In higher vertebrates, immunoglobulin (Ig) gene rearrangements, B-cell development, and the generation of the primary antibody repertoire occur in a variety of ways. In mouse and human, they occur mostly in bone marrow. In other higher vertebrates, Ig gene rearrangements and B-cell development can occur in tissues such as spleen and yolk sac. Many of these species, including chicken, rabbit, cattle, and sheep, also use gut-associated lymphoid tissues (GALT) for development of the primary antibody repertoire. In these species, B-lineage cells migrate to GALT, where B-cell expansion and somatic diversification of Ig genes occurs. Generally in these species, B lymphopoiesis occurs early in development and does not continue throughout life.

Whereas mouse and human use combinatorial joining of multiple V, D, and J gene segments to generate a large anti- body repertoire, most members of the GALT-species utilize only a limited number of Ig gene segments in V(D)J gene rearrangements. Further, unlike mouse and human, which possess VH genes from all three VH groups, the VH genes of the GALT-species generally belong to only one VH group, either group B or C (Sitnikova and Su, 1998). Likewise, the Vk and Vl genes of the GALT-species generally belong to only one group. Another difference between members of the GALT-species and mouse and human is that some of the GALT-species utilize gene conversion as well as hypermu- tation to somatically diversify their Ig genes. We review B- cell development and generation of the antibody repertoire in higher vertebrates, excluding mouse and human, high- lighting the importance of GALT in these processes.

Although the chicken heavy and light chain loci contain multiple V, (D), and a single J gene segments, the loci differ from those of other species in that they contain only one functional V and J gene segment. The sole functional light chain V gene segment,Vl 1, lies 1.8 kb upstream of a single Jl gene segment and about 3.6 kb upstream of a single Cl locus. Twenty-five V pseudogenes lie within 19 kb upstream ofVl 1. Most of these pseudogenes lack recombination signals, and all of them lack upstream promoter and leader sequences. The pseudogenes lie in both transcriptional orientations with respect toVl 1 (Reynaud et al., 1985, 1987).

The only functional heavy chain V gene segment,VH1, lies about 30kb upstream of the Cm locus and 15kb upstream of a single JH gene segment (Reynaud et al., 1989, 1991). Eighty to one-hundred VH pseudogenes spaced an average of 0.8kb apart lie within a 60- to 80-kb region upstream ofVH1. The VH pseudogenes lack recombination signals, as well as promoter and leader sequences, and the 3¢ end is fused to a D-like segment. As discussed below, this structural feature facilitates VDJ gene diversification. As in the Vl locus, VH pseudogenes lying in opposite transcrip- tional orientation with respect toVH1 are interspersed among VH pseudogenes sharing the same orientation asVH1.

Although the heavy chain locus contains only a single functional VH and a single JH gene segment, it contains sixteen functional D gene segments. Fifteen of the D gene segments are, however, highly homologous, with several encoding the same amino acids (Reynaud et al., 1991). All of the D gene segments exhibit a strong bias for usage in

Immunoglobulin Genes and Generation of
Antibody Repertoires in Higher Vertebrates:
A Key Role for GALT

reading frame 1, which encodes primarily hydrophobic and aromatic amino acid residues, particularly Gly, Ser, Ala, Tyr, and Cys. A similar D reading frame bias has been reported in other species, suggesting that the amino acid composition of the D region is important for light chain pairing or selec- tion of developing B cells.

We do not yet understand why multiple, highly homolo- gous D gene segments have been maintained in the chicken heavy chain locus. Because they are utilized with strong reading frame bias, their contribution to antibody repertoire diversity can be only minor. Reynaud et al. (1991) suggested that their major contribution to heavy chain diversity might be through the formation of D–D junctions. They found such junctions in 25% of DJ gene rearrangements and in 20% of rearranged but preselected VDJ genes in bursal B cells. The formation of D–D junctions is unexpected because chicken D gene segments are flanked at both ends by 12-bp recom- bination signal sequences; D–D recombination therefore appears to violate the 12/23 base pair rule. D–D junction formation expands the range of potential heavy chain CDR3 lengths to fifteen to thirty amino acids. There appears to be no requirement in the chicken for matching CDR3 lengths during pairing of the heavy chain with the light chain, in which CDR3 varies little in length (McCormack et al., 1989c; Reynaud et al., 1991).

DHJH gene rearrangements are first detectable by PCR analysis in the yolk sac at day 5 to 6 of embryogenesis and at other hematopoietic sites a few days later (Reynaud et al., 1992). The first fully rearranged V(D)J genes are detectable at embryonic day 8 to 9 in the spleen, blood, and yolk sac, and V(D)J genes are detectable in the bursa 1 to 2 days later (Reynaud et al., 1992). Ig gene rearrangement can, however, occur after B-cell precursors migrate to the bursa. This was demonstrated by the isolation of retrovirally transformed B- cell precursors containing only DJ rearrangements from the bursa at embryonic day 12 (Banatar et al., 1992). The detec- tion of B-cell recombination excision circles (BRECs) in the embryonic bursa is also consistent with ongoing Ig gene rearrangement (McCormack et al., 1989b; Reynaud et al., 1992). The bursa, however, is not required for the induction or completion of Ig gene rearrangement, as B cells express- ing surface IgM were produced in chickens from which the bursa was removed early in embryonic development (60 hours of incubation) (Jalkanen et al., 1983).

The simultaneous appearance of the first detectable VDJ and VJ gene rearrangements at embryonic day 8 to 9 shows that the chicken heavy and light chain loci are not rearranged in the sequential manner characteristic of murine B cells (Reynaud et al., 1992). In murine B cells, VDJ rearrange- ment generally precedes VJ rearrangement by 2 to 3 days (Osmond, 1991). In contrast, analysis of Ig gene rearrange-

ments in individual retrovirally transformed chicken B cells from embryonic day 12 bursa demonstrated that, although heavy chain DJ gene rearrangement appears to occur first, it can be followed by either VJ or VDJ gene rearrangement (Banatar et al., 1992). Chicken Ig genes therefore appear to be rearranged stochastically. Consequently, rearrangement of the chicken light chain locus is not regulated by the expression of a productively rearranged heavy chain gene. Unlike human and murine B cells, chicken B cells therefore probably do not express a pre-B cell receptor (pre-BCR).

Unlike murine B cells, few chicken B cells contain nonproductively rearranged Ig genes. In more than 90% of chicken B cells, the unexpressed light chain locus is in germline configuration, and the unexpressed heavy chain locus generally contains a DJ gene rearrangement. The frequency of productive Ig gene rearrangement, however, appears to be no higher in chicken B cells than in those of mice (McCormack et al., 1989b; Reynaud et al., 1991). Allelic exclusion in chicken B cells is therefore not mediated by the expression of a productively rearranged allele but by an alternative mechanism that is not currently understood. This mechanism probably evolved to limit the possibility of out-of-frame rearrangements being rendered productive by gene conversion. It has been proposed that the restriction of Ig gene rearrangement to a brief period during embryonic development imposes a time constraint that limits the probability that both alleles of the heavy and light chain loci will be rearranged. Lauster et al. (1993) demon- strated that the V-J intron contains a negative regulatory element with strong transcriptional silencing activity. Tran- sient inactivation of this silencer after DJ gene rearrange- ment would provide the recombination machinery with a brief period of access to the locus, with rearrangement pre- sumably restoring silencer activity at the other allele. Such a regulatory mechanism would make it unnecessary to coor- dinate downregulation of the RAG genes with the expres- sion of a functional Ig molecule. Expression of the RAG genes, in fact, continues in the bursa, where RAG-2 is expressed at high levels and RAG-1 at much lower levels (Carlson et al., 1991; Reynaud et al., 1992). It is nonethe- less not known how time-restricted regulation of Ig gene rearrangement could allow the efficient rearrangement of one heavy and one light chain locus while keeping the fre- quency of double rearrangements at both loci low.

Although B lymphopoiesis and Ig gene rearrangement continue throughout life in mice and humans, these processes occur only during embryonic development in the chicken. In allotype suppression experiments in which chicken embryos heterozygous for the IgM allotype were treated with anti-allotype antibodies, severe long-term dele- tion of IgM allotype-expressing B cells was induced, thus

demonstrating that B lymphopoiesis does not continue after hatch (Ratcliffe and Ivanyi, 1981). Similarly, BRECs produced during V(D)J gene rearrangement were readily detectable in the spleen and bursa during embryonic devel- opment but were no longer detectable shortly before hatch (McCormack et al., 1989c; Reynaud et al., 1992). Restric- tion of B lymphopoiesis to an early stage of development implies that long-lived and/or self-renewing B cells main- tain the adult chicken B-cell compartment.

In mice and humans, combinatorial rearrangement of multiple V, (D), and J gene segments generates an enormous number of unique Ig heavy and light chains. In contrast, all chicken B cells utilize the sameVl1,Jl,VH1, andJH gene segments during rearrangement of the light and heavy chain loci (Reynaud et al., 1985, 1989). Although several func- tional D gene segments are present in the chicken heavy chain locus, their contribution to VDJ gene diversity is rel- atively minor, for reasons discussed above. In addition, little junctional diversity is generated during joining of V, D, and J gene segments, further restricting potential heavy chain diversity (McCormack et al., 1989c). In contrast to humans and mice, therefore, little antibody diversity is generated in the chicken during Ig gene rearrangement.

Instead of generating antibody diversity, Ig gene rearrangement in chickens generates substrates for gene conversion, the primary mechanism by which Ig genes are diversified. Gene conversion transfers tracts of nucleotides from upstream V pseudogenes into equivalent regions of the rearranged VJ and VDJ genes (Reynaud et al., 1987, 1989; Thompson and Neiman, 1987). This process introduces novel nucleotide sequences into the rearranged VJ and VDJ genes while leaving the donor V pseudogenes unchanged (Carlson et al., 1990). Although Vl pseudogenes share a high degree of homology withVl1, they differ in nucleotide sequence primarily in and around the CDRs, which ulti- mately contribute to the antigen-binding site of the antibody molecule (Reynaud et al., 1987). A similar concentration of nucleotide differences in the CDRs occurs among VH pseudogenes. VH pseudogenes are homologous to fusedVH1 and D gene segments, and gene conversion events can introduce nucleotide changes into the D region, as well as intoVH1, in rearranged VDJ genes (Reynaud et al., 1989). Pseudogene usage is influenced by several factors, including sequence homology, transcriptional orientation, and prox- imity to the rearranged V-gene segment (McCormack and Thompson, 1990). Each rearranged VJ and VDJ gene can undergo several gene conversion events involving a variety of pseudogene donors, resulting in the generation of con- siderable heavy and light chain diversity. In another avian species, the Muscovy duck, rearrangement of several dif- ferent VL gene segments, in addition to gene conversion,

contribute to Ig light chain diversity. Hence, not all birds are restricted to gene conversion–mediated Ig gene diversifica- tion (McCormack et al., 1989a).

Diversification of chicken Ig genes occurs during B-cell development in the bursa of Fabricius, a lymphoid organ associated with the hindgut (Figure 28.1). Between days 8 and 15 of embryogenesis, the bursa is colonized by a single wave of B-cell precursors (Houssaint et al., 1976). B-cell precursors first populate mesenchymal tissue within the bursa, and 20,000 to 40,000 of these precursors subse- quently migrate across the bursal epithelial basement membrane and begin proliferating in epithelial buds that ultimately develop into bursal follicles (Pink et al., 1985; Ratcliffe et al., 1986). During proliferation, these cells accumulate somatic gene conversion events at their Ig loci, resulting in the generation of a diversified primary antibody repertoire. Shortly before hatch, B cells begin emigrating from the bursa into the periphery, and emigration continues until the bursa involutes several months after hatch.

An important function of the bursa during embryonic development is to selectively expand B-cell precursors that have productively rearranged their Ig genes and are thus able to express surface IgM. Although few V(D)J gene rearrange- ments isolated from the embryonic bursa at days 10 to 13 are productive, their incidence increases markedly over sub- sequent days so that more than 90% of isolated V(D)J gene rearrangements are productive by day 18 (McCormack et al., 1989b; Reynaud et al., 1991). Critical B lineage develop- mental checkpoints include precursor transit across the basement membrane and proliferation within epithelial buds (Reynaud et al., 1992). Both are dependent on surface IgM expression. Because B-cell precursors encounter these checkpoints before diversifying their Ig genes, which is also when they express nearly identical B cell receptors (BCR),