Abstract
VDJ rearrangement at the teleost TCRβ locus leads to a highly diverse repertoire of junctions for each VβJβ combination. From a rainbow trout 5′ RACE library of TCRβ transcripts, 47 clones encompassing a full Vβ-Dβ-Jβ-Cβ sequence were selected and analyzed. A similarity analysis of the sequences evidenced 10 Vβ families, of which 6 were not previously described. Immunoscope and sequence analysis of the Vβ-Dβ-Jβ junctions of the new families confirmed that they create a polyclonal and diverse repertoire. Multiple alignments showed that rainbow trout Vβs possess most of the conserved residues typical of Vβ segments. However, this study revealed a high complementarity-determining region 2 (CDR2) and CDR1 length diversity among rainbow trout Vβ families, suggesting that the spatial orientation of the TCR could fluctuate in the TCR/peptide/MHC complex, depending on the Vβ expressed. Among the new Vβ families, Vβ6 displayed the strongest deviance from typical hypervariable CDR1 and CDR2 loops, with an unusually short CDR2. Moreover, the Vβ6 sequence is overall divergent from typical Vβ sequence, raising the question of its functional relevance. Immunoscope experiments identified a Vβ6-Jβ3 junction, which was amplified during the response against viral hemorrhagic septicemia virus, a fish rhabdovirus. Vβ6 seems therefore to be expressed functionally in a selected TCR. However, the shorter Vβ6 transcripts produced through an alternative splicing lack the C′, C″, D, and E strands of the Vβ domain and are probably nonfunctional.
T cell receptor, the Ag recognition molecule expressed by the T lymphocytes, is a member of the Ig superfamily. It is a heterodimer, composed of an α (or a γ)- and a β (or a δ)-chain, each containing a V and a C domain. The Ag specificity of TCR is determined by sequences corresponding to three hypervariable regions of the V domain, the complementarity-determining regions (CDR).4 The diversity of TCRαβ is generated by the assembly of germline-encoded segments: V, J, and for β-chains D segments. A huge additional diversity derives from the addition and deletion of nucleotides at the junctions between V, D, and J segments. TCR genes, loci, and recombination mechanisms and control have been characterized in detail in humans and mice (reviewed in Ref. 1). TCRαβ recognizes the Ag presented as a peptide by the class I and II MHC molecules. This recognition mediates cytotoxic and Th cell responses.
TCR genes were identified in the different lineages of Gnathostoma (2). In the fish, TCRβ genes were first described in the rainbow trout (Oncorhynchus mykiss) (3) and in the horned shark (Heterodontus francisci) (4). TCR genes α, β, γ, and δ were finally identified in the skate (Raja eglanteria), a Chondrychthian, suggesting that all Gnathostoma should express both αβ and γδ TCR (5). Complete TCRβ sequences are now available for several species of teleosts: in Atlantic salmon (Salmo salar) (6), channel catfish (Ictalurus punctatus) (7), Atlantic cod (Gadus morhua) (8), damselfish (Stegastes partitus) (GenBank AF324813-824), and Japanese flounder (Paralichthys olivaceus) (GenBank AF053407-443). In rainbow trout, four different Vβ families were described (3, 9, 10). In all these species, the sequence of β-chains showed typical V and C1 domains (as defined in 11) with key residues conserved, suggesting that the general three-dimensional structure of this protein was conserved from fish to mammals (12). Phylogenetic studies showed that TCRβ sequences are conserved in mammals (13, 14), but no close similarity was observed between fish and mammalian Vβ sequences. In teleosts, T cell-dependent responses were suggested by the presence of TCR genes, and by the structure of the MHC locus and by the polymorphism of class I and class II molecules (15, 16, 17). However, a functional T cell response has been directly proved in only a limited number of species. A clear allospecific response was described using catfish Ig− cells or catfish T cell lines, showing that fish lymphocytes can display CTL-like activity (18, 19). A specific cell-mediated lysis of virus-infected target cells by primed leukocytes has also been obtained in the cloned goldfish (20). However, in this case, the cells involved in the cytotoxicity had not been unambiguously identified as αβ T cells. More recently, TCRβ Immunoscope methodology (21) was used to study the rainbow trout T cell response to a viral infection. Spectratyping of CDR3 revealed that the rainbow trout naive T cell repertoire is polyclonal and highly diverse. Primary and secondary infection with viral hemorrhagic septicemia virus (VHSV) dramatically skewed the repertoire, and the profiles were reminiscent of the public and private virus-specific T cell responses observed in mammals (22).
In this work, we used a 5′ RACE strategy and Immunoscope analysis to obtain further insight into the Vβ segments expressed in rainbow trout, and to perform a global survey of the diversity of Vβ-Dβ-Jβ junctions. Ten Vβ families (Vβ1–10) were identified from TCRβ transcripts. A comparative study of their primary structure with Vβ sequences from other vertebrates revealed that Vβ6 had unusual characteristics.
Materials and Methods
Fish and leukocyte preparation
Rainbow trout were raised in the fish facilities of Institut National de la Recherche Agronomique (INRA, Jouy-en-Josas, France). The so-called INRA synthetic strain was used throughout, except for infection experiments. It results from successive introductions between 1976 and 1983 of several domestic populations from the United States and France, which were pooled and then maintained as a single population by random mating during four to five generations. For the study of the Vβ6 response to the VHSV, rainbow trout heterozygous clones (22) were used. Trout were sacrificed by overexposure to 2-phenoxyethanol diluted 1/1000. The entire spleen was removed aseptically. Leukocytes from the kidney tissue of a single fish were isolated by centrifugation through a Ficoll gradient (lymphocyte separation medium, d = 1.077; Eurobio, Les Ullis, France), and used for RNA preparation.
RNA isolation and construction of the 5′ RACE library
The 5′ RACE was performed using the SMART RACE cDNA Amplification Kit (Clontech, Palo Alto, CA), according to the instructions of the manufacturer. Total RNA was extracted from spleen and head kidney leukocytes with the TRIzol reagent (Life Technologies, Cergy-Pontoise, France), and treated with DNase (Boehringer Mannheim, Indianapolis, IN) to remove any remaining genomic DNA. The treated RNA was used to generate full-length cDNAs. The 5′ RACE PCR were performed with a Cβ-specific primer (Cβ-RACE: ACACACACTAGGGTCTTCTT) and the universal primers from Clontech. PCR products obtained were purified with Sephacryl S-400 columns (Pharmacia, Peapack, NJ) and cloned into pCR2.1 vector (TOPO TA cloning system; Invitrogen, San Diego, CA).
Screening of the 5′ RACE cDNA library
The clones corresponding to Jβ4-Cβ transcripts were identified by PCR using two primers chosen in the genomic region upstream of the Jβ4 segment (J4C-forward, AGGGAGTTGGTTTGATTCATGTTTG; J4C-reverse, ACATCGTTTTCCTCCTCTTCCAAAA, giving a 172-bp PCR product). This set of primers specifically amplifies cDNA to Jβ4-Cβ sterile transcripts. Bacteria from the colony on the plate were added to the PCR mix. The amplification was performed in the following conditions: 10 min, 94°C; then 30 cycles of 1 min, 94°C; 1 min, 60°C; 2 min, 72°C; then 5 min, 72°C. Colonies that gave PCR product of the expected size were rejected. Colonies giving no PCR product were elected for further analysis. Selected colonies were grown overnight in Luria-Bertani/ampicillin broth, and the plasmid purified with a plasmid miniprep spin kit (Nucleospin; Macherey-Nagel, Durin, Germany). Purified plasmids were subjected to automated sequencing with direct and reverse universal primers.
Immunoscope analysis
The Immunoscope methodology developed for mice or humans (21, 23) was adapted for rainbow trout, using primers specific for trout Vβ, Jβ, and Cβ sequences (22). Vβ family-specific primers were designed for new rainbow trout Vβ families described in this work. We chose the primers in framework region to avoid cross-hybridization between the different Vβ families. Jβ and Cβ primers have been described previously (22), and they were designed to be specific of Jβ1–9 segments except for Jβ4, because Jβ2 and Jβ4 sequences are almost similar. A Jβ2–4 primer was designed in a region similar in Jβ2 and Jβ4 to amplify both kinds of rearrangement with the same efficiency. Primers used were indicated in Table I⇓. Immunoscope analysis was performed essentially as described previously (22, 23). Briefly, a first PCR was performed using Vβ- and Cβ-specific primers, which amplify sequences with a given Vβ, but with different CDR3. In a second step, Vβ-Cβ PCR products were subjected to runoff reactions with different fluorescent C- or J-specific primers. Runoff products were loaded on a polyacrylamide sequencing gel, and size was separated on an ABI-373 automated sequencer (PerkinElmer, Wellesley, MA). CDR3 length distributions were analyzed using the Immunoscope software, as described previously (21). Repertoire editing and comparisons were performed using the ISEApeaks software (24).
Primers for rainbow trout immunoscope analysis
Immunization and virus challenge
The attenuated 25-111 variant of strain 07-71 of VHSV was used to infect fish through i.m. injection of 1–5 × 105 PFU/trout. This infection usually leads to a good protection against a subsequent lethal infection. Four weeks later, fish received a second injection of variant 25-111 (75 × 106 PFU/trout).
Sequence analysis
The Genetic Computer Group package (Madison, WI) was used for sequence assembly. BLAST and FASTA analysis was performed using programs at the National Center for Biotechnology Information and European Molecular Biology Laboratory websites (http://www.ncbi.nlm.nih.gov/blast/, and http://www2.ebi.ac.uk/fasta3/). When useful, different BLOSUM matrices were used, as indicated in the tables. Kabat numbering was used for sequence description in the text and figures, and CDR3 was considered between residues 96 and 106. Immunogenetics database numbering is also provided for all of rainbow trout Vβ families in Fig. 1⇓.
Amino acid sequences from 10 rainbow trout Vβ families. Vβ1, 2, and 3 sequences correspond to GenBank U18123, -4, and -5, respectively. Gaps (noted −) were introduced in the beginning of CDR1 and CDR2 to allow numbering. Above the alignment, + indicates the amino acids, which are important for the three-dimensional structure in the mouse, and present in all rainbow trout Vβ (? is indicated when the conservation is partial)./, Indicates the amino acids, which are important for the three-dimensional structure in the mouse, and not conserved in rainbow trout. ∗, Indicates a position conserved among rainbow trout Vβ, but with a residue different from that of Chothia’s list. The positions of highly variable regions (HVR) and of β strands known for mammalian Vβs are marked above the alignment. The numbering following Kabat, and IMGT systems are given just above and just below the aligned Vβ sequences, respectively. The positions of framework region (FR) and CDR following the IMGT definitions are also indicated below the IMGT numbering.
Results
Identification of new rainbow trout Vβ families
To better characterize the rainbow trout Vβ family composition, we used the 5′ RACE methodology to extend Vβ sequences in the 5′ direction starting from a Cβ-specific primer (Cβ-RACE). Unlike PCR-based methods using specific or degenerate Vβ primers, the 5′ RACE strategy should normally reveal any transcript encompassing a Cβ segment. To take into account the putative genetic diversity of domestic trout populations, we used fish from the so-called INRA synthetic strain, which was generated from successive introductions of several domestic populations from the United States and France, pooled, and maintained by random mating. Spleen and head kidney leukocyte RNA was prepared from three naive rainbow trout, and pooled. cDNA was then prepared from pooled RNA and used as a template for the 5′ RACE experiment. A cDNA library was constructed by cloning the 5′ RACE PCR product into pCR2.1. A first set of 48 random clones was sequenced to analyze the composition of the library. The 5′ RACE library contained three different species of TCR transcripts: completely rearranged Vβ-Dβ-Jβ-Cβ transcripts (11 clones), sterile Dβ-Jβ-Cβ (11 clones), and sterile Jβ-Cβ (26 clones), spliced but not rearranged transcripts. Most of the sterile Jβ-Cβ clones (24 of 26) expressed the Jβ4 segment, which is proximal to the Dβ segment. These sterile Jβ4-Cβ transcripts included a genomic sequence localized upstream of Jβ4. We therefore designed primers specific to this region, which were used to test bacterial colonies of the 5′ RACE library by PCR amplification. Colonies giving a PCR product were excluded from subsequent sequence analysis. Most of 5′ RACE clones subsequently subjected to sequencing therefore corresponded to Dβ-Jβ-Cβ or Vβ-Dβ-Jβ-Cβ transcripts. Finally, we obtained a collection of 47 Vβ-Dβ-Jβ-Cβ sequences expressing a complete Vβ segment.
Pairwise similarity analysis was performed for all Vβ nucleotide sequences using the Bestfit command of the GCG package. We used a 70% overall nucleotide sequence identity to order the sequences into 10 families. Four were already described (Vβ1–4), and six were considered new families (Vβ5–10). Vβ1 (12 clones), Vβ2 (11 clones), and Vβ3 (9 clones) were the most frequent. Vβ6, Vβ7, and Vβ8 were each represented by 3, 5, and 3 clones, respectively. Vβ4, Vβ5, Vβ9, and Vβ10 were found only once. The 5′ RACE PCR protocol should not affect the relative frequencies of the different Vβ segments; thus, these proportions should reflect the actual frequencies of Vβ segments in the available T cell repertoire.
Characteristics of rainbow trout Vβ
Amino acid sequences of rainbow trout Vβs were subjected to multiple alignment using ClustalW and manual adjustment. The V domain typical strands and CDR regions were then identified (Fig. 1⇑). Conserved positions included C23, M(I,V)XWY-R(Q,K)-Q(R,K)37, and G/A-X-Y-F/Y-CA93 (Kabat numbering). We also analyzed the multiple alignment in reference to the list of 29 key residues conserved in mouse and human Vβ sequences established by Chothia et al. (25). Fig. 1⇑ shows that 14 residues from this list are strictly conserved in rainbow trout. In addition, Q6 and P8 residues are present in Vβ1, 2, 3, 4, 5, 9, and 10. In Vβ7, Q and P were present at positions 5 and 7, respectively. In Vβ8, they are at positions 7 and 9, respectively. In Vβ6, there is Q at position 5, but P7 is replaced by L. Thus, 16 of 29 key residues were globally conserved in all rainbow trout Vβ. Vβ canonical features were therefore present in rainbow trout sequences.
It is interesting to note that the C23-W34 region, which contains the CDR1 domain, displayed a large range of length variation. It contains from 12 to 15 aa, compared with 11–12 aa in mice and humans. Similarly, the length of the W34-C92 region, which contains the CDR2 region, varied from 52 to 63 aa compared with 56–59 residues in humans. This part of the V domain is particularly short in three rainbow trout Vβ families, 52 residues for Vβ6, 54 for Vβ9, and 55 for Vβ2.
Vβ6 has unusual primary structure features
Vβ6 displays the most remarkable primary structure features. Vβ6 segment has C23-W34 and W34-C92 regions of 15 and 52 residues, respectively, i.e., the longest CDR1 and the shortest CDR2 among rainbow trout Vβs. Residues corresponding to the C″ strand are absent from Vβ6, and the CDR2 hypervariable loop should be much less protruding toward the peptide/MHC complex than for the other Vβs. Another striking feature of the Vβ6 segment is that the CDR2 (residues 49–57) is highly charged and hydrophilic: 8 or 9 charged residues of 13 (ratio, 0.69:0.61, depending on the clone) instead of 3–6 of 15–24 residues (ratio, 0.18:0.33) in the other rainbow trout Vβ segment.
To search for sequences similar to the rainbow trout Vβ segments and especially to Vβ6, we compared nucleotide and protein sequences to the vertebrate section of GenBank using the BLASTP program (Table II⇓). Rainbow trout Vβ1 had >68% identity with Vβ1 from Stegastes, which was close to the family cutoff. Vβ2, Vβ7, Vβ8, and the group including Vβ3–5 and Vβ10 had significant sequence similarity with other Vβs from different species of vertebrates. By contrast, the best hit for trout Vβ6 was a VH sequence from human Ig with an E value of 9.107. No hit with any Vβ sequence was obtained for E value < 3.104, suggesting that Vβ6 was divergent from other known Vβ sequences. The charged stretch between positions 49 and 57 does not explain completely this result, because blast searches with the 1–34 region led to the same dichotomy between Vβ6 and the other rainbow trout Vβ segments. A FASTAX search in EMBLALL identified a Tetrodon nigroviridis genomic sequence with a highly significant sequence similarity to Vβ6. Conceptual translation of the Tetrodon sequence revealed a V gene of the Ig superfamily, which displays 45% (41 of 92) identity with Vβ6. In addition to sequence similarity, CDR1 and CDR2 regions have the same length in both sequences (Fig. 2⇓A), suggesting that these genes may share the same origin. Noteworthy, the V domain of the nurse shark Ag receptor (NAR), an Ig-like dimer receptor (26, 27, 28), has a short W34-C92 region (48 residues) (Fig. 2⇓A), indicating that a V domain with a short CDR2 region is expressed in a rearranging Ag receptor. In addition, a blastp search for sequences similar to the CDR2-containing K49-G88 region of Vβ6 mainly retrieved NAR sequences. However, most of the identical residues were localized in positions in which amino acids with definite properties are conserved in many Ig V domains, suggesting this similarity pattern probably occurred by chance. Moreover, to the difference of NAR, which has no conserved interdomain residues, Vβ6 has 5 of 6 typical interdomain residues (as defined in 25), indicating it probably interacts with a Vα domain.
Unusual characteristics of the Vβ6 segment. A, Vβ6 primary structure compared with a NAR (28 ), with the Vβ6 homologue from T. nigroviridis (AL184010) (Teni), and with rainbow trout Vβ3 (gb U18124) used as a reference (VB3). An example of CDR3 and G strand was reported from Vβ6 and NAR V. Only strands A–F are shown for the unrearranged genomic sequence from T. nigroviridis, and for the reference. Gaps have been introduced as in Fig. 1⇑ to improve the alignment, and noted −. B, Alignment of the nucleotidic sequences of spliced (VB6s) and unspliced (VB6) form of Vβ6 in TCRβ transcripts. Consensus splicing acceptor and donor motifs are indicated above the alignment in bold characters, and the corresponding nucleotides in the Vβ6 sequence are in italic. C, Distribution of the Jβ segments rearranged to Vβ6 among spliced (black boxes) and nonspliced (gray boxes) TCRβ transcripts. The distribution of Jβ segments among all 5′ RACE Vβ-Dβ-Jβ-Cβ clones is reported in stripped boxes. D, Multiple alignment of amino acid sequences of complete Vβ6 (VB6), spliced Vβ6 (VB6s), complete NKP30 (NKP30), and spliced NKP30 (NKP30s). The regions deleted by alternative splicing in VB6s and NKP30s sequences are indicated by =.
Sequences similar to rainbow trout Vβs as detected by BLASTP program: close relatives and mammalian homologs
Vβ6 has an unusual splicing pattern
Among three Vβ6 clones found in the 5′ RACE collection, the clone BN62 lacked a region corresponding to C′-C″-D-E strands. Alignment of long and short forms of Vβ6 cDNA showed that sequences were otherwise 100% similar, and that the deletion was bordered by putative splicing signals (Fig. 2⇑B). This suggested that these two forms could be produced from the same gene by alternative splicing. To verify that the clone BN62 was not an artifact, we designed a Vβ6-specific primer (VB6) to amplify and clone a representative set of Vβ6 rearrangements from naive spleen cDNA. PCR with VB6/Cβ-specific (CB) primers produced two bands, corresponding to short and long forms of the transcript. These products were cloned, and several clones were sequenced. We obtained only Vβ6-containing rearrangements (24 of 24), showing that amplification was specific of the Vβ6 segment (Table III⇓). Sequence analysis also confirmed that Vβ6 was used as a regular Vβ segment in rearrangements with several Jβ, producing junctions of diverse length and composition. A significant number of clones (9 of 24) corresponded to the short form, and showed exactly the same deleted region as in clone BN62. The clones corresponding to the deleted Vβ6 had rearrangements with different Jβ segments, ruling out the possibility that the BN62 primary structure was a PCR-mediated artifact. The analysis of the VB6-CB clones also shows that Vβ6 rearranged more frequently to Jβ3 and Jβ4 segments, compared with the Jβ usage we observed in the whole collection of Vβ-Dβ-Jβ-Cβ clones (Fig. 2⇑C). There was no difference of Jβ usage between short and long forms of the Vβ6 transcripts.
Distribution of Vβ-Jβ combinations identified among cloned VB-CB PCR products
New Vβ families show CDR3 length diversity
In the TCR binding with the MHC-peptide complex, CDR3 contributes mainly to binding with the antigenic peptide. Consequently, this region should reflect the vast diversity of the TCR recognition site. To assess systematically the CDR3 length diversity of new TCRβ transcripts, we performed Immunoscope experiments with each Vβ1–10-specific primer. We used Vβ1–4-specific primers, as described previously (22), and the Vβ6-specific primer defined above. We designed additional primers to amplify specifically all the new Vβ families (VB5, 7–10). To verify their specificity, PCR amplifications were performed using each of the VB primers in combination with CB. PCR products were cloned, and several clones were sequenced for each VB-CB combination. The Vβ region was identified in each clone, and compared with the reference. This analysis established that the amplification using VB5, VB7, VB8, VB9, and VB10 was specific to the corresponding Vβ family (Table III⇑). We compiled and analyzed 99 Vβ-Dβ-Jβ junctions sequenced in this study to get an overview of TCRβ rearrangements. Structural characteristics of junctions involving the new Vβs were not significantly different from what had been reported previously (12, 22). Neither new Jβ, nor new Dβ segments were identified. Segments Jβ1–9 and Jβ11 were retrieved with different frequencies, but we did not identify the elusive Jβ10 segment (10).
For Immunoscope analysis, spleen cDNA was synthesized from naive rainbow trout, and subjected to PCR amplification using VB primers and a Cβ-specific primer (CB2). Each VB-CB PCR product was then subjected to runoff reaction with an internal Cβ-specific fluorescent primer (CB1). We obtained profiles of 5–8 peaks, each peak corresponding to a pool of TCR transcripts with a given CDR3 length (Fig. 3⇓). The peaks were separated by three nucleotide intervals, corresponding to the size of in-frame transcripts. A bell-shaped distribution of peaks was observed for all Vβs, demonstrating that the T cell populations expressing new Vβs were also polyclonal and diverse in naive fish. The VB6-CB PCR product gave two gaussian profiles, corresponding to the long and short forms of Vβ6 transcripts. As this experiment represents a global survey of CDR3 length profile for each Vβ, it confirmed that both short and long Vβ6 transcripts had highly diverse junctions, distributed in typical bell-shaped spectratypes. In addition, the Vβ6 profiles suggested that spectratypes were generally identical for the short and long transcript in a given fish (compare the slight differences between fish 1 and 2, which were reproduced in short and long form). Vβ6 profiles displayed less Cβ runoff peaks than the profiles observed for other Vβ segments. This distribution is probably linked to the biased Jβ usage in Vβ6 transcripts, because Jβ segments have different lengths (47–53 bp) (10). Interestingly, Jβ3 and Jβ4 segments are both 50 bp long, and the corresponding bell-shaped profiles are centered on the same average length. The bias toward Jβ3 and Jβ4 usage in Vβ6 TCR transcripts should therefore restrict the apparent diversity of corresponding Cβ runoff products, and mask lateral peaks due to rearrangements with other Jβ segments.
Bell-shaped profiles of CDR3 length distribution for all Vβ families. Spleen leukocytes of two naive rainbow trout were amplified using Vβ1–10 and Cβ2 primer, and PCR products were subjected to runoff using the fluorescent Cβ1 primer. After separation of the runoff reactions with an automatic sequencer, the fluorescent profiles were analyzed with the Immunoscope software. Each profile represents the CDR3 length diversity of a given Vβ family. Fragment length is on the x-axis, and fluorescence intensity is on the y-axis. Vβ family is indicated on the top.
Vβ6 participates in the specific response to viral infection
The Vβ6 transcripts had typical CDR3 length diversity in naive rainbow trout (3–10 residues). However, the atypical primary structure of the Vβ6 domain raised the question of its expression at the protein level as part of a TCR. To address this question, we investigated whether Vβ6-expressing T cells were involved in the specific response against a viral infection, and thus subjected to Ag positive selection. To avoid the effects of fish genetic diversity on T cell responses, we used rainbow trout with identical genetic background. Rainbow trout from clone EQ2 were infected with the 25-111 nonvirulent mutant of VHSV, and were given a second injection of the same virus 27 days after primary infection. VB6 and CB2 primers were used to amplify spleen cDNAs from trout sacrificed on days 42 and 48. VB6-CB PCR products were subjected to runoff reaction with Jβ-specific primers. The different segments were resolved on a sequencing gel and analyzed with the Immunoscope software. Typical Immunoscope results for all Jβ are shown in Fig. 4⇓. The Vβ6-Jβ3 profile was strongly altered in one infected individual (infected trout 2), but it was not significantly affected in the other infected fish. This bias most probably corresponded to a private specific response against the virus. In contrast, the Vβ6-Jβ8 profiles from the two infected fish were consistently different from the control. These profiles suggested that expansion of specific Vβ6-Jβ8 clones corresponded to a weak public response against the virus.
CDR3 length distribution of Vβ6-Dβ-Jβ rearrangements in infected and naive rainbow trout. Two individuals from the same clone were infected with the 25-111 attenuated variant of the VHSV on days 1 and 27. Typical Immunoscope analysis of Vβ6 junctions from spleen leukocytes of a naive fish from the same clone is given as a control. Spleen leukocytes were analyzed 14 days (Infected 1) or 20 days (Infected 2) after the secondary infection. cDNA were amplified using primers Vβ6 and Cβ, and PCR products were subjected to runoff reactions with fluorescent Jβ1–9 primers. After separation using an automatic sequencer, profiles corresponding to short, spliced Vβ6 transcripts (shVβ6) or to nonspliced Vβ6 transcripts (Vβ6) were analyzed using the Immunoscope software. Fragment length is on the x-axis, and fluorescence intensity is on the y-axis.
To ascertain that profile modifications were produced by clonal expansion of Vβ6 T cells, we further analyzed Vβ6-Jβ3 and Vβ6-Jβ8 profiles from infected trout 2. Vβ6-Jβ3 and Vβ6-Jβ8 rearrangements were amplified from spleen cDNA, and the PCR products were cloned. Several clones were picked at random, and sequenced (Fig. 5⇓). The junction CAANDPAF was found in 4 of 14 Vβ6-Jβ3 rearrangements, and corresponded to the size of the expanded peak observed in the Immunoscope profile (3 aa). T cell expansion was less evident for Vβ6-Jβ8, as only one junction (CDR3 8 aa) was found twice. Nonetheless, these results strongly suggested that specific Vβ6-expressing T cells were selected in the context of a secondary response to a virus. Vβ6 must therefore be expressed in a TCR at the cell surface, and its structural peculiarities could participate in the diversity of the available T cell repertoire. Similar Immunoscope profiles were systematically observed with spliced and unspliced transcripts for a given Vβ-Jβ combination. In addition, the Vβ6-Jβ3-expanded junction CAANDPAF was found in both spliced (1) and unspliced (3) transcripts. These results therefore suggest that each Vβ6-expressing T cell produced both spliced and unspliced transcripts, of which at least one form is expressed at the cell surface and drives T cell selection.
CDR3 sequences of the Vβ6Jβ3 and Vβ6Jβ8 rearrangements from VHSV-infected fish. Spleen cDNA from infected fish 2 was amplified using Vβ6/Jβ3 or Vβ6/Jβ8 primers. PCR product was purified and cloned, and clones picked at random were sequenced. Nucleotide and amino acid sequence of the in-frame junctions were sorted according to their CDR3 length. Junctions from alternatively spliced Vβ6 transcripts are noted Δ. Underlined nucleotides correspond to the Dβ sequence. Boldfaced nucleotides are not encoded in the germline, and most probably represent N nucleotides. Bold, italicized nucleotides are compatible with P addition mechanism. An asterisk indicates the out-of-frame junctions.
Discussion
In a previous study, we have analyzed the modifications of four Vβ families of the rainbow trout during viral infection. To allow for further insight into the diversity of the rainbow trout T cell repertoire, we have explored in the present work the characteristics of TCRβ transcripts expressed in the spleen of naive fish. A rainbow trout Cβ-anchored cDNA library was constructed using a 5′ RACE strategy, and a total of 47 Vβ-Dβ-Jβ-Cβ clones were identified. A total of 32 coded for Vβ1–4 segments already described, and 15 showed undescribed Vβ sequences, which were grouped into 6 new Vβ families. This diversity was not surprising in cold blood vertebrates, because 9 and 10 Vβ families were reported from the axolotl and from the frog Xenopus, respectively (29, 30). Taken together, our results are consistent with the current view that vertebrate Vβs are organized in many families, each composed of one or several members (12). Rainbow trout Vβ1–3 families were confirmed to contain several members (22 , and this study). In addition, Vβ7 and Vβ8 contain at least three and two different sequences, respectively. However, the diversity observed in rainbow trout Vβ families could represent different genes or different alleles. This issue could only be resolved with the complete sequencing of the rainbow trout TCRβ locus.
Although many fish Vβ sequences became recently available, we could not find in other fish species any sequence that could be assigned to a rainbow trout Vβ family. By comparison, members of the same Vβ family are often found in different species of mammals (14, 31). This difference could simply reflect the longer evolutionary distance separating the branches of teleost species compared with those in mammals. Similarly, it has been reported that the genetic distance among Vβ families was greater in the axolotl than in mammals (29).
Most of the key residues involved in the correct folding of Vβ domain, as defined by Chothia et al. (25), were globally conserved in rainbow trout Vβs. Therefore, rainbow trout Vβ sequences should comply with the general three-dimensional structure of the TCRαβ described in mammals, which is probably mandatory for the expression of a TCR at the cell membrane. This was further confirmed through systematic Immunoscope studies, which de facto probe the functionality of the TCR in T cell selection. Indeed, all the new rainbow trout Vβ families showed typical bell-shaped CDR3 length distribution, proving they were able to generate functional and diverse rearrangements. Residues conserved in rainbow trout were mainly localized in strands AA′BC and F, while the sequences of the strands C′C″DE showed more variability in length and sequence. Indeed, the length of the W34-C92 region, which contains the CDR2 region, varied from 52 to 63 aa, compared with 56–59 residues in humans.
We failed to identify key residues typical of CDR1 or CDR2 mammalian canonical structures, as defined by Al-Lazikani et al. (32). Although the residues at the positions defining the CDR1 mammalian canonical structures were different in rainbow trout, these residues were strikingly conserved among the 10 rainbow trout Vβ families (Table IV⇓). On the contrary, the positions defining the CDR2 canonical structures in humans and mice were not conserved among rainbow trout Vβ families. These observations suggest that in rainbow trout Vβs, the CDR1 loop is more structurally constrained than the CDR2.
Residue distribution found in rainbow trout Vβ sequences at the key positions defining canonical structures of CDR1 and 2 in β-chains of mammalian TCR (as described in Ref. 34 )
In humans and mice, crystallographic studies of several TCR/peptide/MHC complexes identified contacts between CDR regions of the TCR α- and β-chains with the MHC molecule. They revealed that the TCR contacts the MHC molecule mainly through CDR1 and CDR2, while CDR3 is more centered on the peptide (33, 34, 35). If rainbow trout Vβ comply with the same general structure, it is therefore likely that the diversity of CDR2 will affect the interaction between the TCR β-chain and the MHC molecules. Interestingly, rainbow trout MHC class I molecules possess unique features. UAB (rainbow trout classical class I) alleles are much more divergent among each other than in primates (mean difference 23.8% between alleles, compared with 5.4% among HLA-B) (17). In addition, β2-microglobulin genes are also highly diverse in rainbow trout (36). Thus, the diversity of CDR2 length provides an additional level of diversity to teleost T cell repertoire, which probably corresponds to a higher variability of TCR/peptide/MHC interactions.
Among rainbow trout Vβ families, Vβ6 was clearly divergent from typical TCR Vβ sequences known through vertebrates. Concerning the predicted properties of this Vβ domain, the CDR2 region was short, highly charged, and hydrophilic. In fact, Vβ6 had not only the shortest CDR2, but also the longest CDR1, suggesting that CDR1 and CDR2 length may be interdependent. Interestingly, in the shark Heterodontus, the Vβ7 family has also the longest CDR1 (14 residues in C23-W34 instead of 12–13) and the shortest CDR2 (54 residues in W34-C92). Unlike for the other rainbow trout Vβ sequences, comparison of Vβ6 with databases failed to detect high similarity (E value < 1e−08) with known V domains of the Ig superfamily. The most similar sequences were not even TCR V domains, but VH Ig domains. However, Vβ6 was expressed in typical Vβ-Dβ-Jβ-Cβ transcripts and rearranged with various Jβs, leading to a diverse repertoire of junctions. In this context, it was interesting to note that NAR has a very short CDR2 region and is functionally expressed. In such V domains, a short CDR2 would connect the two β-sheets more like in Ig C1 domains.
Because rainbow trout Vβ6 had peculiar features, the question of the expression of Vβ6 in a functional TCR was raised. We therefore focused an Immunoscope analysis on this particular Vβ family in the context of the T cell response against VHSV. As observed for the other Vβ segments (22 for Vβ1–4, and data not shown for Vβ5, Vβ7–10), this experiment identified several Vβ6 responses against the virus. T cells expressing Vβ6 rearrangements have been therefore subjected to Ag-driven amplification, indicating Vβ6 should be expressed at the protein level. Vβ6 appeared as two forms of transcripts with sequence characteristics indicative of alternative splicing within the V domain. Approximately one-third (10 of 27 in naive trout) of Vβ6-Jβ-Cβ transcripts lacked 35 residues corresponding to strands C′, C″, D, and to a part of the E strand. To our knowledge, Vβ transcript with such characteristics has never been described and may have no functional relevance. Because we never found a modification of the Immunoscope CDR3 length profile restricted to the short form of Vβ6 transcript, we have no evidence that the short form of Vβ6 was expressed at the protein level. However, short forms of transcripts due to alternative splicing within a V domain have been already described for another gene of the Ig superfamily, NKP30 (37) (see Fig. 2⇑D). It is interesting to note that the region lacking in NKP30 variant roughly corresponds to the same region that was lacking in clone BN62, although it is shorter. Furthermore, both NKP30 and Vβ6 have long CDR1 regions: C23-W34 region of 16 and 15 residues, respectively. By comparison, this region has 12 of 13 residues in other rainbow trout Vβ, and 11 of 12 residues in human Vβ sequences. It raised the question of the existence of a truncated structure of Ig V domain, which the short forms of NKP30 and Vβ6 may adopt. Intriguingly, the truncation may be compatible with the current three-dimensional model of Vβ domain because the ends of the E and C strands are in close proximity. However, it is probably not safe to use the Ig V domain model to predict the structure of such truncated sequences. Besides, there is no published evidence of the protein expression of the NKP30 short form. We therefore favor the idea that spliced Vβ6 transcripts are byproducts made in each Vβ6+ T cell due to the presence of functional splicing signal in the Vβ6 sequence.
Although Vβ6 is basically a TCRβ segment, its primary structure looks rather like a chimere of features from various types of receptors from the Ig superfamily. The early evolution of Igs and TCRs is still poorly known, but the description of segments with new features suggests that the repertoire of Ag receptors, and the constraints shaping their evolution, may be more complex than previously believed.
Acknowledgments
We thank A. Colette, A. Six, and P.-A. Cazenave for expert advice, and for providing the ISEApeaks software. B. Buteau, F. Coulpier, and the staff of the experimental fish facilities are also acknowledged for their excellent technical assistance. Clonal fish were provided by M. Dorson and E. Quillet. We thank J. Kanellopoulos for helpful discussions, and L. Du Pasquier for many suggestions, and for comments on the manuscript.
Footnotes
-
↵1 This work was supported by the Institut National de la Recherche Agronomique and European Community Project FAIR 98-4026.
-
↵2 Current address: Département de Biologie Cellulaire, Institut Cochin de Génétique Moléculaire, 22 rue Méchain, 75014 Paris, France.
-
↵3 Address correspondence and reprint requests to Dr. Abdenour Benmansour, Institut National de la Recherche Agronomique, Unité de Virologie et Immunologie Moléculaires, 78352 Jouy-en-Josas cedex, France. E-mail address: abdenour{at}jouy.inra.fr
-
↵4 Abbreviations used in this paper: CDR, complementarity-determining region; CB, Cβ-specific primer; NAR, nurse shark Ag receptor; VB, Vβ-specific primer; VHSV, viral hemorrhagic septicemia virus.
- Received July 25, 2002.
- Accepted September 24, 2002.
- Copyright © 2002 by The American Association of Immunologists