Inefficient Assembly and Intracellular Accumulation of Antibodies with Mutations in V_H CDR2¹

Ig genes undergo successive changes in sequence through the physiologic process of somatic hypermutation (1). Thus, Igs provide a natural system in which to study the effects of somatic mutation on Ab function as well as on processes common to most proteins such as folding, assembly, transport, and secretion. As with most other oligomeric proteins, quality control mechanisms exist for retention and elimination of incompletely assembled Igs (2). At present, this quality control is not fully understood. Proteins that fail to pass quality control check points can be disposed of either by dislocation into the cytosol for degradation by cytoplasmic proteasomes (3), degradation in a postendoplasmic reticulum (ER)⁴ intermediate compartment, or via trafficking to lysosomes (autophagy) (4). Proteins exhibiting an improperly folded or assembled state that are not targeted for degradation may also be retained in the ER for extended periods of time (5).

We have a collection of 160 murine Ig mutants that have been produced by in vitro random mutagenesis to mimic somatic hypermutation; of these, 16 are secretion defective (6, 7). The basis for the secretion defect in these mutants is unclear. The secretion-defective mutants have from two to four amino acid replacements in the heavy (H) chain complementarity-determining region 2 (CDR2) or framework 2 regions of either the T15 or PCG1-1 Abs. In this report, we focus on the four T15 HCDR2 low secretor mutants designated M153, M164, M166, and M241. The HCDR2 of T15 is 19 amino acids in length (residues 50–65, numbering based on Kabat et al. (8)) and forms two-loop regions, consisting of residues 50–58 and 59–65. Mutant M153 has four changes that are all confined to the second loop of HCDR2 (Ser⁶⁰ → Thr, Ala⁶¹ → Val, Val⁶³ → Met, and Lys⁶⁴ → Thr). The other three mutants have at least one change in the first loop coupled with one or more replacements in the second loop, M164 (Arg⁵² → Ile and Tyr⁵⁹ → Ser), M166 (Asn^52a → Lys, Glu⁵⁸ → Gly, Ser⁶² → Tyr, and Val⁶³ → Met), and M241 (Ser⁵¹ → Arg, Asp⁵⁴ → Glu, and Tyr⁵⁹ → Ser). Previously we demonstrated that there was little association between the mutant H chains and the T15 light (L) chain based on ELISA (6). When these same mutations were placed in the HCDR2 of another Ab with an identical V_H gene sequence, but differing in HCDR3 sequence and L chain partner, secretion was not impaired. Thus the mutations that affect secretion do not completely destroy function of the HCDR2 (6). Furthermore, deleting four residues in HCDR3 restores secretion in these T15 mutants (9), again supporting the notion that the mutations in HCDR2 do not irreversibly destroy V region structure.

In the present study, we examined the t_1/2 of the four T15 HCDR2 mutant H chains in comparison with T15 wild type (WT). From this analysis, we exclude the possibility that the secretion defect is due to rapid H chain degradation. Rather, we demonstrate that the mutants are retained intracellularly with unusually long half-lives of 10 to 24 h. The site of intracellular accumulation appears to be the ER, based on association with the ER-resident chaperones, BiP and GRP94, and the failure of mutant T15 H chains to gain endoglycosidase H (endo H) resistance. Using metabolic labeling studies we show that only partial assembly of H chain occurs with L chain, while the majority of the mutant H chain pool is arrested at the H₂ and H₂L assembly intermediate stages. In addition, our data further suggest that a minor population of L chain is protected from rapid degradation by an interaction with H chain. This latter finding is novel and suggests that association with the H chain masks or impedes a degradation signal ordinarily expressed on the unpaired Vκ22 L chain.

SP2/0 cells and stable transfectants of these cells were cultured as described previously (6). The transfectants expressing T15 WT H chains with T15 L chains, T15 mutant H chains with T15 L chains, and T15 L chains alone were previously described (6, 10). To construct cells expressing only H chains, T15 WT and mutant VH in the pSV2gptS107γ2b plasmid construct (10) were transfected into SP2/0 (without L chain) using the lipofectin (Life Technologies, Grand Island, NY) method (11). Mycophenolic acid drug-resistant colonies were screened for production of intracellular H chain by ELISA (9).

Washed monolayers of SP2/0 transfectants from subconfluent 35-mm culture wells were depleted of intracellular stores of Met and Cys by 20-min incubation in DMEM without Met, Cys (Sigma Chemical Co., St. Louis, MO). Cells were then metabolically labeled in 0.4 ml labeling medium (DMEM without Met, Cys + 1% FCS + 0.4 mCi/ml [³⁵S] Express Protein Labeling Mix (DuPont NEN, Wilmington, DE)) for a 15-min pulse, except chaperone-binding assays, which were labeled for 4.5 h in 1.0 ml labeling medium with no subsequent chase. Labeling medium was removed and cells were washed in ice-cold PBS (except in BiP-binding experiments in which 130 mM NaCl, 20 mM Bicine (Sigma), pH 8.0, was used), then incubated in Iscove’s modified Dulbecco’s medium + 20% FCS for the duration of the chase. At the end of each chase, supernatants were collected and cells were washed. The monolayers were lysed on the plate in lysis buffer for 3 min, lysates were collected, iced 60 min, and then centrifuged to spin out the nuclei and debris. The lysis buffer used depended on the particular experiment. For the accumulation/degradation analyses: 0.025 M iodoacetamide, 20 μg/ml soybean trypsin inhibitor, 0.25% NP-40, 50 μg/ml PMSF, 1% sodium deoxycholate, and 0.1% SDS in PBS; for assembly and 2-D studies: 0.005 M iodoacetamide, 20 μg/ml soybean trypsin inhibitor, 0.5% Triton X-100, 50 μg/ml PMSF, 1% sodium deoxycholate, and 0.1% SDS in TSA buffer (0.01 M Tris-Cl, pH 8.0, 0.14 M NaCl, 0.025% NaN₃); and for chaperone-binding experiments: 100 μg/ml dithiobis [succinimidyl propionate] (DSP) (Pierce, Rockford, IL), 50 mM Bicine, 40 mM NaCl, 5 mM CaCl₂, 5 mM KCl, 10 mM Na₂MoO₄ · 2H₂O, and 1% NP-40, pH 8.0, as in Reference 12 .

Immunoprecipitations were performed by incubating samples with rabbit anti-mouse κ (Cappel Organon Teknika, Durham, NC) on ice for 60 min, followed by protein A-Sepharose 6MB (Pharmacia, Piscataway, NJ) overnight at 4°C to precipitate the anti-κ Abs as well as the T15 H chains. The anti-κ incubation was not included for all experiments (see figure legends). Immunoprecipitates were washed twice in the appropriate lysis buffer, once in TSA buffer, and then once in 0.05 M Tris · Cl, pH 6.8. Proteins were eluted from the pellet by the addition of 1× SDS-PAGE loading buffer (2% SDS, 10% glycerol, 60 mM Tris, pH 6.8, and 0.005% bromphenol blue) ± 0.1 M DTT, except for the assembly intermediates, in which proteins were first eluted with 0.2 M glycine · HCl, pH 2.5, followed by the same volume of 2× SDS-PAGE loading buffer. Proteins were heated 5 min at 100°C, electrophoresed on SDS/polyacrylamide gels using a 10- × 8-cm SE250 Mighty Small II apparatus (Pharmacia), or for the assembly study, a 16.5- × 22-cm slab gel (CBS Scientific, Del Mar, CA). Gels were subjected to fluorography with Enhance (DuPont NEN), dried, and visualized by autoradiography. Densitometric analyses were performed with NIH Image software on images scanned using a Hewlett/Packard ScanJet IIC.

The intracellular t_1/2 values were calculated from curve fit formulas of percentage of label vs time semi-log plots from individual pulse-chase/immunoprecipitation experiments (in which t_1/2 = 50% of the labeled proteins). In some instances, the t_1/2 extended past 26 h, the last time point, and in these cases a t_1/2 of 26 h was used for all calculations. An equal volume of 2× SDS-PAGE loading buffer (+DTT) was added to the samples.

To test the T15 H chains for sensitivity to endo H, cells were labeled with [³⁵S]Met, Cys and then immunoprecipitated as above through the last wash step. Pellets were resuspended in 25 μl 0.05 M sodium citrate, 0.1% SDS, pH 6.0, ± 100 mU/ml endo H (Boehringer Mannheim, Indianapolis, IN), and incubated at 37°C overnight.

Lysates from unlabeled cell cultures were subjected to immunoprecipitations as above. Proteins were separated by SDS-PAGE and transferred to polyvinyl difluoride membranes (Bio-Rad, Hercules, CA). For identification of GRP94, blots were probed with a 1:1000 dilution of rat anti-GRP94 (StressGen, Victoria, BC, Canada) followed by a 1:2000 dilution of an anti-rat IgG alkaline phosphatase conjugate (Sigma). After chemiluminescent substrate addition and band visualization, blots were stripped with 100 mM 2-ME, 2% SDS, and 62.5 mM Tris-HCl, pH 6.7, at 50°C for 30 min. Blots were washed and then reprobed with rabbit anti-BiP (anti-GRP78, StressGen) and binding was detected with a sheep F(ab′)₂ anti-rabbit IgG-alkaline phosphatase conjugate (Sigma). To confirm the identity of the H chain, blots were probed with rabbit anti-mouse IgG2b (Zymed Laboratories, South San Francisco, CA), followed by alkaline phosphatase-conjugated protein A (Sigma). All reactions were visualized using the AMPPD chemiluminescent substrate (Bio-Rad, Richmond, CA).

We previously found that intracellular levels of T15 WT and mutant H chains were similar, but that Ig secretion was impaired in the mutants (6). To assess the fate of the Ig proteins, cells were pulse labeled, chased for various times, and the H and L chains immunoprecipitated prior to analysis on reducing gels. As seen in Figure 1,A (top), T15 WT Ab proteins were visible in the supernatant by 2-h chase. The Ig bands in Figure 1,A were quantified by densitometry and depicted in Figure 1,B as the percentage of maximum intracellular label for each chain. The concurrent disappearance of T15 WT H and L chains from the lysate, coupled with their concomitant appearance in the supernatant, indicated efficient secretion, which was essentially complete by 4-h chase (Fig. 1, top). In contrast, M241 was not efficiently secreted. At 4-h chase, only 3% was detectable in the supernatant (Fig. 1, bottom). Intracellular H chain was detectable throughout the experiment, up to 26 h, while only low amounts of L chain were present at the later time points. The other mutants (M153, M164, and M166) gave comparable results (data not shown). Low levels of H and L chain bands were visible in some of the mutant supernatants, probably due to release from dying cells. These bands were not detected in all experiments and their appearance correlated with detection of the cytosolic enzyme lactate dehydrogenase (data not shown), a marker of cell lysis (13, 14).

The degree of carbohydrate modification of mutant and WT T15 H chains was assessed by endo H digestion as murine γ2b H chains possess a conserved N-linked carbohydrate addition site at position 297 (8, 15, 16). A pulse-chase/immunoprecipitation analysis at various times was performed on lysates (Fig. 2). By 6-h chase the T15 WT H chains from supernatants had gained endo H resistance, demonstrating that the N-linked oligosaccharide on the T15 H chain had undergone the expected enzymatic alterations in post-ER vesicles (Fig. 2, T15 WT). In contrast, the mutant M241 H chains remained endo H sensitive at 6-h chase, indicating that these H chains had not trafficked to the medial Golgi (Fig. 2). M241 was representative of the behavior of all four mutants (data not shown). Additionally, the T15 WT supernatant (SN) contained an H chain doublet characteristic of asymmetric O-glycosylation of murine IgG2b H chains, a process that occurs primarily in the Golgi (17).

Unassembled or mutant proteins that are retained in the ER associate with BiP and GRP94 for prolonged times compared with their unmutated, secreted counterparts (18, 19, 20, 21, 22, 23). Thus, we hypothesized that the mutant H chains would have an increased association with BiP and possibly other chaperones. Metabolically labeled cells were lysed in the presence of the cross-linking agent DSP, and the H chains immunoprecipitated with protein A-Sepharose 6MB. As shown in Figure 3, H and L chains were readily immunoprecipitated from WT and mutant H chain transfectants. The prominent bands near the 97-kDa and 66-kDa markers (Fig. 3,A) were identified as GRP94 and BiP, respectively, by Western blot analysis with anti-GRP94 and anti-BiP Abs (Fig. 3,B). Both the GRP94 and BiP bands were much stronger in the samples from the low secretor transfectants than from cells expressing T15 WT (Fig. 3,A). The bands were quantified by densitometry, and the relative intensities of the L chain, BiP, and GRP94 bands were normalized to those of the H chain for each lysate from two independent experiments. The levels of BiP and GRP94 associated with the mutant H chains were 7 to 20 times higher than with the T15 WT H chain. Increased levels of BiP were also seen in the absence of cross-linker (data not shown). Conversely, the levels of L chain were 2.5 to 5 times lower in the mutants than in the T15 WT. These data demonstrate that the mutant H chains have increased association with chaperones BiP and GRP94 but decreased association with L chain compared with T15 WT H chain. The major bands at ∼32 kDa and at the dye front in addition to the minor bands at ∼64 and ∼90 kDa have not been identified (Fig. 3 A). It is possible that the 32 kDa band is a H chain degradation product, as several H chains have been found to undergo acid hydrolysis under conditions typically used for SDS-PAGE (24).

A pulse-chase strategy was used to make quantitative comparisons between the intracellular residence time of the T15 WT H chain and each of the mutant H chains. Experiments were performed from two to four times for each H chain (see Fig. 1 for representative gel), and the percentages of labeled intracellular H and L chains remaining after various chase times were measured by densitometry. The results for a representative mutant, M241-are shown in Figure 4,A. It can be seen that for M241 the intracellular H chain disappears very slowly over time. Mutants M153, M164, and M166 H chains also displayed long intracellular residence times with 15 to 60% of H chain protein remaining even after 26-h chase (data not shown). The M153, M164, and M241 half-lives (24.0, 22.9, and 20.6 h, respectively) were significantly longer than the T15 WT H chain, t_1/2 = 2.6 h when secreted as intact Ig (Fig. 5 A). The mutant M166 showed a similar trend with a t_1/2 = 10.8 h, but did not reach statistical significance compared with WT H chain.

To determine whether the T15 WT H chain is intrinsically predisposed to a long intracellular life when denied an assembly partner or if the HCDR2 mutations uniquely affect the t_1/2 of the T15 H chains, we expressed the T15 WT and mutant H chains in the absence of L chains. T15 WT H chain was not secreted (data not shown) and was retained intracellularly (Fig. 4,B) in a manner similar to the mutants (Fig. 4,A). The t_1/2 of T15 WT H chain was significantly longer (t_1/2 = 13 h) than T15 WT in the presence of L chain (p < 0.05), and not significantly different from that of mutants M164, M166, and M241 (Fig. 5 B). M153 had a significantly longer t_1/2 as compared with T15 WT or the other three mutants, suggesting that HCDR2 mutations may have differential effects on degradation. Thus, neither the T15 WT nor the mutant H chains expressed in the absence of L chain are targeted for rapid degradation.

Since the t_1/2 values of three of the four mutant H chains were extended as compared with T15 WT, we examined the possible influence of HCDR2 on the intracellular t_1/2 of the L chains expressed in the low secretor mutants. Unlike most L chains, T15 L is not secreted in the absence of H chain (6, 25, 26). Free L chain was detected in lysates but not in supernatants for up to 26 h (27). However, turnover of intracellular L chain expressed alone was rapid (Fig. 4,C) with a t_1/2 of 1.3 h, demonstrating that without an H chain assembly partner the T15 L chain is targeted for degradation. Interestingly, the decay of L chain was biphasic in the presence of mutant H chains (Fig. 4,A). There was an initial rapid loss of the majority of L chain exhibiting t_1/2 values comparable with that of L chain alone (Fig. 5,C, phase 1). A second subpopulation of intracellular L chains (∼5–20%, phase 2) was degraded significantly slower than phase 1 (p < 0.001, Student’s two tailed, t test with all four mutants in each phase pooled). In the low secretors, L chain was detectable up to 26 h (Fig. 4,A and data not shown). Moreover, decay of this minor L chain population paralleled that of the mutant H chains (Fig. 4,A and data not shown; also compare Fig. 5,A H chains with Fig. 5 C phase 2 L chains).

To determine whether the long-lived L chain was covalently associated with mutant H chain we performed 2-D analysis on mutant M241. Immunoprecipitated samples were separated by nonreducing SDS-PAGE in the first dimension and reducing SDS-PAGE in the second dimension (Fig. 6). The presence of L chain covalently complexed with H chain is apparent in both the T15 WT and M241 transfectants at 2 h and 8 h (short open arrows), and is much longer lived than free L chain (short solid arrows), Fig. 6,A. The experiments above (Figs. 4 and 5) indicate that the presence of H chain is coincident with protection, these 2-D analyses suggest that the L chain is protected because it is complexed with H chain. Interestingly, M241 H chain does not resolve into distinct assembly forms at 2- and 8-h time points as compared with T15 WT H chains (Fig. 6,A). As a control to verify the migration pattern of free and complexed Ig chains, secreted T15 WT Ig was analyzed by 2-D electrophoresis (Fig. 6,B). In the left panel, the open arrow indicates the position of complexed L chain after release from H chain in the second dimension. The electrophoretic mobility of free L chain (Fig. 6 B, right, short closed arrow) was determined by separating secreted T15 WT Ig under reducing conditions in both the first and second dimensions.

To define the extent of oligomerization of the H and L chain complexes seen in the 2-D analysis, the assembly intermediates of the T15 WT and mutant Igs were compared. Assembly intermediates immunoprecipitated from lysates of metabolically labeled cells were analyzed by nonreducing SDS-PAGE. In the T15 WT (Fig. 7), four bands corresponding to H₂L₂, H₂L, H₂, and H were detected (monomeric H chain was barely visible in this gel at 0.25 h). The H₂ intermediate was strongest at 15 min, still present at 1 h, and undetectable by 2-h chase (Fig. 7). The H₂L band was present at all time points and strongest at 1-h chase in the T15 WT lysate. There was a very faint H₂L₂ band at 15 min, which increased in intensity at 1- and 2-h chase times. This band was also noted in T15 WT SN at 6-h chase. The presence of H chains in these bands was confirmed by Western blot (data not shown). In addition, proteins from cells expressing only T15 WT H chains rapidly assembled to the H₂ intermediate (by 0.25-h chase), but the two higher m.w. bands could not be detected (data not shown), consistent with these latter species being L chain-containing oligomers. These data indicate that the T15 WT Ig follows an H₂ → H₂L → H₂L₂ assembly pathway.

The assembly intermediates of the secretion-impaired mutants were examined at 1-h and 6-h chase as shown in Figure 7. An H₂ band was readily visible, and an H₂L band was seen at both time points for all four mutants. The migration of mutant H₂ and H₂L bands was slightly faster than T15 WT, suggesting that mutant H chain conformations are altered. These differences in migration were not observed under reducing SDS-PAGE (see Fig. 3). In addition, mature Ig was not observed as indicated by lack of detectable H₂L₂ bands in the mutants up to 6-h chase. As noted above for T15 WT, the mutant H chains expressed in the absence of L chain formed H₂ intermediates, but lacked bands migrating at positions representing H₂L and H₂L₂ (data not shown). From these results, it was evident that the mutant H chains formed homodimers and that some mutant H chains assembled with L chain, although this interaction had progressed only to the H₂L intermediate by 6-h chase.

Ig CDR regions are targets of the somatic hypermutation process. Despite extensive amino acid variation within these regions, the overall conformations of the CDRs are highly conserved, forming only a small set of canonical structures (28, 29, 30). A prevailing view regarding this structural conservation is that it is required to maintain Ag-binding functions. However, we have shown that random mutations in the V_H of murine anti-phosphocholine Abs may have deleterious effects on protein secretion as well as Ag binding (6, 7). In this study, we investigated the intracellular residence kinetics of mutant H chains and coexpressed T15 L chains in order to delineate the mechanisms of the secretion defect. One possible explanation for the secretion incompetence would be that the mutant H chains are targeted for rapid degradation by the ER quality control system. We found that this is clearly not the case for the HCDR2 mutant Abs. Rather, the mutant H chains displayed extremely long intracellular residence times. In contrast, the bulk of the T15 L chain underwent rapid degradation with essentially the same kinetics as L chain expressed alone. A small amount of L chain (5–20% of total) was protected from rapid degradation and was shown to be associated with mutant H chain.

T15 WT Ab was secreted from stable transfectants within 2 to 4 h, consistent with the secretion kinetics of other Ab-producing cell lines in which Ab appears in the supernatant from 20 to 150 min after protein synthesis (31, 32). In contrast to T15 WT, the mutant H chains were retained intracellularly (independent of L chain coexpression), with a significant fraction of labeled H chain remaining up to 26 h chase. These data suggest that the mutations do not introduce a rapid degradation signal, which would prevent secretion. The T15 WT H chain showed similarly slow kinetics of disappearance when expressed without an L chain partner, compatible with studies that demonstrate Ig H chains are not secreted in the absence of L chains (33, 34, 35) and that such H chains have extended half-lives (20). Our finding that the mutant H chains are not rescued by Vκ22 L chain, implies a lack of proper assembly with L chain. However, we cannot rule out a mechanism in which the mutations cause transport to an area in the ER, which is mostly devoid of L chains. The fact that some H₂L complex is formed argues against this explanation and demonstrates that some overlap in distribution likely exists.

The continued sensitivity of the mutant H chains to endo H digestion confirmed that by 6-h chase these proteins had failed to reach the medial Golgi (36), suggesting retention in the ER. Other analyses of nonsecreted Ig proteins have implicated the ER as the site of accumulation and/or degradation of impaired H and L chains, and many of these studies demonstrate a prolonged interaction with BiP, an ER-resident molecular chaperone (20, 21, 37, 38, 39, 40). Although BiP binds transiently to native chains of many proteins, its binding is enhanced with misfolded, mutant polypeptides, leading to extended retention time in the ER and eventual degradation of the mutant proteins (12, 25, 41, 42). Ig L chains shown to be stably retained by BiP mutants were unable to reach their native conformation suggesting that Igs may not complete essential folding steps until after release from BiP (43). GRP94 is another ER-resident chaperone that associates with unassembled Ig chains (44, 45). The interactions between Ig chains and these two chaperones, BiP and GRP94, are thought to play integral roles in coordinating Ab assembly (12, 23, 46, 47, 48). The finding that H chains from the low secretor mutants could be covalently cross-linked with more BiP and GRP94 than the T15 WT chain was consistent with the inability of the mutant H chains to integrate into transport competent moieties. At present it is unclear whether the mutations directly alter chaperone association as the T15 WT H chain also shows increased BiP and GRP94 association when expressed in the absence of L chain (T.M. Martin and G.D. Wiens, unpublished observations).

Upon examination of the assembly intermediates we found that the T15 WT Ab followed an H₂ → H₂L → H₂L₂ assembly pathway. The H chains formed homodimers rapidly, as the H₂ band was more intense than the H chain monomer immediately after the pulse label (Fig. 7). This observation also held true for the mutant H chains, in which a strong H₂ band was seen at all time points including 0 min chase (Fig. 7 and data not shown) suggesting that a large fraction of the labeled H chains did not progress past this point in assembly, regardless of the presence of L chain. This finding is consistent with an early report showing that full length, mutant H chains dimerize rapidly but persist in the cytoplasm as stable proteins (49). Some assembly to the H₂L intermediate occurred in the low secretor mutants, however, the mutants did not form discrete H₂L₂ oligomers.

When the rates of disappearance of intracellular L chains from the low secretors were examined, it was noted that the curves were biphasic (Fig. 4). The first phase accounted for the rapid degradation of ≥80% of the L chains, with t_1/2 values of 1 to 2 h consistent with other nonsecreted L chains (37). A long-lived minor portion of L chain was present at late chase times in the cells expressing mutant H chains. The decay of this second fraction of labeled L chain paralleled that of the mutant H chains present in the same transfectants and had essentially the same t_1/2 values as the mutant H chains. The observations are consistent with the detection of H₂L bands in the assembly intermediates. Based on the 2-D gel analysis of mutant M241, it is likely that the L chains present in H₂L oligomers are those detected in the second phase, i.e., with long half-lives, indicating that they may be relatively stable over time. One possible explanation for the extended t_1/2 for the minor population of L chain is that a degradation signal ordinarily expressed on the unpaired Vκ22 L chain is masked by association with the H chain. Alternatively, a degradation signal may be constitutively present, but the covalent oligomerization with dimeric H chain is able to prevent transport to the site of degradation. Thus, once the H₂L complex has formed, the fate of the L chain appears to be determined by the H chain unless a second L chain can be added to the complex to allow entry into the secretory pathway. This would have the effect of controlling the formation of secretable complexes. It is intriguing to speculate that the mutant H₂L species exhibits a conformation that prevents interaction with another L chain (or inability to dissociate from chaperones). In addition, a noncovalent assembly of H₂L + L → H₂L₂ may be formed but lost during immunoprecipitation due to weak binding with the second L chain, a possibility that cannot be excluded from the present data.

The assembly patterns of Abs were elucidated more than 20 years ago and follow discrete isotype-specific steps (32, 50, 51). The analysis described here was performed on the T15 WT and mutant V_H domains linked to the murine γ2b C region expressed as stable transfectants of SP2/0 (10). According to published reports on IgG2b assembly, this particular murine isotype typically assembles an HL intermediate in the major assembly pathway, with H₂ usually present in the minority of assembly intermediates (52, 53). Our data demonstrated that the combination of T15 WT VH/γ2b + Vκ22/Cκ proteins assembled via an H₂ homodimer intermediate. Moreover, H chain dimerization occurred rapidly, as it was seen at the earliest time points when the monomeric H chain was only faintly detectable. To our knowledge, this is the only known example of an IgG2b that does not appear to assemble HL heterodimers into H₂L₂ oligomers, with the exception of the aberrant H chains present in the IgG2b-producing MPC 11 tumor (54). It is interesting to note that HL heterodimers do form in the MPC 11 tumor, but they are noncovalently associated, do not participate in the formation of H₂L₂, and actually get secreted from the cell (54). The inconsistency between the findings presented here and other reports of IgG2b assembly, i.e., lack of HL heterodimers, suggests an unfavorable interaction between the nascent H and L chains. Possibly this unique combination of T15 V_H and L chain may somehow disfavor HL formation in IgG2b oligomerization. It was shown previously that the T15 L chain does not compete for association with T15 H chain as well in vitro as the Vκ21 L chain (55). This is consistent with our data suggesting that the T15 H chains may favor H₂ formation over HL. We must also note that the original T15 Ab uses an α H chain, whereas we have expressed it here as a γ2b, which is not unique as other T15 idiotype-positive γ2b Abs have been isolated (56, 57). However, in the reassociation experiments in vitro, the competitive advantage was determined primarily by the Vκ gene segment (55), and other studies in different Ab systems showed that preferential reassociation was not influenced by differences in C_L (58, 59), consistent with the notion that V_H-V_L interactions play a critical role in HL association. Another possible explanation for the lack of HL intermediates lies in the stoichiometry of H and L chains produced in these stable transfectants. Ab-producing cells commonly synthesize an excess of L chain (50) which may drive HL interaction. We do not see an excess of L chain compared with H chain produced from transfected genes. The assembly pathway of T15 H chains with endogenously expressed L chains, which would likely be in excess of the transfected H chains, has yet to be examined.

In summary, biochemical analysis of the T15 low secretor mutants demonstrates that 1) the mutants fail to gain endo H resistance and are associated with chaperones suggesting ER retention, 2) the mutants fail to assemble efficiently into H₂L₂ oligomers explaining the secretory defect, and 3) the mutant H chains and associated L chains exhibit prolonged half-lives. Collectively, these data demonstrate a novel consequence of mutation in HCDR2 and suggest that these mutants will be useful for further elucidation of the mechanisms of H and L chain assembly and degradation.

We thank Dr. Caroline Enns for critical discussions and advice, as well as M. Brown, K. Heldwein, J. Hill, and Drs. E. Whitcomb, T. O’Hare, and M. Stenzel-Poore for critical review of the manuscript.