Within the Variable Domain Binding Protein and Ig Light Chains to Sites

Mapping the Major Interaction Between Binding Protein and Ig Light Chains to Sites Within the Variable Domain

This information is current as David P. Davis, Ritu Khurana, Stephen Meredith, Fred J. of September 29, 2021. Stevens and Yair Argon J Immunol 1999; 163:3842-3850; ; http://www.jimmunol.org/content/163/7/3842 Downloaded from

References This article cites 49 articles, 26 of which you can access for free at: http://www.jimmunol.org/content/163/7/3842.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 29, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1999 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Mapping the Major Interaction Between Binding Protein and Ig Light Chains to Sites Within the Variable Domain1

David P. Davis,2*† Ritu Khurana,2,3*† Stephen Meredith,* Fred J. Stevens,‡ and Yair Argon4*†

Newly synthesized Ig chains are known to interact in vivo with the binding protein (BiP), a major peptide-binding chaperone in the endoplasmic reticulum. The predominant interactions between the light chain and BiP are observed early in the folding pathway, when the light chain is either completely reduced, or has only one disulfide bond. In this study, we describe the in vitro reconstitution of BiP binding to the variable domain of light chains (VL). Binding of deliberately unfolded VL was dramatically more avid than that of folded VL, mimicking the interaction in vivo. Furthermore, VL binding was inhibited by addition of ATP, was competed with excess unlabeled VL, and was demonstrated with several different VL proteins. Using this assay, peptides ␤ derived from the VL sequence were tested experimentally for their ability to bind BiP. Four peptides from both sheets of VL were ␤ shown to bind BiP specifically, two with significantly higher affinity. As few as these two peptide sites, one from each sheet of Downloaded from

VL, are sufﬁcient to explain the association of BiP with the entire light chain. These results suggest how BiP directs the folding of Ig in vivo and how it may be used in shaping the B cell repertoire. The Journal of Immunology, 1999, 163: 3842–3850.

roper folding of their subunits is an essential step in the leasing form of BiP, the CL domain of LC is still capable of ox-

biosynthesis of Ag receptors; misfolded subunits are gen- idizing, but the VL domain remains unfolded (4). Many mutant erally incapable of assembly or subsequent expression on LC, mostly with substitutions in the V domain, bind BiP more P L http://www.jimmunol.org/ the cell surface. While this is true of all secreted and membrane- avidly than wild-type LC (5–8). Several physiological roles for bound proteins, the control of proper folding is even more crucial this interaction can be envisioned: promotion of proper folding, for B and TCRs, since the variable domains of these receptors are coordination of assembly (9), or targeting to degradation (6). Un- a product of several genetic mechanisms that diversify the receptor derstanding how BiP performs these roles, however, requires that repertoire. Given the mechanisms of somatic recombination and the binding between this chaperone and LC be defined. hypermutation, there is heightened likelihood that the resultant Ag Like other members of the hsp70 family, BiP binds incom- receptors will include amino acids that are detrimental to proper pletely folded proteins by recognizing certain peptides within their folding. Thus, it is not surprising that B and TCRs require inter- sequences (10–12). Such binding facilitates the process of protein actions with chaperones early in their biosynthesis. folding, and, in some cases (mitochondrial hsp70 and yeast BiP), by guest on September 29, 2021 5 The chaperone BiP was first identified based on binding to Ig the translocation of polypeptides across membranes (13). The heavy chain (1, 2) and shown to be the endoplasmic reticulum hsp70 proteins, including BiP, share a common structure consist- (ER) member of the hsp70 class of stress proteins. BiP was also ing of a C-terminal 27-kDa substrate-binding domain and an N- shown to interact with the light chain (LC) of Ig. We have previ- terminal 44-kDa nucleotide-binding domain (14). The substrate- ously demonstrated that wild-type LC binds BiP transiently during binding domain consists of two subdomains, one with a shallow its folding in the ER, and that the avidity of binding decreases as groove for transient peptide binding (15). Although the overall the two domains of LC fold (3). When coexpressed with a nonre- conservation of sequence is only about 67%, the residues that contact a bound peptide are well conserved throughout the hsp70 family (15). A second subdomain forms an unusually long ␣-helical segment that encloses the substrate-binding region like a latch and *Department of Pathology and †Committee on Immunology, University of Chicago, Chicago, IL 60637; and ‡Center for Mechanistic Biology and Biotechnology, Ar- may act as a regulatory element (15). The nucleotide-binding do- gonne National Laboratory, Argonne, IL 60439 main contains the binding core (16) and accommodates the Received for publication April 26, 1999. Accepted for publication July 26, 1999. ATP-Mg complex (14, 17, 18). Binding of ATP transmits a con- The costs of publication of this article were defrayed in part by the payment of page formational change (14, 19–22) to the peptide-binding site, de- charges. This article must therefore be hereby marked advertisement in accordance creasing the binding affinity and releasing the peptide (12). Hy- with 18 U.S.C. Section 1734 solely to indicate this fact. drolysis of ATP is thought to convert the protein back to the high 1 D.P.D. was supported by the Baron Fellowship of the Unicorn Foundation. This work was supported in part by the U.S. Department of Energy, Office of Health and affinity-binding state. Individual point mutants of BiP identify the Environmental Research, under Contract W-31-109-ENG-38 (F.J.S.), and by U.S. residues that are involved in each of the separate activities: ATP Public Health Service Grants DK43757 (F.J.S.) and AI30178 (Y.A.). binding, conformational change, and hydrolysis (21). The func- 2 D.P.D. and R.K. contributed equally to this paper. tional linkage of all three is demonstrated by the common conse- 3 Current address: Department of Chemistry and Biochemistry, University of Cali- quence of expressing any of these ATP-binding domain mutants in fornia, 1156 High Street, Santa Cruz, CA 95064. cells. When each of them is expressed in a cell synthesizing LC, 4 Address correspondence and reprint requests to Dr. Yair Argon, Department of Pathology, University of Chicago, 5841 South Maryland Avenue, MC 1089, Chicago, each mutant acts in dominant-negative manner, trapping the LC at IL 60637. E-mail address: [email protected] an early stage of folding, and therefore reducing its secretion (4). 5 Abbreviations used in this paper: BiP, binding protein; ANS, 1,1Ј-bi(4-anilino) Remarkably, folding is arrested at an intermediate stage, with the nahpthalene-5,5Ј-disulfonic acid; ER, endoplasmic reticulum; Gdn-HCl, guanidine hydrochloride; 125I-FYQLALT, 125I-labeled FYQLALT; LC, light chain; SFX, suc- CL domain still oxidized, but with the VL domain trapped in a cinimidyl ester. reduced state (4).

A number of in vitro studies to deﬁne structural features of Peptides hsp70-binding peptides employed either chemically synthesized All peptides were synthesized by the University of Chicago core peptide peptides (12, 24) or afﬁnity panning of a peptide library displayed facility. Stock solutions were prepared in 100% DMSO and stored at on bacteriophages (10). The binding peptides consist of either al- Ϫ80°C. Most of the peptides used were water soluble, except for FTLTISS ternate aromatic or long-chain aliphatic amino acids (10), or sev- and TDFTLTI, which were only water soluble at concentrations below 600 ␮ eral successive hydrophobic residues (24). The optimal peptide is M. Concentrations of peptide solutions were determined with the BCA assay (Pierce, Rockford, IL). seven to eight residues (25) and binds hsp70 in an extended, ␤

strand conformation (26, 27). The predictive value of these studies Iodination of FYQLALT and VL was elegantly demonstrated by Knarr et al. (28) in mapping BiP The peptide FYQLALT has been demonstrated to bind all the members of binding sites on Ig heavy chains, and by Rudiger et al. (24) in hsp70 class of proteins, and its binding to Hsc70 has been studied exten- mapping dnaK binding sites of several proteins. sively (28, 32). FYQLALT was iodinated with Iodo-Beads (Pierce) fol- In contrast to the multitude of peptide-binding studies, the in- lowing the manufacturer’s instructions. Usually, 50 ␮g of peptide in 100 mM NaPO , pH 6.5, was iodinated with 2 mCi of 125I (Amersham, Ar- teraction of BiP with natural occurring substrate proteins has 4 lington Heights, IL) using 3–4 Iodo-Beads. Unincorporated 125I was re- scarcely been studied. Since the above-mentioned data with the moved by passing the peptide through a Sep-Pak C18 cartridge (Waters, dominant-negative BiP and with the point mutations in LC sug- Milford, MA) and eluting with a 30–50% acetonitrile gradient. The sp. act. gested that the predominant interaction with BiP is mediated by the of FYQLALT was typically 0.5–1 ϫ 1016 cpm/mol. V was iodinated also with the Iodo-Beads, following manufacturer’s V domain, we set out to reconstitute this interaction in vitro. L L instructions. The 125I (1–2 mCi) was activated with the Iodo-Beads in 25 Work described in this paper shows that VL, a domain important mM Tris-HCl, pH 7.5, and 0.4 M NaCl, and then incubated with 0.5–1 mg 125 Downloaded from for interaction with BiP in vivo, binds BiP in vitro. This binding is of VL. Unincorporated I was removed by desalting with a prepoured enhanced when the substrate is partly denatured. Furthermore, we Bio-Rad Econo-PacI0DG (10 ml bed vol), and the VL was stored at 4°C in ϫ 16 map potential BiP-binding peptides within V and show, surpris- 20 mM Tris-HCl, pH 7.5, and 150 mM NaCl. The sp. act. was 2 10 L cpm/mol. ingly, that the binding of the entire protein can be accounted for by as few as two BiP-binding peptides. The implications of these Fluorescein conjugation of peptides and proteins results for the progression of folding in the cell and for the selec- The FYQLALT peptide was suspended in 100 mM borate buffer, pH 9, at http://www.jimmunol.org/ tion of the LC repertoire are discussed. 2 mM and coupled to FITC, as suggested by the manufacturer (Molecular Probes, Eugene, OR). The FITC-conjugated peptide was separated from free FITC and unconjugated peptide by chromatography over a C-18 re- verse-phase HPLC column. The fractions were collected and lyophilized. Materials and Methods Fresh stock solutions of labeled peptide were made in 10 mM borate buffer, Protein purification pH 9, and the concentration was calculated from the absorbance at 494 and GST-BiP. The pDS78 plasmid, encoding a GST-hamster BiP fusion pro- 280 nm. The F/P ratio was 1. tein, was obtained from Dr. H. Weissbach (Roche, Nutley, NJ) and grown LEN, SMA, and REC were conjugated to 6-fluorescein-5 (and -6)-car- boxamido hexanoic acid and succinimidyl ester (SFX) (Molecular Probes) in Escherichia coli DH5␣. The bacteria were induced for3hatmid-log by incubating at a ratio of 5 mg V to 2 mg SFX, according to the man- with 1 mM isopropyl ␤-D-thiogalactoside and then resuspended in 20 mM L ufacturer’s instructions. Free SFX was separated from SFX-conjugated by guest on September 29, 2021 sodium phosphate, pH 7.5, and 130 mM NaCl (buffer A) and lysed by protein by passage through a Sephadex G-25 column (Pharmacia, Piscat- sonication in presence of 10% glycerol and 1% Triton X-100, as described away, NJ) and stored at 4°C in 0.1 M NaHCO , pH 8.3. Incorporation of by Carlino et al. (29). rGST-BiP was purified by affinity chromatography 3 SFX yielded an F/P ratio of 3. on glutathione-agarose (Sigma, St. Louis, MO), and eluted with 10 mM reduced glutathione in buffer A. When noted, the GST was cleaved from Gel filtration assay for binding of peptides and V to BiP BiP with 100 U of thrombin per 10 mg of GST-BiP (Calbiochem, La Jolla, L CA) in 20 ml buffer A with 5 mM CaCl2. Aliquots of BiP and GST-BiP Binding assays were routinely performed in 25 ␮l total volume of 50 mM ␮ were stored in 20 mM Tris, pH 7.5, 1 mM EDTA, 100 mM NaCl, and 5% Tris, pH 7.5, 50 mM KCl, 20 mM MgCl2, 0.5% Triton X-100, with 20 M glycerol at Ϫ80°C. BiP, 42 ␮M 125I-FYQLALT, and the appropriate concentration of unla-

His-tagged BiP. His6-tagged mouse BiP was initially obtained from Dr. S. beled VL or peptide for1hat25°C, unless otherwise noted. When peptides Blond-Elguindi (University of Illinois, Chicago). The protein was ex- were used to complete the binding, the reactions included 16% final con- ϩ pressed in E. coli and purified using affinity chromatography on Ni2 - centration of DMSO. This concentration of DMSO in itself had no effect on agarose, as described by the manufacturer (Qiagen, Chatsworth, CA). Fol- the binding of 125I-FYQLALT to BiP (data not shown). BiP-bound 125I- ϩ lowing elution from the Ni2 -agarose column, BiP was further purified FYQLALT was separated from free peptide by centrifugation for 2 min at with an ATP-agarose (Sigma; C-8 linkage) column, as described by Free- 1100 ϫ g through Micro Bio-Spin Chromatography Columns packed with 125 man et al. (30). The column was washed with 2 vol of 2 M NaCl and Bio-Gel P-30 (Bio-Rad). To separate BiP-bound VL from free I-labeled equilibrated with column buffer (20 mM Tris, pH 6.9, 5 mM MgCl2, and VL, similar spin columns were packed with Bio-Gel P-60 beads. The ra- 100 mM NaCl), and the bound BiP was eluted with column buffer con- dioactivity present in the flow-through was determined by scintillation

taining 10% glycerol and 50 mM ATP. His6-tagged yeast BiP (encoded by counting with a Packard Top-Count. Nonspecific binding was determined the Kar2 gene) was obtained from Dr. Jeff Brodsky (University of Pitts- by the addition of 600-fold excess unlabeled FYQLALT, and was routinely burgh) in the form of E. coli strain RR1 transformed with the plasmid 5–10% of the total. pMR2623. It was purified in the same fashion as the murine BiP. The Kd values were calculated from IC50 values (the concentrations of pep- murine version was more active than yeast BiP. tide required to obtain half-maximal inhibition) with the Cheng and Prusoff Variable domains of LEN, SMA, and REC ␬-chains. The pkIVlen004, equation (33): ␬ pkIVsma007, and pkIVrec006 constructs, which target the IV chain (VL) IC50 variable domains of LEN, SMA, and REC, respectively, to the periplasm K ϭ were described before (31). These V proteins were purified according to d L L ϩ the procedure described previously (31) and are referred to in this work 1 Kd* simply as LEN, REC, or SMA. Bacteria were grown at 30°C, and induced ␤ at mid-log growth with 1 mM isopropyl -D-thiogalactoside for 24 h. where Kd* is the dissociation constant of the labeled ligand (derived from Periplasmic proteins were harvested by resuspending the bacterial pellet in the dose-binding analyses, see below), Kd is the dissociation constant of the ice-cold 0.1 M Tris-Cl, pH 8, 0.25 mM EDTA, 0.25 M sucrose, and 0.4 inhibitor peptides, and L is the concentration of the labeled ligand. 125 mg/ml lysozyme, and pelleting the spheroplasts. VL was further purified by To demonstrate saturable binding, dose-binding assays of I-FYQ- 125 two successive ion-exchange chromatography steps, first over Econo-Pac LALT and I-labeled VL were performed by adding increasing concen- High Q cartridge and then a High S cartridge (Bio-Rad, Richmond, CA). trations of the labeled substrate to a fixed concentration of BiP (20 ␮M). Only preparations judged to be Ͼ95% pure based on SDS-PAGE Data obtained from the dose-binding analyses were fit to a Scatchard plot

were used. using the GraphPad Prism computer software to calculate Kd. This program 3844 BiP BINDING SITES ON Ig LIGHT CHAIN

allowed best-ﬁt determinations of the data to either a one- or two-site BiP, mouse BiP (shown below), and yeast BiP (kar-2) was lower binding model. than to human hsp70, but in each case was effectively inhibited by Calculation of the concentration of the BiP-LC complex (C) employed excess cold peptide (Fig. 1A). Interestingly, even the GST-BiP fusion the equation: protein was active in peptide binding, although less active than the ϩ ϩ Ϫ ͱ͑ ϩ ϩ ͒2 Ϫ B L Kd B L K d 4BL authentic BiP released by thrombin cleavage of the fusion protein. C ϭ 2 The activity of the nucleotide-binding domain of either BiP, where B is the concentration of BiP available for LC binding, L is the GST-BiP, or hsp70 was determined in two ways. Bound peptides

concentration of the LC-folding intermediates that can bind to BiP, and Kd were efficiently released by incubation with ATP in a one-cycle is the dissociation constant (34). binding assay (Fig. 1B), in which complexes were first purified by ATPase assays a desalting column, incubated with ATP, and repurified. This suggested that under conditions of continuous incubation (such as The extent of ATP hydrolysis was measured as previously described (35). those used in Fig. 1A), the observed level of peptide binding is the In a final volume of 50 ␮l, 2.5 ␮g of the appropriate preparation of BiP or hsp70 (a generous gift from R. Morimoto) was incubated at 37°C in the product of multiple cycles of binding and release. The peptides presence of 2 ␮Ci [␥-32P]ATP. Aliquots were removed at various time were released effectively by ATP from the various rBiP, despite points and added to activated charcoal slurry. After centrifugation, the their low inherent hydrolytic activity as compared with the recom- amount of 32P in the supernatant was counted in a Packard Top Count. binant hsp70 (Fig. 1C). This is consistent with the notion that nucleotide exchange, not hydrolysis, was important for peptide Gel electrophoresis analysis of the BiP-bound VL release of all members of the hsp70 family of chaperones (22). FITC-V was incubated for1hat25°C with His -tagged KAR-2 immo-

L 6 Downloaded from ϩ When we measured, instead of rates of ATP hydrolysis, the extent bilized on Ni2 -agarose beads. After washing, the bound material was eluted with SDS sample buffer and resolved on a 16% acrylamide Tricine- of stimulation of ATPase activity by FYQLALT, both human

SDS gel (36). Detection of KAR-2-bound FITC-VL was performed by a hsp70 and hamster BiP were stimulated 3–4-fold (data not shown). rapid silver stain method (37). The peptide-binding assay was used to measure the afﬁnity of 125 Prediction of BiP-binding peptides I-FYQLALT to BiP (Fig. 1D). Binding approached saturation at 500 ␮M peptide, and Scatchard analysis showed a biphasic bind- The distribution of putative BiP binding sites on variable domains was

Ϯ ␮ http://www.jimmunol.org/ ing curve, with a calculated Kd of 2 0.8 M for the higher addressed by the use of two computer algorithms. We developed one al- Ϯ ␮ gorithm based on the BiP-binding scores (weights) determined by Blond- affinity site, and a Kd of 163 22 M for the lower affinity site Elguindi et al. (10). These scores were related to the probability of any (n ϭ 4). Since at saturation binding capacity approached a ratio of particular amino acid being located at specific positions within heptameric 1 mol of peptide bound for every mole of BiP, this indicates the peptides that had measurable binding to BiP. For all 20 amino acids, neg- biphasic binding is a manifestation of high and low affinity states ative and positive numerical values were assigned to reflect the preferential exclusion or inclusion of the amino acid at positions 1–7 in BiP-binding for the single binding site of each BiP molecule (15, 38, 39). A peptides. The algorithm analyzes the variable domain sequence seven similar observation was made with the biphasic binding of FYQ- amino acids at a time by summing the position-dependent scores of each LALT by the related chaperone hsp70 (32). Competition-binding amino acid in the heptamer, assigning the resulting cumulative score to the assays (see Materials and Methods), in which increasing concen- first residue of the peptide. The algorithm was used to evaluate individual trations of unlabeled FYQLALT were used to displace the binding by guest on September 29, 2021 variable domains, as well as to determine an average BiP site distribution 125 by automatically analyzing a database of 122 human V␬ domain of a constant concentration of I-FYQLALT, gave a calculated Ϯ ␮ sequences. Kd of 5 2 M for the high affinity site (Fig. 2A). The correspondence between the K values of the direct binding and the Fluorescence measurements d competition assays validates this spin column assay and shows that ␮ Intrinsic tryptophan fluorescence was measured with 3 MofVL in 20 mM iodination of the tracer peptide did not significantly change its Tris, pH 7.5, and 150 mM NaCl in the absence or presence of the appro- binding affinity. Therefore, it was used to measure either direct priate concentrations of Gdn-HCl or DTT (both buffered to pH 7.5). Emis- sion spectra were recorded from 300–400 nm, using a Photon Technology binding of labeled substrates, or the ability of unlabeled substrates Industries (Princeton, NJ) fluorescence spectrophotometer, at excitation to displace the labeled ones. Because GST-hamster BiP and His6- wavelength of 280 nm. Raw data were corrected for the quenching effects mouse BiP were found to exhibit equivalent binding affinities and of the reducing agents. Measurements of binding of the fluorescent dye specificities, these rBiP forms were used interchangeably in all ANS to the native or unfolded V , LEN, were performed by incubating 10 L subsequent experiments. ␮M ANS with 3 ␮M of the appropriate LEN in 20 mM sodium phosphate buffer, pH 7. The emission spectra were then recorded from 420–600 nm Binding of V domains of LC to BiP after excitation at 350 nm. The fluorescence of ANS alone was subtracted L from the spectra of ANS in the presence of LEN. The LEN ␬-chain is a highly soluble Bence Jones protein isolated from a patient who excreted it at a rate of 50 g/day (31). In con- Results trast, the SMA and REC ␬-chains were derived from patients di- BiP-binding assay agnosed with light chain amyloidosis. Recombinant LEN and

Substrate binding to BiP was determined by reacting the two in SMA, in the form of VL polypeptides produced as periplasmic solution, followed by separation of bound and free reactants. The proteins in bacteria, were shown to behave like the respective peptide FYQLALT, which was previously shown to bind to a va- whole ␬-chains: LEN is stable and manifests the same monomer- riety of hsp70 family members (27, 32), was labeled by iodination dimer equilibrium typical of purified L chains, while SMA and or by coupling to FITC, and incubated with recombinant hsp70 REC are less stable proteins (31). Recombinant LEN and SMA (as proteins produced in bacteria. The reactions were applied to gel well as REC, not shown) were tested for their ability to bind rBiP filtration spin columns to separate free from bound reactants. The in a cold inhibition assay and in a direct binding assay (Fig. 2, A validity of this assay is shown in Fig. 1. Of the hsp70 proteins and B). Each protein bound best if it was first deliberately dena- tested, FYQLALT bound best to human hsp70. This binding was tured (Fig. 2, compare A with C). Either treatment with Gdn-HCl inhibited by addition of excess unlabeled peptide and by addition or controlled reduction and subsequent alkylation gave rise to pop-

of ATP (Fig. 1, A and B), consistent with the known mode of ulations of unfolded VL that bound BiP. The highest level of bind- action of hsp70. The iodinated and the ﬂuorescent versions of the ing was observed when LEN, SMA, or REC was both denatured peptide behaved similarly. The binding of FYQLALT to hamster and reduced. That such treatments indeed caused denaturation of The Journal of Immunology 3845

VL was shown by increases in the intrinsic Trp ﬂuorescence and binding of the hydrophobic dye ANS (Fig. 2, E and F). On the other hand, the structure of BiP itself was not affected at the concentrations of denaturant present in the binding assay (Ͻ0.4 M), as judged by the same two criteria (data not shown). Direct binding 125 of I-labeled, denatured, and reduced VL was saturable, with 88 Ϯ 24% of BiP bound at saturation (Fig. 2B, n ϭ 4). The cal- ␮ culated Kd for LEN, SMA, and REC was similar, 3–11 M (Fig. 2, A and B, and Table I). The nature of the label had no signiﬁcant effect, as similar results were observed with either iodinated or

FITC-conjugated VL (data not shown) and the binding of either form of labeled VL was inhibited by unlabeled, denatured VL. No binding of VL to BiP was detected at 25°C, as demonstrated by the inability to compete for BiP binding with 125I-FYQLALT

(Fig. 2C). Interestingly, marginal binding of all three VL could be consistently detected in the absence of denaturant when the incu-

bation was performed at 37°C, resulting in an apparent Kd of 300 ␮ M at best (Fig. 2C). Analysis of VL eluted from BiP showed that

the VL preparations contained a subpopulation that was qualita- Downloaded from tively different from the majority of molecules: the BiP-bound VL exhibited a slower mobility than the input material in Tricine-SDS nonreducing gels. A representative experiment is shown in Fig. 2D, with either native or Gdn-HCl-denatured REC that was incubated at 37°C with BiP. The mobility of the resulting BiP-bound

fractions was the same, i.e., slower than the input REC. Similar http://www.jimmunol.org/ results were observed with LEN and SMA (data not shown). We

surmise, therefore, that a small fraction of the VL population, as isolated from the bacterial periplasm, is nonnative and leads to the

low BiP-binding activity of folded VL. It is likely that this fraction of the protein is larger at 37°C and becomes even larger under unfolding condition. We estimate from the saturation-binding conditions that at least one-half of LEN or SMA can become unfolded enough to interact with BiP under the conditions used. Kinetic analysis of intrinsic Trp fluorescence of the denatured by guest on September 29, 2021 and reduced LEN showed that it did not diminish substantially upon incubation in the binding buffer (Fig. 2E). In contrast, significant reduction in Trp fluorescence was observed upon dilution of Gdn-HCl-denatured LEN (with an intact disulfide bond) into the binding buffer (Fig. 2E). This suggests that the form of LEN used as a substrate in these assays did not refold significantly under the binding conditions, whereas in the case of denatured LEN, whose disulfide bond is oxidized, the fraction of the protein capable of BiP binding decreased with time, mimicking a folding reaction.

Thus, just as seen in vivo, unfolded forms of VL have higher af- ﬁnity for BiP than the more mature, folded species.

Identifying potential BiP binding sites in VL

To map BiP binding sites in VL, we searched the sequence for peptides conforming to published motifs that predict likely binder

C, The kinetics of ATP hydrolysis by four recombinant hsp70 proteins. The ability of bacterially produced proteins to hydrolyze [␥-32P]ATP was measured as the radioactivity that fails to bind to activated charcoal slurry, as described in Materials and Methods. A GST-rab fusion protein was used FIGURE 1. Differential peptide binding and ATPase activity of hsp70 as a negative control for both peptide binding and ATP hydrolysis. D, proteins. A, The ability of human hsp70, hamster BiP, GST-BiP fusion Binding analysis of 125I-FYQLALT to BiP. 125I-FYQLALT at increasing protein, and yeast Kar2 to bind 125I-FYQLALT was compared. Open concentrations was incubated in duplicate for1hat25°C with a constant bars ϭ 42 ␮M 125I-FYQLALT ϩ 20 ␮M hsp70 or BiP; solid bars ϭ concentration of BiP. Shown is a representative experiment utilizing 5 ␮M 125I-FYQLALT bound in the presence of 100-fold excess unlabeled pep- His-mouse BiP; equivalent results were obtained with 20 ␮M GST-hamster tide. B, ATP-induced release of bound peptide. Complexes of the indicated BiP. Bound peptide was separated from free, as described in Materials and hsp70 protein with 125I-FYQLALT were puriﬁed over a spin column, in- Methods. Data obtained from the dose-binding analysis (inset) were ﬁt to

cubated for 10 min with (ﬁlled bars) or without (open bars) 10 mM ATP, a Scatchard plot using the GraphPad Prism program, to calculate Kd. Pep- and repuriﬁed over a spin column. The amount of radioactive peptide re- tide binding at higher input concentrations could not be reliably determined maining in the complex was measured and is presented as relative binding. due to peptide insolubility. 3846 BiP BINDING SITES ON Ig LIGHT CHAIN Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

f ࡗ FIGURE 2. Binding of folded and unfolded VL to BiP. A, Binding of unfolded VL to BiP. LEN ( ) or SMA ( ) was incubated for2hat25°C in the ␮ ␮ 125 presence of 2 M Gdn-HCl and 40 mM DTT. Increasing concentrations of the appropriate VL were then added to 20 M GST-BiP and 42 M I- FYQLALT (resulting in ﬁnal concentrations of 0.8 M Gdn-HCl and 16 mM DTT). As a control for any effect on GST-BiP, unlabeled FYQLALT (F) was 125 used as a competitor in the presence of Gdn-HCl and DTT. B, Dose binding of unfolded VL, showing saturability. Denatured and reduced I-labeled SMA was prepared by treatment with 2 M Gdn-HCl and 40 mM DTT. Increasing doses of this preparation were incubated directly with BiP, and the complexes were separated from the free reactants and counted. The data were ﬁt to a nonlinear regression curve with the GraphPad Prism computer software. The Ϯ Ϯ ␮ ϭ percentage of bound VL relative to the amount of input BiP was calculated as 88 24%, while the Kd was calculated as 6 5 M(n 4). C, Inhibition of 125I-FYQLALT binding to BiP by LEN (f) or SMA (ࡗ) at 25°C, or Len at 37°C (Ⅺ), as compared with the inhibition by unlabeled FYQLALT (F). 2ϩ D, Gel electrophoresis analysis of the BiP-bound VL. FITC-REC was incubated for1hat25°C with His6-tagged Kar2 immobilized on Ni -agarose beads in 96-well plates. After washing, the bound material was eluted with nonreducing SDS sample buffer and resolved on a 16% acrylamide, nonreducing

Tricine-SDS gel. Lane 1, Migration of the input VL; lane 2, mock reaction containing VL and beads without Kar2; lane 3, migration of the Kar2-bound VL in native conditions; lane 4, migration of the Kar2-bound VL with Gdn-HCl denaturation. E, Time course of VL refolding, as measured by intrinsic tryptophan fluorescence. Spectra of native LEN (dashed line), LEN denatured with 2 M Gdn-HCl (Ⅺ), or reduced with 40 mM DTT and denatured with 2 M Gdn-HCl (E) were obtained, and then each sample was diluted into the BiP-binding buffer used in A (but without detergent). Emission spectra from 300–400 nm were recorded upon excitation at 280 nm at the indicated times after dilution, and the peak fluorescence was plotted as a percentage of the ␮ initial Trp fluorescence of each sample. F, Changes in VL structure upon reduction and denaturation, as measured by ANS binding. A total of 10 M ANS was added to 3 ␮M of either folded LEN (dashed line) or unfolded LEN (dotted line) in 20 mM sodium phosphate buffer, pH 7, and emission spectra from 420–600 nm were recorded upon excitation at 350 nm. The spectrum of ANS alone in buffer has been subtracted from each of these spectra.

peptides. Based on the data of Blond-Elguindi et al. (10), derived any VL sequence. A negative score by this program indicates that from BiP panning of dodecameric and heptameric peptide libraries the peptide has a low probability to bind BiP, while scores above displayed by phages, a computer program was written to search ϩ5 indicate a signiﬁcant probability of binding BiP (10). Knarr et The Journal of Immunology 3847

a Table I. Binding of VL peptides to BiP

Peptide Sequence Predicted ␮ (position in VL) BiP Score Kd ( M) FYQLALT ϩ11 5 Ϯ 2 LAVSLGE (11–17) ϩ2 220 Ϯ 108 NTLAWYQ (31–37)b ϩ12 41 Ϯ 16 TLAWYQQ (32–38)b ϩ1 102 Ϯ 38 WYQQKPG (35–41) ϩ9 Ͼ1000 YQQKPGQ (36–42) 0 Ͼ1000 PKLLIYWA (45–52) ϩ7 187 Ϯ 99 DRFSGSG (60–66) ϩ2 Ͼ1000 RFSGSGS (61–67) 0 Ͼ1000 TDFTLTI (69–75) ϩ681Ϯ 1 FTLTISS (71–77) ϩ747Ϯ 15 Ϯ Denatured VL NA 6 5

a Peptides derived from VL are listed with amino acid numbers noted in paren- thesis. The predicted BiP score was calculated using a computer algorithm based on data by Blond-Elguindi et al. (10). The peptide PKLLIYWA was predicted to be the most likely to bind dnaK, based on the data in (24). The related peptide PKLLIYAA, which is the commonly occurring sequence in families other than ␬IV, could not be Downloaded from tested reliably due to its insolubility. Using the IC50, the Kd was calculated for each 125 peptide or unfolded VL that competed with the binding of I-FYQLALT to GST or His-BiP. The values shown are averages of at least three experiments per peptide, except for the peptide TDFTLTI, which was tested twice. b Substituting Thr32 with Tyr did not alter the measured binding to BiP. NA, not applicable. http://www.jimmunol.org/ al. (28) used a similar approach to test predicted BiP-binding peptides in an Ig heavy chain. Our analysis of the LEN sequence, which is a germline ␬IV with one somatic mutation, shows multiple heptameric peptides with the requisite score (Fig. 3A). More recently, a different motif for peptides that bind the bacterial homologue of BiP, dnaK, was deﬁned via an extensive peptide scan (24). This prediction may be applicable to BiP, because all of the amino acids that contact the bound peptides in dnaK are

conserved in BiP (15). The only close match within the LEN se- by guest on September 29, 2021 quence to the latter motif was the peptide P(R,K)LLIY (residues 45–50, by the Kabat nomenclature (40)), which is consistently found in all L chain families.

In addition to analyzing individual VL sequences, we analyzed an entire collection of human ␬ LC, comprising of 122 sequences (F. Stevens, unpublished). The rationale is that scores of individual BiP-binding peptides that are generated by somatic mutation would be dampened by averaging over the entire population,

whereas BiP-binding sequences common to most LC would not, FIGURE 3. Prediction of BiP binding sites within VL. A, A sequence and their identification would be more reliable. Some of the results search of the VL LEN, performed with a computer algorithm based on the of this analysis are shown in Fig. 3B. Two major regions were data of Blond-Elguindi et al. (10). The scores obtained for each peptide found to contain peptides with scores above 5 in several registers, using a moving heptamer window are plotted as a function of sequence indicating significant probability to bind BiP. These regions span position. BiP scores of ϩ5 or higher have been predicted to possess sig- amino acids 31–41 (peptide I, Fig. 3B) and 69–77 (peptide II) in nificant probability of binding to BiP (28). Negative scores indicate that these peptides have little probability of binding to BiP. LEN represents the the framework regions of V . Both of these regions are conserved L ␬IV family and differs from the germline gene in only one position (Ser for in VL sequences. A third region was found in amino acids 102– Asn (29, 31)). B, A similar sequence search of the variable domains of 122 110, but was not tested in this study because it is highly variable ␬ human light chains. The averages of scores obtained for all of these VL in LC sequences due to joining of various VL-J gene segments. sequences using a moving heptamer window are plotted. The threshold score of ϩ5, which was found to be a reasonable predictor in experimental Comparing affinities of peptides for BiP binding tests (28), is marked with a dashed line. Three regions have significant positive scores (denoted I, II, and III) and span the sequences 31–38, 69– To test the above predictions, 10 different VL peptides were synthesized (Table I). Some peptides were derived from the regions 77, and 96–101, respectively, according to the Kabat nomenclature (40). C, Binding analysis of V peptides. V -derived peptides were assayed for predicted to bind BiP (Fig. 3), while others had a fairly low pre- L L their ability to compete with 125I-FYQLALT for binding to BiP. Increasing dicted binding probability. The low probability peptide 61–67 was concentrations of the appropriate competitor were incubated with constant chosen, because previous data from our lab showed that point mu- concentrations of 125I-FYQLALT (42 ␮M) and GST-BiP (20 ␮M) for 1 h tations in this sequence actually enhanced LC binding to BiP in at 25°C. After separating bound peptides from free, Kd were calculated vivo (7, 8). All 10 peptides were tested for their ability to compete from the IC50 values, as described in Materials and Methods. with the binding of the labeled peptide FYQLALT to BiP. Exam- ples of the peptide competition assays are shown in Fig. 3C, and the experiments are summarized in Table I. The apparent dissoci- 3848 BiP BINDING SITES ON Ig LIGHT CHAIN

(Fig. 4, bottom). Thus, each ␤ sheet of the variable domain contains one high affinity BiP binding site. The disparity between the apparent affinity of each peptide and the affinity of the entire protein for BiP could be due to several factors. The heptamers in solution may only contain a fraction of the total population that is in a binding-competent conformation, whereas when they are constrained in the polypeptide they may all be in a similar conformation. The observation that the 11-mer 31–41 bound slightly better than the heptamer 31–37 and consid- erably better than the 32–38 heptamer (Table I) is consistent with

this idea. Alternatively, it is possible that each VL is bound at more than one site and that binding of one BiP molecule to one site increases the likelihood of another BiP binding to the second site. Experiments designed to detect such possible cooperativity are in progress.

Discussion

This work describes the in vitro reconstitution of interactions be- Downloaded from FIGURE 4. BiP binding sites within V . A schematic representation of L tween Ig L chain and one of its known in vivo chaperones, BiP. LC folding, adapted from (23), with the identified BiP-binding peptides This reconstitution mimics the in vivo binding specificity of BiP highlighted. The ␤ sheets in both V and C domains are shown in gray, L L for the immature forms of L chain and the sensitivity of binding to and several key amino acids in VL are indicated by their residue number, for orientation. The disulfide bonds within each domain are shown in black. the nucleotide state of BiP. The reconstituted binding was then The two highest affinity BiP-binding peptides are shaded black, and the two used to map binding sites within the LC, leading to the conclusion

intermediate affinity sites are shaded dark gray. that as few as two peptides are sufficient for this interaction. http://www.jimmunol.org/ One mechanism that is often envisioned for BiP binding to newly synthesized LC is attachment of multiple BiP molecules to ation constant (Kd) calculated for the binding of the denatured VL appropriate peptide sites displayed along the length of the unfolded Ϯ ␮ was 6 5 M, similar to the apparent Kd of the FYQLALT stan- polypeptide. Data for the related hsp70 family protein dnaK show dard. Of the 10 VL peptides, 4 did not bind BiP, including both of that in various random proteins, sites are found on average every the peptides from the predicted negative region 59–67 (DRF- 36 amino acids (24). If these data are extrapolated to BiP, there can SGSG and RFSGSGS). Two heptamers from region I in Fig. 3B be six binding sites per LC. However, this number is almost cer- (NTLAWYQ and TLAWYQQ) bound BiP, while two others tainly an underestimate, because unlike the collection of substrate (WYQQKPG and YQQKPGQ) did not bind, thus mapping the proteins used to generate the data for dnaK, LC consists entirely of by guest on September 29, 2021 binding to residues 31–38. Substituting Thr32 with Tyr, the most ␤ sheets, with no ␣ helices, and BiP binds to peptides in the ex- common residue at this position, did not reduce the binding affinity tended ␤ conformation (26). Attachment to multiple sites along the measurably (data not shown). We also tested BiP binding of an polypeptide is also inferred from the mode of action of BiP or 11-mer peptide 31–41, which contains both binding and nonbind- mitochondrial hsp70 in the translocation of proteins across mem- Ϯ ␮ ing heptamers, and found its Kd to be 33 13 M, essentially the branes: the power stroke of hsp70 is proposed to mediate vectorial same as the Kd of the 31–37 heptamer (Table I). In similar fashion, transport across the membrane, either as a ratchet mechanism or as the two predicted binder peptides from the second region, 69–77 a mechanochemical motor (41–43). (TDFTLTI and FTLTISS), were found to bind BiP experimentally An alternative mechanism for the action of BiP in protein fold-

(Table I and Fig. 3). Apparent Kd were calculated for all the binder ing is attachment to key peptides that must not be left exposed lest peptides from data such as shown in Fig. 3C and ranged from folding take an unproductive pathway. According to this view, 41–220 ␮M (Table I). there are a few dominant binding sites among the many potential Two other heptamers have a lower afﬁnity to BiP. The more sites. Previous work from this lab and others (44–46)6 pointed to

N-terminal is peptide 11–17, which in the folded structure forms a the VL domain as the major determinant in the association of LC ␤ tight turn between strands A and B (Fig. 4). Its apparent Kd with BiP during folding in the cell. In this study, we present direct (Table I) was signiﬁcantly lower than either peptide regions I or II, evidence that the VL domain indeed binds to BiP in vitro, and we in agreement with its barely positive predictive score (Table I). map two likely dominant BiP binding sites.

The fourth potential binding site is the 45–52 peptide, predicted by The data presented in this study demonstrate that two VL pro- Bukau’s algorithm (24) as well as Blond-Elguindi’s algorithm teins, LEN and SMA, bind BiP in vitro with similar affinities. (10). The consensus sequence at this site, PKLLIYAA, could not Furthermore, their binding is a property of partially unfolded in- be tested reliably due to its insolubility. Therefore, we tested termediates and not of the native VL, as shown by the vastly in- PKLLIYWA, which is more soluble and is the exact sequence creased affinities after denaturation of the recombinant proteins. found in the LEN protein. Its Kd was similar to peptide 11–17, This feature recapitulates the in vivo behavior of newly synthe- placing these two peptides at the lower end of the affinity spectrum sized LC that only associate with BiP as long as they have not among all of the tested VL peptides. completed their disulfide bonding (3, 44). LC binding to BiP in The two best BiP binders of all the heptamers tested, peptide vitro is saturable, sensitive to ATP as expected from BiP’s mode 31–37 and peptide 71–77, each form an extended ␤ strand. Peptide of action, and is inhibited by heptameric peptides that themselves

31–37 is in strand C, spanning the diameter of the VL domain (Fig. bind to BiP. 4). It is buried in the core of the native protein and includes Trp35, which packs against the internal disulﬁde bond (Fig. 4, top). The other BiP binding site, peptide 71–77, forms part of ␤ strand E 6S. Aviel et al., submitted for publication. The Journal of Immunology 3849

Our in vitro studies identiﬁed four BiP-binding peptides in the kinetics of LC folding: given that LC bound to BiP cannot com-

sequences of LEN and SMA, out of the collection of predicted plete its folding, between 10% and 14% of the F1 intermediate (the peptides. Of these four peptides, two bind BiP with afﬁnities that one disulﬁde intermediate) (4) must be free at any given time to

are reasonably close to the affinity of the entire VL protein. These account for the rate constant of its conversion to the fully oxidized ␤ two peptides are in strands that are buried in the hydrophobic form (derived from the t1/2 of the F1 intermediate and assuming core of the folded VL and are well conserved among LC se- first order reaction). quences. One site is in strand E of the four-stranded ␤ sheet, and When the human germline V␬ segments are examined, the se- a second site is in strand C of the five-stranded ␤ sheet (Fig. 4). We quences of the two peptides with the highest affinity for BiP are propose that binding of BiP to these two peptides is sufficient to highly conserved. The hydrophobic residues in the 71–77 sequence account for binding of the entire protein. Binding of BiP to one or are invariant, and the variability is limited to a Lys-for-Thr74 sub- both of these dominant sites could maintain the two halves of the stitution in the ␬II family, Arg-for-Ser77 in the ␬II and two ␬III ␤ sandwich in an open conformation, with Cys23 and Cys88 in a genes, and a few other genes with Asn instead of Ser76 or Ser77.In reduced state. Three lines of evidence are consistent with this in- the 31–37 sequence, Trp35 is invariant, as is the tetrapeptide ␬ terpretation. First, maximal binding to BiP occurred when VL was WYQQ, except for the II family again, displaying a few conser- not only denatured, but also reduced. Second, coimmunoprecipi- vative substitutions (Phe or Leu for Tyr36; Leu for Gln37). Position tation data from metabolically labeled cells enabled an estimation 34 is almost always occupied by a residue with a small side chain that 2–4 mol of BiP were bound per mole of LC (3). Third, such (Ala, Gly, or Ser) and occasionally Asn or Asp, while position 31 binding accounts for the in vivo observation that prolonged asso- is always either Ser or Asn. This sequence conservation of the

ciation with BiP retards the oxidation of the VL domain (4). BiP-binding peptides is not due to protection from the somatic Downloaded from The binding to peptide 11–17 is intriguing: in the native state, mutation mechanism: a number of mutations within the sequences this peptide forms a tight turn between strands A and B. This turn have been identiﬁed in LC transcripts from Peyer’s patch B cells, is a conserved feature of the Ig fold, and mutations disrupting this showing that mutations in these residues do occur (52) (C. Mil- turn are associated with various aggregation-prone LC (Ref. 46 stein, personal communication). Therefore, the conservation is due and footnote 6). If the binding of this peptide to BiP persists, the to selection at the level of the expressed protein, presumably be-

formation of the four-stranded ␤ sheet is expected to be delayed, cause these peptides are important in the proper folding of the http://www.jimmunol.org/

leaving the VL domain in a very unfolded state. More likely, given domain. the low affinity of this site, it may be the first of the peptides to BiP binding to framework peptides is most likely an important, fold, enabling the folding of the two ␤ sheets, while BiP binding yet little appreciated, selective force that participates in shaping the to the other two sites delays the formation of the disulfide bond expressed repertoire of Ig V genes. Proper folding of variable do- between Cys23 and Cys88. Like the 11–17 peptide, the 45–50 pep- mains depends heavily on an extensive network of hydrogen bond- tide, whose binding affinity in vitro is intermediate, is largely sol- ing (53), so many somatic mutations disrupt entire constellations vent exposed even in the native structure. In most ␬ families, this of amino acids. Even mutations in the hypervariable loops are peptide is P(K,R)LLIY, a sequence that is predicted (24) to bind known to have global destabilizing effects on the whole domain BiP even less avidly than the PKLLIKY sequence found in the (54). Therefore, mutations in many positions could lead to persis- by guest on September 29, 2021 LEN and SMA proteins. We therefore propose that the 11–17 and tent exposure of the conserved BiP-binding peptides, causing pro- 45–50 peptides are not the important BiP binding sites in vivo. longed BiP-LC interactions. In addition, somatic mutations may

The presence of two dominant BiP binding sites in the VL do- also create new, high afﬁnity, BiP-binding peptides causing pro- main does not rule out the possibility of other BiP binding sites in longed interactions. If the overall avidity for BiP is not too high,

the VL domain or in the CL domain. We predict, however, that the mutant LC may actually benefit from the shielding from water because CL folds and oxidizes even in the presence of a nonre- and other proteins provided by BiP, manage to fold, and be in- leasing BiP (4), only low affinity BiP-binding peptides exist in this cluded in the expressed repertoire. If, on the other hand, the mu- domain. The data presented in this study suggest that binding of tation is catastrophic in terms of proper folding, the persistent as- BiP to a newly synthesized polypeptide in the cell should not nec- sociation with BiP could delay folding, inhibit Ig assembly, and essarily be viewed as a process of coating a string of amino acids ultimately target somatic mutants for degradation. Finally, occu- with multiple BiP molecules. Such binding to multiple low affinity pancy of BiP by mutant proteins is most likely used in an ER-to- peptides may mediate nascent chain translocation across the ER nucleus signal transduction pathway, termed the unfolded protein membrane, one of the physiological functions of BiP (47). This response, and characterized in yeast (55). Thus, BiP could be a role is analogous to that of mitochondrial hsp70 in translocation of sensor that monitors V region diversity and reports back to the proteins across the organelle (42, 48). Instead, we suggest that genetic apparatus whether to undergo further gene rearrangements. once the proteins are in the lumen of the ER, association of BiP In conclusion, our reconstitution of BiP-substrate interactions during folding can be explained by selective binding to higher strongly supports the hypothesis that during folding in vivo BiP affinity sites identified in this work. binds to only few higher affinity sites per substrate that are located How much of the LC in the ER is bound to BiP? The concen- in important elements within the folding module. By binding to tration of the incompletely folded intermediates of LC within the them, BiP retards the folding, thus helping to ensure progress ER is 0.36–0.9 ␮M (49, 50). The total BiP concentration is esti- along the productive pathway. This binding mechanism is a so- mated at 100 ␮M (49), and it is possible that only a fraction of the phisticated tool that may be used by B lineage cells to monitor the

total pool is available to bind LC. Together with the Kd values outcome of gene rearrangements and somatic mutations during the derived in this study, it can be calculated that 32–50% of the LC diversification of the Ig repertoire. intermediates should be bound to BiP when ATP is present, and 48–92% when ATP is depleted. These calculations are consistent Acknowledgments ϳ with our data showing that only 1/3 of the total LC can be im- We thank Helene Auer for synthesizing most of the peptides used in this munoprecipitated with BiP (51), particularly considering that some work, and the BSD Protein-Peptide Core Facility for synthesis of the other substrate is likely to dissociate during the immuno-isolation pro- peptides. N. Christoforou diligently purified the REC and SMA proteins. cedure. The fraction of BiP-bound LC is also consistent with the Drs. J. Brodsky, R. Morimoto, S. Blond, and H. Weissbach generously 3850 BiP BINDING SITES ON Ig LIGHT CHAIN provided constructs and purified hsp proteins, and C. Milstein kindly 26. Landry, S. J., R. Jordan, R. McMacken, and L. M. Gierasch. 1992. Different shared unpublished Ig sequences from mouse Peyer’s patch B cells. We conformations for the same polypeptide bound to chaperones DnaK and GroEL. Nature 355:455. thank Drs. T. Pan and T. Sosnick for the use of the fluorescence spectro- 27. Fourie, A. M., J. F. Sambrook, and M. J. Gething. 1994. Common and divergent photometer, and all members of our group for vigorous discussions and peptide binding specificities of hsp70 molecular chaperones. J. Biol. Chem. 269: comments on this work. 30470. 28. Knarr, G., M. J. Gething, S. Modrow, and J. Buchner. 1995. BiP binding sequences in antibodies. J. Biol. Chem. 270:27589. References 29. Carlino, A., H. Toledo, D. Skaleris, R. DeLisio, H. Weissbach, and N. Brot. 1992. Interactions of liver Grp78 and Escherichia coli recombinant Grp78 with ATP: 1. Haas, I. G., and M. Wabl. 1983. Immunoglobulin heavy chain binding protein. multiple species and disaggregation. Proc. Natl. Acad. Sci. USA 89:2081. Nature 306:387. 2. Bole, D. G., L. M. Hendershot, and J. F. Kearney. 1986. Posttranslational asso- 30. Freeman, B. C., M. P. Myers, R. Schumacher, and R. I. Morimoto. 1995. Iden- ciation of immunoglobulin heavy chain binding protein with nascent heavy tification of a regulatory motif in Hsp70 that affects ATPase activity, substrate chains in nonsecreting and secreting hybridomas. J. Cell Biol. 102:1558. binding and interaction with HDJ-1. EMBO J. 14:2281. 3. Melnick, J., J. L. Dul, and Y. Argon. 1994. Sequential interaction of the chap- 31. Stevens, P. W., R. Raffen, D. K. Hanson, Y. L. Deng, M. Berrios-Hammond, erones BiP and GRP94 with immunoglobulin chains in the endoplasmic reticu- F. A. Westholm, C. Murphy, M. Eulitz, R. Wetzel, A. Solomon, et al. 1995. lum. Nature 370:373. Recombinant immunoglobulin variable domains generated from synthetic genes 4. Hendershot, L., J. Wei, J. Gaut, J. Melnick, S. Aviel, and Y. Argon. 1996. In- provide a system for in vitro characterization of light-chain amyloid proteins. hibition of immunoglobulin folding and secretion by dominant negative BiP AT- Protein Sci. 4:421. Pase mutants. Proc. Natl. Acad. Sci. USA 93:5269. 32. Takeda, S., and D. B. McKay. 1996. Kinetics of peptide binding to the bovine 5. Ma, J., J. F. Kearney, and L. M. Hendershot. 1990. Association of transport- 70kD heat shock cognate protein [abstract]. In Molecular Chaperones and the defective light chains with immunoglobulin heavy chain binding protein. Mol. Heat Shock Response, Cold Spring Harbor Symposium. Cold Spring Harbor Lab Immunol. 27:623. Press, Plainview, NY, p. 289. 6. Knittler, M. R., S. Dirks, and I. G. Haas. 1995. Molecular chaperones involved 33. Cheng, Y., and W. H. Prusoff. 1973. Relationship between the inhibition constant (K1) and the concentration of inhibitor which causes 50 per cent inhibtion (I50) in protein degradation in the endoplasmic reticulum: quantitative interaction of Downloaded from the heat shock cognate protein BiP with partially folded immunoglobulin light of an enzymatic reaction. Biochem. Pharmacol. 22:3099. chains that are degraded in the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA 34. Winzor, D. J., and W. H. Sawyer. 1995. Quantitative Characterization of Ligand 92:1764. Binding. Wiley-Liss, New York. 7. Dul, J. L., and Y. Argon. 1990. A single amino acid substitution in the variable 35. Sadis, S., and L. E. Hightower. 1992. Unfolded proteins stimulate molecular region of the light chain specifically blocks immunoglobulin secretion. Proc. chaperone Hsc70 ATPase by accelerating ADP/ATP exchange. Biochemistry 31: Natl. Acad. Sci. USA 87:8135. 9406. 8. Gardner, A. M., S. Aviel, and Y. Argon. 1993. Rapid degradation of an unas- 36. Schagger, H., and G. von Jagow. 1987. Tricine-sodium dodecyl sulfate-polyac- sembled immunoglobulin light chain is mediated by a serine protease and occurs rylamide gel electrophoresis for the separation of proteins in the range from 1 to in a pre-Golgi compartment. J. Biol. Chem. 268:25940. 100 kDa. Anal. Biochem. 166:368. http://www.jimmunol.org/ 9. Kaloff, C. R., and I. G. Haas. 1995. Coordination of immunoglobulin chain fold- 37. Meril, C., D. Goldman, S. Sedman, and M. Evert. 1981. Ultrasensitive stain for ing and immunoglobulin chain assembly is essential for the formation of func- proteins in polyacrylamide gels shows regional variation in cerebrospinal fluid tional IgG. Immunity 2:629. proteins. Science 211:1437. 10. Blond-Elguindi, S., S. E. Cwirla, W. J. Dower, R. J. Lipshutz, S. R. Sprang, 38. Rudiger, S., A. Buchberger, and B. Bukau. 1997. Interaction of Hsp70 chaper- J. F. Sambrook, and M. J. Gething. 1993. Affinity panning of a library of peptides ones with substrates. Nat. Struct. Biol. 4:342. displayed on bacteriophages reveals the binding specificity of BiP. Cell 75:717. 39. Zhang, J., and G. C. Walker. 1996. Identification of elements of the peptide 11. Gething, M. J., S. Blond-Elguindi, J. Buchner, A. Fourie, G. Knarr, S. Modrow, binding site of DnaK by peptide cross-linking. J. Biol. Chem. 271:19668. L. Nanu, M. Segal, and J. Sambrook. 1995. Binding sites for Hsp70 molecular 40. Kabat, E. A., T. T. Wu, H. M. Perry, K. S. Gottesman, and C. Foeller. 1991. chaperones in natural proteins. Cold Spring Harb. Symp. Quant. Biol. 60:417. Sequences of Proteins of Immunological Interest. U.S. Department of Health and 12. Flynn, G. C., T. G. Chappell, and J. E. Rothman. 1989. Peptide binding and Human Services, Public Health Service, National Institutes of Health, release by proteins implicated as catalysts of protein assembly. Science 245:385. Bethesda, MD. by guest on September 29, 2021 13. Brodsky, J. L. 1998. Translocation of proteins across the endoplasmic reticulum 41. Gambill, B. D., W. Voos, P. J. Kang, B. Miao, T. Langer, E. A. Craig, and membrane. Int. Rev. Cytol. 178:277. N. Pfanner. 1993. A dual role for mitochondrial heat shock protein 70 in mem- 14. Flaherty, K. M., C. DeLuca-Flaherty, and D. B. McKay. 1990. Three-dimensional brane translocation of preproteins. J. Cell Biol. 123:109. structure of the ATPase fragment of a 70K heat-shock cognate protein. Nature 42. Ungermann, C., W. Neupert, and D. M. Cyr. 1994. The role of Hsp70 in con- 346:623. ferring unidirectionality on protein translocation into mitochondria. Science 266: 15. Zhu, X., X. Zhao, W. F. Burkholder, A. Gragerov, C. M. Ogata, M. E. Gottesman, 1250. and W. A. Hendrickson. 1996. Structural analysis of substrate binding by the 43. Horst, M., A. Azem, G. Schatz, and B. S. Glick. 1997. What is the driving force molecular chaperone DnaK. Science 272:1606. for protein import into mitochondria? Biochim. Biophys. Acta 1318:71. 16. Catipovic, B., J. Dal Porto, M. Mage, T. E. Johansen, and J. P. Schneck. 1992. 44. Knittler, M. R., and I. G. Haas. 1992. Interaction of BiP with newly synthesized Major histocompatibility complex conformational epitopes are peptide specific. immunoglobulin light chain molecules: cycles of sequential binding and release. J. Exp. Med. 176:1611. EMBO J. 11:1573. 17. Flaherty, K. M., S. M. Wilbanks, C. DeLuca-Flaherty, and D. B. McKay. 1994. 45. Skowronek, M. H., L. M. Hendershot, and I. G. Haas. 1998. The variable domain Structural basis of the 70-kilodalton heat shock cognate protein ATP hydrolytic of nonassembled Ig light chains determines both their half-life and binding to the activity. II. Structure of the active site with ADP or ATP bound to wild type and chaperone BiP. Proc. Natl. Acad. Sci. USA 95:1574. mutant ATPase fragment. J. Biol. Chem. 269:12899. 46. Stevens, F. J., and Y. Argon. 1999. Pathogenic light chains and the B cell rep- 18. Wilbanks, S. M., C. DeLuca-Flaherty, and D. B. McKay. 1994. Structural basis ertoire. J. Immunol., In press. of the 70-kilodalton heat shock cognate protein ATP hydrolytic activity: I. Ki- 47. Brodsky, J. L. 1996. Post-translational protein translocation: not all hsc70s are netic analyses of active site mutants. J. Biol. Chem. 269:12893. created equal. Trends Biochem. Sci. 21:122. 19. Liberek, K., D. Skowyra, M. Zylicz, C. Johnson, and C. Georgopoulos. 1991. The Escherichia coli DnaK chaperone, the 70-kDa heat shock protein eukaryotic 48. Chauwin, J.-F., G. Oster, and B. S. Glick. 1998. Strong precursor-pore interac- equivalent, changes conformation upon ATP hydrolysis, thus triggering its dis- tions constrain models for mitochondrial protein import. Biophys. J. 74:1732. sociation from a bound target protein. J. Biol. Chem. 266:14491. 49. Wiest, D. L., J. K. Burkhardt, S. Hester, M. Hortsch, D. I. Meyer, and Y. Argon. 20. Gaut, J. R., and L. M. Hendershot. 1993. Mutations within the nucleotide binding 1990. Membrane biogenesis during B cell differentiation: most endoplasmic re- site of immunoglobulin-binding protein inhibit ATPase activity and interfere with ticulum proteins are expressed coordinately. J. Cell Biol. 110:1501. release of immunoglobulin heavy chain. J. Biol. Chem. 268:7248. 50. Melnick, J., and Y. Argon. 1995. Molecular chaperones and the biosynthesis of 21. Wei, J., J. R. Gaut, and L. M. Hendershot. 1995. In vitro dissociation of BiP- antigen receptors. Immunol. Today 16:243. peptide complexes requires a conformational change in BiP after ATP binding 51. Melnick, J. 1995. Folding of immunoglobulin in vivo. Doctoral dissertation, but does not require ATP hydrolysis. J. Biol. Chem. 270:26677. Duke University, Durham, NC. 22. Buchberger, A., H. Theyssen, H. Schroder, J. S. McCarty, G. Virgallita, 52. Gonzalez-Fernandez, A., and C. Milstein. 1993. Analysis of somatic hypermu- P. Milkereit, J. Reinstein, and B. Bukau. 1995. Nucleotide-induced conforma- tation in mouse Peyer’s patches using immunoglobulin ␬ light-chain transgenes. tional changes in the ATPase and substrate binding domains of the DnaK chap- Proc. Natl. Acad. Sci. USA 90:9862. erone provide evidence for interdomain communication. J. Biol. Chem. 53. Chothia, C., J. Novotny, R. Bruccoleri, and M. Karplus. 1985. Domain associa- 270:16903. tion in immunoglobulin molecules: the packing of variable domains. J. Mol. Biol. 23. Janeway, C. A., P. Travers, and S. Hunt. 1997. Immuno Biology, The Immune 186:651. System in Health and Disease. Garland Publishing, New York. 54. Raffen, R., L. J. Dieckman, M. Szpunar, C. Wunschl, P. R. Pokkuluri, P. Dave, 24. Rudiger, S., L. Germeroth, J. Schneider-Mergener, and B. Bukau. 1997. Substrate P. Wilkins Stevens, X. Cai, M. Schiffer, and F. J. Stevens. 1999. Physicochemical specificity of the DnaK chaperone determined by screening cellulose-bound pep- consequences of amino acid variations that contribute to fibril formation by im- tide libraries. EMBO J. 16:1501. munoglobulin light chains. Protein Sci. 8:509. 25. Flynn, G. C., J. Pohl, M. T. Flocco, and J. E. Rothman. 1991. Peptide-binding 55. Chapman, R., C. Sidrauski, and P. Walter. 1998. Intracellular signaling from the specificity of the molecular chaperone BiP. Nature 353:726. endoplasmic reticulum to the nucleus. Annu. Rev. Cell. Dev. Biol. 14:459.