Axolotl MHC class II β chain: predominance of one allele and alternative splicing of the β1 domain

2.1 MHC class II B loci defined by Southern blotting with the β2 and β1 probes

Genomic DNA (gDNA) of blood cells from axolotls of different origins were digested, Southern blotted, and hybridized with the Amme-DB.010 β2 probe. Whatever the restriction enzyme used, BamHI, EcoRV, EcoRI or NcoI, this probe revealed a single hybridizing band (Fig. 1). With BamHI, the single band revealed was polymorphic: 3 kb (lane 1) or 10 kb (lane 2). With EcoRV, a single 16-kb fragment was evidenced whatever the sample origin: wild or Basel stock for A. mexicanum (WAm and BAm, lane 3), or for a wild A. tigrinum (WAt, lane 4). A restriction fragment length polymorphism (RFLP) was observed with EcoRI when the two ambystoma species were compared: a unique 9.5-kb EcoRI fragment was observed for sibs derived from the three BAm families (26 animals tested; lane 5), a unique but different 11-kb fragment was evidenced in A. tigrinum (lane 6). With NcoI, a new RFLP was evidenced between the two species, an 8-kb fragment was in BAm stock individuals or in WAm (lane 7), and a 12-kb fragment in WAt (lane 8).

Then, Southern blots of Eco RI digests were hybridized with the Amme-DB.010 β1 and β2 probes (Fig. 2). Again, even under non-stringent washing conditions, a single 9.5-kb EcoRI fragment was detected with the β2 probe in all BAm (Fig. 2, B2 lane 1), and in wild axolotls (lane 4). More complex restriction profiles were obtained with the β1 probe (Fig. 2, B1). Indeed, under the same experimental conditions, different profiles were obtained, each composed of three to four bands. This feature was shared by the BAm stock (Fig. 2 lanes 2,3 = ten sibs of the A family as an example) and the wild animals (Fig. 2, lanes 5,6 = nine sibs of a WAm family). In all cases, the β1 probe revealed a 9.5-kb band for all BAm and for all WAm axolotls.

The fact that the β1 and β2 probes revealed both a 9.5-kb fragment in all analyzed samples may indicate the presence of β1 and β2 exons on the same 9.5-kb fragment. The additional β1 bands may correspond to MHC class II B loci, unrevealed by our β2 probe.

2.2 Thymic cDNA library screening with both β1 and β2 probes and analysis of full-length class II B clones

In order to look for putative β2 divergent sequences associated to β1 sequences, broad screening of the thymic axolotl cDNA library was carried out with a mix of β1 and β2 probes. Using low stringency hybridization conditions, this analysis allowed isolation of 14 clones. Among these, (1) eight clones hybridized with both β1 and β2 probes: four clones were 5′ truncated (starting at different sites in the β1 exon) and four started in the 5′ untranslated region (UTR). The longer ones (clones 020 and 021) were completely sequenced and are shown in Fig. 3. (2) six clones hybridized with the β2 probe only: one clone started in the β2 exon (5′ truncated) and five clones lacked the β1 exon. Among them, two started in the leader peptide sequence and three in the 5′UTR sequence. Two clones (022 and 023) were entirely sequenced (Fig. 3). (3) No clone hybridized with the β1 probe only.

Clones 020 and 021 were independent since their 5′ and 3prime; extremities were different (Fig. 3). Their sequences overlapped on 1928 bp. Length inequality was due in part to different starts in the 5′UTR (23 to 45 bp long, respectively) and to the presence of two polyadenylation signals separated by 140 bp. Clone 021 (1968 bp) used the first polyadenylation signal, and encoded a mature MHC class II β chain of 240 amino acids (aa), with a 26-aa long putative leader peptide, identical to the coding sequence of the Amme-DB.010 partial splenic cDNA clone already described 15. Both β1 and β2 domains were 93 residues long, the connecting peptide and transmembrane region included 34 residues and the cytoplasmic tail was composed of 20 residues. The coding sequences of the six other clones which hybridized with both β1 and β2 probes were identical to clone 021.

This result pinpoints the fact that no other functional class II B locus can presently be evidenced in thymus, which would hybridize with the β1 probe but not with the β2 probe. These eight β1 and β2 positive clones, and the previously described partial splenic clones 15, were then renamed Amme-DAB.00x to assign the same locus origin.

2.3 Evaluation of PBD sequence variability and allelic polymorphism

The β1 domain sequence is expected to be polymorphic as part of the PBD of the MHC class II molecule. However, all the cDNA clones obtained from the thymus cDNA library encoded identical β1 sequences common with the 5′ truncated sequence previously obtained from a spleen library 15. Thus, we turned to RT-PCR strategy on spleen cDNA to ascertain these unexpected results and to test more thoroughly Amme-DAB variability. RT-PCR were performed using specific primers flanking the β1 domain, and cDNA of three axolotls corresponding to different β1 RFLP profiles obtained from EcoR I digests (from BAm A, C and WAm samples). The amplified products were cloned and six clones were sequenced from each of the initial cDNA samples. All 18 sequences were identical to Amme-DAB.021, thus demonstrating that β1 restriction polymorphism was not linked to an allelic polymorphism. We also performed 5′ rapid amplification of cDNA ends (RACE)-PCR experiments with an antisense primer located in the NGDW conserved region of the β2 domain and a sense primer corresponding to the universal adapter primer.

Finally, out of 52 β1 sequences, obtained from different batches of cDNA (80 animals from Ax 6 stock for the thymic library, three different animals from Ax 6 stock for the splenic library, six others from Ax 6 stock and two wild axolotls WAm for RT-PCR), 51 were identical and only one presented three non-synonymous mutations. This Amme-DAB.004 clone was found once by PCR but was never isolated by library screening 15. It may be a result of PCR artefact or represent a variant, sharing 98.9 % identity with the dominant allele at the DAB locus.

2.4 Analysis of cDNA lacking the β1 sequence

The sequences of two β2-positive and β1-negative clones obtained by thymic library screening were analyzed. Clones 022 and 023 were 1825 bp and 1832 bp long and overlapped on 1825 bp. Length difference between the two clones was due to different starts in the 5′UTR. While lacking 270 bp of the β1 sequence, these two cDNA clones were identical for all the other domains with Amme-DAB.021 clone and included a 5′UTR stretch, the complete presumptive leader sequence, the first three codons, and the first nucleotide of the fourth codon (G) of the β1 domain which is spliced onto the second nucleotide (T) of the first codon of the β2 domain. The β2 domain was followed by the connecting peptide, the transmembrane domain, the cytoplasmic tail, and the 3prime;UTR sequences. The two clones thus encoded the same truncated β chain of 150 aa preceded by a 26-aa putative leader peptide and may share the same common pre-messenger RNA with clone Amme-DAB.021. They were then renamed Amme-DABmB1.022 and Amme-DABmB1.023 (mB1 for minus β1-encoding domain), as coding for a shorter MHC class II B isoform.

2.5 Detection of the two different class II B transcripts

Northern Blot analysis of spleen and thymus has already shown two different transcripts of approximately 2.2 kb and 1.9 kb revealed with the β2 probe, the β1 probe revealing only the 2.2-kb band 15. The size of the longer band is compatible with a full-length Amme-DAB cDNA. The shorter one is compatible with an Amme-DABmB1 cDNA lacking the 270-bp sequence encoding the β1 domain. We carried out a quantitative study on Northern blots performed with splenic mRNA extracted from Ax 6 axolotls (Fig. 4). Autoradiography density and PhosphorImager analyses were made on hybridization profiles obtained with the β2 probe. Using the two techniques, the estimated ratio DAB versus DABmB1 isoforms was 2 / 1 in spleen mRNA. A similar ratio was obtained from mRNA from thymic epithelial cells.

To estimate the presence of the two Amme-DAB transcripts in different wild individuals, their transcription was assessed by RT-PCR using a sense primer located in the leader sequence and an antisense primer located in the β2 domain (Fig. 5). Similar PCR profiles, two DNA bands, were obtained with different splenic cDNA derived from stock A. mexicanum, from wild A. mexicanum and from a wild A. tigrinum. The hybridization profiles obtained from these PCR products confirmed the presence of the β2 sequence in both bands, but no hybridization signal was detected on the shorter band when using the β1 probe (Fig. 5).

These results suggest the natural co-expression in ambystomidae thymic and splenic cells of two MHC class II B cDNA isoforms encoding a complete β chain and a shorter chain lacking the β1 domain.

2.6 Phylogenetic analysis of vertebrate class II β chain

β1 and β2 aa sequences from axolotl, Xiphophorus, Xenopus, pheasant and human were aligned (Fig. 6). Compared to Amme-DAB, the highest identities were obtained with Xenopus Xela-DAB and human HLA-DRB1 (52 % and 55 % for β1 and 48 % and 46 % for β2, respectively). These various vertebrate class II B sequences were also used to build phylogenetic dendrograms for β1 and β2 domains (Fig. 7). The branching order observed for the β2 tree matches the known phylogenetic relationships of major vertebrate classes, with a specific amphibian cluster located between fishes and birds. A particular node was observed with birds (chicken and pheasant) but not with Xenopus β1 sequences. The inconsistent feature for the β1 domain phylogenetic tree may relate to a different evolution timing for each exon: in contrast to the β1 domain subjected to positive selection as part of the PBD, the β2 domain could have evolved at a more constant and slower rate.

Despite the ability of the Amme-DAB locus to perform an alternative splicing, this phylogenetic analysis shows that the axolotl DAB locus is not divergent and belongs to the vertebrate classical class II B family.

4.1 Animals

Axolotls of the Ax 6 strain (Paris-Black) were bred in our laboratory colony in 19 °C tap water and fed with trout pellets. They represent the F5 generation of brother-sister matings (J. Charlemagne, personal communication). Twenty-six BAm animals coming from three F1 families (A, C, D) were analyzed.

Wild animals captured in Mexico were also used in this study, 19 WAm, and four WAt.

4.2 Genomic DNA extractions

Blood cells were incubated in lysis buffer (0.1 M EDTA, 0.05 M Tris-HCl, 0.1 % SDS) at 65 °C for 45 min, supplemented with proteinase K (100 μg / ml lysis buffer) and incubated at 45 °C for 12 h. After salt precipitation of proteins, genomic DNA was recovered from the lysate, precipitated with 0.6 volume isopropanol and ethanol washed.

4.3 RNA extractions and cDNA synthesis

Total RNA were isolated from axolotl spleen cells using the RNA PLUS^TM protocol (Bioprobe, Montreuil-sous-bois, France). PolyA⁺ RNA were extracted on PolydT Sephadex beads (Pharmacia, Piscataway, NJ). The first strand cDNA synthesis was performed as already described 6.

4.4 Southern and Northern blot analyses

gDNA (20 μg) was digested with 240 U of restriction enzyme (40 U / μl, Promega, Madison, WI) in 100 μl for 20 h at 37 °C. After heat denaturation, samples were loaded on a 0.8 % agarose gel and separated for 18 h at 40 V. DNA in the gel was treated with 0.25 N HCl before denaturation with 0.4 N NaOH. Samples (0.5 μg) of mRNA isolated from spleen were separated in a 2 % agarose TAE denaturing gel at 30 V for 18 h in MOPS running buffer. After blotting, membranes were hybridized with Amme-DAB.010 β1 or β2 probes as previously described 7. A final wash was carried out at 65 °C with 2 × SSC, 0.025 % SDS. Autoradiography was performed at − 80 °C using Kodak films (X-O MAT AR, Stuttgart, Germany) or PhosphorImager (Amersham, Freiburg, Germany) exposure.

4.5 Screening of axolotl thymic cDNA library

An unamplified cDNA library (kindly provided by Dr. Fellah) was built from polyA⁺ RNA extracted from thymic lobes of 80 Ax-6 axolotls aged 6 to 12 months and inserted into the λZAPII vector (Stratagene, La Jolla, CA). One million PFU were screened by hybridization using a mix of β1 and β2 probes. Positive clones were purified and excised in vivo with the R408 helper phage (Stratagene).

4.6 DNA sequencing and sequence analysis

Nucleotide sequencing was performed with the sequenase version 2.0 kit (US Biochemicals, Cleveland, OH) on double-strand DNA. All sequences were checked on double-strand at the MWG (Ebersberg, Germany) company. Nucleotide and deduced aa sequence alignments were performed using FASTA, BLAST, LALIGN and CLUSTAL W computer softwares with Infobiogen (www.infobiogen.fr /) facilities.

4.7 RT-PCR, 5′RACE-PCR analyses and labeled PCR probes

RT-PCR was carried out in a total volume of 20 μl containing 1 % of the total cDNA, 50 μM of each dNTP, 1 μM of each primer and 1 U Goldstar DNA polymerase (Eurogentec, Seraing, Belgium) in PCR buffer. Thirty cycles of 1 min at 94 °C, 1 min at 60 °C and 1 min at 72 °C were performed. 5′ RACE-PCR were performed on splenic tailed cDNA 6 using an antisense primer located in the conserved region of the β2 domain NGDW (5′ CTGCCCGGTCCAGTCCCCGTT 3′) and a sense primer corresponding to the universal adapter primer (5′ CGGAATTCTCGAGATCGA 3prime;). β1 and β2 probes were first PCR amplified from the plasmid Amme-DB.010 clone with specific sense and antisense primers of β1 (5′ GATGATTTCGTGACTCAGTGG 3′ and 5′ TCTGCGGCGCTGCATGAAGTC 3′) and of β2 (5′ GTGAAGCCCAAAGTGAAAGTG 3′ and 5′ CCAGTTTCTCCTGAGGGGCTC 3′) domains, respectively. The sense primer used for the β1 probe starts with the fourth codon of β1. One nanogram of amplified and purified DNA fragment was then labeled with [α-³²P] dCTP by PCR with 5 mM MgCl₂, 3 μM of each dATP, dGTP, dTTP, 0.25 μM primers, and 2.5 U Taq polymerase.

4.8 Construction of dendrograms

The inferred protein sequence of the Amme-DAB.021 cDNA clone was aligned with other vertebrate MHC class II sequences using the CLUSTAL W computer program 34. The percentages of aa identity between aligned sequences were used to construct distance matrices 35. The evolutionary relationships were then calculated by the neighbor-joining algorithm 36 and presented by the Njplot software (Manolo Gouy, Laboratoire de Biométrie, Lyon, France). One thousand bootstrap replications were performed to determine the reliability of the branching order.