Ig Light Chain Precedes Heavy Chain Gene Rearrangement during Development of B Cells in Swine

This article is featured in In This Issue, p.1379

Immunological textbooks and reviews (1–4) describe the rearrangement in Ig loci as a tightly controlled sequential process regulated by the surrogate L chain (SLC), which is composed of λ5 (CD179b) and the invariable Ig ι-chain of SLC (CD179a). According to this paradigm, the first step in the rearrangement leading to the formation of the BCR occurs in the IgH locus of proB cells by combinatorial joining of D_H to J_H segments on both chromosomes. The resulting preB-I cells subsequently rearrange certain V_H segments with one of these partial VDJ gene rearrangements for the IgH (DJ_H) combinations to the complete VDJ gene rearrangement for the IgH (VDJ_H) rearrangement, which is then tested for its ability to form a proper pre-BCR by association with SLC. There is no IgL gene rearrangement at this developmental stage. When the pre-BCR fails to fold correctly, the cell has a second chance using the second chromosome. The large preB-II cells die when they fail to produce a productive pre-BCR whereas successful cells survive, expand, and consecutively become small preB-II cells.

The importance of the SLC in selection of IgH gene rearrangements is considerable because as many as 50–70% of productive IgH proteins do not pair with SLC and become apoptotic (4). The IgLκ locus is rearranged in the surviving small preB-II cells until a productive IgLκ gene rearrangement creates an authentic BCR (5, 6). The current paradigm further states that gene rearrangement in the IgLλ locus only begins after all possible Vκ/Jκ segments have been exhausted and/or when Vκ or the recombining element (RE) rearranges to the κ deleting element (KDE), leading to deletion of the Cκ gene segment (7).

There are a few deviations from above described paradigm but none are substantial. For example, V_H to D_H rearrangement may precede D_H to J_H rearrangement in rabbits (8). In chickens, multiple D_H to DJ_H rearrangements may occur before subsequent rearrangement to the V_H gene (9). Also, B cells in transgenic mice can develop through an alternative pre-BCR–independent pathway in which the IgL genes rearrange independently of the IgH genes (10–14). Such findings indicate that sequential rearrangement of IgH before IgL genes need not be essential. This is superimposed in birds in which IgH and IgL gene rearrangements can occur competitively very early in fetal life (9, 15, 16). Currently, no homologs of mouse λ5 have been identified in chickens or swine (17, 18), and SLC may be lost or does not develop in other species (4).

Our previous work disproved that the ileal Peyer’s patches are a site of primary B cell lymphogenesis in swine (19–22). Rather, porcine B cells are developed throughout life in the bone marrow (BM), the primary lymphoid organ also for mice and humans (23). B cell development in swine was characterized according to expression of MHC class II (MHC-II), CD2, CD21, CD25, CD45RC, CD172a, and IgHμ Ags, and seven subpopulations developing from subset 0 to subset 6 were identified (23). These studies also showed that all seven subsets can be unambiguously identified by cell size and decreasing expression of CD172a on MHC-II⁺ BM cells as B cell precursors differentiate (Fig. 1A). Thus, CD172a^bri/hi expression is limited to early precursors until incomplete DJ_H rearrangements, and CD172a^lo expression remains on the surface until complete VDJ_H rearrangement(s) and CD172⁻ cells re-present late preB-II and immature B cells (23). In this study, we characterize the rearrangement and expression of IgL genes in context of IgH genes and show that B cells develop in the porcine BM by a process that is yet another deviation from the textbook paradigm. Analysis of developing B cells in the swine BM revealed that IgL genes rearrange before IgH genes and that there is no dependency on SLC, which is consistent with the failure to recover λ5 from the porcine genome and the observation that the invariable Ig ι-chain of the surrogate L chain is expressed in many nonlymphoid tissues in swine (18, 24). Whereas these initial studies suggested that expression of IgLλ genes appeared before IgLκ, in this study we provide evidence that rearrangement starts with IgLκ genes but then rapidly shifts to IgLλ before IgH gene rearrangements and the formation of a BCR. Therefore, the first Ig⁺ cells are IgH⁺IgLλ⁺ whereas IgH⁺IgLκ⁺ cells are generated later. Phylogenetically it appears that there are two groups of animals, one of which uses a pre-BCR–driven developmental pathway for B cell generation whereas the second group uses a pre-BCR–independent pathway.

Animals used in the study were Minnesota miniature/Vietnamese–Asian–Malaysian crossbred pigs bred in Novy Hradek. Fetuses were obtained by hysterectomy. Germ-free (GF) piglets were recovered from gilts by sterile hysterectomy at day of gestation 112 and were kept in isolator units under GF conditions on sterile formula as previously described (25). GF piglets were used because they have a naive immune system, in which the BM contains a minimum of polymorphonuclear cells, and effector B cells stages including plasma cells are missing. All animal experiments were approved by the Ethical Committee of the Institute of Microbiology, Czech Academy of Science, according to guidelines in the Animal Protection Act.

Cell suspensions were prepared essentially as previously described (26). Briefly, blood was obtained by intracardial (piglets) or umbilical (fetuses) puncture. Cell suspensions from the spleen and liver were prepared in PBS by teasing apart the tissues using a forceps. Erythrocytes were removed from these tissues using hypotonic lysis that also destroys erythroblasts. BM cells were directly flushed from the tibia and/or femur, and leukocytes were purified using a Histopaque-1077 (Sigma-Aldrich, St. Louis, MO) gradient centrifugation (26). All cell suspensions were filtrated through a 70-μm-mesh nylon membrane. Cell suspensions for flow cytometry (FCM) were finally washed twice in PBS containing 0.1% sodium azide and 0.2% gelatin from cold water fish skin (PBS-GEL) whereas those for cell cultures were transferred to cultivation medium (see below). Cell numbers were determined by a hemacytometer.

The following mouse anti-pig mAbs were used as primary immunoreagents: anti-IgHμ (swine IgM H chain, M160 or M154, IgG1), anti–MHC-II (swine MHC-II leukocyte Ag type DR, 1038H-12-34, IgM or MSA3, IgG2a), anti-CD172a (74-22-15, IgG1 or IgG2b), anti-IgLκ (27.2.1, IgG1), anti-IgLλ (27.7.1 or 1g6, IgG1 or K139.3E1, IgG2a), anti-CD14 (MIL-2, IgG2b), and anti-SWC8 (MIL-3, IgM). Specificities of anti-porcine IgHμ Abs (M160 or M154) were confirmed by comparative staining with pan-specific mouse anti-human IgM Ab (CM7, IgG1; Sigma-Aldrich), which gave the same staining profile, and also by Western blot (see below). Goat polyclonal Abs specific for mouse Ig subclasses labeled with FITC, R-PE, PE/Cy7 tandem complex, allophycocyanin, allophycocyanin/Cy7 tandem complex, or PE/Texas Red tandem complex were used as secondary immunoreagents (all secondary reagents were from SouthernBiotech, Birmingham, AL). All immunoreagents were titrated for optimal signal/noise ratios. Primary isotype-matched mouse anti-rat mAbs were used as negative controls. Secondary polyclonal Abs were tested for cross-reactivity (no primary mAbs) and also for cross-reactivity with primary isotype-mismatched mouse anti-pig mAbs.

Staining of cells for FCM was performed as described previously (25–28) by indirect subisotype staining. Briefly, multicolor staining was done using cells that had been incubated with a combination of three or four primary mouse mAbs of different subisotypes. Cells were incubated for 15 min and subsequently washed twice in PBS-GEL. Mixtures of goat secondary polyclonal Abs conjugated with different fluorochromes were then added to the cell pellets in appropriate combinations. After 15 min, cells were washed three times in PBS-GEL and analyzed by FCM. In the case of intracellular staining for IgH and IgL, cells that had been indirectly stained for cell surface molecules were subsequently intracellularly stained using an IntraStain kit according to a protocol recommended by the manufacturer (DakoCytomation, Glostrup, Denmark). In experiments designated for cultures, PBS-GEL was replaced by cultivation medium and/or PBS.

Samples were measured or sorted on standard FACSCalibur or FACSAria III flow cytometers, respectively (BD Immunocytometry Systems, Mountain View, CA). Sorted cells were collected to 1) 1 ml of inactivated FBS (PAA Laboratories, Pasching, Austria) if for cultivation, or 2) empty tubes if for PCR amplification from bulk-sorted cells. Electronic compensation was used to eliminate residual spectral overlaps between individual fluorochromes. Forward side scatter (FSC)–area/FSC-width parameters were used for elimination of doublets. The PCLysis or FACSDiva software (BD Immunocytometry Systems) was used for data processing.

Cell cultures of sorted cells (27) were done in RPMI 1640 medium supplemented with l-glutamine and 25 mM HEPES, 10% FBS, 100 U penicillin, and 0.1 mg/ml streptomycin. Final concentration of cells was always set to 1 × 10⁶ cells/ml. Stromal cultures were established using freshly isolated BM suspensions by a similar approach as described for thymic stromal cells (29). After 7–14 d of incubation when the nonadherent cells were removed during passaging, the monolayers were extensively washed and then used as feeder cells. Allogeneic cultures were used in most of the studies because the origin of stromal cells did not affect the behavior of examined cells in cultures. Syngeneic stromal cultures were used only for testing whether allogeneic cultures produce the same effect. In that case, BM suspensions from the same donor were frozen during the time of preparation of stromal cells. Input stromal cultures were always tested for the unwanted presence of B cell lineage cells by both FCM and CDR3 spectratyping. Pure stromal cells were also involved in each experiment as a negative control. In FCM analyses, stromal cells were gated out from all analyses by their high side scatter and low MHC-II expression.

Detection of different transcripts and gene segments was done on different populations of 30,000–100,000 sorted cells that were dissolved in 5 μl of TRI Reagent per thousand cells. In a particular experiment, only the same amount of sorted cells was used for isolation of total RNA and DNA according to a protocol recommended by the manufacturer (Sigma-Aldrich). Total cDNA was prepared using random hexamer primers. Each cDNA preparation was amplified in six concurrent analyses: 1) TdT, 2) recombination activation gene (RAG), 3) VDJ_H rearrangement, 4) VJ gene rearrangement for the IgL κ (VJκ rearrangement), 5) VJ gene rearrangement for the IgL λ (VJλ rearrangement), and 6) β-actin. Each DNA preparation was amplified in four concurrent analyses: 1) signal joint circle (SJC) for DJ_H rearrangement, 2) SJC for VDJ_H rearrangement, 3) partial DJ_H rearrangement, and 4) nonrearranging portion of IgM constant CH4 region. VDJ_H, VJκ, and VJλ rearrangements were occasionally detected at the DNA level. In some cases, samples were analyzed for the presence of Vκ to KDE rearrangements and RE to KDE rearrangements in their DNA. Each analysis consisted of two rounds of PCR. Amplification of a portion of β-actin from cDNA was used as a control for sample quality and to determine the relative transcript expression and efficiency whereas amplification of a Cμ exon from DNA was used as a positive control. All PCR amplifications were constantly checked on agarose gels. All primers used for amplifications are listed in Supplemental Table I, and their positions in genomic sequences are shown in Fig. 1B–D. The Ig gene complex in swine is simple compared with mice and humans (reviewed in Refs. 30, 31). All porcine V_H genes (IGHV) belong to the ancestral V_H3 family sharing the same leader and framework sequences, and only one J_H segment is functional (Fig. 1B). Thus, a single nondegenerate primer set is needed to recover all VDJ_H rearrangements. The porcine IgLκ and IgLλ loci are also restricted and simplified. Although there are multiple gene families for both IgL isotypes, only two IgLκ families (IGKV1 and IGKV2) containing nine genes and a single Cκ are functional (Fig. 1C). Moreover, the Jκ2 segment constitutes >90% of all IgLκ rearrangements because Jκ3–Jκ5 have noncanonical recombination signal sequences. Similarly, only two IgLλ families (IGLV3 and IGLV8) with 10 genes and two identical Jλ-Cλ cassettes are functional (Fig. 1D). Therefore, few primer sets are needed to recover and study all IgL rearrangements in swine.

Electrophoretic analysis of the CDR3 regions of IgH and IgL rearrangements on polyacrylamide sequencing gels provides a clonotypic analysis of the BCR repertoire. This procedure is called CDR3 length analysis or spectratyping, and it was performed on DNA and cDNA prepared from sorted B lineage populations to show their level of BCR diversification and whether there has been selection for in-frame rearrangements (27–29). Technically, the CDR3 segments of the amplified VDJs or VJs were reamplified in the third round of PCR (see Supplemental Table I for primers) where one of the two primers was ³²P labeled (28). The sequencing gels were subsequently dried and their images were obtained by a fluorescent image analyzer (FLA-7000; Fujifilm, Tokyo, Japan).

Proteins from equal numbers of cells were isolated by RIPA buffer supplemented by the Halt protease and phosphatase single-use inhibitor cocktail according to a protocol recommended by the manufacturer (Thermo Fisher Scientific, Rockford, IL). As a control, a normal porcine adult serum was used. The samples were supplemented by 5% 2-ME and were heated at 95°C for 5 min prior to loading onto Tris-glycine SDS 5% stacking and 8% separating polyacrylamide gels in sample buffer (50 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 15.5 mM EDTA, 0.02% bromophenol blue) under reducing conditions. After electrophoresis, proteins were transferred to a Protran BA83 0.2-μm nitrocellulose membrane (Whatman, Dassel, Germany) by semidry blotting, and membranes were blocked for 1 h with 5% BSA in TBST (50 mM Tris, 150 mM NaCl, 0.1% Tween 20) and incubated for 1 h with primary anti-IgHμ Abs. After washing in TBST, the blots were incubated for 45 min with secondary peroxidase-conjugated horse anti-mouse Abs (Cell Signaling Technology, Danvers, MA). The signal was developed with an ECL Plus Western blotting substrate kit (Thermo Fisher Scientific) detected by a Gel Logic 2200 Pro imaging system (Carestream, Rochester, NY). In some experiments, control staining of mouse IgG and IgM mAbs was used. In that case, mouse IgG was detected by the same secondary peroxidase-conjugated horse anti-mouse Abs only, whereas mouse IgM was detected by primary goat anti-mouse IgM followed by secondary peroxidase-conjugated mouse anti-goat Abs.

The earliest precursors of B cells in swine are characterized by the high expression of CD172a on large MHC-II⁺ BM cells (23) (Fig. 1). The present studies were initiated after finding that FCM surface staining cannot detect IgH on these large precursors (Fig. 2A) whereas IgL can be detected (Fig. 2B). Screening of fetal liver and BM during ontogeny revealed similar precursors, and these were not found in circulation or in secondary lymphoid tissues (Supplemental Fig. 1). Intracellular staining for IgH and IgL confirmed that early precursors do not contain IgH (Fig. 2C) but contain IgL (Fig. 2D). However, immature B cells express IgH (Fig. 2E) together with IgL (Fig. 2F). Intracellular staining showed that some portion of the CD172a^lo precursors still contains IgL (Fig. 2H) without IgH (Fig. 2G). The existence of IgL⁺IgH⁻ precursors in the BM (Fig. 2I) but not in the periphery (Fig. 2J) was also confirmed by simultaneous IgL/IgH staining. Western blot analysis further proved that sorted IgL⁺IgH⁻ precursors do not contain IgH (Fig. 2K, black arrow) whereas sorted IgL⁺IgH⁺ immature B cells contain IgH (Fig. 2K, open arrow). Western blot analysis was also used to show that the anti-IgHμ Abs are IgHμ-specific, and that the secondary Abs did not recognize swine Igs (Fig. 2K, rest of the gel).

Previous studies characterized B cell development in swine and identified seven subpopulations developing from subset 0 to subset 6 (23). These subsets can be distinguished by cell size and expression of CD172a on MHC-II⁺ BM cells (Fig. 1A). Because of the detection of free IgL proteins on the surface of subsets 1 and 2, we have subdivided these subsets and designated them 1a, 1b, 2a, and 2b in this study (Fig. 3A). FCM sorting of individual subsets (Fig. 3B) followed by PCR amplification of DNA and cDNA revealed that precursor subset 0 does not contain any rearranged IgH genes, but transcripts for RAG and both types of IgL rearrangements were detected both in DNA and cDNA levels (Fig. 3C). At this stage of development, the level of IgLκ rearrangements was always higher than for IgLλ, and in some experiments only IgLκ rearrangement was detected. Subsets 1a and 1b showed also both types of IgL rearrangements as well as transcripts for RAG and, additionally, evidence of partial DJ_H rearrangements (Fig. 3C, DJ_H), including SJC for these partial DJ_H rearrangements (Fig. 3C, SJC DJ_H). In subsets 2a and 2b, the first signs of IgH gene rearrangements were identified along with SJC (Fig. 3C, SJC VDJ_H). From subset 3 and onward, IgH and IgL gene rearrangements and transcripts were routinely detected. Interestingly, no apparent downregulation of RAG or TdT expression between preB-I (subsets 2–3) and preB-II cells (subsets 4–5) was seen as has been reported in mice (4).

Our finding of preferential IgL rearrangements (Figs. 2, 3) prompted in vitro cultivation of cells at different stages of development. Individual subsets were sorted by FCM under aseptic conditions, as were those used for PCR studies (Fig. 3B). The sorted cell populations were then cultivated for 4 d at 4°C or at 37°C in medium only or at 37°C in medium on a bed of BM stromal cells (Fig. 4). Stromal cells were included because previous findings revealed that development of subset 3 is not supported in medium alone (23). Stromal cells were always checked for their phenotype (large MHC-II^−/lo cells) and an absence of MHC-II^hi B cell lineage cells (data not shown). As might be expected, cultivation at 4°C did not lead to any phenotypic changes (Fig. 4, all dot plots for 4°C) whereas cultivation at 37°C caused a variable decrease in expression of CD172a (Fig. 4B–E). This is consistent with the developmental pathway for porcine B cells (23). However, cultivation of subsets 0–2b did not result in complete loss of CD172^bri cells (Fig. 4A–D). Rather, even cells expressing CD172^hi/lo after sorting generated a portion of CD172^bri cells (Fig. 4B–D). These observations correspond with our earlier findings that a portion of early B cell lineage cells can revert to myeloid cells (23). A third observation was that cultivation of subsets 0–2b generated few CD172⁻IgL⁺ immature B cells (Fig. 4A–D). This is also the case for subset 3, but only when cultivated with stromal cells (Fig. 4E). This indicates that some of the early subsets can generate immature B cells directly but that subset 3 cannot develop further without help from stromal cells. This finding confirms the presence of a developmental checkpoint associated with the subset 3 described earlier (23). The developmental block cannot be reached by all upstream subsets after 4 d. This concerns mainly very early populations (0–1a) that develop downstream but not into population 3 (Fig. 4A, 4B). Further developmental stages (populations 1b, 2a, and 2b) develop partially into subset 3 but do not develop further (Fig. 4C, 4D). More time is probably needed for the earliest subsets to develop into subset 3. Unfortunately, prolonged cultivation could not be done because of the excessive dying of cells (data not shown). With regard to the stromal cells, there was no apparent effect on the number of CD172⁻IgL⁺ immature B cells generated by cultivation of subsets 0–2b (Fig. 4A–D). Alternatively, stromal cells seem to have a positive effect on the genesis of CD172⁻IgL⁺ immature B cells in subsets 3–5 (Fig. 4E–G). Fourth, cultivation of subsets 0–2a with stromal cells generated significant amounts of CD172⁺IgLλ⁺ cells but almost no CD172⁺IgLκ⁺ cells (Fig. 4A–C). This preferential generation of IgLλ⁺ cells was especially apparent for subsets 1b plus 2a (Fig. 4C). Thus, subsets 1b plus 2a remain generally IgLλ⁺ during cultivation (5 plus 44% in medium alone and 4 plus 47% with stromal cells) whereas IgLκ expression is mostly lost (2 plus 2% in medium alone and 2 plus 3% with stromal cells). These results indicate that IgLλ⁺ cells accumulate during early phases of B cell development whereas IgLκ⁺ cells disappear. Finally, subset 4 (Fig. 4F) generated significantly more IgLκ⁺ immature B cells (18% in medium alone and 21% with stromal cells) than IgLλ⁺ immature B cells (2% in medium alone and 5% with stromal cells). In contrast, subset 5 (Fig. 4G) generated significantly more IgLλ⁺ immature B cells (10% in medium alone and 20% with stromal cells) than IgLκ⁺ immature B cells (4% in medium alone and 13% with stromal cells). These results indicate that IgLκ⁺ immature B cells are preferentially generated from subset 4 whereas IgLλ⁺ immature B cells can be also generated from subset 5.

Sorted and thereafter cultivated subsets (Fig. 4) were also subsequently examined for their CDR3 spectratypes to characterize the rearrangements (Fig. 5). This length analysis offers 1-bp resolution and allowed determination of whether a particular subset had undergone selection for productive rearrangement, indicated by almost exclusive presence of in-frame bands, or whether no selection had occurred such that many out-of-frame bands would be present. Moreover, results can also show whether the rearranged repertoire is expanded, resulting in a diversified distribution of bands, or whether diversification is restricted as represented by a scattered distribution of bands. We choose to analyze transcripts rather than DNA because: 1) both productive and nonproductive rearrangements are transcribed (33), 2) the yield of PCR amplification is much higher for cDNA so that smaller amount of cells can be used for population studies, and 3) the analysis of cDNA reflects the actual transcription activity. In any case, comparison of DNA and cDNA amplifications did not reveal considerable differences (Supplemental Fig. 2).

Staining, sorting, and cultivation were carried out by the same procedure as described in Fig. 4. CDR3 spectratypic analyses confirmed that IgH transcripts were absent in sorted subset 1a–1b and infrequent in subsets 2a–2b when cells were cultivated at 4°C, that is, the equivalent of freshly sorted cells (Fig. 5C, 5F, strip 4). However, both IgLκ (Fig. 5A, 5D, strip 4) and IgLλ (Fig. 5B, 5E, strip 4) transcripts were recovered from these early subsets. A comparison of the IgLκ and IgLλ transcripts in early subset 1a showed that the initial IgLκ repertoire was more diverse and contained many out-of-frame transcripts (Fig. 5A, 5D, strip 4). The initial IgLλ repertoire in subset 1a also showed prominent out-of-frame transcripts but the repertoire was more restricted (Fig. 5B, 5E. strip 4). Further data showed that the CDR3 diversity of IgLκ transcripts decreases as the cells develop to subsets 1b–3 (Fig. 5A, 5D, strip 4), whereas during the same transition, the diversity of IgLλ transcripts increases (Fig. 5B, 5E, strip 4). These results show that in an absence of IgH gene rearrangement, the initially diversified IgLκ repertoire becomes restricted as B lineage cells develop whereas the initially restricted IgLλ repertoire is becoming diversified. This trend becomes more prominent when sorted subsets 1a–2b were cultivated in medium alone (compare Fig. 5A and 5D, strip M for IgLκ spectratyping, with Fig. 5B and 5E, strip M for IgLλ spectratyping, respectively).

Apparently, there is little transcription of IgLκ (Fig. 5A, 5D, stars), IgLλ (Fig. 5B, 5E, stars), and IgH rearrangements (Fig. 5C, 5F, stars) when subset 3 is cultivated in medium alone. However, this apparent transcriptional block is overcome when subset 3 was cultivated together with stromal cells (Fig. 5, S strips). These results agree with the FCM analyses (Fig. 4) and demonstrate that subset 3 has a transcriptional block in the absence of stromal cells. These findings also correlate with the earlier described developmental checkpoint in subset 3 (23). Interestingly, cultivation of subset 3 in medium alone also resulted in considerable differences in IgH spectratypic profiles between cDNA and DNA that were not observed in any other samples (Supplemental Fig. 2). The apparent lack of IgH transcripts was accompanied by an entirely unselected IgH repertoire in DNA. The finding that in-frame IgH rearrangements had the same intensity as out-of-frame rearrangements indicates a loss of selection for productive rearrangements and subsequent transcription blockade in the absence of stromal cells.

The analyses of later developmental stages in subsets 4–6 reveal that they contain a selected IgH repertoire (Fig. 5C, 5F, strip 4) and that selection for in-frame rearrangements was similar when cells were cultivated in medium alone (Fig. 5C, 5F, strip M) or with stromal cells (Fig. 5C, 5F, strip S). These findings indicate that subsets 4–6 went through selection for productive IgH gene rearrangements, which is consistent with earlier findings (23). However, analyses of IgL transcripts disclose that selection for in-frame rearrangement is dependent on the type of IgL for which cells were sorted. Specifically, Fig. 5A shows that CDR3 spectratypes for IgLκ transcripts were always selected in subsets 4 and 5 but unselected in subset 6 for IgLλ sorting (Fig. 5A, arrow). The same is true for IgLλ transcripts in a case of IgLκ sorting (Fig. 5E, arrow). Oppositely, Fig. 5B shows that IgLλ transcripts were always selected in subset 6 but unselected in subsets 4 and 5 for Igλ sorting (Fig. 5B, arrows). A similar situation was found for IgLκ transcripts in the case of IgLκ sorting (Fig. 5D, arrows). These results indicate that only productive IgL gene rearrangements can be expressed on the surface of cells. Furthermore, both IgLλ⁺ and IgLκ⁺ late B lineage cells with IgH gene rearrangements also contain IgL transcripts for the second type of IgL that is mostly nonproductive. To further elucidate these conclusions we have performed a more detailed sorting described below.

To investigate late B lineage precursors with IgH gene rearrangement, more stringent sorting conditions for subsets 4 and 5 were established. Importantly, note that all earlier sorting procedures were set to acquire strictly IgL⁻ cells for subsets 4 and 5. However, subsets 4 and 5 also contain IgL^lo cells. For this reason, we have set up a new sorting strategy (Fig. 6) to sort IgLλ⁻ and IgLλ^lo cells for large and small MHC-II⁺CD172a⁻ BM cells (Fig. 6A), and the same strategy was done for IgLκ sorting (Fig. 6B). As is evident from CDR3 analyses of sorted cells, IgL^lo populations always contained selected in-frame transcripts for the same type of IgL (Fig. 6D, 6E, populations I and III) whereas depleted IgL⁻ populations always contained a selected CDR3 repertoire for the second type of IgL (Fig. 6C, 6F, populations II and IV). This is a clear demonstration that the surface staining discriminates two types of IgL in an allelic-exclusive manner. Thus, IgLλ⁺IgLκ⁻ cells have productive IgLλ gene rearrangements in IgLλ⁺ and IgLκ⁻ fractions but nonproductive gene rearrangements in IgLλ⁻ and IgLκ⁺ fractions. Similarly, IgLλ⁻IgLκ⁺ cells have productive IgLκ gene rearrangements in IgLλ⁻ and IgLκ⁺ fractions but nonproductive gene rearrangements in IgLλ⁺ and IgLκ⁻ fractions. Importantly, these results also indicate that IgLκ⁺ as well as IgLλ⁺ B cells may contain rearrangements for the second type of IgL, which is different from the mouse/human paradigm that IgLλ genes rearrange only when IgLκ genes are consumed (5, 6).

Detection of IgLλ and IgLκ on the surface of precursors in the absence of IgH prompted us to sort MHC-II⁺ BM cells by gating on those bearing IgLλ and/or IgLκ (Fig. 7A), and then to analyze these for size and CD172a expression (Fig. 7B–F). As shown earlier, CD172a expression decreases as B cell lineage cells develop further (23). Unexpectedly, we found three populations of IgL^lo BM cells that were composed of IgLλ⁻IgLκ^lo (R1), IgLλ^loIgLκ^lo (R2), and IgLλ^loIgLκ⁻ (R3) subsets (Fig. 7A). Analysis of the sorted IgLλ⁻IgLκ^lo subset (Fig. 7A, R1) showed that they were composed almost exclusively of large CD172a^bri/hi early precursors (Fig. 7B) that contained only IgLκ and RAG transcripts (Fig. 7H).

The sorted IgLλ^loIgLκ^lo subset (Fig. 7A, R2) also contained large CD172a^hi precursors, but CD172a^bri cells were almost missing (Fig. 7C). Interestingly, cells with surface IgLλ and IgLκ proteins lacked detectable IgLκ transcripts but contained IgLλ transcripts and carried Vκ to KDE and RE to KDE rearrangements (Fig. 7I). An absence of IgLκ transcripts in cells with Vκ–KDE and RE–KDE rearrangements indicates that IgLλ^loIgLκ^lo cells had undergone IgLκ gene rearrangement and the Cκ gene segment had been deleted. Therefore, the presence of IgLκ proteins on the surface of IgLλ^loIgLκ^lo precursors is a remnant of a previous translation event. We interpret these findings as evidence that IgLκ precedes IgLλ gene rearrangement as has been described in mice (5–7), but the initial IgLκ locus activity is quickly replaced by rearrangement in the IgLλ locus prior to the onset of IgH gene rearrangement. Neither IgLλ⁻IgLκ^lo (Fig. 7H) nor IgLλ^loIgLκ^lo (Fig. 7I) precursors carry partial DJ_H or full VDJ_H rearrangements. To further investigate the loss of IgLκ expression, IgLλ⁻IgLκ^lo and IgLλ^loIgLκ^lo cells were sorted and subsequently cultivated for 1–5 d (Fig. 7M). Both populations were shown to lose IgLκ expression and differentiate into IgLλ^loIgLκ⁻ cells. Kinetic studies demonstrated that whereas IgLλ⁻IgLκ^lo precursors develop into >90% IgLλ^loIgLκ⁻ cells in ∼3 d, IgLλ^loIgLκ^lo precursors do so in <1 d. These results indicate that the loss of surface IgLκ expression in IgLλ^loIgLκ^lo cells after IgLλ gene rearrangement occurs in hours.

In accordance with cultivation studies (Fig. 7M), the freshly sorted IgLλ^loIgLκ⁻ subset (Fig. 7A, R3) did not contain IgLκ transcript but did contain IgLλ, and also contained partial DJ_H and full VDJ_H rearrangements as well as Vκ–KDE and RE–KDE rearrangements (Fig. 7J). FCM analysis showed that these IgLλ^loIgLκ⁻ cells consisted of some large CD172a^hi precursors but mainly of their progenies, including small and large CD172a^lo and CD172a⁻ cells (Fig. 7D).

Finally, we analyzed the sorted IgLλ⁻IgLκ^hi (Fig. 7A, R4) and IgLλ^hiIgLκ⁻ (Fig. 7A, R5) BM cells and found that they represent exclusively CD172a⁻ immature B cells (Fig. 7E and 7F, respectively). Both populations contained IgH and IgL transcripts and other products of rearrangements (Fig. 7K, 7L). Combined with other data presented above, these studies indicate a developmental transition from IgLκ to IgLλ transcription that is followed by the second wave of IgLκ gene rearrangement.

Our findings indicate that in swine the order of IgH and IgL loci recombination is reversed from that in mice and as described in textbooks (3, 4). This results in the expression of IgL without IgH in early B cell precursors (subsets 0–1). The observation that IgL gene rearrangement can be initiated before IgH gene rearrangement and pre-BCR expression in nonengineered swine supports the conclusion that IgH and IgL gene rearrangements are independent. The same was predicted from studies done using IgH-deficient mice (2, 10, 13, 34) and humans (35).

The reverse order of IgH and IgL gene rearrangements brings into question the concept of allelic exclusion because IgL gene rearrangement is not controlled by productive IgH gene rearrangement. However, an authentic IgL gene rearrangement has been shown to serve in selection of the IgH repertoire when SLC is deleted from the genome (14, 36). Presumably, this occurs naturally in swine that lack a SLC (18). Only targeted disruption of the membrane exon of IgH genes causes allelic inclusion (34). Studies in chickens demonstrate that allelic exclusion occurs when IgL and IgH genes are competitively rearranged (15, 16, 37). Furthermore, sharks display allelic exclusion even though they have many functional IgL genes that are prerearranged in the germline and subsequently expressed (38). Thus, allelic exclusion appears to be independent of SLC and occurs irrespective of the IgH/IgL recombination order.

Although IgL precedes IgH gene rearrangement in the pig, IgL gene rearrangements occur in the same stepwise order as in humans and mice, starting with IgLκ genes and proceeding to IgLλ genes (5–7). However, initial IgLκ gene rearrangements in swine are rapidly replaced by IgLλ gene rearrangement before recombination of IgH genes, giving the impression that IgLλ precedes IgLκ rearrangement in swine (18). The first substantial IgH gene rearrangement occurs in subset 3, and these IgLλ^loIgLκ⁻ precursors can generate only IgLλ⁺ B cells. Most IgLκ⁺ B cells are generated subsequently from subsets 4 and 5 (Fig. 4). This indicates that IgLκ gene rearrangement occurs in two successive waves in different developmental stages with IgLλ gene rearrangement in between. Thus, most IgLλ⁺ B cells are generated earlier whereas most IgLκ⁺ B cells arise later. Moreover, the second wave of IgLκ gene rearrangement occurs in the presence of IgLλ gene rearrangement(s). This differs from the mouse-based paradigm in which IgLλ genes rearrange only when IgLκ genes are consumed, which results in >90% of IgLκ⁺ B cells (6).

The present study confirmed the developmental checkpoint in subset 3 when the selection for a productive IgH gene rearrangement occurs (23). These small MHC-II⁺CD172a^lo precursors cannot develop further without BM stromal cells whereas it is possible for subsequent developmental stages (Figs. 4, 5). This closely resembles mouse preB-II cells that do not need the BM and/or cytokines for expansion once they express functional pre-BCR (1). FCM showed that subset 3 almost exclusively expresses IgLλ proteins. The lack of stromal cells (or BM environment) should therefore lead to accumulation of IgLλ⁺ cells. This is exactly what we described in vivo during early ontogeny (18) when BM is not functional (32), which led us to the incorrect conclusion that IgLλ may precede IgLκ gene rearrangement in swine (18). The differences in the ability of fetal liver and BM to support B cell development (4) appear to be related to the developmental checkpoint in subset 3 in which the IgLλ⁺ B cell accumulates. Differences in this checkpoint or its timing might explain why certain species express >90% of IgLλ proteins such as in horses, cows, sheep, cats, or dogs (39) rather than being simply attributed to a disproportionate repertoire of germline IgLκ versus IgLλ genes (30, 31). Thus, it would be interesting to further study these species to establish the order of IgL and IgH gene rearrangement and whether they possess a consensus SLC and the IgLλ/IgLκ ratio throughout B cell development. Studies in transgenic mice that have prolonged time for successive IgL gene rearrangement also showed a shift for higher usage of IgLλ over IgLκ (40).

One of the striking observations of this study was the presence of free IgL on the surface of B cell precursors in the absence of IgH gene rearrangement. Only productive IgL were expressed on the cell surface (Figs. 5, 6), suggesting that nonproductive IgL are not translated and/or are unable to be expressed on the surface. The expression of free IgL on a surface of precursor probably has no functional role because IgL cannot make signal without association with IgH (41), although they can serve for further selection of the IgH repertoire (14, 36). The expression of productive IgL gene rearrangement in mice is controlled by productive IgH gene rearrangement (35, 36), whereas our findings in this study thus indicate that productive IgL are able to escape from the endoplasmatic reticulum without chaperoning by IgH. A similar phenomenon was observed in humans where IgL produced in excess of IgH are secreted as free IgL (42) and can then associate with the outer membrane of cells via interaction with phospholipids such as sphingomyelin A (43). Moreover, free IgL associate only on the surface of cells that produce these IgL (43). Our results also exclude the possibility that free IgL on a surface can be adventitiously acquired from other sources because intracellular staining showed the same IgL isotype inside of cells, and sorting revealed rearranged IgL in DNA and in transcripts of the same IgL isotype. In any case, further analyses are needed to explain the attachment of free IgL on the cell surface and their possible role in B cell selection.

Our data indicate that swine do not use an invariable SLC but rather authentic IgL. This could be an advantage because each type of IgL could serve as a different type of SLC without excessive skewing of IgH repertoire (36). Usage of authentic IgL instead of SLC for selection of the IgH repertoire may not be limited to swine because inactivation of SLC in mice does not prevent an initiation of IgL gene rearrangement or development of mature B cells, and it causes only a temporal decrease in a number of B cells, which is compensated for latter in life (14, 36). Furthermore, mice deficient in SLC also have normal IgM serum levels (14) and immune responses (36). One may speculate that SLC is only important for species in which IgH precede IgL gene rearrangement. Another group of species may lose (or did not develop) λ5 gene (37), including swine (18, 24). It remains to be investigated whether a pre-BCR–independent developmental pathway is evolutionarily more ancient than the pre-BCR–driven pathway. Our current belief is that IgH before IgL gene rearrangement is the most ancestral because this is used in amphibians (44). In any case, one group of animals might increase an efficiency of B cell generation by employing components of SLC whereas others might invert the order of IgH and IgL gene rearrangement or employ yet other mechanisms such as gene conversion in chickens (37) or usage of prerearranged IgL genes in sharks (38).

We thank Lucie Poulova, Mirka Kratochvilova, Sarka Pfeiferova, and Blanka Dusankova for excellent technical assistance. Our warm gratitude also goes to John E. Butler and Nancy Wertz (University of Iowa, Iowa City, IA) for discussions and critical reading of the manuscript.

The authors have no financial conflicts of interest.