V(D)J Recombination: Of Mice and Sharks

Origins of the Adaptive Immune System

V(D)J recombination is the process by which antigen receptors, immunoglobulin (Ig) and T-cell receptor (TCR), are assembled for expression during development of the respective B- and T-lymphocytes. Somatic rearrangement of the V (variable), D (diversity) and J (joining) gene segments¹ is initiated by the recombinase RAG (recombination-activating gene)^2,3 in a cut-and-paste process that entails joining of these separate gene components to encode the V region, the N-terminus of the receptor polypeptide. The V region is 100-120 amino acid long and forms the ligand-binding site in heterodimers of heavy (H) and light (L) chains of Ig, the alpha and beta chains of TCRαβ and the gamma and delta chains of TCRγδ.

RAG and lymphocytes expressing Ig and TCR are present in all jawed vertebrates (Fig. 1), from cartilaginous fishes to mammals. Neither RAG nor the rearranging receptors are found in protochordates or lamprey and hagfish, which suggests that the present RAG function became established in a vertebrate ancestor sometime in the 80 million years between the divergence of jawless fishes and cartilaginous fishes.^4,5 Extensive duplication events, either two whole-genome duplications or one genome-wide duplication and multiple segmental duplications occurred before and after divergence of jawless fishes.^6-9 The incipience and evolution of the adaptive immune system took place during this period of extensive genomic restructuring.¹⁰

The origin of the rearranging genes was first suggested by Sakano and coworkers,¹¹ who remarked that the recognition motifs (recombination signal sequences, RSS) adjacent the V(D)J gene segments were reminiscent of signals found at the termini of integrated transposable elements. Subsequently the chemistry of the RAG-mediated pathway was found to resemble those described for transpositional recombination by mobile elements.¹² With the growing availability of genome information from different species, it became feasible to attempt delving into the origins of RAG and its recognition sequences. Terminal inverted repeats with motifs and spacer interval similar to RSS were observed in the Transib transposon in nematodes, insects and sea urchin.¹³ RAG is composed of two components and both RAG1- and RAG2-like sequences were detected in the sea urchin genome, although their function is not yet clear.¹⁴ These discoveries, together with demonstration of latent transposase activity in RAG,^15,16 argue for RAG having been part of a DNA transposon that was introduced early into the vertebrate lineage, evolving to its role of V(D)J recombinase by retaining the excision component of transposase activity.^17-19 The presence of RAG sequences in echinoderms could indicate entry of the transposon at a far earlier time and lost in certain phyla and classes (protochordates, jawless fishes) but retained in others, or else a separate horizontal transfer in jawed vertebrates.

It is hypothesized that in the ancestral vertebrate the RAG transposon became integrated into a V-like gene, splitting it into two components that can rejoin after RAG-induced double-strand breakage and removal of the intervening DNA.¹¹ The cleavage occurs in the same place due to RSS recognition, but because of the nucleotide loss and/or gain arising from the repair process, the new joints would be varied in sequence (next section). Breakage and repair of DNA induced by RAG thus generates molecular heterogeneity, which may arguably have been the selecting factor if the original V gene had an immune function that was enhanced by diversified sequences.

V(D)J recombination became established in early vertebrates about 500 million years ago and is the process that assembles Ig and TCR genes in all species and the species-specific receptor genes like the IgNAR (new antigen receptor) and NAR-TCR in sharks^20,21 and TCRμ in marsupials.²² This chapter deals mainly with comparative studies on the Ig gene system. There is considerably more information on antibody in early vertebrates, due to the much longer history of studies of Ig protein and to the relative ease of detecting VH sequences across species with heterologous probes. The IgM molecule is very well conserved from sharks to mammals^23,24 in overall sequence and structure, being the antigen receptor on naive B-cells that in plasma cells is secreted as a polymeric antibody, usually a pentamer. TCR cDNA sequences characterized in all animals show that they are cell surface receptors only. TCRαβ and TCRγδ have been cloned from all classes of jawed vertebrates,^25-29 including all three major groups of mammals (marsupials such as opossum,³⁰ monotremes such as duckbill platypus³¹ and placentals of most orders, including rodents, rabbits, ruminants and primates).

There are two evolutionarily conserved features of V(D)J recombination: the mechanism of RAG action and the regulation of this process to ensure one end result—that only one kind of antigen receptor is expressed per cell (for a review, see ref. 32). This restriction is called allelic exclusion. Although the recombination pathway mediated by RAG is well conserved, the regulated accessibility of the antigen receptor genes to RAG has evolved with the organizational structure, which differs most significantly between cartilaginous fishes and tetrapods (Fig. 2).

RAG Recognition and Joint Resolution

The rearranging elements—the gene segments V, D and J with their adjacent RSS—are present in all classes of jawed vertebrates, as are the key enzymes involved in DNA nicking and modification, RAG1/RAG2 and terminal deoxynucleotidyl transferase (TdT). Although these lymphocyte-specific enzymes have been studied almost entirely in mouse or in vitro systems,³³ their highly conserved mode of action in other animals may be deduced. First, pairwise recognition of the RSS is required and the RSS pair to be recombined must consist of one RSS containing a 12-bp spacer and the other a 23-bp spacer (“12/23 rule”).¹ In all vertebrates where the genomic organization of the gene segments has been determined, the RSS that flank potentially recombinogenic gene segments reflect this pairing relationship.

RAG initiates the pathway that leads to double-strand breaks at either gene segment and the coding ends being subsequently joined by the cell's DNA repair processes (Fig. 3). Double-strand breakage is obtained in a transesterification reaction that results in a covalently closed hairpin on the coding end and a free blunt RSS at the other. The hairpin is opened asymmetrically, creating an overhang with inverted repeat, some of which is occasionally retained (P region) as part of the ligated joint.

The presence of P region is thus indicative of a hairpin intermediate created during the double-strand break and joining process. Examination of the VD and DJ junctions in Ig H chains and TCR β and δ chains, or VJ junctions in Ig L chains and TCR α and γ chains, the portion of the V sequence called CDR3 (complementarity-determining region 3), shows germline contribution (V and J gene segment flanks, portions of D gene sequence) and occasional P region in all animals, suggesting that V(D)J recombination at different loci and in various species undergo the same unique process involving hairpinned coding ends.

Selection for Junctional Diversification

A second category of somatically-generated additions at the junction is N region, which consists of non-templated, mostly GC-rich sequences catalyzed by TdT^34,35 that, together with coding end-processing mediated by exonucleases, are the main contributors to generating the diversification at the junctions of the rejoined gene segments.

The other members of the mammalian Family X DNA polymerases, to which TdT belongs, are not restricted to precursor lymphocytes, but two of them, Pol μ and Pol λ, are also involved at different points during coding end-processing and appear to modulate the extent of coding end nucleolytic processing.³⁶ TdT and Pol μ are very closely related³⁷ and their presence in fishes,³⁸ in contrast to the single-copy ancestral form in the urochordate Ciona (tunicates, Fig. 1), suggests that the lymphocyte-specific TdT evolved to its current role in the immune system by the time of divergence of cartilaginous fishes. The early involvement of TdT in the evolution of V(D)J recombination reflects its importance in amplifying the selected attributes conferred by gene rearrangement: sequence and sequence length variation as a result of RAG-induced breakage.

Diversification mechanisms like mutation or gene conversion may exist in invertebrates³⁹ and predate the rearranging Ig gene system, but these processes do not generate sequence length diversity repeatedly and reliably in one location that will tolerate a loop size spectrum of 2-23 amino acids in the human H chain⁴⁰ or 9-13 in the shark L chain.⁴¹ The greatest contribution to the combining site topology is thus made by the variable CDR3. From crystallographic studies of antigen-antibody complexes H chain CDR3 appears to play the most significant role, not only in the number of contacts with antigen but also in its potential for conformational changes for “induced fit” upon binding ligand.^42,43

Novel Rearranging Genes in Sharks and Marsupials

Some species require antigen receptor diversity additional to that provided by heterodimer specificities of the Ig/TCR repertoire. They are (1) shark IgNAR²⁰: neither Ig nor TCR but an early divergent gene, (2) shark NAR-TCR²¹: TCR isoforms, produced by grafting an additional V region onto an existing TCRδ rearrangement by splicing, forming two successive V regions and (3) marsupial TCRμ²²: a hybrid of Ig and TCR components whose product also contains two joined V regions. IgNAR is a secreted serum protein and the other two presumably are active in cell-mediated processes.

IgNAR is a dimer but the V regions are not paired; the ligand-binding site is thus a single V region. The IgNAR V region is generated by four rearrangements—V, three D and J gene segments—providing highly variable and exceptionally plastic CDR3 that are postulated to adopt multiple conformations for induced-fit binding.⁴⁴ Ig H chain dimers with single-domain V regions (VHH) are also expressed in camels,⁴⁴ although these gene segments are part of the IgH locus.⁴⁵ Shark IgNAR and the camel VHH are the result of convergent evolution, as are shark NAR-TCR and opossum TCRμ. NAR-TCR is part of the shark TCRδ locus, whereas TCRμ in opossum are encoded by independent gene clusters.⁴⁶

TCRμ, like IgNAR, involves rearrangement of 2-3 D elements. Because TCRμ and IgNAR are encoded by a few miniloci, their repertoire is based solely on CDR3 junctional diversity. The use of a single V domain, some with longer CDR3, in cartilaginous fishes and in mammals suggests that there exists some category of antigens that require ligand-binding sites perhaps more flexible than provided by the classical Ig or TCR heterodimer.

One Receptor Per Cell

Both the strength and the weakness of V(D)J recombination is its random nature. An immensely diverse, anticipatory repertoire is generated concomitant with cell and resource wastage. The process cannot ensure that the V becomes joined in-frame with respect to the J (and C region) sequence, so that least two of three rearrangements attempts are nonfunctional. Moreover, randomly-generated specificities also include those that recognize self components and these are eliminated at the immature lymphocyte stage when triggered by a self ligand. Selection for self-tolerance or mounting an immune response is most efficaciously (i.e., specifically) mediated when only one species of receptor is expressed per cell.⁴⁷ The last phenomenon, known generally as allelic exclusion, results from regulated RAG access to the recombinogenic elements. V(D)J recombination is lineage- and cell stage-specific, meaning that DNA from nonlymphoid cells or from cells of the incorrect developmental stage are not acted upon by RAG.⁴⁸ There are, however, some interesting exceptions in cartilaginous fishes and these are described in a later section (“Rearrangement of Ig genes in non B-cells”).

Mouse

In the mouse IgH gene rearrangement takes place in a set order and in step-wise fashion.^59,51 At the pro B-cell stage the chromatin domain encompassing the D, JH and Cμ genes become activated, probably through the intronic enhancer and allow D to JH recombination on both chromosomes. This is followed by activation of the chromatin domain containing the upstream VH genes. Because of the very large distance between the VH gene segments and the DJ, locus contraction and looping of the DNA^52-54 are required to bring them into close proximity for rearrangement (Fig. 2). H chain exclusion is the outcome of the staggering of the V to DJ step between the two alleles. If the first VDJ is not viable, rearrangement continues on the homologous chromosome.

The initiation and maintenance of allelic exclusion involves relocation of the genes in nuclear compartments.^53,55 In pro B-cells IgH repositions away from the nuclear periphery and this may have to do with its activation; in pre B-cells the nonrearranged allele is recruited to the pericentrometric heterochromatin, an interaction thought to be repressive for recombination. How rearrangement begins at one allele before the other is not clear and the basis may differ at the IgH, TCRβ and Igκ genes. An explanation for asynchronous rearrangement at the TCRβ locus has been recently proposed after finding that in rearranging T-cells both alleles of TCRβ interacted with repressive nuclear compartments at equal and high frequency.⁵⁶ This observation suggests a limited window of opportunity to achieve the V to DJ step and that any rearrangement is consequently a very low frequency event. Two simultaneous rearrangements in a cell are thus unlikely to occur and allelic inclusion is avoided.

Chicken

The earliest recombined cells are in the yolk sac at day 5 and 6 of incubation and carry DJ only; VDJ is found on day 9.⁶⁴ Rearrangement occurs exclusively to the D-proximal VH gene, the only functional gene out of multiple VH elements; the other VH act as donor templates during the gene conversion process in the bursa. The primary Ig repertoire in chicken, as in rabbits, is generated by postrearrangement gene conversion.^65,62 There is a distinct DJ/DJ step that is B-lineage specific in chicken and this is followed by simultaneous V rearrangement at the H and L chain loci.^64,66 Clones carrying only the VJ or only the VDJ could be observed,⁶⁶ showing that there is no ordered H and L chain rearrangement, as there exists in mouse and rabbit.⁶¹ Thus, in the chicken, L chain rearrangement is not dependent on the success of H chain rearrangement and there is no pre-B-cell stage as in mammals.

More than 90% of bursal follicles contain the VDJ/DJ configuration and none carried VDJ rearrangements on both chromosomes. Similarly, only one allele of the L chain recombined. It was suggested that the V rearrangement occurs after removal of repression from one allele randomly and that this is an event of such low efficiency that there is little probability of its occurrence on both alleles.⁶⁷

Ig Rearrangement in the Shark

The IgH minilocus organization in cartilaginous fishes, representatives of the earliest vertebrates, is considered primitive and ancestral to the classical IgH locus in other vertebrates. Sharks, rays and skates carry 15-200 miniloci (“clusters”) each consisting of a few gene segments (VH-D1-D2-JH-Cμ)^4,23 as shown in (Fig. 2). In most species the rearranging elements are located within a total span of 2 kb. The clusters themselves are located far apart from each other,⁷² >120 kb and can be situated on different chromosomes.⁷³ V(D)J recombination takes place among the four gene segments of the minilocus; there is no evidence for intercluster rearrangement in B-cells and hence no need for locus contraction in such a system. The close proximity of the gene segments (400 bp apart) also makes unlikely any separately activated chromatin domains within a cluster. In fact, there is no strict order of rearrangement of the VH, D1, D2 and JH. Once an IgH gene is activated in a precursor B-cell, its gene segments recombine all at once and to completion.⁷⁴

In single B-cell studies, few Ig transcripts⁷⁵ and few genomic rearrangements⁷⁴ were observed per lymphocyte. In the nurse shark there are 9-12 functional IgH genes and in any B-cell there are 1-3 VDJ genomic rearrangements of which only one appeared to encode a viable receptor. Less than 10% of the cells carried any partially rearranged genes and the rest of the IgH genes were in germline configuration. This suggests that once initiated, recombination occurs efficiently between the four gene segments. These data show that H chain exclusion exists in the shark, despite its unique IgH organization. As in higher vertebrates, H chain exclusion in sharks is based on limitation of rearrangement, but the mechanism of repression (or activation) must accommodate the large and varied numbers of IgH loci in different cartilaginous fish species.

The process producing monoallelic Ig H chain expression at the murine IgH locus evolved with and is a consequence of the complex gene organization, whose multiple gene segments are scattered over 2 Mb. If you take away the locus contraction and the separately activated domains, what shark and mouse have in common is that initiation of rearrangement is an inefficient, low frequency event. Whether there are regulatory features in common between shark and tetrapod IgH gene systems remains to be established. However, a few conclusions can be extracted. Because of the large number of IgH loci and their dispersed locations, it is unlikely that H chain exclusion in the shark is based on any mechanism that predetermines^77,78 rearrangement preference at homologous chromosomes. In nurse shark at least two IgH genes are adjacent⁷² and the model for kappa L chain exclusion based on rearrangement preference evinced by the earlier replicating chromosome will not distinguish multiple, linked genes. It is not clear whether the 1-3 rearrangements in a B-cell occurred simultaneously or sequentially and we suggest that their activation was probably stochastic. If it happened that one rearrangement at a shark IgH gene is nonfunctional it seems unlikely that its allele is more apt to be the one next (or simultaneously) targeted for recombination than an adjacent or any other IgH in the genome.

Rearrangement of Ig Genes in Non B-cells

There exist pre-rearranged Ig genes in the germline of cartilaginous fishes, catfish and chicken.^78-80 In sharks, skates and rays some IgH clusters carry partially or fully recombined VD-J or VDJ and the IgL clusters joined VJ.⁴ Examination of nurse shark L chain junctions in some germline-joined VJ showed P region sequence that may indicate a one-time hairpin formation. This evidence and the fact that the “12/23 rule” is always obeyed, suggest that there was RAG activity in germ cells of some animals.^81,19 It was hypothesized that RAG-mediated changes in germline Ig genes produced the VD templates used in chicken H chain gene conversion or perhaps generated D elements during antigen receptor gene evolution.

The function of recombined genes in the shark antibody repertoire is not known; it appears that many are pseudogenes. In a species with many pre-rearranged VDJ there would be a strong likelihood for allelic inclusion if more than one IgH is activated at a time, but at the moment the germline genes in these animals have not been fully characterized. Nurse shark is an instance where all its IgM clusters have been characterized and none are pre-rearranged, showing that germline-joined genes are particular to the species.⁷²

Once initiated, somatic rearrangement in B-cells leads to VDJ. Partially rearranged IgH on the other hand have been observed in abundance in nurse shark thymocytes and 3-7 can be isolated per cell.⁷⁴ Thymic H chain transcripts could not be detected, implying that availability of DNA to RAG does not require transcription. That many thymocyte rearrangements are incomplete as VD-D-J, V-DDJ, etc., suggests that transcription may be part of the process that recruits⁸² RAG to its target for efficient recombination. This IgH rearrangement-permissive state in thymocytes may have characteristics in common with that in germ cells enabling RAG, when present, to effect recombination. However, the state of the IgH chromatin in either cell type has yet to be characterized.

About L Chain

In the course of evolution, whole-locus duplications produced the multiple cluster organization of cartilaginous fish IgH and IgL, whereas successive tandem duplications of the gene segments V, (D) and J generated the “translocon” organization that exists in tetrapods. While H chain genes are organized either as translocon or multiple clusters, the evolution of L chain genes⁸³ is more complex.

The number of L chain isotypes varies among vertebrates. In chicken there is only the one locus, Igλ; in mammals there are two, Igλ and Igκ. In Xenopus there are three: Igσ (sigma) and the homologs of Igκ (Igρ, called rho) and Igλ (called type III). In shark there are four: cartilaginous fish-specific “Igσ-cart” (called type I/NS5) and the homologs of sigma, Igκ (called type III/NS4) and Igλ (called type II/NS3). Igκ is thus present in all animals except birds and its organization varies considerably. In tetrapods Igκ is one locus. In nurse shark the Igκ homolog exists as >60 miniloci, with one V, one J and one C exon and tend to be separated by some distance. However in a bony fish like zebrafish, Igκ genes (type 1/3)⁸⁴ are arranged closely in serial arrays (examples in Fig. 4) and on at least four different chromosomes.⁸⁵

It is not clear how L chain expression is regulated in zebrafish (or any bony fish). In cod it was shown that multiple enhancers existed in the serial clusters but not every IgL C region was associated with downstream enhancer activity.⁸⁶ It cannot be anticipated from mere distance how regulatory control is exercised. Because there can be additional possibilities for intracluster rearrangement following an initial V to J attempt (Fig. 4), we have suggested that the bony fish organization allows for correction not only of nonproductive VJ but also in-frame VJ that contribute to forming a self-reactive specificity.⁸⁴ In other words, there exists a potential for receptor editing⁸⁷ in fishes, since the organizational set-up appears to allow for secondary rearrangements.

In zebrafish, the IgH organization is translocon like tetrapods⁸⁸ so that both types of arrangement exist for its Ig genes. It is clear that H and L chain gene organizations do not have to co-evolve—as they did not in bony fish⁸⁹—and information from this and the other model systems suggest they can be regulated independently. L chain exclusion is not as stringent as H chain, but the mechanism for restricting their expression in zebrafish must manage a large array of clusters, many of which carry multiple recombinogenic elements on either side of the C exon. How V(D)J recombination is sorted out in zebrafish will elucidate those aspects of RAG accessibility that evolve with individual species' immune system requirements.