Key Points
A new, MHC-linked class I gene (UDA) was found in cartilaginous fishes.
UDA is monomorphic and in single/low copy number in elasmobranchs.
Chondrichtyans have at least four class I lineages, one classical and three nonclassical.
Abstract
Cartilaginous fishes, or chondrichthyans, are the oldest jawed vertebrates that have an adaptive immune system based on the MHC and Ig superfamily–based AgR. In this basal group of jawed vertebrates, we identified a third nonclassical MHC class I lineage (UDA), which is present in all species analyzed within the two major cartilaginous subclasses, Holocephali (chimaeras) and Elasmobranchii (sharks, skates, and rays). The deduced amino acid sequences of UDA have eight out of nine typically invariant residues that bind to the N and C termini of bound peptide found in most vertebrae classical class I (UAA); additionally, the other predicted 28 peptide-binding residues are perfectly conserved in all elasmobranch UDA sequences. UDA is distinct from UAA in its differential tissue distribution and its lower expression levels and is mono- or oligomorphic unlike the highly polymorphic UAA. UDA has a low copy number in elasmobranchs but is multicopy in the holocephalan spotted ratfish (Hydrolagus colliei). Using a nurse shark (Ginglymostoma cirratum) family, we found that UDA is MHC linked but separable by recombination from the tightly linked cluster of UAA, TAP, and LMP genes, the so-called class I region found in most nonmammalian vertebrates. UDA has predicted structural features that are similar to certain nonclassical class I genes in other vertebrates, and, unlike polymorpic classical class I, we anticipate that it may bind to a conserved set of specialized peptides.
Introduction
The chondrichthyes or cartilaginous fish, including all its modern representatives (i.e., elasmobranchs [sharks, skates, and rays] and holocephalans [chimaeras]), comprise the oldest class of extant jawed vertebrates possessing an adaptive immune system grounded on Igs, TCRs, and the MHC (reviewed in Refs. 1–3). Thus, cartilaginous fish are essential for studying the evolution of the adaptive immune system as well as for identifying ancestral and derived features of adaptive immunity. Previously, it was thought that chondrichthyans had a very simple immune system; however, many studies of shark immunology conducted in the last three decades have shown their immune system to be quite complex [(4–7); reviewed in Refs. 2, 8, 9] but still preserving several ancestral features, such as the cluster-type organization of Ig genes (10), close genetic linkage of MHC class I processing and presenting genes (11), and the MHC linkage of β-2 microglobulin (12).
MHC class I molecules play the central role in Ag presentation to CD8-positive T cells. Classical class I (class Ia) has been best studied for its function of presenting peptide Ags to trigger the activation of cytotoxic T cells, whereas nonclassical class I (class Ib) molecules have a myriad of functions (review in Ref. 13), mostly examined in mammals. The classical class I protein from most vertebrates generally shows conservation of 9 aa residues that bind to the N and C termini of bound peptide in the peptide-binding region (PBR). Class Ia genes show high levels of polymorphism, linkage to the MHC, and ubiquitous tissue expression. In contrast, nonclassical class I genes are generally monomorphic or minimally polymorphic, have a limited tissue distribution, may or may not be linked to the MHC, and have been recruited for multiple functions, some not even involved in immunity (3, 14–17).
Classical MHC class I genes (UAA) have been identified in several cartilaginous fishes, with similar structural and genetic features as are found in other vertebrates (5, 18). Several nonclassical class I genes and lineages also have been described in cartilaginous fish: UBA in nurse shark (Ginglymostoma cirratum) and horn shark (Heterodontus francisci) (5, 19) and a highly divergent gene, UCA, so far found only in spiny dogfish (Squalus acanthias) (20). To date, no functional study has been performed on any class I molecules in cartilaginous fish. Note that we designate these previously described nonclassical class I genes in cartilaginous fish for the first time in this report, following the nomenclature criteria for MHC genes in nonmammalian species (21).
In this study, using predominantly the chondrichthyan genome and transcriptome databases, we have identified a new nonclassical class I lineage in all cartilaginous fish examined and assigned it as UDA. We characterized this new lineage, focusing on its expression, level of polymorphism, deduced structural features, presence or absence in cartilaginous fish species, and linkage to the MHC.
Materials and Methods
Database searches
Several representative MHC class I sequences from GenBank (http://www.ncbi.nlm.nih.gov), namely nurse shark (G. cirratum) and horn shark (H. francisci) class Ia (Gici UAA: AAF66110 and Hefr UAA: AAC60349) and class Ib (Gici UBA: AAC60347 and Hefr UBA: AAC60348), spiny dogfish (S. acanthias) class Ib (Sqac UCA: AAN78091), and human CD1 (AAX49405), were used as templates to search for MHC class I sequences using a blastp in nonredundant protein databases and tblastn in the transcriptome shotgun assembly and short read archive databases using default parameters in public databases (http://www.ncbi.nlm.nih.gov; http://skatebase.org) (Supplemental Table I). After searching the nurse shark transcriptome (SRX219865; SRX219866), we assembled the UDA gene using the Geneious software 6.0 (22). A 42-bp gap (positions 612–653 bp) in this assembly was filled with PCR in the same nurse shark individual used for the transcriptome (23). The primers α2 forward (Fw) 5′-GGTGCTGCAGTACTGAATCG-3′and α3 reverse (Rv) 5′-GTATCTCCTTCGGTGCAGG-3′ PCR was performed at 95°C for 2 min, followed by 35 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s, with a final extension of 72°C for 10 min using GoTaq Master Mix (Promega). The PCR products were cloned into pGEM-T Easy Vector and sequenced.
Sequence alignments and phylogenetic tree analyses
Deduced amino acid sequences were aligned using the ClustalX program in the Geneious software 6.0 (22) with manual adjustments. The neighbor-joining (NJ) phylogenetic tree of the class I peptide-binding domains (α1 and α2) was constructed in MEGA 6.06 (24) using p-distances, uniform rates among sites, pairwise deletions, and 10,000 bootstrap replicates. The U and Z lineages of MHC class I from bony fish were also included, and we chose the human and chicken CD1 molecules as outgroups because they are the most divergent class I lineage in vertebrates (16).
Northern blotting
Lineage-specific tissue expression was assessed and compared between UDA and UAA using Northern blotting. Ten micrograms of total RNA from various nurse shark tissues (brain, epigonal, gill, gonad, liver, muscle, pancreas, spiral valve, spleen, stomach, thymus, and WBCs) were electrophoresed on denaturing 1% agarose gel electrophoresis and subsequently transferred onto nitrocellulose membranes as previously described (5). Hybridization was done using 32P-labeled regions encoding the α3 domains of UAA and UDA as well as a loading control using the nucleoside diphosphate kinase probe (4) under high-stringency conditions (5).
In situ hybridization
To detect cell types expressing UDA within organs, we performed in situ hybridization on those tissues with the highest expression on the Northern blotting (epigonal, spiral valve, gill, and spleen). Nurse shark tissues were collected and fixed in 4% paraformaldehyde in 1× SPB solution (0.06 M phosphate buffer [Na2HPO4/NaH2PO4]/3% sucrose/0.15 mM CaCl2 [pH 7.4]) for 6 d at 4°C overnight. Tissues were rinsed gradually in SPB containing an increasing amount of sucrose from 10 to 30% and infiltrated overnight at 4°C. The fixed tissues were then embedded in OCT medium (Sakura Finetek) and frozen in a liquid nitrogen/2-methylbutane bath. Frozen tissues were sectioned (8 μm in thickness) and mounted onto glass slides. Nurse shark UAA and UDA
Southern blotting
To estimate the presence/absence and number of UDA genes in various chondrichthyan species (including those with no transcriptome or genome sequences), we performed Southern blotting. For the cartilaginous fish blot (“Chondroblot”), we digested 10 μg of genomic DNA (gDNA) extracted from erythrocytes, with BamHI for 48 h and electrophoresed in a 0.8% agarose gel. The digested gDNA was transferred onto a nitrocellulose membrane via capillary transfer, and a 32P-labeled α3 domain probe of nurse shark UDA was hybridized to the membranes under low-stringency conditions (5). The membrane was exposed to x-ray film for different periods to obtain the optimal signal strength. For the nurse shark family blot, the same α3 probe was used for hybridization but under high-stringency conditions (5).
Statistical analysis of linkage
We validated the linkage status of UDA to the MHC using parametric linkage analysis. We calculated the odds of the likelihood of whether two loci are linked versus nonlinked using an MHC-typed family of 39 siblings (19, 25). Family-based linkage analysis makes use of the information of at least one of the parents (e.g., in this case, the mother) and a large number of descendants to detect cosegregation of markers. We compared the RFLP banding pattern of UDA to the MHC haplotypes and determined concordance or nonconcordance patterns between the UDA and MHC haplotypes. The log of the odds (LOD) score was calculated as previously described (12):where θ is the recombination frequency, R is the number of recombinant offspring, and NR is the number of nonrecombinant offspring. Because we do not know whether the parental phase is linked or nonlinked, we calculated both phases (i.e., phase 1 and phase 2: switching R and NR) and averaged. Because the recombination frequency is not known, we calculated LOD scores with the recombination frequency (θ) ranging between 0 and 0.5 to obtain recombination frequency at maximum LOD score (26). The p value was further calculated using a one-sided χ2 test at the maximum LOD score (27).
Determination of the degree of polymorphism
To amplify the UAA and UDA genes from nurse shark gDNA, we performed a PCR using the GoTaq Master Mix (Promega) and the following primers: UAA primers α1 Fw 5′-GGTCTCACAGTCTCCGGT-3′and α2 Rv 5′-GGTCTCAGTTCCACATTTCC-3′ PCR for UAA was performed at 95°C for 2 min, followed by 35 cycles of 95°C for 45 s, 50°C for 30 s, and 72°C for 45 s, with a final extension of 72°C for 10 min. UDA primers were based on the new sequence obtained from the nurse shark (GenBank accession no. MN339476) and were as follows: α1 Fw: 5′-CTGAGGTATTACTACACCTC-3′ and α1 Rv: 5′-GTTCTCCTTGTTGCTATCTG-3′ and α2 Fw: 5′-CAGGTTTGAACTACCTGCA-3′ and α2 Rv: 5′-CGATTCAGTACTGCAGCACC-3′. PCR was performed using GoTaq Master Mix (Promega) at 95°C for 2 min, followed by 35 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s, with a final extension of 72°C for 10 min. All PCR products were cloned into pGEM-T Easy Vector (Takara Bio), and at least three clones per individual (in a total of four individuals) were sequenced to eliminate PCR errors. Note that we used α1–5′ and α2–3′ primers for amplification of gDNA for UAA because of the short intron size between the α1 and α2 encoding exons, but the same strategy did not work for UDA, and we had to use intraexonic primers and amplified the α1 and α2 regions separately. To further confirm the monomorphic feature and eliminate PCR errors of UDA alleles, we set up separate PCR reactions, extracted the corresponding bands from the gel, cleaned, and sequenced them directly.
Results
Identification of a new cartilaginous fish class I lineage
Our initial approach was to identify all annotated MHC class I genes in the available cartilaginous fish genomes and transcriptomes using published MHC class Ia and class Ib sequences from different shark species as well as human CD1 [the most divergent vertebrate class I gene (16)] (Supplemental Table I) as templates for blast searches. One group of sequences (representative sequences in Fig. 1, entire list of sequences in Supplemental Figs. 1, 2, Supplemental Table I) was found to be quite different from the three previously published class I lineages, UAA, UBA, and UCA, at the nucleotide (51, 50, and 41%, respectively) (data not shown) and protein level (37, 36, and 23%, respectively) (Supplemental Table II), and thus, we named it UDA following the MHC nomenclature for nonmammalian vertebrates (21). The deduced amino acid sequence of UDA has the basic features of all class I molecules with the typical peptide-binding domains (α1 and α2) and the Ig superfamily–based α3 domain. Residues in dark shade in Fig. 1 are highly conserved in all class I proteins, primarily for maintaining structural integrity (16, 28). The α2 domain has the canonical disulfide bridge found in most class I molecules, and there is a “typical” Asp/Glu-containing connecting piece and a hydrophobic transmembrane region (Fig. 1, Supplemental Figs. 1, 2). The Asn-linked glycosylation site at the C terminus of the α1 domain, present in all classical and most nonclassical class I molecules, is not found in all UDA sequences (Fig. 1, Supplemental Fig. 1). Most conspicuously, UDA was found to have a much longer cytoplasmic tail (Fig. 1, Supplemental Fig. 2) than most class I lineages; this characteristic is similar to the bony fish typical Z class I lineage (29) but unlike the highly conserved length of classical class I in most vertebrates (30, 31). The UDA lineage does not have conserved Tyr (Y320, position based in HLA-A2 sequence) in the cytoplasmic tail found in the classical class I (Fig. 1, Supplemental Fig. 2). The conserved Ser (S335, position based on the HLA-A2 sequence) in the classical class I cytoplasmic tail may be present in UDA, although this is highly speculative because of the high Ser content in all UDA cytoplasmic sequences and the high divergence between the UDA sequences across cartilaginous fish, making the alignment of the cytoplasmic region and the ascertainment of the conserved Ser position ambiguous (Fig. 1, Supplemental Fig. 2). No distinct signaling motifs were identified in the UDA cytoplasmic tail using ScanProsite (http://prosite.expasy.org/scanprosite) nor were there long stretches of conservation among UDA in the different cartilaginous fish species.
Multiple amino acid alignment of the four distinctive MHC class I lineages in cartilaginous fish, UAA, UBA, UCA, and the new lineage UDA. GenBank accession numbers are listed in Supplemental Table I. Dots indicate amino acids identical to Gici UAA, and dashes indicate gaps, respectively. Highly conserved residues in all class I proteins are shaded black (16, 28), s and h indicate the β-strands and α-helices, and the line connecting the two Cys in the α2 domain indicates the class I canonical disulfide bridge. P marks the invariant residues that bind to the N and C termini of the bound peptide in the classical class I molecules, and p indicates the other 28 peptide binding residues. The Asn marks the asparagine-linked glycosylation site, Asp/Glu indicates the typical aspartic acid and glutamic acid residues found in the connecting piece (Conn, light shade), and Tyr and Ser mark the conserved positions of Tyr and Ser in the cytoplasmic tail (also in light shade) of classical class I molecules.
In contrast to the poor conservation of the cytoplasmic tail between UAA and UDA, UDA shared eight of the nine invariant residues that bind to the N and C termini of bound peptides in almost all classical class I PBR, which lock peptide into the class I groove (28) (Table I). This feature is also found in the bony fish typical nonclassical Z lineage (32). Furthermore, interestingly, the other 28 predicted peptide-binding residues (16) of UDA are perfectly conserved in elasmobranch sequences, and among them, 18 out of 28 residues are hydrophilic (polar or charged), a feature shared with classical class I molecules but not the lipid-binding CD1 protein (only 3 out of 28 hydrophilic residues). Note that the peptide-binding residues of the typical Z lineage in all bony fish sequences are also nearly perfectly conserved, but these Z residues are different from those in UDA. Based on the conservation of these peptide-binding residues (but not the rest of the PBR between UDA in all of the cartilaginous fish species; percentage of identities between the different domains shown in Supplemental Table II), we predict that UDA (and Z) binds to a set of peptides with the same or similar anchor residues, unlike UAA. In contrast, preliminary results suggest that UCA, like CD1, has a strongly hydrophobic binding groove, suggesting that it also may bind to glycolipids.
UDA is an ancient class I lineage present in all cartilaginous fish
To understand the evolutionary relationship between all chondrichthyan class I sequences, we constructed class I NJ phylogenetic trees using the PBR (α1 and α2 domains) and rooted with the nonclassical CD1 protein (Fig. 2; alignment in Supplemental Fig. 1). As was suggested by Wang et al. (20), the shark UCA lineage is highly divergent, clustering outside of all vertebrate class I sequences besides CD1 (Fig. 2). UDA class I sequences were found in all holocephalans and elasmobranchs tested, and they formed a single clade supported with high bootstrap values (99%). The UDA lineage is more closely related to the groups of UAA, UBA, and U sequences and to the bony fish typical Z lineage when compared with UCA lineage. We also found that the cartilaginous fish UBA sequences (5) were only found in elasmobranch taxa, showing on average ∼50% similarity to UAA at the amino acid level between all cartilaginous fish sequences (Supplemental Table II), and most likely arose via duplication of the UAA lineage based on the tree topology (Fig. 2).
Phylogenetic tree of cartilaginous fish and other selected vertebrate class I molecules. The tree was constructed using the neighbor-joining (NJ) method with the α1 and α2 domains (PBR) and rooted with CD1. Bootstrap support values are shown as percentages on the branches. The boxes indicate the four distinctive MHC class I lineages in cartilaginous fish. GenBank accession numbers for all the sequences are listed in Supplemental Table I, and the corresponding alignment is found in Supplemental Fig. 1. Common names are shown followed by the abbreviation of the scientific name (two letters of the genus and two letters of species). The scale bar indicates the number of amino acid differences per sequence.
Tissue-specific expression of UDA
We examined the UDA expression in different nurse shark tissues and compared it to that of UAA by Northern blotting (Fig. 3). UDA was most highly expressed in the epigonal organ and spiral valve, moderately expressed in the gill, spleen, stomach, and WBCs and poorly expressed in the thymus, gonad, and brain. UAA exhibited the typical ubiquitous classical class I expression in many tissues, with the highest expression in immune organs, such as the spleen, thymus, and mucosae. To detect cell types expressing UDA within organs, we performed in situ hybridization on various tissues with the highest expression by Northern blotting (Fig. 3). In the gill, we detected expression of UDA within the filaments (especially in the so-called pillar cells) and weak expression on the gill epithelium, whereas UAA was highly expressed only by the epithelium (Fig. 4). We did not detect UDA expression in the other tissues with a high expression by Northern blot (epigonal, spiral valve, and spleen; data not shown), suggesting that UDA is expressed broadly but at low levels by many different types of cells or regions in these tissues. We should add that we have only studied the baseline expression of UDA, and many nonclassical class I molecules are induced under a variety of stimulatory conditions (33).
Unique tissue distribution of UDA lineage in nurse shark compared with UAA via Northern blotting. Nucleoside-diphosphate kinase (NDPK) was used as a loading control (ubiquitous expression), although we found it to have a lower expression in muscle and pancreas in the presence of similar amounts of 28S and 18S rRNA across samples. RNA marker size (kilobase) is shown on the left of the blot.
Differential expression in nurse shark gill section between the classical UAA and UDA lineages using in situ hybridization of full-length UAA and UDA region nurse shark riboprobe. (A) H&E-stained tissue section with the gill filament structures highlighted in the remainder images. (B–D) UAA antisense riboprobe. (E and F) UAA sense riboprobe. (G–I) UDA antisense riboprobe. (J and K) UDA sense riboprobe. (B, E, G, and J) DAPI contained staining. (C, F, H, and J) Riboprobe with streptavidin–Alexa Fluor 647 signal detection and (D and I) riboprobe with signal detection into streptavidin–alkaline phosphatase and NBT/BCIP substrate. Note the highest expression of UAA in epithelial cells (white arrows) and of UDA in some pillar (blue arrows) and epithelial cells (white arrows). All images were taken at original magnification ×20 except the H&E image that was taken at original magnification ×10.
UDA is single or low copy in elasmobranchs
In agreement with our bioinformatic searches, UDA genes were present in all chondrichthyan species tested by Southern blotting (Chondroblot in Fig. 5). Similar to UAA (5, 19), a single or low copy number of UDA genes was detected in all elasmobranchs. In contrast, several UDA genes were found in the Southern blot for the holocephalan spotted ratfish (Hydrolagus colliei). However, bioinformatic searches of the genome of another holocephalan, the elephant shark (Callorhinchus milii), revealed only one UDA gene. The C. milli genome assembly is incomplete (and unfortunately, no gDNA was available for Southern blotting), so our gene estimate of C. milii for this species might be an underestimate.
Presence of the UDA lineage in chondrichthyans via Southern blotting. BamHI-digested DNA from different species was hybridized with the nurse shark MHC class I UDA α3 domain probe under low-stringency conditions. The chondrichthyan species used were one chimaera (spotted ratfish H. colliei), three rays (cownose ray Rhinoptera bonasus, little skate Leucoraja erinacea, and thornback ray Raja clavata), and eight sharks (nurse shark G. cirratum, bamboo shark Chiloscyllium punctatum, horn shark H. francisci, sand tiger shark Carcharias taurus, catshark Scyliorhinus canicula, lemon shark Negaprion brevirostris, spiny dogfish S. acanthias, and blue shark Prionace glauca). The marker size (kb) is shown on the left. Bands in the gel are marked with arrows.
UDA seems to be monomorphic in nurse sharks
Classical class I genes are highly polymorphic in almost all vertebrates, whereas nonclassical class I genes are generally mono- or oligomorphic. To test the degree of UDA polymorphism, we selected four MHC-disparate nurse shark individuals collected from the wild and sequenced the regions encoding the α1 and α2 domains of UAA and UDA genes. The UDA sequences for all four individuals were identical (Fig. 6), further confirming UDA’s nonclassical status.
UDA is monomorphic in wild nurse shark individuals. (A) Alignment of nucleotide sequences obtained from cloning of PCR products of α1 and α2 UAA and UDA from four wild nurse shark individuals (1–4). (B) Trace files of nucleotide sequences obtained by direct sequencing of PCR products of α1 UAA (one individual) and of α1 and α2 UDA for the same four wild nurse shark individuals (1–4). Dots indicate nucleotides identical to nurse shark no. 1 (highlight).
UDA is linked to the MHC
Nonclassical class I genes in all vertebrates are found in various genetic regions, some encoded outside of the MHC. Thus, we examined linkage status of UDA using an MHC-typed nurse shark family of 39 siblings (19, 25). Three segregating UDA RFLP bands were obtained with the nurse shark α3 probe by Southern blotting (marked U, M, and L in Fig. 7). Previous studies showed that this family comprises 13 MHC groups (indicated groups a to m in Fig. 7, Table II) because of the multiple paternity of at least seven fathers (19, 25). These groups are the combination of maternal and paternal MHC haplotypes that were previously identified (Table II). We observed a strong correlation between m1 (UAA L band) and UDA U band and m2 (UAA U band) and UDA L band. However, there was no correlation in three of the siblings, 2, 8, and 24, suggesting that they are recombinants between UDA and MHC (indicated by R in Table II). Paternal haplotype p2 also had a perfect correlation with the UDA M RFLP band in all 13 siblings (Table II). Based on the correlations observed in m1, m2, and p2, we calculated the corresponding LOD score (Table III). Because we cannot distinguish maternal versus paternal bands when the siblings have the U/L band of UDA, we eliminated these siblings from our calculations to be conservative. We used a total of 30 maternal alleles, including three potential recombinants, and 13 paternal alleles for the calculations and obtained the LOD score 7.617 (odds are 4,139,967 to 1 with a p value of 3.169 × 10−09) (Table III) with a recombination frequency (θ) of ∼0.07. We anticipate that UDA is located at a relatively great distance from UAA (as well as from TAP and LMP, which are closely linked to UAA), perhaps to preclude recombination or other gene exchanges between UAA and UDA.
UDA is linked to the nurse shark MHC. The Southern blot was performed with gDNA from 39 siblings in a nurse shark family (animal number shown above the blot) and their mother using BamHI restriction fragments hybridized with nurse shark α3 domain of UDA probe. The previously identified 13 MHC groups a–m identified with an UAA probe are indicated below (19, 25). Marker size (kb) is shown on the left.
Discussion
The amino acid sequence alignments and phylogenetic tree analyses clearly indicate that the UDA lineage is a unique MHC class I lineage. UDA’s tissue distribution, low polymorphism, and unusual cytoplasmic tail show it to be a nonclassical class I molecule, likely having a specialized function. Based on its sharing of the classical class I canonical peptide-binding residues and the general high conservation of the hydrophilic PBR in all species, we propose that UDA binds to a specific set of peptides different from the polymorphic UAA, which binds to sets of peptides in an allele-specific fashion. Indeed, the UDA system may be similar to HLA-E (and mouse ortholog Qa1) that binds to leader peptides of classical HLA and peptides from some other self and foreign peptides in a “nonallele-specific” manner (34).
We showed that UDA is linked to UAA in the shark MHC, and similar scenarios of linked classical and nonclassical class I genes are found in many other nonmammalian species. For example, there are two class I genes identified as classical in the chicken, BF1 and BF2; BF2 is highly polymorphic with ubiquitous and high expression, whereas BF1 is oligomorphic and poorly expressed, more like a class Ib molecule. Kim et al. (35) found that BF1 is recognized by NK cells, whereas BF2 is recognized primarily by CTL. Their results suggest that BF1 may bind only to specialized peptides and not the diverse array of peptides presented by the highly expressed BF2 proteins. Another comparison to consider is the classical U and nonclassical typical Z class I lineages in bony fish. Like UAA and UDA, both bony fish class I lineages are MHC linked (but note that in bony fish, classical class I and class II genes are not linked; however, Z is linked to U) (29, 36). Also, like UDA, class I proteins of the Z lineage are predicted to have long cytoplasmic tails, preserve the classical class I canonical peptide-binding residues, and appear to be monomorphic within a species. However, our phylogenetic analysis does not support that UDA and the typical Z lineage are derived from a recent common ancestor.
Unlike the multigene family nonclassical UBA (5) but similar to UAA [and most vertebrate classical class I genes (17, 30)], UDA has a low copy number for all elasmobranchs examined. Because UDA is linked to the MHC and preserved in all cartilaginous fish, it emerged over 400 million years ago. The evolutionary conservation of UDA suggests that it plays an important role in the cartilaginous fish immune system or even another essential function. We detected three recombinants between UAA and UDA in our family study, suggesting that unlike UAA, UDA is not closely linked to the TAP and immunoproteasome (LMP) genes (no recombinants have ever been detected between these genes). Ancient lineages of TAP/LMP/UAA are found in many nonmammalian vertebrates, including sharks (25, 30, 37–39); UDA likely uses a different Ag-processing pathway, perhaps like mouse Qa1 and human HLA-E binding to peptides generated in the endoplasmic reticulum (34).
The Southern blot suggests that the holocephalan spotted ratfish (H. colliei) has many UDA gene copies as opposed to the elasmobranch species. However, only one copy was found in the genome sequence of its close relative, elephant shark C. milii. Whether the high copy number of spotted ratfish H. colliei is specific of this species remains to be confirmed, but it raises the question on whether the UDA copy number may be tied to the biological and ecological features of the different holocephalan taxa or is rather a general aspect of the holocephalans.
One important aspect to be considered in future studies is that previous work on chondrichthyan immunity has focused primarily on a model species for immune studies, the nurse shark (G. cirratum). Although this cartilaginous fish is an important animal model for many different immunological studies, its subtropical range, benthic habit, and perhaps unique exposure to pathogens might have resulted in specific features in its adaptive immune system. Indeed, lifestyle/habit may be a driver of copy number of MHC class I genes in cartilaginous fish. This feature can be found for other immune genes like Igs: the nurse shark G. cirratum has a relatively low number of Ig genes (∼15 IgM cluster) (40, 41) compared with horn shark H. francisci (∼200 IgM clusters) (42). Cartilaginous fish species differing in spatial range and ecological features may provide important and useful comparisons to assess the influence of habit and habitat on the genetic architecture of immune genes in this ancient vertebrate group. As suggested by Wilson (32), the lifestyle of different members of a vertebrate class, especially one as diverse as the chondrichthyans, should be reflected in the structure and function of different physiological systems, including the immune system.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
The authors are indebted to L. F. Castro, A. Machado, and A. μñoz-Mérida for help in preliminary data collection and bioinformatic analyses and for useful criticism of the data.
Footnotes
This work was supported by Portuguese funds through Portuguese Foundation for Science and Technology Grant PD/BD/114542/2016 (to T.A.) and Contracts IF/00376/2015 (to P.J.E.) and DL57/2016 (to A.V.). Y.O. and M.F.F. were supported by National Institutes of Health Grants AI140326-26 and AI02877. This work is also financed by FEDER Funds through the Operational Competitiveness Factors Program COMPETE and by national funds through the Foundation for Science and Technology within the scope of Project PTDC/ASP-PES/28053/2017.
The sequence presented in this article has been submitted to GenBank under accession number MN339476.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- Fw
- forward
- gDNA
- genomic DNA
- LOD
- log of the odds
- NJ
- neighbor-joining
- PBR
- peptide-binding region
- Rv
- reverse.
- Received August 21, 2019.
- Accepted December 5, 2019.
- Copyright © 2020 by The American Association of Immunologists, Inc.