N-Linked carbohydrates are frequently found in the V region of Ig H chains and can have a positive or negative effect on Ag binding affinity. We have studied a murine anti-α(1→6) dextran VH that contains a carbohydrate in complementarity-determining region 2 (CDR2). This carbohydrate remains high mannose rather than being processed to a complex form, as would be expected for glycans on exposed protein loops. We have shown that the glycan remained high mannose when murine-human chimeric Abs were produced in a variety of cell types. Also, when another carbohydrate was present in CDR1, CDR2, or CDR3 of the L chain, the VH CDR2 glycan remained high mannose. Importantly, we found that when the anti-dextran VH CDR2 replaced CDR2 of an anti-dansyl VH, the glycosylation site was used, but H chains were withheld in the endoplasmic reticulum and did not traffic to the Golgi apparatus. These results suggest that inappropriate V region glycosylation could contribute to ineffective Ab production from expressed Ig genes. In some cases, a carbohydrate addition sequence generated by either V region rearrangement or somatic hypermutation may result in an Ab that cannot be properly folded and secreted.
All Igs are glycoproteins, with N-linked oligosaccharides attached in the Fc region. Human IgG contains one carbohydrate at Asn297 in the CH2 domain; however, the average number of carbohydrates on serum IgG is 2.8 (1), reflecting the fact that carbohydrate is frequently attached to the V region of the Ab. The consensus sequence for attachment of carbohydrate is: Asn-X-Ser/Thr, where X may be any amino acid except proline (and if Ser is in the third position, may only rarely be Asp, Glu, Trp, or Leu) (2, 3, 4, 5). Although only a few germline V region genes contain the sequon in complementarity-determining region 2 (CDR2)4 or framework region 3 of the H chain (6), it may be generated during the process of affinity maturation, when somatic hypermutation of V region genes occurs (7). Studies have shown that carbohydrate in the Ag-binding site can influence the affinity of the Ab-Ag interaction (8, 9). Carbohydrate found at the Ag-binding interface may increase affinity by increasing hydrophilic interactions between Ab and carbohydrate epitopes; alternatively, it may decrease affinity through stearic hindrance of the interaction of the Ab with Ag.
Asparagine- or N-linked carbohydrates are added to nascent proteins as they enter the lumen of the endoplasmic reticulum (ER). A dolichol (lipid)-linked precursor carbohydrate consisting of a branched GlcNAc2Mann9Glc3 structure is transferred from the ER membrane-bound dolichol pyrophosphate to a target asparagine residue. It then undergoes enzymatic modification by glycosidases and glycosyltransferases in the ER and Golgi apparatus. Mature carbohydrates may be high mannose, hybrid, or complex. High mannose glycans undergo the least processing, yielding a trimmed version of the precursor, with only terminal mannose units. Hybrid carbohydrates are trimmed to a greater extent and contain some terminal mannose residues, as well as other terminal sugars. Complex carbohydrates are processed to the greatest degree as they migrate through the ER and Golgi complex. They often contain terminal fucose and/or sialic acid molecules and may take one of several different structural forms: biantennary, branched, or bisected.
The factors governing the extent of carbohydrate processing at each site on a glycoprotein have not been fully elucidated. It has been proposed that glycans located on exposed protein loops, rather than sequestered within β-pleated sheets, will be more highly processed, because they are more accessible for enzymatic modification (10). However, additional factors such as primary sequence information, which might regulate the degree of processing, remain unclear at this time.
The murine anti-α(1→6) dextran VH gene 14.6b.1, used in previous investigations of V region glycosylation, was found to contain a complex N-linked carbohydrate at Asn58 in CDR2 (9). The VH gene from another hybridoma, 19.22.1, is identical except that it contains Asn at position 60 instead of Thr and lacks the glycosylation sequon. Site-directed mutagenesis of CDR2 (Lys62→Thr) of 19.22.1, resulted in a glycosylation site at Asn60 (11). Interestingly, the carbohydrate attached at Asn60 was high mannose. Asn58 and Asn60 are near each other on the exposed loop of CDR2, and consequently, the glycans attached to both would be expected to be complex. That they differ suggests that other factors in addition to surface exposure play a role in determining the extent of processing of carbohydrate precursors.
Using the V region gene of the 19.22.1 mutant as a model (referred to here as Asn60 VH), we addressed the question of whether processing of the glycan attached to Asn60 is influenced by its context. The environment was altered in several ways. The Asn60 VH H chain was expressed with three anti-dextran L chains containing carbohydrate in their CDRs, and in three different cell lines. The CDR2 sequence of the Asn60 VH was also used to replace CDR2 of a murine V region gene 27.44 (12) specific for the hapten dansyl, to determine whether the V region context would influence the processing of the carbohydrate within VH CDR2. We found that the carbohydrate attached to Asn60 within the anti-dextran V region was, regardless of cell type or surrounding environment, processed only to the high mannose form. Surprisingly, we found that placing the CDR2 from the Asn60 VH in the anti-dansyl VH resulted in an Ab that was not secreted. This was shown to be a consequence of the presence of the V region carbohydrate, because aglycosylated Abs produced in the presence of tunicamycin (Tm) were secreted.
Materials and Methods
Mutagenesis and cloning of anti-dextran Abs
The murine anti-α(1→6) dextran VH chain gene obtained from hybridoma 19.22.1, mutated to bear carbohydrate at Asn60 (11), was amplified by PCR as a NotI-NheI insert to be ligated with the NheI-Not GCGGCCGCCCACCATGGAATGGAGCGGG-3′ and 5′-CCCGCTAGCTGCAGAGACAGTGACCAGA-3′, encoding NotI and NheI sites in bold, respectively. All primers used in this study were prepared by Life Technologies/BRL (Grand Island, NY).
A hybrid VH was made by preparing two cassettes, which together encoded a VH anti-dansyl gene with a CDR2 sequence from the anti-dextran Asn60 VH. The murine 27.44 anti-dansyl V region gene (12EcoRV-ScaI fragment and the second was a ScaI-NheI fragment, and the hybrid VH was assembled by cutting with ScaI and ligating the two VH fragments. The hybrid VH was then placed as an EcoRV-NheI fragment in the human IgG1 expression vector described above. The primers used were as follows: 5′-GGGGATATCCACCATGTACTTGGGACTGAAC-3′ and 5′-GGGAGTACTACCACTTCCAGGTAAAATTTCAGCAACCCACTCAAGC-3′; and 5′-GGTAGTACTAACTACAACGAGACGTTCAAAGGGAGGTTCACCATCTCAAGAG-3′ and 5′-GGGGCTAGCTGCAGAGACAGTGACCAGAGT-3′, with EcoRV, ScaI, and NheI sites in bold. Thus, the following amino acid sequence (from the anti-dextran Asn60 VH) replaced aa 50–65 of the anti-dansyl VH: EILPGSGSTNYNETFKG, with the N-linked glycosylation sequence underlined.
A murine/human chimeric anti-dansyl κ L chain expression vector was constructed by first removing the hybridoma 27.44 κ C region gene from vector pSV184ΔHneo-Huκ (13) and placing it into pcDNA3.1(−). PCR using the template 27.44 Vκ gene as template yielded most of the V region and the beginning of the C region as an EcoRV-BbsI fragment. This partial Vκ/partial Cκ product was joined to the remainder of the Cκ gene in pcDNA3.1(−). The PCR primers used were as follows: 5′-GGGCTAGCGATAT-3′ and 5′-ACTGTGGCTGCACCATCTGTCTTCATCTTC-3′, with EcoRV and BbsI sites, respectively, in bold (the EcoRV site is complete when joined with a cytosine in the template).
Previously engineered mutants of the VL anti-dextran, coding for N-linked glycan addition in each of the CDRs (8), were inserted as EcoRV-HindIII fragments into the pcDNA3.1(−) vector containing the κ C region gene.
Electroporation of murine Sp2/0 cells was performed as previously described (8) with 10 μg each of H chain and L chain expression vectors, and 10 μg of pSV2his for selection, all linearized using PvuI. Selection was performed with 5 mM histidinol, and clones were screened for Ab expression by ELISA, as described (8, 14).
Lipofection was performed as described previously (15) with 10 μg of uncut H and/or L chain vector per tissue culture dish containing ∼5 × 106 6 cells for screening by biosynthetic labeling and immunoprecipitation of cell lysates, followed by SDS-PAGE. Cell populations selected by either ELISA or SDS-PAGE were subcloned by limiting dilution, and the clone producing the highest amount of Ab was selected.
Transient transfection of human kidney epithelial 293T cells was performed using reagents from 5 Prime-3 Prime (Boulder, CO) or from Edge Biosystems (Gaithersburg, MD) following the manufacturers’ protocols. Briefly, 293T cells were grown to 60% confluence in 100-mm tissue culture dishes, and the reagent mix containing 3 μg each of H and L chain vectors was added to cells in 10 ml of fresh IMDM plus 5% calf serum. After a minimum of 4 h of incubation, cells were washed, and then incubated in IMDM plus 5% calf serum overnight before biosynthetic labeling.
Biosynthetic labeling and immunoprecipitations
Transfectant cells were washed twice with sterile PBS and grown overnight at ∼106 cells/ml in DMEM lacking methionine and cysteine (Life Technologies/BRL), supplemented with 1× GlutaMax glutamine analog (Life Technologies/BRL) and EasyTag [35S]methionine/[35S]cysteine mix (Amersham, Arlington Heights, IL) at 25 μCi/ml. Supernatants were collected and cytoplasmic lysates were prepared and immunoprecipitated, as previously described (16), using rabbit anti-human Fab (both κ and λ L chain specific) and rabbit anti-human IgG, both prepared in the laboratory, or rabbit anti-human Fab and rabbit anti-human IgA (Sigma-Aldrich), followed by a 10% Staph A solution (IgSorb; The Enzyme Center, Malden, MA). Samples were then analyzed by SDS-PAGE under nonreducing conditions on 5% phosphate gels or under reducing conditions on 12.5% Tris-glycine (TG) gels, as described (16). Samples prepared for Endoglycosidase H (Endo H) analysis were resuspended in Endo H buffer (see Endo H hydrolysis) rather than electrophoresis sample buffer.
Tm treatment of transfectants
Cells were preincubated with fresh IMDM containing 5% calf serum and 8 μg/ml Tm for ∼3 h. Cells were washed and then resuspended in labeling medium, as above, with Tm at 8 μg/ml. Supernatants were collected for immunoprecipitation after overnight incubation. For cytoplasmic lysates, cells were preincubated with Tm for 30 min, washed, incubated in labeling medium with Tm for 3 h, and then lysates were prepared.
Endo H hydrolysis
After immunoprecipitation, Staph A pellets with bound, labeled Abs were resuspended in 100 μl of Endo H buffer containing 50 mM sodium citrate (pH 5.5), 2 mM PMSF, and 100 mM 2-ME. A volume of 25–40 μl of each sample, prepared from the supernatants or cell lysates of ∼1–2 × 106 cells, was incubated overnight at 37°C with 6–9 U of Endo H, which cleaves the bond between the two proximal N-acetylglucosamine (GlcNAc) residues of high mannose, but not complex carbohydrates (17). Duplicate control samples were incubated without enzyme. Sample buffer (5×) (125 mM Tris (pH 6.7), 1.5% SDS, 50% glycerol, and 20 μg/50 ml bromphenol blue) was added to each tube, and samples were placed in a boiling water bath for 3 min to elute labeled molecules before analysis by SDS-PAGE. Ten to 20% gradient TG gels were purchased from BioWhittaker Molecular Applications (Rockland, ME).
Immunofluorescent staining and confocal microscopy
The wells of eight-chamber Permanox slides (Nalge Nunc, Naperville, IL) were seeded with adherent cells in IMDM containing 5% FBS and incubated overnight. Cells were gently washed with PBS and then fixed in a freshly prepared solution of 0.01 M sodium metaperiodate, 0.075 M lysine, 0.0375 M NaH2PO4 (pH 7.4), and 2% paraformaldehyde for 2–3 h at room temperature. Wells were washed three times with a solution of 0.5% OVA (Sigma-Aldrich) in PBS (pH 7.4) (buffer A) for 10 min. Cells were then permeabilized by incubation with buffer A containing 0.05% saponin (Sigma-Aldrich) (buffer B) for 5 min at room temperature. Primary Abs were diluted in buffer B: rabbit anti-major ER glycoprotein (MERG; 1/30 dilution; a generous gift from Dr. D. Meyer (University of California, Los Angeles, CA)), rabbit anti-Golgi β-coatomer protein (β-COP; 1/75 dilution; Affinity Bioreagents, Golden, CO), and goat anti-human IgG (1/75 dilution; Zymed, South San Francisco, CA). Abs were added to relevant wells, slides were incubated with shaking overnight at 4°C, and the cells were later washed three times with buffer B. Texas Red-conjugated swine anti-goat IgG (EY Laboratories, San Mateo, CA) and FITC-conjugated swine anti-rabbit IgG (Nordic Immunology, Tilburg, The Netherlands) were diluted in buffer B at 1/25 and 1/100, respectively, and centrifuged to remove Ab aggregates, which bind to cells directly, increasing background fluorescence. The cells were stained with secondary Ab at room temperature for 1 h, washed three times with buffer B and one time with PBS, and chambers were removed from each slide. A drop of Prolong (Molecular Probes, Eugene, OR) was placed over each group of cells to reduce quenching of dyes, and then a coverslip was overlaid and sealed to the slide. Data were collected using a Nikon (Melville, NY) Eclipse E800 system with a Bio-Rad (Hercules, CA) MRC1024ES confocal microscope equipped with a krypton/argon laser.
Portions of the V regions used for the present studies are depicted in Tables I⇓ and II⇓. The different VH genes include the murine anti-α(1→6) dextran gene 19.22.1, a 19.22.1 mutant with a Lys62Thr substitution within CDR2 (referred to as Asn60 VH), the anti-dansyl gene 27.44, and 27.44 with the CDR2 sequence from Asn60 VH. This CDR2 was exchanged in place of CDR2 in the anti-dansyl VH by PCR mutagenesis and cloning, creating a hybrid VH. All VH were joined to human IgG1 and expressed in pcDNA3.1(−) under the control of the CMV promoter.
The VL of the murine anti-dextran hybridoma 19.22.1 was mutated in each of the three CDRs to introduce single N-linked glycosylation sites (8). Each of these three mutants, as well as wild-type anti-dextran VL and hybridoma 27.44 anti-dansyl VL genes, was joined to a human κ C region.
Crowding of the V region in an anti-dextran Ab does not alter carbohydrate processing
To determine whether the presence of multiple glycans within the Ag binding site of the anti-dextran would influence carbohydrate processing, 293T cells were transiently cotransfected with the Asn60 VH H chain and anti-dextran κ L chain expression vectors. Wild-type aglycosylated, or mutant κ L chain genes with glycosylation sequences in each of the CDRs, were used (see Table I⇑). Abs with both H and L chain V region glycosylation sites should contain carbohydrates in close proximity within the folded protein, given the large size of the glycans. Transfected cells were biosynthetically labeled with or without Tm, an inhibitor of N-glycosylation. IgG4 Asn60 VH Abs used in earlier studies were included as a positive control for Endo H digestion. Secreted Abs were immunoprecipitated, and then incubated overnight under reducing conditions with or without Endo H and analyzed on 12.5% SDS-PAGE gels (Fig. 1⇓). H chains expressed with wild type and all of the mutant L chains were susceptible to Endo H cleavage, indicating that H chain carbohydrates at Asn60 remained high mannose in all cases. The carbohydrates attached to Asn28 and Asn50 in the L chains were processed to complex form, demonstrated by identical migration of bands with or without Endo H treatment. Judging from the band intensity, it appears that almost all Asn28 residues were used for carbohydrate addition (Fig. 1⇓, lanes 6 and 7), whereas a smaller percentage of Asn50 sites were occupied (lanes 10 and 11), with bands representing aglycosylated and glycosylated L chains almost equal in intensity. VL Asn91 in the CDR3 was consistently glycosylated, with only one L chain band seen in Fig. 1⇓, lane 13. However, the Asn91 glycan appeared to be variably processed, because two bands were visible below 30 kDa in Fig. 1⇓, lane 14, with a minority of the L chains susceptible to Endo H. The L chain of the IgG4 control Ab (Fig. 1⇓, lanes 1 and 2) appears smaller, because it contains murine rather than human C regions.
The cell line used for expression does not influence the processing of the Asn60 VH carbohydrate
Vectors for the Asn60 VH IgG1 H chain and anti-dextran κ L chain were expressed in human kidney epithelium 293T cells, murine myeloma Sp2/0 cells, and CHO cells. Transfectants were biosynthetically labeled in the presence or absence of Tm. Secreted Abs were immunoprecipitated and a glycosylated sample from each transfectant was incubated overnight with Endo H, which cleaves the high mannose form of N-glycans from glycoproteins. Reduced samples were analyzed by SDS-PAGE (Fig. 2⇓). An anti-dansyl IgG1 lacking V region carbohydrate produced in Sp2/0 cells was included as a marker of H chain size and as a negative control for Endo H digestion. The size of the aglycosylated H chains produced in the presence of Tm (Fig. 2⇓, lanes 3 and 8) can be compared with that of H chains with both VH and Fc glycans (lanes 4, 6, and 9) and with only one glycan in the Fc region after Endo H treatment (lanes 5, 7, and 10). In every case, the H chain carbohydrate at Asn60 is susceptible to Endo H digestion, indicating that it is high mannose in form. Thus, there do not appear to be differences in oligosaccharide processing of this glycoprotein in these cell lines.
The presence of CDR2 from Asn60 VH in an anti-dansyl VH inhibits Ab secretion
It is unclear why the exposed VH carbohydrate at position 60 in CDR2 remains in the high mannose form, whereas a carbohydrate attached nearby at position 58 is processed to complex form. To further explore the contribution of surrounding sequences to glycan processing, we created a hybrid VH in which CDR2 of the Asn60 VH, with its N-linked glycosylation site, replaced CDR2 of an anti-dansyl VH, which bears no glycan (see Table II⇑). A Pro-5 cell line stably expressing anti-dansyl κ L chain was transfected with the hybrid VH IgG1 expression vector. Interestingly, when transfectants were biosynthetically labeled, it was discovered that no Ab was present in the secretion. To determine whether the H chain was made and whether the carbohydrate attachment site was used, cells were biosynthetically labeled with and without Tm treatment. Both secretions and cytoplasmic lysates were immunoprecipitated, reduced, and analyzed on a 12.5% SDS-PAGE gel (Fig. 3⇓). It was apparent that H chain is synthesized by these transfectants, but is not secreted (Fig. 3⇓, lanes 1 and 2, respectively). However, when glycosylation was inhibited by growing cells in the presence of Tm, H chain was found in both lysates and secretions, indicating that Ab lacking N-linked glycans is secreted (Fig. 3⇓, lanes 5 and 6) and that it was not the amino acid changes per se that interfered with secretion. The attachment site at Asn60 appeared to be used, as shown by the greater molecular mass of the band in Fig. 3⇓, lane 1, compared with the anti-dansyl IgG1 H chain in lane 3. The intracellular H chain is identical in size to secreted Asn60 VH IgG1 H chain, which bears carbohydrates in the V region and in the Fc (Fig. 3⇓, lane 4). Therefore, it seems that the presence of V region glycan in the hybrid Ab inhibits its ability to be secreted.
Hybrid VH IgG1 H chains do not traffic to the Golgi apparatus
To determine the intracellular location of hybrid VH Abs, transfectants expressing either the VH region hybrid or wild-type anti-dansyl Abs were examined using confocal microscopy. Cells grown on microscope slides were stained with goat anti-IgG and rabbit antiserum specific for either an ER glycoprotein or the Golgi β-COP, followed by secondary Abs conjugated with either Texas Red or FITC (Fig. 4⇓). Wild-type IgG1 colocalized with the ER (Fig. 4⇓C) and with the Golgi apparatus (F), whereas the hybrid Ab localized to the ER (I), but not to the Golgi apparatus (L). As has been observed with other misfolded proteins, it appears that the hybrid Ab is withheld in the ER and does not proceed any further along the secretory pathway.
Previous studies from this laboratory using a murine V region specific for α(1→6) dextran have shown a role for V region glycosylation in Ab affinity, half-life, and immunogenicity (8, 9, 11). Abs differing in the location of a glycan addition site within CDR2 were shown to contain carbohydrates that were processed differently. The VH carbohydrate at Asn58 of hybridoma 14.6b.1, within the naturally occurring Asn58-Tyr59-Thr60 sequon, is complex. This is consistent with the observation that carbohydrates located on exposed loops, such as CDR2, are accessible to glycosidases and glycosyltransferases as proteins migrate through the ER and Golgi apparatus, and therefore tend to be processed to a greater extent than those that are shielded or sequestered (10). Surprisingly, when a carbohydrate addition site was introduced into the same VH at position 60, the glycan was located on the surface of the molecule (11) but was found to remain high mannose. Therefore, stearic accessibility to processing enzymes during protein maturation (18) may not be solely responsible for the degree of glycan processing, and structural information near an addition site may influence enzyme activity. For example, differences in glycan processing between closely related strains of influenza virus (19, 20) and murine leukemia virus (21) have indicated that primary sequence may play a role in directing oligosaccharide processing. Local conformation and domain interactions have been reported to contribute to processing at individual glycosylation sites on soluble variants of rat and human CD4 expressed in CHO cells (22). In this study, we have investigated glycan processing by expressing the anti-dextran Asn60 VH in different contexts.
Expression of carbohydrate-processing enzymes is species and tissue specific (23, 24). To address the issue of possible cell line- and species-specific differences in glycan processing, anti-dextran Asn60 VH-IgG1 was produced in the murine myeloma Sp2/0, in CHO cells, and in human kidney epithelial 293T cells. We found no evidence for either species-specific or cell type-specific differences in processing of the Asn60 VH carbohydrate (Fig. 2⇑).
It has long been known that not all N-linked carbohydrate addition sites are used (2, 25, 26), and that some sites may be variably occupied (27, 28, 29). It has also been observed that processing at one site may be affected by the presence of a glycan at another site. For example, tissue plasminogen activator is a protease glycoprotein consisting of five protein domains, with two main classes of glycosylation variants. Type I has three N-linked carbohydrates, at Asn117, Asn184, and Asn448, whereas type II has only two, at Asn117 and Asn448. Fine structural analysis showed that the glycan at site 448 is different if glycan is present at site 184 (30). When we expressed the anti-dextran Asn60 VH with the C region genes for human IgA1, which has two exposed N-linked glycans in the Fc and O-linked carbohydrates in the hinge, or with IgA1 mutated to lack N-linked Fc glycosylation sites, we found that both IgA1 Abs bore high-mannose VH carbohydrate (data not shown). This indicated that neither isotype sequence differences between IgA1 and IgG1 nor different Fc region glycans caused detectable differences in Asn60 glycan processing. We also found that the isotype of the L chain (κ vs λ) did not influence the processing of the carbohydrate at Asn60 (data not shown), although the L chain isotype has been found to influence some of the properties of IgG (31).
Our results showed that crowding of the V region did not alter carbohydrate processing. When Abs possessed oligosaccharides in the V region of both H and L chains, the H chain carbohydrates remained in the high mannose form, whereas L chain carbohydrates were processed almost exclusively to complex form (Fig. 1⇑), consistent with results obtained in an earlier study (8). It is interesting that the presence of two glycans in close proximity did not alter their processing. The dimensions of an Ag-combining site are ∼22 × 32 Å (32). N-Linked carbohydrates are very large, and the length of just their first three sugar residues (GlcNAc, GlcNAc, and mannose) is 16 Å (33); therefore, two carbohydrates would occupy a relatively large area of the protein surface. It should be noted that this study merely addressed whether the carbohydrates were complex or high mannose. It is still possible that smaller structural differences exist between crowded and uncrowded states.
Anti-dansyl H chains do not normally bear V region glycans. When Asn was substituted for His55 in an anti-dansyl VH, creating the consensus sequence Asn55-Ala56-Thr57 in CDR2, the site was not used and Abs bearing only the Fc carbohydrate were secreted (11). In this study, we made a more radical change, replacing the whole CDR2 of the anti-dansyl VH with CDR2 of the anti-dextran VH. Our goal was to determine whether the environment of the carbohydrate addition sequence would influence either site use or carbohydrate processing. We found that, in the context of the anti-dansyl VH, the site was used, but that the hybrid VH Abs were not secreted (Fig. 3⇑). Because Abs were secreted when transfectants were grown in the presence of Tm, it was evident that the carbohydrate and not the amino acid changes made with the CDR2 graft interfered with the trafficking of the Ab.
A model of the folded Fd (VH plus CH1) fragment of the hybrid VH, based on the murine anti-tumor Ab R24, which bears 84% identity to the hybrid VH, was obtained from SWISS-MODEL (data not shown); it places the Asn side chain nitrogen atom used for carbohydrate attachment on the surface of the molecule, where it should be readily available for glycosylation. Indeed, our results indicated that carbohydrate was attached to the hybrid VH; however, the presence of glycan appeared to impede secretion of the Abs. It is known that newly made proteins associate with chaperones and other proteins in the ER, which assist the protein in proper folding (34, 35, 36, 37, 38, 39). In particular, H chains transiently associate with the ER chaperone BiP/GRP78 during folding; BiP dissociates with L chain binding (40). If folding is incorrect after multiple attempts, the protein is eventually degraded by the proteosome after retrograde translocation into the cytosol from the ER (41, 42). It is possible that the carbohydrate in the hybrid VH of the H chain prevented association with the L chain so that BiP remained associated, and the H chains were held in the ER and, unlike the wild-type anti-dansyl IgG1, did not traffic to the Golgi apparatus (Fig. 4⇑).
N-Linked carbohydrate addition sites can arise spontaneously during somatic hypermutation. In some cases, the introduction of N-linked carbohydrate addition sites in the V region results in an increase in affinity for Ag (11). Affinity may be raised either through direct hydrophilic interaction with Ag, especially a carbohydrate Ag, or through altering the combining site geometry for a better cognate fit. In other cases, introduction of a carbohydrate can result in lowered affinity for Ag or can entirely block its binding (8, 11). We have now shown that addition of carbohydrate may also result in an Ab that is not secreted. Therefore, generation of carbohydrate addition sites during somatic mutation may sometimes result in cells not capable of producing functional Ab. As we increase our knowledge of the factors influencing carbohydrate processing and our predictive ability regarding the effect of glycan positioning within the Ag combining site on assembly and function, we will be better equipped to more effectively design Abs with the desired binding and functional capabilities.
We thank Ryan Trinh for constructing the anti-dansyl κ L chain expression vector, and Dr. David Meyer for kindly providing the rabbit anti-MERG Abs.
↵1 This work was supported by National Institutes of Health Grants AI 39187, AI 29470, and CA 16858 (to S.L.M.), and Institutional National Research Service Award T32 GM08375 (to F.A.G.).
↵2 Current address: National Genetics Institute, 2440 South Sepulveda Boulevard, Los Angeles, CA 90064.
↵3 Address correspondence and reprint requests to Dr. Sherie Morrison, Department of Microbiology, Immunology, and Molecular Genetics, University of California, Los Angeles, 405 Hilgard Avenue, Los Angeles, CA 90095-1489. E-mail address:
↵4 Abbreviations used in this paper: CDR, complementarity-determining region; ER, endoplasmic reticulum; GlcNAc, N-acetylglucosamine; Tm, tunicamycin; TG, Tris-glycine; MERG, major ER glycoprotein; β-COP, β-coatomer protein.
- Received October 7, 2003.
- Accepted February 13, 2004.
- Copyright © 2004 by The American Association of Immunologists