring specifically duringstress.

We have established in vitro reactions that recapitulate ATF6dissociation from BiP and engagement with COPII during stress.These data show that during stress ATF6 undergoes alterationsthat deliver it to COPII vesicles for transport.

ResultsATF6 Transport Is Recapitulated in Vitro in a Cell Line Stably Express-ing FLAG-ATF6. To assess ATF6 transport, we generated a stableCHO cell line, CHO-ATF6, expressing full-length ATF6 withthree copies of the FLAG epitope at the N terminus (12).CHO-ATF6 cells were used to measure the response of ATF6 toER stress (Fig. 1A). Full-length ATF6 at 110 kDa represents theER-localized protein, and the band at 65 kDa represents ATF6that has trafficked to the Golgi and been cleaved. Cells weretreated with the ER stress inducers dithiothreitol (DTT), tuni-camycin (Tm), or thapsigargin (Tg) for varying times. DTTbreaks disulfide bonds to unfold proteins, Tm blocks N-glycosylation of nascent polypeptides, and Tg inhibits ER cal-cium pumps. As a control, cells were treated with hydrogenperoxide (H2O2), an oxidative stress agent. DTT induced a rapidshift in ATF6 to the processed form, causing cleavage of greaterthan half of cellular ATF6 within 30 min. Tm and Tg were bothslower-acting, requiring 2–4 h for maximum effect. H2O2 didnot cause ATF6 cleavage (Fig. 1 A). The localization of ATF6was examined by immunofluorescence. ATF6 in unstressedor H2O2-treated cells was diffuse in the ER. During stress,ATF6 partially localized with a Golgi protein, Gos28, anddisplayed a more prominent nuclear localization, consistent withits trafficking (Fig. S1).

To assess the mechanisms of ATF6 transport, we used an invitro assay that recapitulated ER stress-induced vesicle budding.CHO-ATF6 cells were permeabilized with digitonin to allowaccess to intracellular organelles. This technique has been shownto generate functional ER membranes that maintain proteintopology and resist protease treatment (13, 14). In vitro COPIIvesicle budding from the ER was induced by addition of GTP,rat liver cytosol, ATP, and an ATP regeneration system. Afterincubation to allow vesicle formation, membranes were removed

Author contributions: A.J.S. and R.S. designed research; A.J.S. performed research; A.J.S.contributed new reagents/analytic tools; A.J.S. and R.S. analyzed data; and A.J.S. and R.S.wrote the paper.

The authors declare no conflict of interest.

1To whom correspondence should be addressed. E-mail: schekman@berkeley.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0910342106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0910342106 PNAS � October 20, 2009 � vol. 106 � no. 42 � 17775–17780

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

June

22,

202

0

by sedimentation at low speed, and budded vesicles were collectedby sedimentation at high speed. The packaging of 3�-FLAG-ATF6was compared with a positive control protein, ERGIC-53, alectin that cycles between the ER and ER-Golgi intermediatecompartment, and a negative control, Ribophorin-I, part of theoligosaccharyl transferase complex and a resident ER protein.As a second positive control, we examined the budding ofSec22b, a v-SNARE that transports constitutively between theER and Golgi.

ATF6 budded poorly in the standard reaction (Fig. 1B, lanes 4and 5). When DTT was added into the reaction, ATF6 was enrichedin vesicles (lanes 6–8). DTT did not affect ERGIC-53 or Sec22bbudding, and did not cause significant Ribophorin-I release. Noother toxin caused ATF6 budding. The inability of Tm and Tg topackage ATF6 may seem in conflict with data from the cleavageassay (Fig. 1A). However, DTT can act on folded proteins directlyby reducing disulfide bonds, whereas Tm and Tg may act onlyindirectly on proteins as they are synthesized, conditions that likelydo not occur in the cell-free transport reaction. Other reports havealso shown that DTT, but not Tg, can induce ATF6 transport in theabsence of protein synthesis (15).

ATF6 packaging in the presence of DTT was 2- to 5-foldhigher than in untreated membranes. The overall efficiency ofATF6 packaging in the presence of DTT was �5–10%. Theefficiency of ATF6 transport was lower in vitro than in vivo after1 h treatment with DTT. The decreased efficiency may beaccounted for by the inaccessibility of the majority of nativelyfolded, disulfide-linked proteins to reducing agents (16). DTTwill likely generate higher levels of unfolded proteins in cellsengaged in protein synthesis, conditions that are not reproducedin our cell-free reaction.

We next examined the budding of endogenous ATF6 todetermine whether the cell-free reaction reproduced transportof physiological levels of ATF6 (Fig. 1C). We found thatendogenous ATF6 in CHO.K1 cells was packaged with similarefficiency in response to DTT as FLAG-tagged ATF6 in stablytransfected cells, demonstrating that the effect is valid for ATF6expressed at normal levels.

In Vitro Packaging of ATF6 Requires Luminal Reducing Activity andCOPII Proteins. We probed the requirement for a reducing agentusing permeabilized cells treated with two membrane-permeable reducing agents, DTT and �-mercaptoethanol(BME), and a membrane-impermeable reducing agent, Tris(2-Carboxyethyl) phosphine hydrochloride (TCEP) (Fig. 2A) (17).Both DTT and BME stimulated packaging, whereas TCEP hada small stimulatory effect at lower concentration and an inhib-itory effect at higher concentration. The reducing potential ofTCEP is greater than that of DTT at neutral pH (18); therefore,the levels of all chemicals were comparable. These results showthat reducing activity in the lumen of the ER is necessary forATF6 packaging into vesicles.

Fig. 1. ATF6 transports in vivo and in vitro in stably expressing cells. (A) CHO.K1 cells stably expressing 3�-FLAG-ATF6 were treated in 24-well culture disheswith toxins for indicated times. Full-length protein is 110 kDa, cleaved protein is 65 kDa. (B) Permeabilized CHO-ATF6 cells were incubated with GTP, ATP, andan ATP-regenerating system (ATPr), and 4 mg/mL rat liver cytosol for 1 h at room temperatur. Lane 1, 20% load of starting membranes; lanes 2 and 3, controls,in which either nucleotides or rat liver cytosol were excluded; lanes 4 and 5, duplicate incubations of the full budding reaction; lanes 6–17, full reactions withthe addition of indicated toxins. Control proteins analyzed in parallel with ATF6 were ERGIC-53, a constitutively transported membrane lectin; Sec22b, aconstitutively transported v-SNARE, and Ribophorin-I, a component of the oligosaccharyl transferase complex and resident ER protein. Proteins were analyzedby SDS/PAGE and immunoblotting. Quantification was performed by comparing the signal in vesicle lanes to the 20% starting membrane signal in lane 1. (C)Budding reaction in CHO.K1 cells with immunoblotting to endogenous ATF6. *, Nonspecific band.

Fig. 2. In vitro budding reaction recapitulates ER-stress induced ATF6 trans-port. (A) Budding reaction in the presence of membrane permeable reducingagents DTT and BME, or membrane-impermeable reducing agent TCEP.(B) Budding reactions conducted in the presence of COPII inhibitors GTP�Sand Sar1H79G, or controls GDP and Sar1, or in the presence of the COPIinhibitor Brefeldin A or the proteasome inhibitor ALLN. All lanes contained2 mM DTT.

17776 � www.pnas.org�cgi�doi�10.1073�pnas.0910342106 Schindler and Schekman

Dow

nloa

ded

by g

uest

on

June

22,

202

0

Selective inhibitors were used to assess the role of COPII inATF6 budding. A nonhydrolyzable GTP analog, GTP�S, and adominant negative Sar1, Sar1H79G, both blocked budding ofATF6 and ERGIC-53 (Fig. 2B). GDP had no effect, and WTSar1 enhanced budding. Brefeldin A, which prevents retrogradetransport from Golgi to ER, had a moderately inhibitory effecton all cargo, and the proteasome inhibitor ALLN had no effect.We conclude that ATF6 packaging depends on COPII.

Budded vesicles convey ATF6 to the Golgi membrane in vivo.Fusion of vesicles at the Golgi exposes ATF6 to Site-1 and Site-2protease processing. We examined whether this fusion eventoccurred in our in vitro assay. We found no processing of ATF6,suggesting that the full transport event was not sustained in thisreaction (Fig. S2).

ATF6 Physically Interacts with COPII for Transport. We examined thephysical interaction of ATF6 with COPII on ER membranesusing a prebudding assay in which purified COPII proteins arerecruited to membranes and COPII-bound ER proteins isolated.The Sec23/24 complex is hypothesized to bind to a GTP-restricted Sar1 mutant, Sar1H79G, but not to a GDP-restrictedSar1, Sar1T39N (Fig. 3A). Cargo is captured by Sec23/24 onmembranes.

In the experiment, GST-Sar1 and Sec23/24 were mixed withmembranes and the bound proteins on sedimented membranessolubilized with detergent to permit isolation of cargo complexesby immobilized glutathione. Salt-washed, permeabilized cellswere untreated or pretreated with 2 mM DTT for 15 min inculture to mobilize ATF6 to ER exit sites. Immunoblot results(Fig. 3B) showed that both forms of GST-Sar1 were recruited tomembranes, with Sar1T39N binding slightly better to mem-branes than Sar1H7G (Fig. 4B, ‘‘membrane bound’’ lanes 3 and4, and 7 and 8). Sec23 bound to membranes incubated with

Sar1H79G, but less well to membranes incubated with Sar1T39N,consistent with the requirement for GTP (lanes 1 and 2, and 5and 6). ERGIC-53 served as a control protein for specific cargobinding, and Ribophorin-I served as a control for nonspecificbinding. ERGIC-53 was recruited to Sar1H79G/Sec23/24, whereas

Fig. 3. ATF6 forms a physical association with COPII proteins and transports without additional cytoplasmic proteins. (A) Diagram of the prebudding assayscheme to detect associations between ATF6 and COPII. (Left) GDP-restricted form of Sar1, Sar1T39N, is not able to recruit the Sec23/24 complex to membranes,and cargo proteins (such as ATF6) are not precipitated by glutathione pulldown to Sar1. (Right) GTP-restricted Sar1, Sar1H79G, recruits Sec23/24 and cargoproteins. ATF6 binding to COPII complexes is hypothesized to require DTT stimulation. Adapted from (25). (B) Prebudding assay in the presence of 10 �g of theSec23/24 complex and 5 �g of GST-Sar1. (Left) Absence of 2 mM DTT pretreatment; (Right) After DTT pretreatment. Lanes 1 and 2, and 7 and 8 containsupernatants that did not bind to membranes in the reaction. These lanes demonstrate that Sec23/24 preferentially binds to Sar1�GTP, because unbound Sec23/24is elevated in the presence of Sar1T39N and depleted in the presence of Sar1H79G. Lanes 3 and 4, and 9 and 10 contain proteins bound to membranes and solubilizedby detergent. These lanes contain resident ER proteins and soluble proteins that adhered to membranes. Lane 5 and 6, and 11 and 12 are proteins extracted afterglutathione pulldown of GST-Sar1 complexes. Cargo proteins ERGIC-53 and ATF6 are elevated in the Sar1H79G condition. ATF6 is present specifically afterpretreatment with DTT (arrow). Bound fractions were 30% of the total; lysates were 1% of the total. (C) Quantification of four assays as in B. Binding efficiencywas determined by subtracting the T39N signal from the H79G signal in the beads lanes, and calculating the bound amount as a percentage of total protein inlysates. n � 4; *, P � 0.05. (D) Budding reaction in HeLa-ATF6 cells in the presence of COPII and absence of cytosol. First four lanes represent the standard buddingassay with 4 mg/mL cytosol. Remaining lanes had cytosol excluded and COPII added. Values for COPII indicate amounts each of Sar1, Sec23/24 complex, andSec13/31 complex.

Fig. 4. ATP and DTT act synergistically to dissociate BiP from ATF6. (A)Immunoprecipitation of FLAG-ATF6 in permeabilized CHO-ATF6 cells incu-bated for 30 min in the presence of 1 mM ATP or ATP�S, or 5 mM DTT. Afterincubation, cells were washed, lysed, and immunoprecipitated by �-FLAGfollowed by immunoblotting to FLAG and BiP. (B) Quantification of tworeplicates of assay as in A. BiP:ATF6 ratios were determined for individualreactions and normalized to the value for untreated cells. (C) Budding reactionto detect the presence of BiP in vesicles.

Schindler and Schekman PNAS � October 20, 2009 � vol. 106 � no. 42 � 17777

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

June

22,

202

0

Ribophorin-I was not (lanes 5 and 6, and 11 and 12). ATF6 wasrecruited in the Sar1H79G/Sec23/24 condition after DTT treat-ment (lane 12). In four assays, DTT-stimulated binding of ATF6to COPII was detected at 3-fold above untreated samples (Fig.3C). The efficiency of ATF6 binding was low compared withERGIC-53 (0.5% ATF6; 2.5% ERGIC-53), a difference com-parable with that seen in the budding assay. These resultsdemonstrate that ATF6 forms physical association with COPIIproteins under conditions of ER stress.

We wished to assess the role of COPII as the sole cytosolicrequirement for ATF6 packaging in our cell-free reaction. Cellswere washed with 1 M KoAc to remove peripheral proteins, andpurified COPII was added in a titration experiment. ATF6 andother cargo proteins were packaged in a COPII-concentrationdependent manner (Fig. 3C). ATF6, although less efficientlypackaged than other cargo, was most stimulated by increasingamounts of COPII. These data demonstrate that ATF6 pack-aging requires no cytosolic components other than COPII.

ATP and DTT Synergistically Induce BiP Dissociation from ATF6 in Vitro.ATF6 dissociation from BiP is necessary for transport (12). Totest whether stress reproduces this effect in vitro, permeabilizedcells were treated with combinations of ATP and DTT, followedby FLAG-ATF6 immunoprecipitation and immunoblot of BiP(Fig. 4A). Treatment with ATP or DTT alone caused approx-imately a 30% reduction in the ratio of BiP:ATF6 signal relativeto untreated cells (Fig. 4B). Although BiP has been found tostably associate with ATF6 (8), the high levels of ATP in theassay may allow some release of BiP. The combination of ATPand DTT reduced the association to �50%, and ATP�S, anonhydrolyzable analog of ATP, reduced the complex to �30%of the untreated control. These data demonstrate that DTT andATP act together to release BiP from ATF6. This dissociationmay trigger access of ATF6 to COPII.

BiP is normally retained in the ER. We examined the effectof DTT on the retention or transport of BiP (Fig. 4C). BiP wasfound only at low levels in vesicles, and its packaging was notinduced by DTT. Thus, it appears that ATF6 is packaged in theabsence of BiP.

ATF6 Is Primed to Transport After in Vivo Treatment. We tookadvantage of the ability of DTT to mobilize ATF6 rapidly in vivoto assess whether DTT was required in the in vitro assay.CHO-ATF6 cells were pretreated in culture for 15 or 60 min with2 mM DTT. This short time period was designed to mobilizeATF6 for transport while maintaining it in the ER. Permeabil-ized cells were washed to remove DTT and incubated in thebudding reaction in parallel with nonpretreated controls (Fig.5A). ATF6 packaging was enhanced after pretreatment even inthe absence of added DTT. In contrast, cells that were notpretreated displayed the standard DTT sensitivity, with littleconstitutive activity. A similar effect was seen after 60 mintreatment, although budding efficiency was reduced from 15min. Thus, the COPII-ATF6 interaction in vitro does not requirethe presence of reducing agent if ATF6 is primed for transport.

Treatment of membranes with high salt removes peripherallyattached complexes. We examined the effect of such a treatmenton DTT-induced packaging of ATF6 in vitro. Cells were eitheruntreated or pretreated for 15 min with DTT before harvest, andboth sets of permeabilized cells were salt washed with 1 M KoAcfor 15 min (Fig. 5B). This treatment reduced the overall buddingefficiency, but did not alter the effects of pretreatment seen inFig. 5A. These results demonstrate that the ATF6 complexformed during ER stress is stable to high salt treatment thatremoves most peripheral proteins.

ATF6 Is Actively Retained in the ER via Its Luminal Domain. It has beenestablished that the luminal domain of ATF6 controls its

localization independent of the cytoplasmic domain (12). Weexamined the contribution of the ATF6 luminal domain in thecontext of a constitutively transported protein, the v-SNARESec22b. Both Sec22b and ATF6 are single-pass, type II mem-brane proteins. The Sec22b sorting signal for COPII bindinghas been mapped to a structural motif near the ER membrane(19).

We conducted cleavage assays on ATF6 and on constructs inwhich the ATF6 cytoplasmic domain was replaced with that ofSec22b. A constitutive transport mutant, ATF6 1-500 (�431-475)(12), was cleaved in the presence or absence of ER stress,whereas a transport-deficient mutant ATF6 1-670 (�468-500)(12), missing the luminal transport signal, remained intact withor without DTT (Fig. 6A). These same constructs were gener-ated with the Sec22b cytoplasmic domain (Fig. 6B) and alsoexamined by cleavage assay. Sec22b-ATF6TM migrated at thepredicted 22-kDa size (lanes 9 and 10). The chimeric proteincontaining the full ATF6 luminal domain generated a cleavedspecies predominantly in the presence of DTT (lanes 3 and 4).The luminal domain mutants (lanes 5–8) displayed the samephenotype as seen for ATF6 mutants (Fig. 6C).

We used the COPII budding reaction to examine Sec22b-ATF6TM and Sec22b-ATF6 transport activity. The transmem-brane-only construct budded in the absence or presence of DTT,analogous to wild-type Sec22b. Sec22b-ATF6 transport showedhigh constitutive transport, likely because the Sec22b domainconferred some degree of transport. Still, it was stimulated2.4-fold by DTT, showing that it was restrained in the absenceof stress. Thus, in cells and in the cell-free budding reaction, theATF6 retention signal overrode at least in part the constitutiveSec22b sorting signal.

DiscussionWe report a cell-free vesicle budding reaction that recapitulatesthe ER-stress induced transport of the transcription factorATF6. The addition of a reducing agent, DTT, mobilized ATF6into COPII vesicles from the ER in the absence of proteinsynthesis. No other cytosolic adaptor proteins are required topackage ATF6 into COPII vesicles.

Fig. 5. Pretreatment in culture overcomes the requirement for DTT in vitro.(A) CHO-ATF6 cells were pretreated in culture with 2 mM DTT for indicatedtimes and washed to remove DTT. Budding reactions were conducted in theabsence or presence of DTT. (B) Cells were pretreated as in A, permeabilizedand salt washed with 1 M KoAc for 15 min to remove peripheral membraneproteins, followed by the budding reaction.

17778 � www.pnas.org�cgi�doi�10.1073�pnas.0910342106 Schindler and Schekman

Dow

nloa

ded

by g

uest

on

June

22,

202

0

ATF6 dissociates from the luminal chaperon BiP (GRP78) inresponse to exogenous reducing agent in cells and in our cell-freereaction. The trigger for BiP dissociation may be proteins unfoldedin response to reducing agent or a specific �SH reactive sensorprotein, either of which may expose BiP to ATP binding and releasefrom ATF6. By analogy, antibody light chain binding to heavy chaincauses ATP release of BiP from heavy chain (20).

We observed a basal level of constitutive transport of ATF6 inthe absence of DTT, whereas the control protein Ribophorin-Iwas entirely excluded. ATF6 may have low levels of basaltransport that serve to maintain cellular homeostasis. Reportshave shown a role for ATF6 in development (21) and in thesurvival of dormant tumor cells (22).

The experiments with Sec22-ATF6 fusion proteins show thatATF6 does not transport constitutively in the context of a strongCOPII sorting signal (19). ATF6 is present in a complex thatdisplays slow diffusional mobility in the ER, possibly suggestinga higher-order complex (8). ATF6 may interact with additionalproteins besides BiP that restrain it from transport. Alterna-tively, the slow diffusion of ATF6 in the ER under unstressedconditions may limit its accessibility to COPII at ER exit sites.

These data suggest that BiP release from ATF6, triggered byunfolded proteins or a stress-sensor protein, is followed byrecognition of ATF6 or a partner protein by COPII. Pastexperiments showed that ATF6 still transported when its cyto-plasmic domain was replaced with the cytoplasmic domain of anER-resident protein (11), suggesting that transport involves aprotein that mediates association between the luminal domain ofATF6 and the cytoplasmic COPII proteins. There are manyexamples of such receptors required to convey a membrane

cargo protein into COPII vesicles (23). Some receptors accom-pany membrane cargo proteins into the vesicle, whereas othersare retained in the ER (24). In the case of regulated transport,the best understood example is sterol control of SREBP, whichdepends on COPII binding to the cargo receptor SCAP (25). Aproteomic analysis of COPII vesicles formed under conditionspermissive for ATF6 packaging may facilitate the identificationof its sorting receptor.

MethodsPlasmids and Cell Lines. The WT, �431-475, and �468-500 human ATF6� cDNAwith a 3� N-terminal FLAG tag were gifts of R. Prywes (Columbia University,New York, NY) and are described elsewhere (12). The original Sec22b cDNAwas a gift of J. Hay (University of Montana, Missoula, MT). Chimeric Sec22-ATF6 proteins were made by amplifying Sec22b to include a 5� FLAG tag andexclude the 3� transmembrane region. PCR fragments were cloned into thepIRES2.neo (Clontech) backbone at StuI and ClaI sites. Transmembrane andluminal regions of ATF6 were PCR amplified to include a 5� ClaI site and a 3�XbaI site, and were cloned into the Sec22b.pIRES2 vector. All plasmids weresequenced before use.

Transfections were with Lipofectamine 2000 (Invitrogen) according tomanufacturer’s protocols. Transient transfections were conducted for 18–24h. For stable transfections in HeLa and CHO cells, plasmids were subcloned intothe pIRES.hyg3 vector. Cells were kept in a 37 °C incubator at 5% CO2.

Antibodies. Commercial antibodies were monoclonal �-FLAG M2 (Sigma),monoclonal �-Actin C4 (ICN Biomedicals), and rabbit polyclonal �-ATF6 andrabbit polyclonal �-Grp78 (Santa Cruz). Anti-Gos28 polyclonal antibody isdescribed elsewhere (26). Anti-ERGIC-53 antiserum was raised against anEscherichia coli purified GST fusion of amino acids 49–484 in the rat protein.�-Ribophorin I was raised against C-terminal peptides corresponding to hu-man protein (residues 588–605) and mouse protein (residues 576–605). The

Fig. 6. The ATF6 luminal domain controls its localization. (A) Cleavage assay of transiently transfected ATF6 constructs in HeLa cells, either untreated or treatedwith 2 mM DTT for 1 h. Red arrow indicates size of the cleaved N-terminal domain. (B) Schematic representation of chimeric proteins with the human Sec22bcytoplasmic domain (194 aa; yellow), ATF6 transmembrane domain (light blue), and varying ATF6 luminal domains (teal). Proteins contained an N-terminal FLAGtag. (C) Cleavage assay in HeLa cells transiently transfected with constructs from B. Treatment was for 1 h with 2 mM DTT. (D) Budding reactions on HeLa cellstransiently transfected with Sec22b-ATF6 chimeras. (Upper) Sec22b-ATF6TM; (Lower) Sec22b-ATF6.

Schindler and Schekman PNAS � October 20, 2009 � vol. 106 � no. 42 � 17779

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

June

22,

202

0

�-Sec22b was raised against a peptide corresponding to human residues91–116. The �-Sar1 was raised against full-length, thrombin cleaved GST-fusion protein purified from E. coli.

Cleavage Assay. CHO-ATF6 cells or transfected HeLa cells were grown in24-well plates to 90% confluency. Media was changed, and 1 h later cells weretreated with toxins for indicated times. For harvesting, wells were washed 1�

with PBS, followed by addition of heated SDS sample buffer

In Vitro Budding Assay. The in vitro vesicle budding assay was performed asdescribed (27). For description, see SI Methods.

Prebudding Assay. Prebudding complex pulldowns were performed as de-scribed previously (25) using permeabilized, salt-washed CHO-ATF6 cells. Fordescription, see SI Methods.

BiP Dissociation and Immunoprecipitation. Permeabilized cells were incubatedfor 30 min at 22 °C in the presence of 1 mM ATP or ATP�S, or 5 mM DTT. Cellswere washed three times and lysed for 1 h in 50 mM Tris, pH 8.0, 150 mM NaCl,10% glycerol, 2 mM EDTA, and protease inhibitors. Lysates were pelleted at14,000 � g for 20 min, and supernatants cleared for 1 h with IgG Sepharose.ATF6:BiP complexes were isolated by incubation with 1 �g �-FLAG M2 anti-

body conjugated to protein A Sepharose. Control reactions used 1 �g mono-clonal �-GFP (Roche).

Immunoblotting and Protein Quantification. Immunoblotting was carried outaccording to standard procedures, using either ECL plus enhanced chemilu-minescence (Amersham) or an Odyssey infrared imaging system (Li-Cor Bio-sciences) with Alexa Fluor 680 �-rabbit (Invitrogen) or IR-dye 800 �-mouse(Molecular Probes) secondary antibodies.

Quantification of protein signal was performed on scanned membranes usingOdyssey software or ImageJ. Budding efficiency was calculated by comparing theamount in each vesicle lane with the amount in the starting membrane lane.Prebuddingefficiencywascalculatedbysubtractingthebackgroundsignal in thebound Sar1-GDP lane from the bound Sar1-GTP lane, and dividing by the amountpresent in a comparable aliquot of starting lysate. Lysates were loaded at 1% oftotal, and bound fractions loaded at 30% of total.

Immunofluorescence Microscopy and Protein Purification. For descriptions, seeSI Methods.

ACKNOWLEDGMENTS. We thank Ron Prywes and Jesse Hay for plasmids;J. Christopher Fromme, Jinoh Kim, and Bertrand Kleizen for protein purifica-tion and antibody production; Ann Fischer and Michelle Yasukawa for tissueculture; and all members of the R.S. laboratory who assisted with experimen-tal design, data analysis, and manuscript preparation.

1. Rutkowski DT, Kaufman RJ (2004) A trip to the ER: Coping with stress. Trends Cell Biol14:20–28.

2. Schroder M, Kaufman RJ (2005) The mammalian unfolded protein response. Annu RevBiochem 74:739–789.

3. Ye J, et al. (2000) ER stress induces cleavage of membrane-bound ATF6 by the sameproteases that process SREBPs. Mol Cell 6:1355–1364.

4. Yoshida H, Haze K, Yanagi H, Yura T, Mori K (1998) Identification of the cis-actingendoplasmic reticulum stress response element responsible for transcriptional induc-tion of mammalian glucose-regulated proteins. Involvement of basic leucine zippertranscription factors. J Biol Chem 273:33741–33749.

5. Bickford LC, Mossessova E, Goldberg J (2004) A structural view of the COPII vesicle coat.Curr Opin Struct Biol 14:147–153.

6. Miller EA, et al. (2003) Multiple cargo binding sites on the COPII subunit Sec24p ensurecapture of diverse membrane proteins into transport vesicles. Cell 114:497–509.

7. Nadanaka S, Yoshida H, Kano F, Murata M, Mori K (2004) Activation of mammalianunfolded protein response is compatible with the quality control system operating inthe endoplasmic reticulum. Mol Biol Cell 15:2537–2548.

8. Shen J, Snapp EL, Lippincott-Schwartz J, Prywes R (2005) Stable binding of ATF6 to BiPin the endoplasmic reticulum stress response. Mol Cell Biol 25:921–932.

9. Credle JJ, Finer-Moore JS, Papa FR, Stroud RM, Walter P (2005) On the mechanism ofsensing unfolded protein in the endoplasmic reticulum. Proc Natl Acad Sci USA102:18773–18784.

10. Kimata Y, et al. (2007) Two regulatory steps of ER-stress sensor Ire1 involving its clusterformation and interaction with unfolded proteins. J Cell Biol 179:75–86.

11. Chen X, Shen J, Prywes R (2002) The luminal domain of ATF6 senses endoplasmicreticulum (ER) stress and causes translocation of ATF6 from the ER to the Golgi. J BiolChem 277:13045–13052.

12. Shen J, Chen X, Hendershot L, Prywes R (2002) ER stress regulation of ATF6 localizationby dissociation of BiP/GRP78 binding and unmasking of Golgi localization signals. DevCell 3:99–111.

13. Wilson R, et al. (1995) The translocation, folding, assembly and redox-dependentdegradation of secretory and membrane proteins in semi-permeabilized mammaliancells. Biochem J 307:679–687.

14. Kleizen B, van Vlijmen T, de Jonge HR, Braakman I (2005) Folding of CFTR is predom-inantly cotranslational. Mol Cell 20:277–287.

15. Haze K, Yoshida H, Yanagi H, Yura T, Mori K (1999) Mammalian transcription factorATF6 is synthesized as a transmembrane protein and activated by proteolysis inresponse to endoplasmic reticulum stress. Mol Biol Cell 10:3787–3799.

16. Thornton JM (1981) Disulphide bridges in globular proteins. J Mol Biol 151:261–287.17. Cline DJ, et al. (2004) New water-soluble phosphines as reductants of peptide and

protein disulfide bonds: Reactivity and membrane permeability. Biochemistry43:15195–15203.

18. Han JC, Han GY (1994) A procedure for quantitative determination of tris(2-carboxyethyl)phosphine, an odorless reducing agent more stable and effective thandithiothreitol. Anal Biochem 220:5–10.

19. Mancias JD, Goldberg J (2007) The transport signal on Sec22 for packaging intoCOPII-coated vesicles is a conformational epitope. Mol Cell 26:403–414.

20. Vanhove M, Usherwood YK, Hendershot LM (2001) Unassembled Ig heavy chains donot cycle from BiP in vivo but require light chains to trigger their release. Immunity15:105–114.

21. Yamamoto K, et al. (2007) Transcriptional induction of mammalian ER quality controlproteins is mediated by single or combined action of ATF6alpha and XBP1. Dev Cell13:365–376.

22. Schewe DM, Aguirre-Ghiso JA (2008) ATF6alpha-Rheb-mTOR signaling promotes sur-vival of dormant tumor cells in vivo. Proc Natl Acad Sci USA 105:10519–10524.

23. Malkus P, Graham LA, Stevens TH, Schekman R (2004) Role of Vma21p in assembly andtransport of the yeast vacuolar ATPase. Mol Biol Cell 15:5075–5091.

24. Kuehn MJ, Schekman R, Ljungdahl PO (1996) Amino acid permeases require COPIIcomponents and the ER resident membrane protein Shr3p for packaging into transportvesicles in vitro. J Cell Biol 135:585–595.

25. Sun LP, Seemann J, Goldstein JL, Brown MS (2007) Sterol-regulated transport of SREBPsfrom endoplasmic reticulum to Golgi: Insig renders sorting signal in Scap inaccessibleto COPII proteins. Proc Natl Acad Sci USA 104:6519–6526.

26. Nagahama M, et al. (1996) A v-SNARE implicated in intra-Golgi transport. J Cell Biol133:507–516.

27. Kim J, Hamamoto S, Ravazzola M, Orci L, Schekman R (2005) Uncoupled packaging ofamyloid precursor protein and presenilin 1 into coat protein complex II vesicles. J BiolChem 280:7758–7768.

17780 � www.pnas.org�cgi�doi�10.1073�pnas.0910342106 Schindler and Schekman

Dow

nloa

ded

by g

uest

on

June

22,

202

0