Immunity, Vol. 15, 105–114, July, 2001, Copyright 2001 by Cell Press

Unassembled Ig Heavy Chains Do Not Cyclefrom BiP In Vivo but Require Light Chainsto Trigger Their Release

Ig heavy chains (HC) and light chains (LC) are com-prised of a series of Ig domains that can fold indepen-dently of each other in vitro (Goto et al., 1979). In theabsence of LC synthesis, HCs are efficiently retained inthe ER due to their association with BiP (Haas and Wabl,

Marc Vanhove,2 Young-Kwang Usherwood,3

and Linda M. Hendershot1

Department of Tumor Cell BiologySt. Jude Children’s Research Hospital332 North LauderdaleMemphis, Tennessee 38105 1983; Bole et al., 1986). Although BiP interacts tran-

siently with multiple HC domains (Kaloff and Haas,1995), only its interaction with the CH1 domain is persis-tent and responsible for ER retention (Hendershot et al.,Summary1987). Like all hsp70 proteins, BiP binds ADP and ATPtightly (Wei and Hendershot, 1995) through its N-termi-Unassembled Ig heavy chains are retained in the ERnal domain, which is essential to its chaperone activityvia the binding of BiP to the CH1 domain, which remains(Wei et al., 1995). BiP recognizes extended conforma-unoxidized. Interestingly, this domain folds rapidly, al-tions on unfolded substrates (Blond-Elguindi et al.,beit nonproductively, when heavy chains are released1993), which can be accommodated in a cleft in the C-ter-from BiP in vitro with ATP. The in vivo cycling of BiPminal peptide binding domain (Zhu et al., 1996). Thefrom heavy chains was monitored using BiP ATPaseidentification of putative BiP binding sequences on HCsmutants as kinetic traps. Our data suggest that BiPand LCs (Knarr et al., 1995) suggests that BiP binds todoes not cycle from the CH1 domain of free heavyhydrophobic stretches on nascent or unassembled Igchains. However, heavy and light chain assembly oc-chains that become inaccessible after folding and as-curs rapidly and requires the ATP-dependent releasesembly are complete.of BiP. We propose that BiP’s ATPase cycle is stalled

In the absence of LCs, the BiP-bound CH1 domain ofor nonproductive when it is bound to free heavy chains.the HC remains unfolded (unoxidized) in the cell (Lee etThe binding of light chains to the complex reactivatesal., 1999). However, when BiP is released from HC inthe cycle and releases BiP.vitro with ATP, this domain is able to fold. This led usto examine the nature of BiP’s interaction with the CH1

Introduction domain and to determine the requirements for CH1 do-main folding. Using BiP ATPase mutants as “kinetic

The development of B lineage cells is characterized in traps,” we measured the dynamics of the BiP-HC com-part by the ordered expression of the heavy and light plex. Surprisingly, we found that in vivo the HCs bindchain subunits of the Ig receptor. This asynchronous stably to BiP and do not continuously undergo cyclesexpression is critical for monitoring the mechanisms of binding and release. Instead it appears that LCs trig-used to generate maximum diversity from the limited ger the ATP-dependent BiP release from HCs.number of V, D, and J region gene segments. For mostgenes, great care is taken to ensure that mutations are

Resultsrapidly detected and repaired to maintain primary aminoacid sequences that fold efficiently into biologically ac-

The CH1 Domain Folds Rapidly When HCs Aretive proteins. Instead, the rearrangements that give riseReleased from BiP In Vitro but Aggregateto variable regions are characterized by imprecise join-Readily at 37�Cing of the V, D, and J gene segments, addition of nontem-We recently constructed a simplified HC containing theplated bases to their ends by TdT, and finally directedVH and CH1 domains but lacking the hinge region, whichhypermutation of the variable region (Maizels, 1995; Gel-contains the cysteines involved in HC dimerization, thelert, 1992). While these mechanisms are essential forCH2, and the CH3 domains. Using this simplified two-generating diversity and allowing affinity maturation ofdomain HC, we showed that in vivo the CH1 domain ofthe immune response, they clearly increase the likeli-HCs bound to BiP remains unfolded, thus providing ahood of producing a protein incapable of folding prop-persistent site for BiP interaction (Lee et al., 1999). Thiserly or assembling with its heavy or light chain partnercreated a paradox, since � HCs have a long half-life andand the necessary accessory molecules. The failure tothe CH1 domain is capable of folding in vitro if BiP iscorrectly execute and monitor these steps would wreakreleased with ATP (Lee et al., 1999; this study). We rea-havoc on the proper development of the B cell repertoiresoned that either BiP cycling must be faster than theand functioning of the immune system. Thus, B lineagerate of CH1 domain folding or BiP must not cycle fromcells are particularly dependent on endoplasmic reticu-HCs in vivo. To determine how fast BiP would need tolum (ER) quality control (Hammond and Helenius, 1994).cycle, we first measured the kinetics of CH1 domainfolding (i.e., oxidation) after its release from BiP in vitro.

1 Correspondence: linda.hendershot@stjude.org Cells were lysed and incubated at 22�C with ATP for2 Present address: Laboratoire d’Enzymologie and Centre d’Ingen- various times before N-ethylmaleimide (NEM) wasierie des Proteines, Institut de Chimie B6, Universite de Liege, Sart-

added to stop oxidation of the CH1 domain. BiP wasTilman, B-4000, Liege, Belgium.released immediately at 22�C; �50% of the truncated3 Present address: Trudeau Institute Inc., 100 Algonquin Avenue,

Saranac Lake, New York 12983. HCs were fully oxidized within 5 min of release, and by

Immunity106

Figure 1. HCs Released from BiP In Vitro Ei-ther Fold or Aggregate

(A and B) In vitro release of simplified HCsfrom BiP and the effect on their oxidationstatus. COS cells cotransfected with WThamster BiP and simplified HCs (HA-�CH1)were metabolically labeled for 1 hr, lysed inbuffer containing ATP, and incubated at ei-ther 22�C (A) or 37�C (B) for the indicatedamount of time before NEM was added tostop further oxidation. ImmunoprecipitatedHCs were analyzed under nonreducing con-ditions.(C) HCs do not fold in the absence of BiPrelease. Cells expressing truncated HC andmutant BiP were incubated at 22�C with ATPfor the indicated times, and HCs were iso-lated as above.(D) Molecular crowding agents interfere withHC oxidation. COS cells expressing WT BiPand simplified HCs were lysed in the pres-ence of ATP and 30% Ficoll 70 at 22�C forthe indicated times and HCs were analyzedas above.(E) Unassembled full-length HCs aggregatewhen BiP is released in vitro with ATP. Meta-bolically labeled Ag8(8) cells were lysed in thepresence of NEM and apyrase (1), ATP (2 and4), or apyrase (3) and incubated at either 22�C(1–3) or 37�C (4). After 30 min, NEM was addedto the samples (2–4), and HCs were precipi-tated and analyzed under nonreducing condi-tions.(F) Unassembled full-length HCs remain in-completely oxidized in vivo. Ag8(8) cells werelabeled for 30 min and chased for indicatedtimes. One of the pulse labeled samples waslysed in the presence of ATP (0 � ATP) todemonstrate the mobility of completely oxi-dized HC. The remaining samples were lysedin the presence of NEM and apyrase. All wereelectrophoresed under nonreducing condi-tions.

15 min most of them had completed folding (Figure 1A). of hen lysozyme (van den Berg et al., 1999, 2000), oxida-tion of the released HCs was impeded, and they wereA portion of the released HCs formed aggregates, but

this amount did not appear to increase over time. When somewhat more prone to aggregation, a situation thatwas even more pronounced at 37�C (data not shown).BiP was released at 37�C, over half of the truncated HCs

were oxidized within 5 min of BiP release, but after 15 To ensure that the oxidation and aggregation of BiP-released HCs was not an artifact of using the simplifiedmin most of the released HCs no longer entered the gel

and instead formed large disulfide-linked aggregates at tagged-HCs or expressing them in COS cells, we exam-ined full-length � HCs from Ag8(8) cells. The mousethe top of the gel (Figure 1B). Oxidation of the CH1 do-

main appeared to be dependent on BiP release, because plasmacytoma cells were lysed in the presence of ATPand incubated at either 22�C or 37�C for 30 min (Figurecoexpression of a BiP ATPase mutant inhibited com-

plete oxidation of this domain (Figure 1C). Other in vitro 1E). In both cases, BiP was released; the HCs incubatedat 22�C migrated more rapidly, which is consistent withprotein folding studies showed that molecular crowding

agents, which are used to approximate the more con- the oxidation of the CH1 domain. Consistent with resultsobtained with the simplified HC, the full-length � HCscentrated cytosolic conditions that exist in vivo, can

either inhibit the release of nonnative proteins from the incubated at 37�C formed large aggregates that wereunable to enter the gel. The significance of this aggrega-GroEL chaperone (Martin and Hartl, 1997), decrease the

rate of folding (van den Berg et al., 2000), or lead to tion is not entirely clear, since these samples have un-dergone significant dilution and the concentration ofaggregation of nonnative intermediates (van den Berg

et al., 1999). When BiP release experiments were done ATP in the ER is not known. We did find the HCs aggre-gated in vitro even when ATP was added at concentra-in the presence of 30% Ficoll 70, we found that unlike

the rhodanese experiments where GroEL release was tions below the Km, which should have provided aslower, more regulated release of HCs (data not shown).inhibited in the presence of Ficoll 70 (Martin and Hartl,

1997), BiP was still released from the nonnative HC with When cell lysates were incubated at 22�C without ATPor NEM, the HCs remained bound to BiP and did notATP (Figure 1D). However, similar to the in vitro refolding

Regulated Release of BiP from Ig Heavy Chains107

oxidize or aggregate (Figure 1E, lane 3), again sug- not limiting. The mutant BiP was expressed at slightlygreater amounts than WT BiP (�2-fold greater), but HCsgesting that complete oxidation of the HCs requires BiP

release. To determine if in vivo a small portion of the bind both equally. This suggests that WT BiP binds witha slightly greater efficiency than mutant BiP, which couldfull-length HCs was continuously being released from

BiP, folding, and then aggregating over time, we per- cause our measured cycling rate to appear a bit slowerthan it actually is. However, since equivalent amountsformed pulse-chase experiments and examined the iso-

lated HCs under nonreducing conditions. The HCs were of each BiP species bind during a 10 min pulse, theeffect should be very modest.quite stable over the 6 hr course of this experiment and

showed no indication of folding or aggregating (Figure To ensure that the cycloheximide treatment used forthis experiment was not somehow interfering with BiP1F). Together, these data demonstrate that HCs re-

leased from BiP in the absence of LCs are susceptible cycling, we examined this in a different way. HCs wereisolated from COS cells coexpressing WT and mutantto aggregation even after complete oxidation, a situation

that would be disastrous inside the ER. There is no BiP and subjected to Western blotting to estimate theratio of WT to mutant BiP bound to HCs at equilibrium.evidence of either complete folding or aggregation oc-

curring in vivo. HCs coexpressed with only WT or mutant BiP servedas a reference for their mobility (Figure 2B, lanes 1 and2). In agreement with the pulse-chase data, HCs wereBiP Does Not Cycle on and off the Unfolded CH1bound to both WT and mutant BiP at equilibrium (FigureDomain of Unassembled HCs2B, lane 3) and at levels very similar to those observedIn vitro studies have revealed that chaperone-mediatedafter a 10 min pulse (Figure 2A, lane 4). To determinefolding occurs via multiple cycles of ATP-dependentthe relative amount of free BiP versus BiP bound to HCs,binding and release (Bukau and Horwich, 1998; Hartl,the supernatant of the HC immunoprecipitation was col-1996). The finding that unassembled HCs fold rapidlylected, and 1/20, 1/10, and 1/5 of the total volume wasand aggregate when BiP is released in vitro but remainimmunoprecipitated with anti-BiP (Figure 2B, lanes 4–6).incompletely oxidized and soluble in the ER for longAgain, we found that BiP was present in large excessperiods of time is hard to reconcile if BiP is repeatedlyof the HC and that cycling from WT to mutant BiP wascycling from heavy chains in vivo. This led us to measureoccurring very slowly if at all.the rate of BiP cycling from the unassembled HC in vivo.

It was possible that WT hamster BiP was not fullyWe coexpressed � HCs with both wild-type (WT) andactive in COS monkey cells and therefore unable toa mutant (G37) hamster BiP, which binds polypeptidecycle off the HCs in the above experiment. Hamster BiPsubstrates but is unable to release them (Hendershotmay be unable to use the host cell’s cofactors or mightet al., 1996). We reasoned that initially the nascent HCsbe expressed in excess of them. This was an unlikelywould bind in part to WT BiP and in part to the mutantexplanation, since WT hamster BiP expressed in COSBiP. Those HCs bound to WT BiP would cycle off andcells is fully active in the context of LC folding (Hender-rebind to either WT or mutant BiP, whereas those boundshot et al., 1996) and Ig assembly (this study). Alterna-to mutant BiP would remain trapped. Eventually the HCstively, it was possible that most cells expressing HCwould “shift” entirely to the mutant at a rate dependentwere only coexpressing one of the hamster BiP proteinson the kinetics of the cycling reaction. This experimentinstead of both. We could not rule this out, since the twois conceptually similar to the kinetic trap experimentsBiPs cannot be distinguished by immunofluorescenceused in vitro to demonstrate cycling of molecular chap-staining. To eliminate all of these possibilities, we co-erones (Weissman et al., 1994; Burston et al., 1996).transfected COS cells with only HC and mutant BiP, andCells coexpressing HCs, along with both WT and mu-the endogenous BiP served as our source of WT protein.tant BiP, were pulse labeled for 10 min and then chasedConditions were found in which transfected and endog-in complete media containing cycloheximide to preventenous BiPs were synthesized at similar rates and couldnew protein synthesis, which would complicate the in-be distinguished on SDS gels (Figure 3). Immunofluores-terpretation of the data. At various times during thecence staining revealed that �95% of cells expressingchase, cells were lysed and HCs were immunoprecipi-� HC were also expressing the G37 mutant BiP (datatated. Transfected WT BiP and the G37 mutant can benot shown), and all of these would also express thedistinguished on SDS gels (Figure 2A, lanes 2 and 3).endogenous wild-type monkey BiP. Cells were pulseThe endogenous BiP has an intermediate mobility butlabeled, chased in complete media containing cyclohex-is not readily detected in this experiment because itimide, and HCs were precipitated at the indicated times.gives a weaker signal compared to transfected BiPs.Mutant BiP was able to compete with endogenous BiPHCs bound to both WT and mutant BiP when each isto bind HC (Figure 3, lane 2). This time, very short inter-expressed alone (lanes 2 and 3) or together (lane 4).vals were examined to ensure that cycling had notHowever, even after chasing for 6 hr, the relative ratiosreached equilibrium before our experiment commenced.of the two BiP proteins remained very similar, revealingAgain, there was no evidence of a shift from WT endoge-there was no apparent shift of HC to the mutant BiPnous to mutant BiP during the 2 hr chase (lanes 2–8).(lanes 4–8). To ensure that this inability to shift from WTThis was not because the amount of BiP in the cellto mutant BiP was not because BiP protein was limiting,was limiting, nor was it due to the inability of wild-typewe measured the relative amount of HC and BiP in thehamster BiP to interact with an essential cofactor in thecells. HCs were precipitated from 9/10 of the lysate, andmonkey cells. Together, these data strongly suggestthe remaining 1/10 was precipitated with polyclonal anti-that unassembled HCs do not constantly cycle on andBiP serum. This revealed that BiP was present at �10-

fold excess of HCs (lanes 9 and 10) and therefore was off BiP in cells.

Immunity108

Figure 2. HCs Do Not Appear to Cycle from BiP In Vivo

(A) Unassembled HCs do not shift from WT to mutant hamster BiP over time. COS cells transfected with the indicated DNAs were pulselabeled for 10 min, chased for the indicated times (hr) in complete media containing 50 �g/ml cycloheximide, and lysed in buffer containingapyrase (lanes 4–8). Cells transfected with DNAs encoding HC and either WT or mutant BiP were used as controls for the mobility of the twoBiPs (lanes 2 and 3). In lane 1, cells were treated with cycloheximide just before labeling and lysed immediately after labeling to demonstratethat protein synthesis was efficiently blocked. HCs and associated BiP proteins were isolated with PA sepharose. Cell lysate from oneadditional dish was divided; 9/10 were incubated with PA sepharose beads to precipitate the HCs and 1/10 was reacted with anti-BiP (lanes9 and 10). Proteins were electrophoresed on an SDS 5%–12% gradient gel under reducing conditions, transferred to a nitrocellulose membrane,and exposed to X-ray film.(B) HCs are associated with equal amounts of wild-type and mutant transfected BiP under steady state conditions. Unlabeled COS cellstransfected with HC and either WT BiP (lane 1), G37 mutant BiP (lane 2), or both WT and mutant BiP DNAs (lane 3) were lysed in thepresence of apyrase, and HCs were precipitated with PA sepharose. Alternatively, 1/20, 1/10, or 1/5 of the triply transfected sample wasimmunoprecipitated with an anti-rodent BiP antibody (lanes 4–6). Samples were analyzed on an SDS 5%–12% gradient gel under reducingconditions, transferred, and blotted with the anti-rodent BiP antibody.

WT but Not Mutant BiP Is Released from Nascent folded LCs (Figure 4A). When the cells were washed andreincubated in the absence of DTT, WT BiP released,Proteins In Vivo

Because this was an unanticipated finding, we wished allowing a portion of the LCs to fold, but mutant BiPremained bound to a pool of LCs and blocked theirto identify conditions in which WT BiP was released

from a substrate protein in vivo but mutant BiP remained folding, consistent with previously reported data (Hen-dershot et al., 1996). To examine mutant VSV-G associa-bound. For these experiments, we chose a � LC that

binds BiP when it is synthesized in the presence of tion with BiP, cells were labeled at 39�C, the nonpermis-sive temperature, to block folding and maximize BiPDTT (Hendershot et al., 1996) and a mutant vesicular

stomatitis virus (VSV) G protein (ts045) that binds to binding. Again, both mutant and WT BiP bound to theunfolded VSV-G protein, but in this case the ratio ofBiP at the nonpermissive temperature (Machamer et al.,

1990). COS cells were transfected with DNAs encoding mutant to WT BiP was increased to 4:1 to obtain equiva-lent binding. When cells were chased at 39�C for 2 hr,either LCs or mutant VSV-G along with WT and mutant

BiP. In the case of LCs, cells were pulse labeled in the there was no apparent change in the ratio of WT tomutant BiP associated with VSV-G (Figure 4B). Whenpresence of DTT to prevent oxidation of the LC variable

region and provide a BiP binding site. Both WT and the cells were chased at 32�C, WT BiP was readily re-leased, but mutant BiP remained associated. Thus, wemutant BiP coprecipitated with the pulse labeled un-

Figure 3. Endogenous BiP Behaves asTransfected WT BiP in Its Interaction withHCs

COS cells transfected with DNAs for HC andmutant BiP were labeled for 10 min andchased for the indicated times in completemedia containing 50 �g/ml cycloheximide(lanes 2–8). Cells were lysed in buffer con-taining apyrase, and HCs were precipitated.One dish of cells was pulse labeled and im-munoprecipitated with a monoclonal anti-BiPantibody (lane 1). Proteins were separated onan SDS 10% polyacrylamide gel run until HCsnearly reached the bottom of the gel, trans-ferred to a membrane, and visualized by auto-radiography.

Regulated Release of BiP from Ig Heavy Chains109

Figure 4. WT but Not Mutant BiP Is Releasedfrom Substrates in COS Cells

(A) COS cells transfected with �LC along withmutant and WT BiP were pulse labeled for 10min in the presence of 10 mM DTT and eitherlysed immediately or chased for 1 hr in com-plete media without DTT. LCs were immuno-precipitated and examined on nonreducingSDS gels. Mobilities of partially folded (F1),and completely folded (Fox) are indicated.(B) Three dishes of cells expressing the VSV-Gmutant along with WT and mutant BiP werelabeled for 15 min at 39�C. One was lysedimmediately in the presence of apyrase, andthe remaining two were chased for 2 hr incomplete media at either 39�C or 32�C beforelysing. VSV-G was immunoprecipitated, andcomplexes were analyzed by reducing SDS-PAGE.

can readily observe WT BiP release from substrates pressing both WT and mutant BiP showed anintermediate effect on Ig assembly and secretion thatin COS cells, and under these conditions mutant BiP

remains bound and prevents the folding of substrates. was not quite as severe as with the mutant alone (Figure5A). Examination of similarly transfected cells under re-Hence, our inability to observe cycling of BiP is not

because COS cells do not provide an appropriate envi- ducing conditions revealed that WT BiP (both trans-fected and endogenous) was readily released from HCsronment for measuring BiP release, nor is it due to the

fact that WT BiP is behaving as a mutant or conversely during their assembly with LCs, allowing their secretionfrom the cells, whereas mutant BiP remained bound tothat mutant BiP is not defective in releasing from sub-

strates in these cells. the HC, preventing their assembly and secretion (Figure5B). Again, the assembly experiments were done in thepresence of ongoing protein synthesis, and, thus, newlyHC � LC Assembly in the Cell Is Efficientsynthesized HCs continue to contribute to the WT BiPTo begin to understand how LCs assemble with HCssignal in the chase sample. The kinetics of Ig assemblythat are stably bound to BiP, we measured the kineticsin these experiments are similar to those observed in aof Ig assembly in our system. COS cells expressing HCs,mouse plasmacytoma cell line that synthesizes HC andLCs, and WT and/or mutant hamster BiP were pulseLC (Bole et al., 1986) and indicates that the transfectedlabeled for 10 min and chased in complete media forWT hamster BiP is efficient in supporting Ig assembly.the indicated times. HCs were immunoprecipitated andFurthermore, these data reveal that Ig assembly in COSanalyzed on SDS gels run under nonreducing conditionscells is much more rapid than the release of BiP fromto monitor the formation of the completely assembledunassembled HCs appears to be (Figures 2 and 3) andH2L2 complex. As in plasmacytomas and myelomassuggests that LC may serve to trigger BiP release from(Scharff et al., 1970), assembly of the �1 HC proceededHC during Ig assembly in an ATP-dependent manner.through an H → H2 → H2L → H2L2 pathway in COS cells

and only H2L2 molecules were secreted (Figure 5A). TheHCs and LCs assemble readily when synthesized with ATPase Activity of BiP Is Required for Efficient

LC-Induced HC ReleaseWT BiP and are secreted, with the amount of H2L2 mole-cules in the cell reaching a maximum at 1.5 hr. Cyclohex- To separate any effects of BiP mutants on LC folding

from those on BiP release from HC and Ig assembly,imide was not included in the media for this experiment;therefore, new HCs continued to be introduced into the we used an in vitro vesicle fusion protocol (Watkins et

al., 1993). In their experiments, ER vesicles were derivedER and bind the labeled pool of BiP, which is why thesignal for BiP remained high. When Ig assembly was from plasmacytoma cells expressing only �1 HCs and

from cells expressing the complementary � LCs andmonitored in the presence of mutant BiP alone, we foundit was blocked at the H2 intermediate, and very little Ig incubated together in the presence of cytosol and ATP.

This allowed vesicle fusion, resulting in HC and LC as-was secreted from these cells. However, unlike Ag8(8)plasmacytoma cells, the H2 intermediate does not ap- sembly (Watkins et al., 1993). We tried a similar approach

using transfected COS cells. On average, only 15%–20%pear to be very stable in these cells and must be targetedfor intracellular degradation, as very little remains in the of the cells expressed the transfected cDNAs (data not

shown). Thus, only a fraction of the ER vesicles prepared3 hr time point. This is in contrast to the relative stabilityof the HCs observed in Figure 2A, when the cells were from these cells would express the Ig protein(s). To

determine if this was an obstacle for developing thechased in the presence of cycloheximide, and is consis-tent with data showing that cycloheximide inhibits deg- assay, vesicles containing unlabeled HCs were fused

with vesicles containing labeled LCs, both in the pres-radation of ER proteins (Amshoff et al., 1999). Cells ex-

Immunity110

blocked. This observation is consistent with the effectsof mutant BiP expression on Ig assembly and secretionin COS cells (Figure 5). Thus, it appears that the abilityof LCs to assemble stably with HCs and release BiPrequires the ATP binding and hydrolysis activity of BiP.

This same vesicle fusion system was used as an inde-pendent method of assaying for any possible cycling ofBiP from HCs. Cells were transfected with either HCsand WT BiP or HCs and mutant BiP, and ER vesicleswere isolated. A third set of cells transfected with onlyWT BiP was metabolically labeled, and ER vesicles wereprepared from these. Vesicles were fused for 15 min,and the fusion reaction was stopped by adding GTP�s.GTP is required for fusion but does not bind to BiP(Watkins et al., 1993). These vesicles were then incu-bated for the indicated times after GTP�s addition tomonitor changes in the amount of labeled BiP associ-ated with the HCs. Here, we anticipated that in the ab-sence of cycling, no labeled BiP would coprecipitatewith the HCs. Surprisingly, we found that small amountsof labeled WT BiP coprecipitated with the HCs from bothsets of vesicles (Figure 7A). The fact that the amount oflabeled BiP was similar for HCs bound to WT or mutantBiP and, more importantly, that this amount did notincrease significantly over time (Figures 7A–7C) againsupports the conclusion that BiP is not cycling fromnonnative HCs in the absence of LCs. The reason whysmall amounts of labeled BiP were detected in this ex-Figure 5. Ig Assembly Requires ATP-Dependent BiP Releaseperiment is still unclear, but it is possible that the fusion(A) Transfected HCs and LCs assemble efficiently in COS cells ex-procedure is somehow transiently opening up BiP bind-pressing WT but not mutant BiP. COS cells transfected with the

indicated DNAs were metabolically labeled for 10 min and chased ing sites on a small number of HCs. Amounts of HCs infor 0, 1.5, or 3 hr. Cell lysates from each time point and culture each of the vesicle preparations were determined bysupernatant from the 3 hr chase were incubated with Protein A Western blotting (Figure 7B) and used to quantitate thesepharose beads to precipitate HC, and samples were analyzed by relative amounts of labeled BiP bound to the HCs atnonreducing SDS-PAGE.

each time point (Figure 7C).(B) HCs are released from WT but not mutant BiP during assembly.COS cells transfected as indicated were pulse labeled for 10 minand chased for 0 or 2 hr. HCs were precipitated and examined on Discussion10% SDS gels under reducing conditions.

We demonstrate that the CH1 domain of unassembledHC is capable of oxidizing very quickly when BiP isence and in the absence of ATP (Figure 6A). Our ability

to coprecipitate labeled LCs with HCs only when ATP released in vitro with ATP (t1/2 � 5 min). However, whenrelease occurs at 37�C, the HCs rapidly form large aggre-was included in the reaction confirmed that vesicle fu-

sion had occurred and could be monitored (Figure 6B). gates. In contrast, in vivo free HCs enjoy a long half-lifebut neither fold completely nor aggregate. The combina-Examination of precipitated proteins under nonreducing

conditions revealed a 150 kDa protein, indicating that tion of data presented here suggests this is becauseBiP does not cycle from unassembled HC in vivo.the labeled LCs had assembled with HCs to form H2L2

complexes after vesicle fusion. However, most of the Hsp70s participate in protein folding by binding dena-tured or incompletely folded proteins to prevent aggre-labeled LCs did not assemble with the HCs (Figure 6B,

anti-� lanes), indicating that the fusion reaction is fairly gation and then releasing them to provide an opportunityfor folding to occur. This cycle is controlled by ATPinefficient under our experimental conditions.

ER vesicles were isolated from unlabeled cells coex- binding, hydrolysis, and nucleotide exchange. As bestdescribed for DnaK, the binding of ATP to the N-terminalpressing HCs and either WT or mutant BiP (G37). ER

vesicles were simultaneously prepared from cells ex- domain converts the hsp70 protein into a fast peptidebinding and releasing, low-affinity form, whereas hydro-pressing LCs that had been metabolically labeled. The

vesicles were mixed and allowed to fuse, and the assem- lysis to ADP stabilizes substrate binding by convertinghsp70 to a slow peptide binding and releasing, high-bly of HC and LC was determined by immunoprecipitat-

ing HCs and monitoring the association of labeled LCs. affinity form (Schmid et al., 1994). A second round of ATPbinding then leads to the release of hsp70 from the boundHC and LC assembly was significantly lower when G37

mutant BiP was present (Figure 6C), even though similar peptide. Typically, productive chaperone-assisted pro-tein folding is thought to occur through multiple cyclesamounts of HCs were present in the different prepara-

tions of ER vesicles (Figure 6D). The transfected cells of ATP-mediated binding and release (Szabo et al., 1994;Lilie and Buchner, 1995). In general, this is because onlyalso contain endogenous WT BiP, and, therefore, the

assembly of HC and LC should not be completely a fraction of the unfolded proteins reach the native state

Regulated Release of BiP from Ig Heavy Chains111

Figure 6. ER Vesicles from COS-TransfectedCells Can Fuse In Vitro and Allow HCs andLCs to Assemble

(A and B) ER vesicles were prepared fromunlabeled cells expressing � HCs as well asfrom metabolically labeled cells expressing �

LCs as indicated in the schematic (A). Thetwo preparations were mixed and incubatedfor 90 min in the presence (f-fused) or ab-sence (m-mock) of ATP. Vesicles were thenlysed, HCs or LCs were immunoprecipitated,and proteins were electrophoresed on SDS-polyacrylamide gels under reducing and non-reducing conditions (B).(C) BiP’s ATPase activity is required for LC-induced HC release. ER vesicles preparedfrom unlabeled cells expressing HCs and ei-ther WT or mutant BiP (G37) were fused withvesicles prepared from labeled cells express-ing LCs. After 90 min, vesicles were lysed,HCs were immunoprecipitated, and copreci-pitating LCs were detected by reducing SDS-PAGE.(D) To detect HC in the unlabeled ER vesicles,proteins were electrophoresed on SDS re-ducing gels, transferred to a membrane, andblotted with a goat anti-human IgG antibody.

in vitro after a single round of release from the chaper- ATPase mutant that binds to substrate proteins but doesnot release them (Wei et al., 1995). We expected thatone. The abortive species are rebound by the chaper-

one, providing these molecules with another chance WT BiP would continuously bind and release the HCmolecules, whereas the mutants would act as a kineticto fold. Cycling is thought to be central to chaperone-

assisted protein folding. trap, resulting in the HCs gradually shifting to an associ-ation with the BiP mutants. Instead, our data suggestIn our experiments, we coexpressed HCs with a BiP

Figure 7. Labeled BiP Combines Poorly withBoth WT BiP-Associated and Mutant BiP-Associated HCs

(A) ER vesicles were isolated from COS cellscoexpressing either HCs with WT BiP or withmutant BiP and incubated with vesicles con-taining labeled WT BiP in the presence of nonucleotide (lanes 1 and 7), both nucleotides(lanes 2 and 8), or only ATP for the first 15min (lanes 3–6 and 9–12). The fusion reactionswere then stopped by adding GTP�s (*), andthe vesicles were further incubated for theindicated times before lysis and HC immuno-precipitation.(B) To determine the amount of HC presentin each sample, an aliquot was analyzed byWestern blotting with anti-human � anti-serum.(C) The signals for HC and labeled BiP werequantitated by densiotometry, and the rela-tive amount of BiP bound to HC was deter-mined with the value at 150 min set to 100.

Immunity112

that BiP binds stably to the CH1 domain without undergo- a complex with the HCs and that some component ofthis complex interferes with ATP-mediated release ofing cycles of binding and release and argue that in vivo

cycling (or at least the rate of cycling) is not implicit to BiP (L. Meunier, et al., submitted).Alternatively, LCs could promote binding of a nucleo-hsp70s but may be controlled by cofactors whose ac-

cess to hsp70 is regulated by the substrate. Thus, it tide exchanger to the complex. In bacteria and mito-chondria, nucleotide exchange is facilitated by the ac-is conceivable that the cycling (or release) rate of BiP

represents a point of control to accommodate nascent tion of a GrpE protein (Liberek et al., 1991; Bolliger et al.,1994), and in mammalian cytosol, the hsp70-organizingER proteins with various folding rates and propensities

to aggregate during their biosynthesis. Despite the protein (Hop) accelerates nucleotide exchange in vitro(Gross and Hessefort, 1996; Chen and Smith, 1998).seemingly stable interaction of HCs with BiP, their as-

sembly with LCs is quite rapid and efficient and requires Addition of large amounts of ATP to the BiP-HC complexin vitro could be sufficient to drive nucleotide exchangeBiP’s ATPase activity (Bole et al., 1986; Figure 5). We

propose that binding of LCs to HCs triggers the release and HC release. Again, no ER homologs of Hop havebeen identified in mammalian cells. Finally, it is possibleof BiP. However, it is not clear from the present study

whether the stable binding of BiP to HCs is the exception that binding of ATP to BiP does not result in HC releasein vivo because the BiP/ATP-HC complex is sufficientlyor the rule. When the VSV-G temperature-sensitive mu-

tant was maintained at the nonpermissive temperature stable to allow ATP hydrolysis to occur before dissocia-tion. This stability could explain why BiP was first identi-for an additional 2 hr, there was no evidence that BiP

was cycling from it either (Figure 4). This might imply fied with Ig HCs and why, unlike many other substrates,there is no need to inhibit ATP upon cell lysis to maintainthat mutant proteins or proteins that cannot fold do not

cycle from BiP but that instead the interaction of another this interaction (Haas and Wabl, 1983; Bole et al., 1986).LC binding would alter the affinity of HC for BiP. At-region of the protein (or a subunit) with amino acids near

the BiP binding site in some way triggers the release of tempts to demonstrate the existence of a BiP-HC-LCternary complex by coprecipitation after treatment withBiP. Further experiments are needed to clarify this point.

However, it should be noted that unassembled heavy cross-linking agents were unsuccessful, suggesting thatthis complex must be extremely transient if it existschains are not mutant proteins, are able to fold, and

exist as a normal stage of B cell development. Interest- (data not shown).One thing that is implicit in all these models is thatingly, when the GroEL kinetic trap experiments were

performed in the presence of molecular crowding the unfolded substrate itself plays a role in the hsp70ATPase cycle via its ability (or inability) to interact withagents, the release of nonnative intermediates of rho-

danese was inhibited (Martin and Hartl, 1997), raising the other cofactors. The idea that substrates have differentaffinities or duration of binding to chaperones has beenpossibility that repeated cycling may not be a universal

feature of chaperone:substrate complexes. demonstrated before (Ewalt et al., 1997). A correlationbetween the stability of BiP binding and the kinetics ofThe stability of the BiP-HC complex observed here

suggests that BiP is maintained in the ADP-bound state degradation has been reported for a number of proteins,including transport-incompetent Ig LCs (Knittler et al.,when bound to HC in vivo. This could be achieved in

one of several ways. First, BiP release from HC could 1995; Skowronek et al., 1998), unassembled Na,K-AT-Pase � subunits (Beggah et al., 1996), and an assembly-be prevented by a regulatory protein(s) that is displaced

by LC. We would argue that the binding of this inhibitor defective form of the vacuolar storage glycoproteinphaseolin (Pedrazzini et al., 1997). Free � HCs are verywas not stable to detergent lysis even in the presence

of molecular crowding agents, thus explaining the ready stable, with a half-life of about 8 hr. It is plausible thatthe stable interaction of BiP with the CH1 domain deter-release of BiP from HCs in vitro. Several cytosolic pro-

teins have been identified that regulate the ATPase cycle mines the remarkable stability of the HC molecule. It istempting to speculate that the immune system has takenof hsp70. DnaJ proteins interact with both the unfolded

substrate protein and hsp70s and stimulate the rate of advantage of this variability in cycling (or release) ratesto design the CH1 domain to be at the extreme end ofATP hydrolysis, switching the hsp70 protein into the

ADP bound form (Liberek et al., 1991; Szabo et al., 1994; the spectrum (i.e., no cycling). This would ensure thatthe HC does not leave the ER without assembling withMinami et al., 1996). A DnaJ-related protein could be

part of the BiP-HC complex, which maintains BiP in the LC or that it does not aggregate before it can do so. Ifunassembled HCs cycled off BiP in pre-B cells, it isADP bound form and thus keeps it tightly bound to HC.

Recently, a mammalian ER-targeted DnaJ protein was possible that they would be transported without bindingto surrogate light chains. This could circumvent the roleidentified (Brightman et al., 1995) that can stimulate the

ATPase activity of BiP (Chevalier et al., 2000), but no of the surrogate light chain in controlling HC transportand of the pre-B receptor in signaling the end of furtherdata are available to suggest that it is part of the BiP-

HC complex. The cytosolic hsp70-interacting protein HC gene rearrangement and the beginning of LC re-arrangement (Shaffer and Schlissel, 1997; Melchers et(Hip) inhibits the release of ADP from hsp70 (Hohfeld

et al., 1995), and Bag1 stabilizes the hsp70-substrate al., 1993). Clearly, the secretion of incompletely assem-bled Ig molecules by plasma cells would impact on thecomplex by uncoupling ATP binding from substrate re-

lease (Hohfeld and Jentsch, 1997; Bimston et al., 1998). effectiveness of the immune response. Together, thesetwo detrimental outcomes could provide the drivingEither of these could be envisioned to stabilize BiP-HC

binding if they were part of the complex. Although no force for producing a CH1 domain that did not cycle fromBiP in the absence of LC (or surrogate LC) binding.ER homologs of Hip or Bag1 have been identified, we

recently isolated HCs by nondetergent methods and In conclusion, we demonstrate here that unassembledIg HCs remain incompletely folded even though they aredemonstrated that a number of other ER proteins form

Regulated Release of BiP from Ig Heavy Chains113

were washed once in 20 volumes of 25 mM HEPES-KOH, 125 mMquite long lived. If they are released from BiP in vitroKCl (pH 7.2) (25/125), resuspended in 4 volumes of the same buffer,with ATP, the CH1 domain oxidizes rapidly, and the HCsand homogenized in a dounce homogenizer. Crude homogenatesaggregate. Our data suggest that BiP does not cyclewere centrifuged for 10 min at 500 g, and supernatants containing

from the CH1 domain of unassembled HC in vivo, thus ER vesicles and cytosol were collected. Fusion was performed byensuring that the HC remains in the ER and preserving mixing the appropriate ER vesicle/cytosol preparations and incubat-

ing them at 32�C in the presence of 2 mM ATP, 4.6 mM phosphocre-it in a soluble form capable of assembling with LC. Weatine, 5 U/ml creatine phosphokinase (all from Sigma), and 2.5 mMsuggest that it is the LC, via interactions with the HC ormagnesium acetate. For Ig assembly experiments, vesicles werewith BiP cofactors, that participates in BiP release.fused for 90 min and centrifuged at 16,000 g for 10 min. Thevesicle pellets were washed once with 25/125 buffer and then lysedExperimental Proceduresfor immunoprecipitation. For the BiP cycling experiment, vesicleswere incubated at 32�C for 15 min, and then further fusion wasCell Linesstopped by adding 100 �M GTP�s. The vesicles were incubated forAg8(8) cells are derived from the murine IgG1 plasmacytoid line P3Kthe additional indicated times at 32�C before treating them as above.and express � HCs but no longer synthesize LCs (Bole et al., 1986).

COS cells are derived from the African green monkey kidney fibro-Western Blotsblast line CV1 and express the SV40 large T antigen (Gluzman, 1981).Samples were electrophoresed on SDS-polyacrylamide gels andtransferred to nitrocellulose membranes. HCs were probed with aCOS Cell Transfectiongoat anti-human IgG followed by a rabbit anti-goat IgG antiserum.DNAs encoding the �I LC (Bernard et al., 1978), a chimeric mouse:hu-Hamster BiP was probed with a polyclonal rabbit anti-rodent BiPman Ig � HC (Liu et al., 1987); a truncated Ig HC comprised of theantiserum (Hendershot et al., 1995) followed by a goat anti-rabbitVH and CH1 domains plus a C-terminal 9 residue epitope from theIgG antiserum. Membranes were subsequently incubated withinfluenza hemagglutinin (HA-�CH1) (Lee et al., 1999); ts045, a temper-horseradish peroxidase-conjugated protein A (EY Laboratories, Sanature-sensitive mutant of VSV-G protein (Gallione and Rose, 1985),Mateo, CA) and developed with the ECL reagent (Amersham, Arling-kindly provided by Dr. John Rose, Yale University; and WT andton Heights, IL) according to the manufacturer’s specifications.ATPase mutants of hamster BiP (Wei et al., 1995) have been de-

scribed. COS monkey fibroblasts were transfected with DNAs usingAcknowledgmentsthe DEAE-dextran procedure and analyzed 24–40 hr post trans-

fection.We wish to thank Mr. Ron Ryall for assistance in antibody preserva-tion. This work was supported by National Institutes of Health GrantMetabolic Labeling and ImmunoprecipitationGM54068, the Cancer Center CORE Grant CA21765, and the Ameri-COS transfectants were incubated for 45 min in cysteine- and methi-can Lebanese Syrian Associated Charities of St. Jude Children’sonine-free media and biosynthetically labeled with [35S] methionineResearch Hospital.and cysteine (Translabel, ICN, Irvine, CA) as detailed in the legends.

Cells were lysed, and 10 U apyrase (Sigma, St Louis, MO) was addedReceived April 27, 2000; revised May 17, 2001.when indicated. For those experiments in which samples were ana-

lyzed under nonreducing conditions, N-ethylmaleimide (NEM) wasReferencesincluded in the lysing buffer. Lysates were incubated with either a

rodent-specific rabbit anti-BiP polyclonal serum (Hendershot et al.,Amshoff, C., Jack, H.M., and Haas, I.G. (1999). Cycloheximide, a1995), a monoclonal anti-BiP antibody (Bole et al., 1986), a goat anti-new tool to dissect specific steps in ER-associated degradation ofmouse � LC-specific antibody (Southern Biotechnology Associates,different substrates. J. Biol. Chem. 380, 669–677.Birmingham, AL), a monoclonal antibody specific for the HA epitope

(Lee et al., 1999), or a rabbit polyclonal antiserum to VSV-G protein Beggah, A., Mathews, P., Beguin, P., and Geering, K. (1996). Degra-kindly provided by Dr. Michael Whitt, University of Tennessee, Mem- dation and endoplasmic reticulum retention of unassembled alpha-phis, TN, followed by binding to Protein A Sepharose beads (Sigma). and beta-subunits of Na,K-ATPase correlate with interaction of BiP.Mouse � HCs were directly precipitated with protein A Sepharose J. Biol. Chem. 271, 20895–20902.beads. Immunoprecipitated proteins were electrophoresed on SDS- Bernard, O., Hozumi, N., and Tonegawa, S. (1978). Sequences ofpolyacrylamide gels and visualized by autoradiography using the mouse immunoglobulin light chain genes before and after somaticamplify reagent (Amersham, Arlington Heights, IL) or transferred to changes. Cell 15, 1133–1144.nitrocellulose (NC) membranes and exposed to X-ray film.

Bimston, D., Song, J., Winchester, D., Takayama, S., Reed, J.C.,and Morimoto, R.I. (1998). BAG-1, a negative regulator of Hsp70In Vitro ATP-Mediated BiP Releasechaperone activity, uncouples nucleotide hydrolysis from substrateLabeled cells were lysed in the presence of apyrase and NEM whererelease. EMBO J. 17, 6871–6878.indicated to prevent BiP release and post lysis oxidation. WhenBlond-Elguindi, S., Cwirla, S.E., Dower, W.J., Lipshutz, R.J., Sprang,BiP was released with ATP, cells were scraped from the plates,S.R., Sambrook, J.F., and Gething, M.J. (1993). Affinity panning of atransferred to an Eppendorf tube, pelleted, and lysed at the indicatedlibrary of peptides displayed on bacteriophages reveals the bindingtemperature in the presence of 2 mM Mg-ATP and 25 mM KCl. Atspecificity of BiP. Cell 75, 717–728.various times after ATP addition, further oxidation of the CH1 domain

was prevented by adding NEM (20 mM). For molecular crowding Bole, D.G., Hendershot, L.M., and Kearney, J.F. (1986). Posttransla-experiments, the ATP-lysing buffer was made to 20% Dextran 70 tional association of immunoglobulin heavy chain binding proteinor 30% Ficoll 70 (Sigma). After stopping further oxidation at the with nascent heavy chains in nonsecreting and secreting hybrido-indicated time with 20 mM NEM, the sample was incubated on ice mas. J. Cell. Biol. 102, 1558–1566.for 10 min and then diluted with an equal volume of ATP-lysing Bolliger, L., Deloche, O., Glick, B.S., Georgopoulos, C., Jeno, P.,buffer containing apyrase and NEM to facilitate clarification of the Kronidou, N., Horst, M., Morishima, N., and Schatz, G. (1994). Alysate and immunoprecipitation. mitochondrial homolog of bacterial GrpE interacts with mitochon-

drial hsp70 and is essential for viability. EMBO J. 13, 1998–2006.Preparation and Fusion of ER Vesicles

Brightman, S.E., Blatch, G.L., and Zetter, B.R. (1995). Isolation of aCOS cells grown in 100 mm dishes were transfected as described.

mouse cDNA encoding MTJ1, a new murine member of the DnaJFor each fusion, only one set of vesicles was pulse labeled for 30–45

family of proteins. Gene 153, 249–254.min with 150 �Ci [35S] Translabel per dish. ER vesicles were then

Bukau, B., and Horwich, A.L. (1998). The Hsp70 and Hsp60 chaper-prepared as described previously (Watkins et al., 1993) with theone machines. Cell 92, 351–366.following modifications. The cells were scraped 40 hr after transfec-

tion in 1 ml PBS and centrifuged 5 min at 2000 rpm. Cell pellets Burston, S.G., Weissman, J.S., Farr, G.W., Fenton, W.A., and Hor-

Immunity114

wich, A.L. (1996). Release of both native and non-native proteins structured folding intermediate: iterative binding cycles do not in-volve unfolding. Proc. Natl. Acad. Sci. USA 92, 8100–8104.from a cis-only GroEL ternary complex. Nature 383, 96–99.

Liu, A.Y., Mack, P.W., Champion, C.I., and Robinson, R.R. (1987).Chen, S., and Smith, D.F. (1998). Hop as an adaptor in the heatExpression of mouse::human immunoglobulin heavy-chain cDNA inshock protein 70 (Hsp70) and hsp90 chaperone machinery. J. Biol.lymphoid cells. Gene 54, 33–40.Chem. 273, 35194–35200.

Machamer, C.E., Doms, R.W., Bole, D.G., Helenius, A., and Rose,Chevalier, M., Rhee, H., Elguindi, E.C., and Blond, S.Y. (2000). Inter-J.K. (1990). Heavy chain binding protein recognizes incompletelyaction of murine BiP/GRP78 with the DnaJ homologue MTJ1. J.disulfide-bonded forms of vesicular stomatitis virus G protein. J.Biol. Chem. 275, 19620–19627.Biol. Chem. 265, 6879–6883.Ewalt, K.L., Hendrick, J.P., Houry, W.A., and Hartl, F.U. (1997). In vivoMaizels, N. (1995). Somatic hypermutation: how many mechanismsobservation of polypeptide flux through the bacterial chaperonindiversify V region sequences? Cell 83, 9–12.system. Cell 90, 491–500.Martin, J., and Hartl, F.U. (1997). The effect of macromolecularGallione, C.J., and Rose, J.K. (1985). A single amino acid substitutioncrowding on chaperonin-mediated protein folding. Proc. Natl. Acad.in a hydrophobic domain causes temperature-sensitive cell-surfaceSci. USA 94, 1107–1112.transport of a mutant viral glycoprotein. J. Virol. 54, 374–382.Melchers, F., Karasuyama, H., Haasner, D., Bauer, S., Kudo, A.,Gellert, M. (1992). V(D)J recombination gets a break. Trends Genet.Sakaguchi, N., Jameson, B., and Rolink, A. (1993). The surrogate8, 408–412.light chain in B-cell development. Immunol. Today. 14, 60–68.Gluzman, Y. (1981). SV40-transformed simian cells support the repli-Minami, Y., Hohfeld, J., Ohtsuka, K., and Hartl, F.U. (1996). Regula-cation of early SV40 mutants. Cell 23, 175–182.tion of the heat-shock protein 70 reaction cycle by the mammalianGoto, Y., Azuma, T., and Hamaguchi, K. (1979). Refolding of theDnaJ homolog, Hsp40. J. Biol. Chem. 271, 19617–19624.immunoglobulin light chain. J. Biochem. 85, 1427–1438.Pedrazzini, E., Giovinazzo, G., Bielli, A., de Virgilio, M., Frigerio, L.,

Gross, M., and Hessefort, S. (1996). Purification and characterizationPesca, M., Faoro, F., Bollini, R., Ceriotti, A., and Vitale, A. (1997).

of a 66-kDa protein from rabbit reticulocyte lysate which promotesProtein quality control along the route to the plant vacuole. Plant

the recycling of hsp 70. J. Biol. Chem. 271, 16833–16841.Cell 9, 1869–1880.

Haas, I.G., and Wabl, M. (1983). Immunoglobulin heavy chain bindingScharff, M.D., Bargellesi, A., Baumal, R., Buxbaum, J., Coffino, P.,

protein. Nature 306, 387–389.and Laskov, R. (1970). Variations in the synthesis and assembly of

Hammond, C., and Helenius, A. (1994). Quality control in the secre- immunoglobulins by mouse myeloma cells: a genetic and biochemi-tory pathway: retention of a misfolded viral membrane glycoprotein cal analysis. J. Cell. Physiol. 76, 331–348.involves cycling between the ER, intermediate compartment, and Schmid, D., Baici, A., Gehring, H., and Christen, P. (1994). KineticsGolgi apparatus. J. Cell. Biol. 126, 41–52. of molecular chaperone action. Science 263, 971–973.Hartl, F.U. (1996). Molecular chaperones in cellular protein folding. Shaffer, A.L., and Schlissel, M.S. (1997). A truncated heavy chainNature 381, 571–579. protein relieves the requirement for surrogate light chains in earlyHendershot, L., Wei, J., Gaut, J., Melnick, J., Aviel, S., and Argon, B cell development. J. Immunol. 159, 1265–1275.Y. (1996). Inhibition of immunoglobulin folding and secretion by Skowronek, M.H., Hendershot, L.M., and Haas, I.G. (1998). The vari-dominant negative BiP ATPase mutants. Proc. Natl. Acad. Sci. USA able domain of non-assembled Ig light chains determines both their93, 5269–5274. half-life and binding to BiP. Proc. Natl. Acad. Sci. USA 95, 1574–Hendershot, L., Bole, D., Kohler, G., and Kearney, J.F. (1987). Assem- 1578.bly and secretion of heavy chains that do not associate posttransla- Szabo, A., Langer, T., Schroder, H., Flanagan, J., Bukau, B., andtionally with immunoglobulin heavy chain-binding protein. J. Cell. Hartl, F.U. (1994). The ATP hydrolysis-dependent reaction cycle ofBiol. 104, 761–767. the Escherichia coli Hsp70 system DnaK, DnaJ, and GrpE. Proc.Hendershot, L.M., Wei, J.-Y., Gaut, J.R., Lawson, B., Freiden, P.J., Natl. Acad. Sci. USA 91, 10345–10349.and Murti, K.G. (1995). In vivo expression of mammalian BiP ATPase van den Berg, B., Ellis, R.J., and Dobson, C.M. (1999). Effects ofmutants causes disruption of the endoplasmic reticulum. Mol. Biol. macromolecular crowding on protein folding and aggregation.Cell. 6, 283–296. EMBO J. 18, 6927–6933.Hohfeld, J., and Jentsch, S. (1997). GrpE-like regulation of the hsc70 van den Berg, B., Wain, R., Dobson, C.M., and Ellis, R.J. (2000).chaperone by the anti-apoptotic protein BAG-1. EMBO J. 16, 6209– Macromolecular crowding perturbs protein refolding kinetics: impli-6216. cations for folding inside the cell. EMBO J. 19, 3870–3875.Hohfeld, J., Minami, Y., and Hartl, F.U. (1995). Hip, a novel cochaper- Watkins, J.D., Hermanowski, A.L., and Balch, W.E. (1993). Oligomer-one involved in the eukaryotic Hsc70/Hsp40 reaction cycle. Cell 83, ization of immunoglobulin G heavy and light chains in vitro. A cell-589–598. free assay to study the assembly of the endoplasmic reticulum. J.

Biol. Chem. 268, 5182–5192.Kaloff, C.R., and Haas, I.G. (1995). Coordination of immunoglobulinchain folding and immunoglobulin chain assembly is essential for Wei, J.-Y., and Hendershot, L.M. (1995). Characterization of thethe formation of functional IgG. Immunity 2, 629–637. nucleotide binding properties and ATPase activity of recombinant

hamster BiP purified from bacteria. J. Biol. Chem. 270, 26670–26677.Knarr, G., Gething, M.J., Modrow, S., and Buchner, J. (1995). BiPbinding sequences in antibodies. J. Biol. Chem. 270, 27589–27594. Wei, J.-Y., Gaut, J.R., and Hendershot, L.M. (1995). In vitro dissocia-

tion of BiP:peptide complexes requires a conformational change inKnittler, M.R., Dirks, S., and Haas, I.G. (1995). Molecular chaperonesBiP after ATP binding but does not require ATP hydrolysis. J. Biol.involved in protein degradation in the endoplasmic reticulum: Quan-Chem. 270, 26677–26682.titative interaction of the heat shock cognate protein BiP with par-

tially folded immunoglobulin light chains that are degraded in the Weissman, J.S., Kashi, Y., Fenton, W.A., and Horwich, A.L. (1994).endoplasmic reticulum. Proc. Natl. Acad. Sci. USA 92, 1764–1768. GroEL-mediated protein folding proceeds by multiple rounds of

binding and release of nonnative forms. Cell 78, 693–702.Lee, Y.-K., Brewer, J.W., Hellman, R., and Hendershot, L.M. (1999).BiP and Ig light chain cooperate to control the folding of heavy Zhu, X., Zhao, X., Burkholder, W.G., Gragerov, A., Ogata, C.M., Got-chain and ensure the fidelity of immunoglobulin assembly. Mol. Biol. tesman, M.E., and Hendrickson, W.A. (1996). Structural analysis ofCell. 10, 2209–2219. substrate binding by the molecular chaperone DnaK. Science 272,

1606–1614.Liberek, K., Marszalek, J., Ang, D., Georgopoulos, C., and Zylicz,M. (1991). Escherichia coli DnaJ and GrpE heat shock proteins jointlystimulate ATPase activity of DnaK. Proc. Natl. Acad. Sci. USA 88,2874–2878.

Lilie, H., and Buchner, J. (1995). Interaction of GroEL with a highly