Asymmetric Contribution to Ig Repertoire Diversity by Vκ Exons: Differences in the Utilization of Vκ10 Exons

The recombination of an estimated 140 germline Vκ genes (1) with multiple Jκ genes and 100 to 200 V_H (2, 3) genes with multiple D_H and J_H genes in the mouse confers an enormous potential preimmune repertoire of Ab specificities. Diversity is further enriched by the contributions of junctional additions or deletions of nucleotides, the combinatorial pairing of individual heavy and light chain proteins, and finally, somatic mutation that occurs in the periphery. By these mechanisms it has been estimated that 10⁹ to 10¹¹ distinct Ab specificities are theoretically possible. The primary Ab repertoire is more limited in size because it is shaped by the number of functional V genes in the genome, the frequency of individual V gene recombination, the effect of promoter efficiency on transcription (4, 5, 6), the ability of specific V_H and V_L chains to pair, and positive or negative selection of B cells during ontogeny (7, 8, 9).

The precise mechanisms governing the selection of individual V_H and Vκ exons for rearrangement and expression are unclear. Several studies have examined the use of V_H or Vκ genes at the family level. In some instances, the utilized repertoire was reported to be nonrandom. For example, early in ontogeny, V_H gene families that lie proximal to DJ genes are utilized to a greater extent than those distal to DJ (10, 11, 12, 13), while in the adult mouse, V_H utilization does not appear to be positionally biased, but correlates with the size of a given family (12, 14, 15, 16, 17, 18). A positional bias is not evident at the κ locus during early development, and furthermore, the frequency of Vκ utilization in adult mice is not dependent on family size (19, 20, 21). That approximately 40% of Vκ genes lie in an opposite transcriptional orientation relative to Jκ and rearrange by inversion (22) may explain some of the observed differences in Vκ and V_H utilization. As a byproduct of inversional recombination, reciprocal products consisting of the fused VJ recombination signal sequences (RSS)² and the VJ intervening DNA are retained on the chromosome, allowing secondary recombinations between what were once distant Vκ genes with Jκ gene segments. Rearrangement at the κ locus does not shut down immediately after VJ recombination and, thus, secondary rearrangements may mask detection of any inherent positional bias in the κ locus.

Several investigators have reported the preferential use of certain Vκ families early in ontogeny. Medina and Teale showed that the early κ repertoire is dominated by Vκ families that undergo inversion-type rearrangements (23). It was shown that the repertoires of day 18 fetal liver and day 15 fetal omentum were restricted to five and six families, respectively, with a predominant usage of the Vκ4,5, Vκ9, and Vκ10 families, all of which undergo recombination by inversion. Kaushik et al. (19) reported a preference for the Vκ1 and Vκ9 families in B cell colonies derived from splenic B cells of 6- to 8-day-old neonatal mice. In contrast, Ramsden et al. demonstrated that Vκ usage from day 14 and 16 fetal livers represented 14 of the 18 known Vκ families (24).

In the preimmune repertoire of B lymphocytes, the precise mechanisms favoring selection of one Vκ gene over another for recombination and expression are not known. Several factors, however, are known to be necessary for recombination and transcription of V genes. For example, Ig gene rearrangement has been correlated with DNA hypomethylation and chromatin accessibility (25, 26, 27). Often germline transcripts from V genes (4, 28, 29, 30) or the C region locus (31, 32) are detected. It is not known whether these germline transcripts play a role in locus accessibility or if they are a byproduct of an open chromatin configuration. Expression of a V gene requires a functional RSS and a functional promoter. It has been shown that RSS strengths can directly affect the frequency of recombination (33, 34). In addition, differences in promoters of both V_H and Vλ genes are known to influence transcription efficiencies (4, 6, 35).

Previous studies have compared levels of expression of different V_H or V_L families, or have examined the functionality of the promoters or RSS of individual genes. In this study, we have compared the utilization of individual, but closely related, Vκ genes and the ability of their regulatory sequences to drive efficient recombination or expression.

Members of the same Vκ family share >80% homology at the DNA level (36) and, for the most part, lie close together within the κ locus on chromosome 6 of the mouse. We chose to study the Vκ10 family, which is small, containing three members (37), Vκ10A,³ Vκ10B, and Vκ10C; resides in the middle of the κ locus (1); and has been shown to rearrange by inversion (22). Vκ10A and Vκ10B are utilized in response to a wide variety of T-dependent and T-independent Ags and have been isolated and sequenced from numerous hybridomas and myelomas derived from several inbred mouse strains. The third family member, Vκ10C, has not been seen in functional Abs and has been isolated only once as a reciprocal product of an aberrant VJ recombination (22, 38). Since Vκ10A and Vκ10B are expressed in response to a wide variety of Ags and appear to contribute to early repertoire diversity (20), while Vκ10C is not utilized, it was of interest to determine the factors limiting a Vκ gene’s contribution to diversity.

The Vκ10C germline gene (GenBank accession AF029261) was isolated from a genomic library constructed from BALB/c kidney DNA in the λdash II vector (Stratagene, La Jolla, CA). Plaques were screened by hybridization with a ³²P-labeled 0.9-kb PC3386 EcoRI/HindIII fragment (22) on nitrocellulose. Hybridizing fragments from EcoRI and BamHI digests of positive plaque DNAs were subcloned into pBluescript (Stratagene) for sequencing. Genomic clones were cycle sequenced with primers produced in Core Facility for Biotechnology Resources at the Center for Biologics Evaluation and Research, Food and Drug Adminstration, and a dsDNA cycle sequencing kit (Life Technologies, Gaithersburg, MD).

Oligonucleotides were synthesized in the Core Facility for Biotechnology Resources facility at the Center for Biologics Evaluation and Research and are listed in Table I. All PCR reactions were performed in a DNA thermal cycler 480 (Perkin-Elmer, Norwalk, CT) in 100-μl reactions with 1.5 mM MgCl₂, 0.05 mM dNTPs, and 50 pmol of each primer. Specific PCR conditions were determined for the Vκ10A, Vκ10B, and Vκ10C primers in cross-priming experiments using hybridoma 226.1 (39) cDNA (Vκ10A), hybridoma H24C2 (40) cDNA (Vκ10B), and a Vκ10C spleen clone 4C17 (see below) as templates. Vκ10A and Vκ10B primers were specific for their templates under the following conditions: 95°C, 5 min; 73°C, 2 min/95°C, 1 min (30 cycles); and 73°C, 10 min, 4°C hold. The Vκ10C primer was specific under identical conditions, but with an annealing/extension temperature of 74°C. As shown in Figure 1, the Vκ10C primer differs from Vκ10A and Vκ10B template by 2 and 4 bases, respectively, Vκ10B primer differs from Vκ10A and Vκ10C template by 2 and 3 bases, respectively, and Vκ10A primer differs from Vκ10B and Vκ10C template by 5 and 6 bases, respectively. The Cκ region primers 5′KC and 3′KC were specific for their target under the following conditions: 95°C, 5 min; 65°C, 1 min/72°C, 1 min/95°C, 1 min (30 cycles); and 72°C, 10 min, 4°C hold. PCR products obtained from a BALB/c spleen cDNA with Vκ10A, Vκ10B, Vκ10C, and Cκ region primers under specific conditions were cloned into the PCRII vector (Invitrogen, San Diego, CA). Twenty clones each for Vκ10A, Vκ10B, and Vκ10C were cycle sequenced using Sp6 and T7 primers under the following conditions: 95°C, 3 min; 55°C, 30 s/70°C, 30 s/95°C, 30 s (20 cycles); and 58°C, 1 min/95°C, 30 s (10 cycles), 4°C hold. Clones A5, 4B16, 4C17, and K7 were used as standards in Vκ10A, Vκ10B, Vκ10C, and Cκ semiquantitation assays, respectively.

Total RNA from four BALB/c spleens was isolated using the Trizol method (Life Technologies), according to the manufacturer’s instructions. RNA was reverse transcribed with oligo(dT) primer and the superscript preamplification system (Life Technologies), according to the manufacturer’s instructions. A total of 1 μl of the cDNAs and 1 μl of log dilutions of A5, 4B16, 4C17, and K7 standards (10⁻¹-10⁻⁸ for A5, 4B16, and 4C17; 10⁻¹-10⁻¹⁰ for K7) was amplified using the specific PCR conditions described above. κ constant 3′KC primer was the 3′ primer for all PCRs, while Vκ10A, Vκ10B, and total κ PCRs utilized Vκ10A-specific, Vκ10B-specific, and κ constant 5′KC as 5′ primers, respectively. Thirty-five-microliter samples from each reaction tube were electrophoresed in 3% agarose gels at 170 V in 1× TAE for 3 to 4 h. Gels were stained with ethidium bromide (0.5 μg/ml) in 1× TAE for 30 min, and band intensities were quantitated on a fluorimeter. Band intensities of standards were plotted versus number of standard molecules at each dilution. The number of target structures in each cDNA sample was estimated by interpolating its band intensity into the standard curve. Vκ10C PCR sensitivity was determined by performing Vκ10C PCRs on serially diluted 4C17 standard.

Nested PCRs were performed as described above, with the following exceptions: the 3′KC and Gen7 primers were used to amplify 1 μl of cDNA in the primary PCR, as follows: 95°C, 5 min; 55°C, 1 min/72°C, 1 min/95°C, 1 min (30 cycles). Primary PCR products were purified with Wizard PCR preps columns (Promega, Madison, WI), and 1 μl was used as the template for a secondary PCR with the Vκ10A-, B-, and C-specific primers, and the 3′KC2 primer under the cycling conditions described for Vκ10A, B, and C PCRs above.

Volumes for all PCR reactions were 100 μl with 1.5 mM MgCl₂, 0.05 mM dNTPs, and 50 pmol each primer. The Vκ10AS (142-bp) promoter fragment was PCR amplified from a BALB/c liver DNA Vκ10A genomic clone (22) with the 5′AJ1/10C short HindIII and AJ13′ HindIII primers. The Vκ10CS (142-bp) promoter fragment was PCR amplified from the λdash II (Stratagene) BALB/c kidney DNA Vκ10C genomic clone 91-3 with the 5′AJ1/10C short HindIII and 10C3′HindIII primers. The intronic κ enhancer (537 bp) was PCR amplified from pECK DNA (containing the intronic κ enhancer and germline κ constant DNA) with the Enhanc.5′KpnI+Enhanc.3′KpnI primers. Vκ10AS, Vκ10CS, and κ enhancer PCRs were performed, as follows: 95°C, 5 min; 50°C, 1 min/72°C, 1 min/95°C, 1 min (30 cycles); and 72°C, 10 min, 4°C hold. Vκ10AS and Vκ10CS PCR products were digested with HindIII, and κ enhancer product was digested with KpnI. The κ enhancer fragment was ligated into KpnI-linearized pGL3 basic vector (Promega) and labeled pGL3-κen. Digested, purified Vκ10AS and Vκ10CS promoter fragments were ligated into pGL3-κen linearized with HindIII. Orientation and promoter sequence confirmation were determined by sequencing. Plasmids, including control vector pCMV-β (Clontech, Palo Alto, CA), were purified by double banding in CsCl.

All electroporations were performed using a Bio-Rad (Richmond, CA) gene pulser (0.22 kV, 960 μFd) and 0.2-cm path-length cuvettes containing 6 × 10⁶ cells in 250 μl electroporation media (RPMI 1640, 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, nonessential amino acids 1× final concentration, 50 μM 2-ME, and 2 mM glutamine). pCMVβ (0.32 μg) was cotransfected with either 5 pmol of pGL3-κen/Vκ10AS, pGL3-κen/Vκ10CS, or pGL3κen-only into the pre-B cell lines 18-81 and NFS-467, the immature B cell line Wehi 231.4, and the plasmacytoma cell line Sp2/0. The pre-B cell lines NFS-5 and 70Z/3 were each cotransfected with 10 pmol of the promoter plasmids and pGL3-κen plasmids and 0.32 μg of pCMV-β as a control. The mature B cell line A20 was cotransfected with 0.5 pmol pGL3-κen/Vκ10AS, pGL3-κen/Vκ10CS, or pGL3-κen-only, and 1.7 μg pCMV-β. Following transfection, cells were transferred to T75 flasks containing 15 ml complete RPMI and incubated for 24 h at 37°C in 5% CO₂. After 24 h, cells were harvested by centrifugation, washed once in PBS, and lysed for 15 min in 70 μl 1× reporter lysis buffer (Promega). Cellular debris was pelleted by centrifugation at 11,000 rpm for 2 min at 4°C. Supernatants were transferred to new tubes and stored at −70°C.

For luciferase assays, 20 μl of lysate was combined with 100 μl of luciferase assay substrate (Promega) in the wells of a microlite I 96-well tray (Dynatech, Chantilly, VA), and light production was measured in a luminometer. For β-galactosidase assays, 5 μl of lysate was added to the wells of a microlite I 96-well tray and combined with 50 μl of galacton plus (Tropix, Bedford, MA) diluted 1/100 in 0.1 M sodium phosphate, pH 7, and incubated for 15 min at room temperature. A total of 50 μl of emerald enhancer (Tropix) diluted 1/10 in 0.2 N NaOH was added to each well, and light production was measured in a luminometer. Luciferase activity was calculated by dividing the luciferase luminometer value by the β-galactosidase value for each well. Statistical differences in Vκ10 promoter efficiencies were analyzed using one-way ANOVA.

BALB/c genomic DNAs (100 ng) from spleen and ThB-enriched bone marrow B cells (see below) were amplified with the Gen7 and Jκ 5–3 primer pair to amplify all Vκ10 rearrangements. PCRs were performed in duplicate 100-μl reactions (pooled before gel loading) with 1.5 mM MgCl₂, 0.05 mM dNTPs, and 50 pmol each primer under the following conditions: 95°C, 5 min; 60°C, 1 min/72°C, 2 min/95°C, 1 min (30 cycles); and 72°C, 10 min, 4°C hold. Thirty-seven microliters of each sample were electrophoresed on triplicate 2.3% agarose gels for 3 h at 150 V. Included on each gel as specificity controls were genomic clones for Vκ10A (2.1-kb EcoRI/BamHI fragment from pC13-13 (22)) and Vκ10C (2.4-kb EcoRI fragment from the pBluescipt 91-3), and an RT-PCR clone for Vκ10B (4B16 EcoRI digest). EcoRI digests of RT-PCR clones A5 and 4C17 were included on the Vκ10A and Vκ10C gels, respectively, as additional positive controls. EcoRI-digested 4B16 was used as both the specificity and positive control on the Vκ10B gel, since a genomic clone of Vκ10B was not available. DNA was transferred to positively charged nylon membranes in 10× SSC overnight by capillary action, and UV cross-linked to filters in a Bio-Rad gene linker (150 mJ). Oligonucleotide labeling and washing methods were described by Pennycook et al. (41). Membranes were prehybridized in separate roller tubes with 20 ml 5× SSC, 2.5% skim milk powder, 0.1% N-laurylsarcosine, and 0.02% SDS at 42°C for 4 h. Oligonucleotides Vκ10A3, Vκ10B, and Vκ10C were end labeled with [γ-³²P]dATP and purified on Select-D G25 spin columns (5 prime→3 prime, Boulder, CO). A quantity amounting to 3 × 10⁷ total cpm of either the Vκ10A3, Vκ10B, or Vκ10C probe was added to the tubes containing the prehybridization solution and hybridized overnight at 42°C. Blots were washed twice for 5 min with 2× SSC, 0.1% SDS at room temperature and twice for 15 min at 42°C (Vκ10A3), 51°C (Vκ10B), or 57°C (Vκ10C). The Vκ10C primer differs from Vκ10A and Vκ10B template by 2 and 4 bases, respectively, Vκ10B primer differs from Vκ10A and Vκ10C template by 2 and 3 bases, respectively, and Vκ10A3 primer differs from Vκ10B and Vκ10C template by 7 bases. The specificity of each probe was determined experimentally by washing nylon strips containing the Vκ10A and Vκ10C genomic clones and the Vκ10B RT-PCR clone 4B16 at increasing temperatures until each probe hybridized to its own Vκ10 family member and not the other two. Membranes were placed in Molecular Dynamics (Sunnyvale, CA) PhosphorImager cassettes overnight. Screens were developed using a Molecular Dynamics PhosphorImager and Image Quant software.

Bone marrow cells collected from both femurs and tibiae were resuspended at 1 to 4 × 10⁷ cells/ml PBS and incubated with biotin-labeled anti-ThB at 4°C with gentle shaking for 30 min. Cells were washed twice in HBSS, pH 7.4, and mixed with Dynabeads M-280 streptavidin at a ratio of 4:1 beads/cell, with the beads at 1 to 2 × 10⁷ beads/ml. Cells and beads were incubated at 4°C with rotation for 30 min. The cells bound to the beads were collected with a Dynal (Great Neck, NY) MPC-1 or MPC-2 magnet. Genomic DNA was prepared from the ThB⁺-selected population. Flow-cytometric analysis of cells before enrichment showed ThB⁺ cells ranging from 12 to 24% of total bone marrow cells. After enrichment, 4 to 5% of unselected cells were ThB⁺ (data not shown).

The Vκ10 family contains three members (37). Southern blots of BALB/c BamHI-digested or EcoRI-digested kidney genomic DNA with a 0.9-kb PC3386 probe labeled with ³²P revealed two and three strongly hybridizing bands, respectively (data not shown and (37)). The smaller (5.2-kb) BamHI band is a doublet that is known to contain both Vκ10A and Vκ10B. The Vκ10A gene resides on a 5.2-kb EcoRI fragment, while the Vκ10B gene resides on a 3.7-kb EcoRI fragment (42). Although unconfirmed, the 7.4-kb BamHI and the 2.6-kb EcoRI bands most likely contain the Vκ10C gene.

The Vκ10C germline gene was isolated from a BALB/c kidney genomic library and sequenced beginning from ∼600 bp 5′ of the transcription start site to 1700 bp 3′ of the heptamer/nonamer. Analysis of the Vκ10C sequence (Fig. 1) shows it is structurally functional. There are no obvious defects such as frameshifts or missense mutations to easily explain the lack of detection of Vκ10C in functional Abs. Comparison of the germline sequence of Vκ10C with germline Vκ10A and Vκ10B sequences reveals that it is most closely related to Vκ10B, sharing 97% homology in the coding region and 94% homology with Vκ10A. The majority of base substitutions among the Vκ10 family members occur in the CDRs. The promoter region, splice site, and RSS of Vκ10C are also intact. Vκ10C differs from Vκ10A in the last position of the nonamer; however, this difference is shared with Vκ10B, which is known to be expressed, and most likely does not account for a decrease in recombination efficiency.

The Vκ10C amino acid sequence (Fig. 2) is 91 and 94% homologous to Vκ10A and Vκ10B, respectively, differing primarily in the CDRs. The Vκ10C sequence contains two unusual substitutions: a Cys to Tyr substitution at position −1 of the leader peptide, and a Thr to Ala substitution at position 69 in framework 3. Neither of these substitutions is thought to interfere with light chain processing, folding, or the ability to pair with heavy chain (see Discussion).

A search of DNA databases revealed that both Vκ10A and Vκ10B light chains are utilized in response to a wide variety of T-dependent and T-independent Ags in different inbred mouse strains, while Vκ10C has not yet been detected in a functional Ab. We wanted to determine whether this was due to the lack of appropriate Ags selecting for Vκ10C or if Vκ10C is underexpressed in the spleen. It has been estimated in murine spleen that individual Vκ exons represent approximately 0.6% of total κ mRNA (43).

RT-PCR experiments to detect the presence of Vκ10A, B, and C message in an adult BALB/c mouse were performed with the 5′ Vκ10A, B, and C primers shown in Figure 1 and Table I and a 3′ primer (3′KC) from the Cκ region (Table I). Specific conditions were determined in cross-priming experiments in which cDNA from both Vκ10A (226.1)- and Vκ10B (H24C2)-producing hybridomas was amplified with the three gene-specific Vκ10 primers and the Cκ region primer 3′KC (data not shown). Using specific conditions established in the hybridoma cross-priming experiments, we were able to detect Vκ10A, B, and C mRNA in the spleen of an adult mouse (Fig. 3). Vκ10A and Vκ10B appeared as intense bands, while Vκ10C was barely visible. This was the only mouse tested that had visible Vκ10C bands after a single PCR (see below). The RNA from this mouse was consumed during assay development and was not quantitated. Nineteen clones each of Vκ10A, Vκ10B, and Vκ10C RT-PCR products were sequenced to establish the specificity of each reaction (data not shown). Mispriming events were not observed.

The levels of Vκ10A, B, and C mRNA in the spleens of four adult BALB/c mice were measured semiquantitatively using cloned Vκ10 products from adult spleen RT-PCRs to construct standard curves. Total κ mRNA levels were determined similarly using a standard curve for the Cκ region. The concentration of each mRNA product was calculated by interpolation of the intensity of each sample band into the curve (Fig. 4). Three mice tested were 3 mo of age, while a fourth mouse was 14 mo of age. Figure 5 shows that Vκ10A and B are expressed at equivalent levels in the 3-mo-old mice and represent from 0.3 to 3.4% of total κ message, which is consistent with the estimate that individual Vκ genes are expressed at a frequency of 0.6% in preimmune B cells (43). While no statistically significant difference between the levels of Vκ10A and Vκ10B mRNA was observed in the 3-mo-old mice, the 14-mo-old mouse (mouse 4) had significantly lower levels of Vκ10B compared with Vκ10A (p < 0.05). In all four mice tested, the level of Vκ10C mRNA was below the detection limit (1.14 × 10³ targets) of the assay. As is common in a PCR-based quantitative assay, variations in the levels of Vκ10A, Vκ10B, and total κ were evident for all mice tested. However, compared with the average level of expression for Vκ10A and B mRNA derived from multiple assays and, based on the sensitivity of the Vκ10 PCR, we estimate that Vκ10C mRNA is present at least 1000-fold less than that of Vκ10A and Vκ10B in the 3-mo-old mice. In the 14-mo-old mouse, Vκ10C is present at approximately 1000-fold less than Vκ10A, and 10- to 100-fold less than Vκ10B. Splenic B cells were stimulated with LPS to expand underrepresented clones, and Vκ10C message was still undetectable by RT-PCR. Vκ10C message could be detected by a nested PCR from the spleen of a 3-mo-old mouse, indicating that Vκ10C mRNA is present at low levels (data not shown).

It has been shown that Vκ10 genes contribute substantially to the repertoire in early B cell development and that the Vκ10 sequences detected in fetal liver, fetal omentum, and bone marrow all derived from the Vκ10A or Vκ10B genes (23, 24). Our examination of the expression of Vκ10 genes in bone marrow and fetal liver revealed that Vκ10A mRNA was readily detectable using a single RT-PCR reaction in both adult bone marrow and fetal livers. In fetal liver, Vκ10A was detectable as a faint band at day 16 and increased in intensity through day 19 (data not shown). Neither Vκ10B nor Vκ10C was detectable using a single RT-PCR reaction in either tissue. Using a nested PCR reaction, both Vκ10B and Vκ10C mRNAs could be detected in the bone marrow, and by day 18 in fetal livers (data not shown).

Since Vκ10C is poorly expressed in splenic B cells, we next compared the Vκ10C promoter efficiency with that of Vκ10A. The major elements of a Vκ gene promoter are the octamer (44, 45) and a TATA box. These elements are contained within a ∼70- to 100-bp region that has been defined as the minimal promoter necessary for driving κ transcription (44, 46, 47). A comparison of the Vκ10A and Vκ10C gene promoters is shown in Figure 6 A. Both promoters contain identical octamers that differ from the consensus octamer by 2 bases. Neither Vκ10A nor Vκ10C contain a consensus TATA box, but both genes have the sequence TAATT at position −27. Vκ10A and Vκ10C also contain the pentadecamer and κY sites, which have also been implicated in activation of κ transcription (44, 48). The pentadecamers for both genes are identical, lie upstream of the octamer, and differ from the consensus pentadecamer by a single base (TGCAGCTGTGCTCAG). The κY site, unlike those reported by Atchison (48), is downstream of the octamer and differs from the consensus by 2 bases (CTTCCTAT). Overall, the 142-bp Vκ10A promoter differs from that of Vκ10C at three positions, an A to C substitution at position −6, a C to G substitution at position −42, and an additional A nucleotide at position −97 just upstream of the octamer. Two of these substitutions, the C to G at position −42 and the A to C at position −6, lie within a potential transcription-regulatory element (CANNTG), the E-box (49, 50). At position −42, the Vκ10A E-box sequence (CAGATT) differs from the Vκ10C sequence by a single base (CACATT). In the downstream E-box beginning at position −10, both Vκ10A (CAGCCTG) and Vκ10C (CAGCATG) contain unique insertions into an E-box sequence.

The 142-bp promoters of Vκ10A and Vκ10C shown in Figure 6,A were cloned into pGL3κen (the pGL3 luciferase reporter vector containing the κ intronic enhancer). These promoter clones or pGL3κen as a control were used to transfect B cell lines representing the pre-B cell, immature B cell, mature B cell, and plasma cell stages of B cell development. The pre-B cell lines used were 18-81, 70Z/3, NFS-5, and NFS-467. Other cell lines examined include the immature B cell line Wehi 231.4, the mature B cell line A20, and the plasmacytoma cell line Sp2/0. All cells were cotransfected with pCMV-β for normalization purposes. Vκ10A and Vκ10C promoter activity in Wehi 231.4, A20, and SP2/0 was not significantly different (Fig. 6,B). In all of the pre-B cell lines tested, the Vκ10A promoter construct significantly outperformed that of Vκ10C (Fig. 6 C) (p < 0.05, one-way ANOVA). Constructs containing longer promoters of 300 and 500 bp from Vκ10A and Vκ10C were also tested and yielded similar results (data not shown). Three of these pre-B cell lines have active Ig loci: 18-81 undergoes VDJ recombination in vitro, but its κ locus is silent (51); 70Z/3 contains an unexpressed, but rearranged, κ locus that can be induced by stimulation with LPS (52); and in NFS-5, both κ- and λ-expressing clones can be purified from the bulk culture, and the rearranged V_H gene has been shown to undergo V_H gene replacement (53, 54). NFS-467 has not been examined for the ability of its Ig loci to rearrange in culture.

The RSS of Vκ10A and Vκ10C differ at the last position of the nonamer, a T to G substitution. The Vκ10B and Vκ10C RSS, however, are identical, and Vκ10B mRNA is present at similar levels as Vκ10A in adult spleen, suggesting that the RSS difference between Vκ10A and Vκ10B/C does not influence the efficiency of Vκ10C recombination. While the efficiency of the Vκ10C RSS is likely to be equivalent to that of Vκ10A and Vκ10B, we do not know whether the frequency of Vκ10C recombination is also equivalent. To examine this question, PCR was performed on genomic DNA using a 5′ probe and an annealing temperature that would amplify all three family members, and a 3′ probe lying downstream of Jκ5. Thus, all amplified rearrangements would be in the context of the Cκ region and not as a part of a reciprocal joint to a VJ recombination. We examined DNA isolated from both spleen and ThB-enriched bone marrow cells. ThB is a differentiation Ag expressed on thymocytes and on B cells, with its expression first seen on pre-B cells, the differentiation stage at which light chain rearrangement begins (55, 56). Thus, enriching for ThB-expressing cells would capture all B-lineage cells in the bone marrow undergoing the transition to or already having made the transition to an immature B cell.

Vκ10 recombination products were hybridized with ³²P-labeled Vκ10A3, Vκ10B, and Vκ10C oligonucleotide probes. Recombinations to all four Jκ genes were detectable for all Vκ10 family members in both ThB-enriched bone marrow and spleen (Fig. 7). This assay is not quantitative, and the less intense Vκ10C bands most likely result from the higher wash temperature needed for specificity of the Vκ10C probe. In preliminary experiments done for probe specificity, specific band intensities (for all probes) were shown to decrease as wash temperatures were elevated (data not shown). Additionally, although the Vκ10 PCR annealing temperature was adjusted to amplify all Vκ10 products, the Gen7 primer differs from Vκ10C by a single base and may influence yield of Vκ10C products. Because Vκ10C recombinations are detected in both tissues, these data suggest that most Vκ10C-rearranged cells do not undergo negative selection and elimination in the bone marrow. It is not known whether these Vκ10C rearrangements are productive or nonproductive.

We analyzed the translated sequences of cloned Vκ10A and Vκ10C RT-PCR products derived from the initial control assays. Nineteen sequences for each gene were analyzed. Of these 19 sequences, 10 of 19 Vκ10A and 13 of 19 Vκ10C sequences were established as unique based on their nucleotide sequences.

Of the 10 Vκ10A rearrangements, 7 recombined with Jκ1, and 3 with Jκ2 (Table II). Three of these rearrangements were nonproductive due to out-of-frame rearrangements (2 involving Jκ1, and 1 with Jκ2). Of the 13 Vκ10C sequences examined, 9 were to Jκ1, 2 to Jκ4, and 1 each to Jκ2 and Jκ5 (Table II). As with Vκ10A, three of the Vκ10C rearrangements were also nonproductive due to out-of-frame rearrangements (2 to Jκ 1, and 1 to Jκ4 rearrangement).

An interesting observation of the remaining seven Vκ10C sequences rearranged to Jκ1 is that five lack the proline at position 95. Proline 95 is an invariant residue. In each of these clones, leucine, the penultimate amino acid of the three Vκ10 coding regions, is the amino acid immediately preceding the Jκ region. Proline 95 is seen in all of the functional Vκ10A junctions examined and in all Vκ10A and Vκ10B sequences selected from the protein database. It is possible that the loss of this invariant residue from the Vκ-Jκ junction in Vκ10C rearrangements leads to improper folding of the Vκ10C light chain or impedes its ability to pair with heavy chain, effectively resulting in nonproductive rearrangements.

Vκ10 gene segments, as well as up to 40% of all Vκ genes, lie in opposite transcriptional orientation relative to the Jκ locus and rearrange by inversion (22). Recombination by inversion results in the formation of reciprocal products that are retained on the chromosome. The Vκ10 family and other families that rearrange by inversion, such as Vκ4, Vκ8, and Vκ12,13, contain a conserved BamHI site approximately 1 kb downstream of the coding region (22, 57, 58). Rearrangement of one of these genes with Jκ1 results in an 8-kb BamHI reciprocal product. An analysis of splenic DNA for 8-kb BamHI reciprocal products resulted in an estimate that approximately 25% of Vκ alleles retain this BamHI site. Since this site is conserved downstream of many different Vκ genes, it has been postulated that this site may play a role in regulation of Vκ rearrangement (57). Sequences ≥1 kb downstream from the germline Vκ10A and Vκ10C genes and a germline Vκ8 gene were determined. Both Vκ10A and Vκ8 germline sequences contain the 1-kb BamHI site, whereas the Vκ10C sequence does not. It is known from the restriction maps of other investigators (42) that the Vκ10B germline gene also has a 3′ BamHI site. The BamHI sites for Vκ10A and Vκ8 are contributed by LINE elements (59) that lie in an inverted orientation 600 and 850 bp downstream of the Vκ10A and Vκ8 coding regions, respectively. The LINEs are truncated at different points in their respective 3′ regions, but both contain the BamHI site at position 6989 of the LINE sequence. Like Vκ10A, Vκ10C harbors a LINE element 600 bp downstream of the coding region, but the region containing the BamHI site at position 6989 is absent due to a truncation at position 5690 of the LINE sequence.

Previous studies examining the generation of Ab diversity have compared the expression of different V_H or V_L families (10, 11, 12, 14, 15, 16, 17, 18). Others have looked at promoters or RSS of individual genes and identified differences that lead to more or less efficient transcription (4, 6, 35) or recombination (60). Recently, Buchanan et al. compared the promoters of V_H families and showed that there may be differential transcription and regulation of V_H promoters in vivo (35). In this study, we have examined and compared both the promoters and RSS of individual members of one Vκ family. This study demonstrates that a structurally functional Vκ gene, with no apparent defect in its promoter sequence or RSS, is underutilized in both the emerging and adult repertoires.

The Vκ10 family contains three members, Vκ10A and Vκ10B, whose sequences have been previously published (42, 61), and Vκ10C, whose sequence is reported in this work. It was of interest to study this family because Vκ10A and Vκ10B are utilized by a variety of inbred strains in response to a diverse array of Ags, including T dependent, T independent, and autoantigens, while Vκ10C has not been detected in a functional Ab. All of the Vκ10 sequences in the GenBank database could be assigned to either the Vκ10A or Vκ10B genes. Indeed, a Vκ10C sequence has only been reported once, as part of a reciprocal element to a VJ join that has undergone a Vκ10C to Jκ1 rearrangement (22).

The Vκ10C gene is structurally intact and does not contain stop codons, deletions, or insertions that might explain its underrepresentation in splenic B cells. It is ≥94% homologous at the nucleic acid level and ≥90% at the amino acid level with Vκ10A and Vκ10B. The Vκ10C promoter, splice sites, and RSS are intact. Vκ10C contains a rare Thr to Ala substitution at position 69 in framework 3. Threonine is highly conserved at this position, but an Ala residue has been detected at this position in a Vκ10A sequence (62), and is therefore not believed to impede light chain folding or pairing with heavy chain. Another rare substitution in Vκ10C is Cys to Tyr at position −1 of the leader peptide. Leader peptides are variable at this position (63), and one human light chain signal peptide is known to contain a Tyr at position −1 (64). Therefore, the Cys to Tyr substitution at position −1 of the leader peptide is unlikely to interfere with light chain processing.

We have shown that Vκ10A and Vκ10B are transcribed at equivalent levels in adult spleen, while Vκ10C is present at levels at least 1000-fold lower. Vκ10A mRNA was readily detectable in day 16 to 19 fetal livers and in bone marrow. Vκ10B and Vκ10C were only detectable in day 18 to 19 fetal livers and bone marrow using a nested PCR, indicating that both Vκ10B and Vκ10C mRNAs are present at significantly lower levels than Vκ10A in early development. Clones expressing Vκ10B may be expanded in the spleen in response to environmental Ags. Low levels of Vκ10C mRNA in spleen may be due to the lack of a suitable environmental stimulus to expand Vκ10C-expressing clones. We observed no increase, however, in Vκ10C mRNA levels after stimulation of splenic B cells with LPS.

To further examine the low levels of Vκ10C mRNA expression, the sequences of the Vκ10A and C promoters and their ability to drive expression of reporter constructs were compared. Vκ promoters contain several motifs that have been demonstrated to play a role in transcription. The most important of these sequences is the octamer (44, 45) element to which bind the Oct-1 and Oct-2 proteins. Other regulatory elements of the κ promoter include the κY site, which has been shown to lie upstream of the octamer and can compensate for a mutated octamer in a Vκ19 gene (48), and the pentadecamer (44), which is highly conserved among Vκ promoters. E-boxes, originally described by Ephrussi (49), are contained within the pentadecamer sequence (44) and function as promoter elements. Evidence that these sites can function as promoter elements was provided by Hogbom et al. (50), who demonstrated that a range of nuclear proteins bound to the SP6 κ promoter pentadecamer. Interestingly, these sites had variable efficiency as promoter elements depending on the cells/cell lines used in the experiments.

A comparison of the Vκ10A and C promoter sequences shows there are no differences between Vκ10A and C in the major promoter elements, the octamer, pentadecamer, κY site, and TATA box (Fig. 6 A). Vκ10A and Vκ10C promoters each contain one E-box motif within the pentadecamer and two potential E-boxes downstream of the octamer. There are three nucleotide differences between the Vκ10A and C promoters. At position −97, the Vκ10C promoter contains a string of four As immediately upstream of the octamer, while the Vκ10A promoter contains five A nucleotides. Baumruker et al. (65) have shown that octamer-flanking sequences can alter the affinity of Oct-1 binding to the V_H octamer. Likewise, Sigvardsson et al. (66) have shown that 3′ flanking sequences of the SP6 κ promoter can affect the affinity of Oct-2A binding to the octamer. The other differences are in two potential E-box sites at positions −42 and −6. At position −42, the Vκ10A (CAGATT) and Vκ10C (CACATT) E-box sequences differ by a single base. At position −6, Vκ10A and Vκ10C each have a unique insertion into the E-box sequence.

To determine whether differences in the Vκ10A and Vκ10C promoters could explain the difference in mRNA levels between these genes, we examined the Vκ10A and Vκ10C promoter efficiencies in transiently transfected B cell lines representing different developmental stages. While the Vκ10C promoters worked as efficiently as Vκ10A promoters in immature B cells, mature B cells, and plasma cells, a statistically significant difference was observed in pre-B cells (Fig. 6 C). This phenomenon was seen in all four pre-B cell lines tested. The biologic significance of this difference is unclear. It is possible that a complex of pre-B cell nuclear transcription factors binds with less affinity to the Vκ10C promoter due to one or all of the three nucleotide changes in this promoter, resulting in lower levels of Vκ10C transcripts. Experiments are in progress to address this question. The late pre-B cell stage in development is when light chain rearrangement begins. If the Vκ10C promoter is inefficient in pre-B cells, one possibility might be that insufficient amounts of Vκ10C protein are produced at this critical point in the B cell developmental pathway to express surface IgM. In such a scenario, this cell could undergo further light chain recombination, receptor editing, or be eliminated by apoptosis.

We next examined the ability of the Vκ10C gene to rearrange. The Vκ10C RSS differs from that of Vκ10A by one nucleotide, the terminal base in the nonamer. This position has not been shown to affect recombination efficiency. Since Vκ10C and Vκ10B have identical RSS, this change is not likely to be responsible for the underutilization of the Vκ10C gene. Indeed, we demonstrated that Vκ10C is capable of recombination, as rearrangements are detected in both spleen and bone marrow of adult mice. The Jκ 5–3 primer used in the PCRs is located downstream of Jκ5; thus, the recombination products are in the context of the Cκ region and are not the result of secondary recombinations with displaced Jκ resulting from inversional recombinations of other Vκ genes. Since small amounts of Vκ10C mRNA are detectable in spleen, bone marrow, and fetal liver, some of the rearrangements may result in productive V-J joins, although it has been shown that nonproductive rearrangements can be transcribed at levels comparable with or even greater than productive rearrangements (38).

Analysis of 13 unique Vκ10C and 10 unique Vκ10A junctions revealed some interesting differences. Seven of the nine in-frame Vκ10C-Jκ1 junctions were missing the proline residue at position 95. This was not seen for any in-frame Vκ10A junctions. Proline 95 is invariant in murine κ-chains and is present in all of the Vκ10A or Vκ10B sequences obtained from GenBank. It is possible that the loss of proline 95 results in light chains that cannot fold properly or pair effectively with heavy chains.

Finally, it is not known how overall chromatin configuration affects the expression of Vκ genes. LINE elements are commonly found in the Vκ and V_H loci of mice (67, 68, 69). In some cases, LINEs have been involved in aberrant rearrangements of these genes (70). LINEs are transposable elements that contain two open reading frames, one of which codes for a reverse transcriptase (59), and recently, LINE proteins have been identified (71, 72). LINE elements are present at 10⁴ copies per haploid genome, and many of these are truncated, usually at the 5′ end (59). It has been shown that at least 25% of Vκ genes contain a common BamHI site 3′ of the coding region (57). We have shown that this BamHI site is contributed by LINE elements. Both Vκ10A and Vκ10C contain LINE elements in this position, but only Vκ10A has the conserved BamHI site. We have not sequenced a Vκ10B gene to see whether there is a LINE element present, but, from published restriction maps (42), it does contain the conserved BamHI site. It is not known whether the truncation of the Vκ10C LINE affects the overall expression of this gene or interferes with proper recombination, resulting in a high frequency of Vκ10C rearrangements that lack the invariant proline 95.

Because the Vκ10C gene is structurally functional, the lack of hybridomas and myelomas utilizing Vκ10C suggests that Vκ10C-expressing cells are negatively selected. Alternatively, Vκ10C-rearranged cells may not be actively selected, either negatively or positively. Positive selection has been demonstrated for a specific VHCDR3 sequence to make the transition from a pre-BI cell to a pre-BII cell (73) and for certain V_H-VL pairs to move immature cells from the bone marrow into the immunocompetent B cell pool in the periphery (8). Because Vκ10C rearrangements are detected in both the spleen and bone marrow and the Vκ10C promoter is inefficient in pre-B cells when κ rearrangement begins, Vκ10C-rearranged cells may be unable to undergo selection. There may be a threshold of surface Ig expression for selection to occur. Inefficient Vκ10C transcription may not produce enough Vκ10C protein to reach this threshold. Similarly, an intrinsic defect in recombination of Vκ10C, resulting in proteins unable to pair efficiently with heavy chain, would have the same outcome. Because such cells are not selected, they can rearrange the other κ allele, the λ locus, or undergo receptor editing. Such cells would then have a chance to undergo positive or negative selection, but this selection would not be due to Vκ10C expression. Regardless of the reason for the underutilization of Vκ10C in the Ab repertoire, it is clear that not all structurally functional genes contribute to Ab diversity.

We thank Drs. Fred Mushinski and Walter Gerhard for gifts of cell lines, Dr. Kevin Holmes and Larry Lantz for biotin-labeled ThB, and Drs. Stuart Rudikoff, Steve Bauer, and Mark Brunswick for critical review of the manuscript.