Adaptive immunity in jawed vertebrates is mediated by B cells and T cells, two lymphocyte subsets characterised by the expression of highly diverse repertoires of heterodimeric antigen-specific receptors – immunoglobulins (IG or B cell receptors) and T cell receptors (TR). Immunoglobulins are formed from heavy (IgH) and light (either Igκ or Igλ) chains and TR from either α and β chains (αβ T cells) or γ and δ TR chains (γδ T cells).

The variable domains of individual IG and TR chains are formed in lymphocyte precursors by somatic recombination of single discontiguous variable (V), diversity (D – in IgH, TRB and TRD chains only) and joining (J) genes selected from multiple genomic copies of these genes. This recombination is directed by the products of the RAG1/2 genes which introduce double-strand breaks at recombination signal (RS) sequences flanking V, D and J genes [1] that are subsequently repaired by non-homologous joining of the cleaved V(D)J elements. After recombination the V(D)J product is spliced to a relatively invariant constant (C) gene. The different permutations of recombined V(D)J genes used to form individual IG and TR chains generate diversity (combinatorial diversity) that is further amplified by the activities of exonuclease and terminal deoxynucleotide transferase which together modify the germline nucleotide sequence at the V(D)J junction (junctional diversity). Together these mechanisms enable the generation of hugely diverse IG and TR repertoires – for example, the potential repertoire of unique human αβ TR has been calculated to be 10¹⁵[2]. Highly diverse IG and TR repertoires are crucial to facilitating recognition of the vast array of antigens to which hosts may be exposed and are therefore considered critical to an effective adaptive immune system [3].

The constituent genes used to construct TR chains are located in 3 distinct chromosomal loci - TRB for TRB, TRG for TRG and TRA/TRD for both TRA and TRD chains. Sequencing of murine and human TR loci has enabled the full complement of the TR genes in these species to be described (e.g. [4]) and allowed the organisation and regulation of these immunologically significant loci to be analysed (reviewed in [5, 6]). Additionally, inter-species and inter-loci comparisons have provided insights into how these loci have evolved (e.g. [7]). Between mammalian species the orthologous TR loci generally demonstrate conservation of overall structure – for example the organisation of genes in the TRB locus is similar in humans [4], mice [7], cattle [8] and dog [9]. Despite this conservation there are marked inter-species differences in both the total number of V genes and the relative representation of individual V gene subgroups (genes sharing ≥75% nucleotide identity) located in orthologous loci, indicating that dynamic evolutionary development of V gene germline repertoires has occurred post-speciation. TR V genes follow a ‘birth and death’ model of evolution whereby new genes are generated by repeated duplication (and form the substrate for diversification by mutation), of which some are retained whilst others are deleted or become pseudogenes following acquisition of deleterious mutations [10–12]. Dramatic evolutionary modifications of functional V gene germline repertoires are assumed to reflect selective pressures exerted on the adaptive immune system and in turn result in beneficial changes in the expressed TR repertoires.

The TRA/TRD locus is unique in containing the TR gene elements for 2 TR chains. In humans the TRA/TRD locus spans ~1 Mb and contains 49 TRAV genes, 5 dual usage TRAV/TRDV genes, 3 TRDV, 3 TRDD, 61 TRAJ, 4 TRDJ and single TRAC and TRDC genes, whilst in mice the TRA/TRD locus occupies ~1.7 Mb and contains 88 TRAV, 10 dual usage TRAV/TRDV genes, 6 TRDV, 2 TRDD, 60 TRAJ, 2 TRDJ and single TRAC and TRDC genes [13]. Comparative genomics show that the human and murine TRA/TRD loci share a similar gene organisation with most of the V gene segments located at the 5′ end of the locus and the other TRA/TRD genes and a single TRDV gene with an inverted orientation located at the 3′ end [7]. Two studies have analysed the genomic sequence of the bovine TRA/TRD locus [14, 15]. However, these studies were largely restricted to cataloguing the repertoires of either TRD [14] or TRAV/TRDV [15] genes, and an annotation of all of the TR genes within the locus has yet to be completed. Furthermore, although a main finding in both studies was the dramatic expansion of the V gene repertoires (a total of >400 TRAV/TRDV genes were identified), analysis of the duplication events that generated this hugely expanded germline repertoire, which could inform our understanding of how the bovine TRAV/TRDV repertoire has evolved, have not been conducted.

In this study, we use the most recent assembly of the bovine genome (UMD3.1 [16]) to re-examine the genomic sequence of the TRA/TRD locus. We sought to complete a comprehensive annotation of the TRA/TRD gene repertoire and use this as a basis to fulfil our primary aim of examining the evolutionary development of the massive expansion of bovine TRAV/TRDV genes. By combining analysis of the genome assembly and TRA/TRD transcripts we present data indicating that i) the majority of the TRAV/TRDV gene expansion is attributable to a complex series of duplications of multiple homology units (i.e. regions of DNA incorporating multiple genes), ii) expansion of some TRAV subgroups may have occurred ‘collaterally’, as a consequence of selection for expanded functional repertoires of other subgroups (most notably TRDV1), and iii) the expressed TRA and TRD repertoires preferentially use different TRAV/TRDV subgroups, suggesting TRA and TRD selection have made distinct contributions to the evolutionary development of the genomic TRAV/TRDV repertoire.

Gene duplication has been fundamental to the evolution of antigen-specific receptors, leading to the formation of the separate IG/TR loci [7, 27, 28] and the expansion of the repertoire of D, J and most notably V genes within these loci. Inter-loci and inter-species comparisons demonstrate that the extent of duplication that has occurred in different IG/TR loci during their evolution varies dramatically. The large number of bovine TRAV/TRDV genes identified in previous genomic analyses [14, 15] demonstrates that duplication within the bovine TRA/TRD locus has been prolific. In this study we have used genomic and cDNA analyses to investigate the evolutionary basis for this dramatic expansion and its relationship with the expressed TRA and TRD chain repertoires.

The availability of an improved bovine genome assembly was critical to the study. The UMD3.1 assembly is more accurate than the Btau versions with a greater degree of completion, more contiguous sequence, corrections of SNPs and removal of erroneous segmental duplications [16]. Thus, although the catalogue of genes defined in our annotation is in general agreement with previous analyses [14, 15] there are substantive differences. Most significant of these are the identification of genes belonging to 5 novel bovine TRAV/TRDV subgroups (TRAV6, 34 and 37 and TRDV2 and Y) and a reduction in the total number of TRAV/TRDV genes identified (371 vs. 402 in [15]). Comparative analyses indicated that Btau_4.0 included many more TRAV/TRDV genes showing 100% nucleotide identity, suggesting the discrepancy in TRAV/TRDV gene number is largely attributable to the annotation of erroneous duplications in Btau_4.0. However, more pertinent to our study was the improved mapping of genes to the TRA/TRD locus on chromosome 10 and the larger fragments of contiguous TRA/TRD locus sequence. Despite these improvements, the UMD3.1 assembly, like the previous Btau_4.0 and _3.0 assemblies, is i) derived from DNA taken from a father-daughter pairing, which due to the highly polymorphic nature of TRAV/TRDV genes, means that the potential for some proportion of the TRAV/TRDV expansion to be artefactual cannot be excluded, and ii) highly fragmented, with cDNA analysis suggesting that a substantial number of TRAV/TRDV genes are still missing. Consequently, definition of the total bovine TRAV/TRDV repertoire will be dependent on further work to obtain a completed assembly of the bovine TRA/TRD locus.

The high level of similarity in the organisation and gene content of the 3′ end of the bovine, murine and human TRA/TRD loci, which contains the TRAJ, TRDJ, TRAC, TRDC genes and the main regulatory elements, indicates that since the divergence of the primate, rodent and artiodactyl lineages 100 million years ago (MYA) [29] there has been strong conservative evolution of this region. In direct contrast, the huge disparity in the numbers of TRAV/TRDV genes located in the 5′ end of the human, murine and bovine TRA/TRD loci describes a divergent evolutionary development. Expansion of the human TRAV/TRDV repertoire (54 genes) has been limited with only 7 of the 44 TRAV/TRDV subgroups having multiple-members and only TRAV8 (7 genes) having >3 members. In mice a higher proportion of subgroups are multi-membered (14 out of 28) and these are in general larger (TRAV6/TRDV9 is the largest subgroup, comprising 12 members) but the repertoire of 104 TRAV/TRDV genes is still modest in comparison to the 371 genes in cattle. Duplication appears to be a general feature of bovine TRAV/TRDV subgroups, with the majority (34 out of the 42) being multi-membered; however, the more remarkable feature is the extent of duplication within some of the subgroups – TRDV1 has 60 members, and TRAV22, 23, 25, 26 and X all have >20 members. In this respect there are clear parallels with the bovine TRB locus where amplification of the TRBV repertoire is characterised by the prodigious expansion of certain subgroups (TRBV6 and 9) [8].

The location of 77.6% of bovine TRAV/TRDV genes in homology units demonstrates that homology unit replication has been the principal mechanism for expansion of the repertoire. In IG/TR loci expansion of the V gene repertoire by duplication of homology units incorporating genes from multiple subgroups is a common feature [7–9, 30, 31], although the extent to which this has occurred and its complexity differs markedly between loci. The identification in the UMD3.1 assembly of 10 putative homology units, ranging in size from 15 to 157 Kb and containing 2–17 TRAV/TRDV genes, many of which had undergone repeated replication, shows that the expansion of the bovine TRAV/TRDV repertoire has involved a particularly complex series of duplication events. In direct contrast, the majority of the murine TRAV/TRDV expansion is accounted for by a recent single duplication event in which a ~410 Kb segment containing 40 TRAV/TRDV genes was replicated, whilst in the human TRA/TRD locus homology unit duplication has not been documented. Similarly, homology unit replication in other mammalian TR loci is much simpler than that described here. Due to the fragmented nature of the TRA/TRD locus in the UMD3.1 assembly, many of the homology unit copies analysed were incomplete and, with the exception of homology unit 2 (TRAV2-TRAV3), it was not possible to define the full extent of the homology units. Therefore, although our data clearly demonstrate the key role homology unit duplication has had in evolution of the bovine TRAV/TRDV repertoire, analysis of improved, more contiguous assemblies of the bovine TRA/TRD locus will be required to fully appreciate the extent and complexity of homology unit duplication.

In addition to a prolific ‘birth’ rate, bovine TRAV/TRDV genes also exhibit a high ‘death’ rate (37.2% pseudogenes) compared to humans (19%) and mice (16.5%). There was marked variation in the percentages of non-functional genes between the multi-membered subgroups; for example amongst the 6 largest subgroups it ranged from 83.3% (TRAV23) to 20% (TRDV1). Although observed in IG V subgroups [32], the high proportion of pseudogenes seen in several of the bovine TRAV/TRDV subgroups, is rare in TR V subgroups. Based on the assumption that within the context of ‘birth-and-death’ evolution, homology unit replication is utilised to expand the functional TRAV/TRDV gene repertoire, the frequent co-localisation within homology units of genes from low (e.g. TRDV1 or TRAVX) and high (TRAV23 or TRAV18) ‘death’ rate subgroups, suggests that expansion of some high ‘death’ rate TRAV/TRDV subgroups may have occurred as a ‘collateral’ consequence of selection for juxtaposed low ‘death’ rate subgroups. Such ‘collateral’ expansion of V gene repertoires appears to be a feature of the evolution of other bovine IG/TR loci. Genomic analysis of the bovine IGLH locus suggests that the non-functional IGHV2 subgroup may have been expanded due to co-localisation of IGHV2 and functional IGHV1 genes in a duplicated homology unit [33] and similarly in the IGL locus many members of the expanded but predominantly non-functional IGLV5 subgroup are found juxtaposed to functional IGLV1 subgroup 1 genes [34]. Many bovine TRAV/TRDV pseudogenes in equivalent/analogous locations within different replicates of a homology unit share a common mutational lesion (e.g. TRAV8 genes in replicates of homology unit 9 all have an identical 4 bp deletion in the V exon) whilst conversely there are examples of analogous genes in homology unit replicates being a mixture of pseudogenes with non-identical lesions and functional genes (e.g. TRAV24 genes in homology unit 9). This indicates that ‘collaterally’ expanded genes may either be pseudogenes prior to duplication or may, due to absence of selection, lose functionality following replication. Notably, TRDV1 genes are co-located in 4 different homology units with different permutations of genes from 4 of the other massively expanded subgroups - TRAV22, 23, 25 and 26. Whilst the percentage of TRDV1 pseudogenes (20%) is much lower than the bovine TRAV/TRDV average (37.2%) that of the other subgroups is higher, ranging from 43.4-83.5%. Thus, under the model of ‘collateral’ expansion, replication of homology units 7–10 and the consequent expansion of subgroups TRAV22, 23, 25 and 26 may have largely been due to evolutionary selection for a large functional TRDV1 repertoire. As these 5 subgroups together account for nearly half of the annotated bovine TRAV/TRDV genes, TRDV1 selection may have been a critical force in the evolution of the bovine TRAV/TRDV repertoire.

A unique feature of the TRAV/TRDV locus is the co-localisation and sharing of V genes available for rearrangement of 2 distinct antigen-specific receptor chains expressed by 2 distinct lineages of lymphocytes. Consequently, to understand the evolution of the bovine TRAV/TRDV gene repertoire in the context of the TR expressed by αβ and γδ T cells we undertook an analysis of the TRA and TRD transcripts expressed by these T cell subsets. Phylogenically defined TRAV/TRDV gene orthologues don’t necessarily have the same capacity to partake in TRA and TRD rearrangements – for example mTRDV3 and hTRAV40 are phylogenetic orthologues but rearrange exclusively with TRDC and TRAC respectively. It was therefore necessary to first empirically evaluate TRAV/TRDV subgroup usage in rearranged TRA and TRD transcripts using TRAV/TRDV subgroup-specific and TRAC/TRDC specific primers. This analysis confirmed that each of the subgroups assigned as TRAV based on orthology to human TRAV subgroups were represented in rearranged TRA chain transcripts (with the exception of TRAV39) and that genes from all of the subgroups assigned as TRDV were demonstrated to be represented in TRD chain transcripts. Only genes from subgroups TRAV33, TRDV1, TRDV3 and TRDVb3 were expressed in both rearranged TRA and TRD chains, extending the previous observations of dual usage of these subgroups [17]. Thus, in contrast to humans and mice, dual usage of TRAV/TRDV genes in cattle makes a minimal contribution to diversifying the available TRDV germline repertoire but instead has been employed to further diversify the germline repertoire of V genes available for TRA rearrangement.

Commensurate with the number of TRAV/TRDV subgroups available for rearrangement, ‘unbiased’ transcript analysis demonstrated that the TRA repertoire uses genes from a diverse range of TRAV/TRDV subgroups, most of which are represented at <5%, whereas the TRD repertoire predominantly expressed TRDV1 genes, with minor contributions from a limited number of additional TRAV/TRDV subgroups. The dominance of TRDV1 in the expressed TRD repertoire suggests a functional advantage conferred by TRD chains utilising TRDV1 genes may have selected for the expansion of the TRDV1 genomic repertoire and consequently been a major determinant in the replication of homology units 7–10. Conversely the under-representation of TRAV26 (and TRDV1) genes in the expressed TRA repertoire suggests functional selection through TRA chain expression has not been a primary factor in replication of these homology units, but rather supports the concept that some TRAV subgroups have been expanded ‘collaterally’ as a consequence of being within the same homology units as TRDV1. However, over-representation of TRAV3 and TRAVX gene usage in the expressed TRA repertoire indicates that TRA selection has driven replication of other homology units (2, 5 and 6). As such the transcript data indicate that TRA and TRD selection may have driven expansion of different components of the genomic bovine TRAV/TRDV gene repertoire, with TRD selection playing an apparently disproportionately large role. The significance of TRD selection in shaping evolutionary development of the functional bovine TRAV/TRDV genomic repertoire is suggested by the marked increase in the proportion of Group 4 genes (preferential used in TRD chains) and concomitant decrease in the proportion of Group 1 genes (preferentially used by TRA chains) seen in cattle compared to mice and humans.

Although our studies have provided an insight into the potential mechanisms of the evolution of the bovine TRA/TRD locus, the biological factors that have led to the massive expansion of the TRAV/TRDV repertoire remain a matter of conjecture. Cattle, along with other artiodactyl species have proportionally higher numbers of circulating γδ T cells than humans and mice; this is assumed to reflect a greater immunological role for γδ T cells in these species [35]. A parallel but apparently evolutionarily independent expansion of the TRDV1 subgroup in pigs (Sus scrofa) may signify that in artiodactyls common selective forces have led to high frequencies of γδ T cells and TRDV1 gene duplication. Incomplete knowledge on the functional roles of γδ T cells and the ligands recognised by γδ TR limit our ability to construct a hypothesis of the evolutionary advantage conferred to artiodactyls by the expansion of TRDV1 genes. However, it is well recognised in mice and humans that subsets of γδ T cells defined by specific TRDV (and TRGV) gene expression have distinct functional phenotypes (reviewed in [36]). In addition to roles such as cytotoxicity and IFNγ production, human TRDV1⁺ γδ T cells have been shown to express an immune-regulatory phenotype following stimulation with a TR-independent agonist [37]. Recent evidence demonstrating that γδ T cells rather than FoxP3⁺CD4⁺T cells contain the dominant regulatory T cell subset in cattle [38] raises the intriguing possibility that the requirement for a substantial immune-regulatory T cell population may have been a factor leading to the expansion of TRDV1⁺ γδ T cells in cattle and other artiodactyls.

A comprehensive understanding of the evolutionary development of complex multi-gene families, such as the V genes of IG/TR loci requires integration of a variety of data. Previous studies cataloguing the bovine TRAV/TRDV genes have demonstrated the unprecedented size of this repertoire but have not addressed the underlying evolutionary processes leading to its expansion. By combining data on gene characterisation, function, organisation and expression we have shown that the expansion of bovine TRAV/TRDV genes has been achieved primarily through a highly complex series of homology unit replications. We provide evidence indicating that as a consequence of this mode of replication a substantial proportion of the expanded TRAV gene repertoire may have occurred ‘collaterally’ as a consequence of positive selection for duplication of adjacent TRAV/TRDV segments, most notably TRDV1 genes. As TRD selection appears to be the main driver of TRDV1 expansion, this suggests that in bovine (and perhaps other artiodactyls) TRD selection may have been a dominant factor in the genomic expansion of a number of TRAV subgroups. Such data provide an important novel perspective of the evolutionary dynamics that have led to the development of the dramatically expanded bovine TRAV/TRDV repertoire and may have relevance to other IG/TR loci with enlarged V gene repertoires.

Without a fully assembled TRA/TRD locus it was not possible to implement the approved nomenclature system which requires knowledge of the genomic order of genes from the 5′ to the 3′ end of the locus [46]. Bovine TRAV/TRDV subgroups have been named according to the orthologous human subgroups where appropriate and an alphabetic designation where this was not possible. Members in the V gene subgroups have been given alphabetic rather than numeric designations, similar to systems employed in previous work [8, 14], although there is no correspondence between the TRDV1 nomenclature employed here and in [14]. The TRDD and TRDJ nomenclature is consistent with previously published work and TRAJ genes have been assigned numbers according to their 5′ to 3′ order in the genome.