Abstract
The gastrointestinal tract is populated by a multitude of specialized immune cells endowed with receptors for classical (class Ia) and nonclassical (class Ib) MHC proteins. To identify class I products that engage these receptors and impact immunity/tolerance, we studied gut-transcribed class Ib loci and their polymorphism in inbred, outbred, and wild-derived mice. Intestinal tissues enriched in epithelial cells contained abundant transcripts of ubiquitously expressed and preferentially gut-restricted Q and T class I loci. The latter category included the “blastocyst Mhc” gene, H2-Bl, and its putative paralog, Tw5. Expression of H2-Bl was previously detected only at the maternal/fetal interface, where it was proposed to induce immune tolerance via interactions with CD94/NKG2A receptors. Analysis of coding region polymorphism performed here revealed two major alleles of H2-Bl with conserved residues at positions critical for class I protein folding and peptide binding. Both divergent alleles are maintained in outbred and wild mice under selection for fecundity and pathogen resistance. Surprisingly, we found that alternative splicing of H2-Bl mRNA in gut tissues is prevalent and allele-specific. It leads to strain-dependent expression of diverse repertoires of canonical and noncanonical transcripts that may give rise to distinct ligands for intestinal NK cell, T cell, and/or intraepithelial lymphocyte receptors.
The balance between immune tolerance to harmless commensal flora and dietary products and immune response to harmful pathogens is highly regulated in the mammalian gastrointestinal tract. Although the relevant, gut-specific pathways of the innate and adaptive immune system are not yet fully understood, they likely depend on the immune effectors residing in, or migrating into, the intestinal tissues. Many gut-localized effector cells, including intraepithelial lymphocytes (IELs),3 T cells, and/or NK cells, express TCR, NKG2D, CD94/NKG2, and other receptors which, in vitro or in the context of systemic immunity, engage MHC class I Ags or related proteins and signal change in cytotoxicity, cytokine secretion, or other immune effector functions (1, 2). To date, only a handful of class I MHC Ags and class I-like proteins has been implicated in the regulation of mucosal immunity in the gastrointestinal tract (3, 4, 5, 6, 7).
In the mouse, the extended family of MHC class I-related proteins is very large and includes two to three highly polymorphic “classical” class I MHC Ags (class Ia) encoded by H2-K, D, and L loci, multiple oligomorphic “nonclassical” MHC proteins (class Ib) mapping to H2-Q, T, and M regions, and additional, structurally divergent class I-like proteins found outside of the Mhc (8, 9, 10). Although most mouse strains encode dozens of Q, T, and M class I genes, only a few of these genes have been studied in any detail. (For a full review on mouse class Ib Mhc organization, see Kumanovics et al. (8).) We performed a survey of the known mouse class Ib genes and identified those that are preferentially expressed in gut tissues and therefore may impact immune regulation in this tolerance-dominated milieu. Unexpectedly, we found that T region-encoded “blastocyst MHC” (11), the class I MHC that was previously proposed to be a homologue of tolerance-inducing, trophoblast-restricted human HLA-G (12), as well as its putative paralog, Tw5 (13), are abundantly transcribed in the gastrointestinal tract of many mouse strains. “Blastocyst Mhc,” designated as H2-Bl, has been shown recently to give rise to two alternatively spliced mRNAs in mouse placenta and a teratocarcinoma cell line, F9 (14). Furthermore, in a model system of a transfected RMA-S tumor and in the presence of Qa-1 class Ib (15), the two H2-Bl isoforms acted to protect target cells from NK cell-mediated lysis (14). The mechanism responsible for this pathway was hypothesized to involve binding of H2-Bl encoded leader peptides to Qa-1 molecules and their presentation to CD94/NKG2A inhibitory receptors on NK cells. This pathway was proposed to play an important role in enhancing reproductive success of the species by maintaining NK cell tolerance at the maternal/fetal interface in a manner reminiscent of human HLA-G (12).
We find that in intestinal tissues H2-Bl is alternatively spliced into multiple, truncated isoforms, in addition to the two detected in the mouse placenta (14). Furthermore, analysis of H2-Bl sequences from several inbred, outbred, and wild mouse strains revealed that the putative Ag-binding contact sites and the residues required for maintenance of class I structure are generally conserved. Intriguingly, in some cases, the limited polymorphism of the H2-Bl locus is associated with pronounced differences in the gut-expressed profile of mRNA isoforms. This observation suggests that allelic variation of H2-Bl genes controls the potential impact of its leader peptide-containing products during H2-Bl/NK cell receptor interactions in the gut. The findings are discussed in the context of the proposed relationship of H2-Bl to trophoblast-restricted HLA-G and other class Ib genes active in immune-tolerance dominated tissues.
Materials and Methods
Mice
129/SvJ and C57BL/6J mice were obtained from The Jackson Laboratory. CD1 (ICR)BR (henceforth referred to as ICR) mice were obtained from Charles River Laboratories. C57BL/6NCr mice used for DNA sequence analysis were obtained from the National Cancer Institute Animal Production Program. Kb−/− Db−/− mice (16) were generously provided by Dr. J. Forman (University of Texas Southwestern, Dallas, TX). C3H/HeJ, FVB/NJ, and BALB/cJ mice were generously provided by Dr. M. Bennett (University of Texas Southwestern). C57BL/6J, ICR, and Kb−/− Db−/− mice were housed under specific pathogen-free conditions at the University of Texas Southwestern Medical Center; C3H/HeJ, FVB/NJ, BALB/cJ, and C57BL/6NCr mice were housed in the University of Texas Southwestern conventional mouse facility. Analysis of gut-restricted class Ib genes from wild-derived mouse strains (17) was performed on DNA samples provided by Drs. A. Orth and F. Bonhomme from mice held at the genetic repository of Laboratoire Génome Populations Interactions Adaptation lab (Université de Montpellier, France; 〈www.univ-montp2.fr/∼genetix/souris.htm〉). All experiments involving animals were performed according to institutional review board guidelines.
Cell lines
The B16-derived B78H1 melanoma line (18) was provided by Dr. H. I. Levitsky (The Johns Hopkins School of Medicine, Baltimore, MD). B78H1 was propagated in a 1/1 mixture of high-glucose DMEM and RPMI 1640 (Mediatech) with 0.1 mM nonessential amino acids (Invitrogen Life Technologies) and 10% FBS (Atlanta Biologicals) at 37°C and 5% CO2.
RNA and DNA isolation
Tissues were harvested from 8- to 12-wk-old adult mice. Gastrointestinal tissues were washed of luminal contents. To enrich for intestinal epithelial cell (EC) populations, all visible Peyer’s patches were removed from small intestinal samples. All tissues were then flash frozen in liquid nitrogen and pulverized. Total RNA and genomic DNA were individually isolated using Tri Reagent (Molecular Research Center) according to the manufacturer’s protocols for both flash frozen/pulverized tissues and cultured cells. All RNA and DNA samples were quantitated spectrophotometrically. RNA quality was determined by visualizing rRNA bands on a 1% agarose gel.
DNA primers
Oligonucleotide primers were purchased from Integrated DNA Technologies. Primer specificity was determined by blastn analysis (National Center for Biomedical Information; http://www.ncbi.nlm.nih.gov/BLAST/). The sequences of the primers used in this study are listed in Table I. Primers designated H2-Blbc ex2 R2 and H2-Blbc ex5 R were first described by Sipes et al. (11) under the designations E3R and E5R, respectively. The “b” or “bc” superscripts in primer names indicate that the primers were designed from DNA sequences of the H2b (C57BL/6) or H2bc (129/SvJ) haplotypes, but this designation does not preclude cross-reactivity on the same or very similar genes in other haplotypes.
Primer Name . | Sequence (5′ to 3′) . |
---|---|
Cyclophilin F | CCA TCG TGT CAT CAA GGA CTT CAT |
Cyclophilin R | TTG CCA TCC AGC CAG GAG GTC T |
GAPDH F | TGA AGG TCG GTG TGA ACG GAT TTG |
GAPDH R | GGC CTT CTC CAT GGT GGT GAA GAC |
β2m F | ATG GCT CGC TCG GTG ACC CTG |
β2m R | ATT GCT CAG CTA TCT AGG ATA |
Kb ex2 F | GAG CCC CGG TAC ATG GAA |
Kb ex3 R | CAG GTA GGC CCT GAG TCT |
Db ex2 F | GAG CCC CGG TAC ATC TCT |
Db ex3 R | CAG GTA GGC CTT GTA ATG |
Xho I/Q1b ex1 F | TGG CAG CTC GAG TGA CCC TGA CCA AAA CCG GA |
EcoR I/Q1b ex3 R | ACC CCA GAA TTC AGC CAG ACA ACT TCT GGA AG |
Q1b/Q2b 5′UTR/ex1 F | GCC TCA GAT GCC CTG TAT TCC |
Q1b/Q2b 3′UTR R | CTC AGT CTA CTC CAG GCA GCT GTC |
Q2b 5′UTR/ex1 F | TCA GAT GCC CTG TAT CCC AGA TGG |
Xho I/Q2b ex2 F | CCT GGC CTC GAG AGG AGC CCC GGT TCA TTA TC |
EcoR I/Q2b ex3 R | GTA ATC GAA TTC ATC GTA GGC AGA CTG CTC A |
Q2b 3′UTR R | CAC CAG AGT GTC ACC TTT ACA ATT C |
T3b 5′UTR/ex1 F | CTT CAG ATT TCC CTA ACA TGA GG |
T3b ex2 F | GTA CAT AGC TGT GGG CTA CC |
T3b ex3 R | CCA TCA TAG CCA TGC TGC TC |
T3b 3′UTR R | GAA GAA GTA ACA AGA CAT TGT CAG G |
T23b ex2 F | CAG AGT AAA CCT GAG GAC CC |
T23b ex3 R | AGG CCT CCT GAC AAT ACC CG |
H2-Blbc 5′UTR/ex1 F1 | CAG ATG CCC TGT ATT CCA AAT GG |
H2-Blbc 5′UTR/ex1 F2 | TAT TCC AAA TGG GGC AAT GGC GCA |
H2-Blbc int3 F | GCA GTC GGG TGC TCT TAC C |
H2-Blbc ex3 R1 | CGC TAC CAG ATC CGC CGC CA |
H2-Blbc ex3 R2 | CGC ACT CGC CCT CTA GGT AGA A |
H2-Blbc ex5 R | CTC TTC AAC ACA AAA GCC AC |
H2-Blbc 3′UTR R | ACT CCA CTA ATC AAC CCT CAG |
Tw5bc 5′UTR F | TCA GAC ATC CAG GAT CCC AG |
Tw5bc 5′UTR/ex1 F | ACA TCC AGG ATC CCA GAT GG |
Tw5bc ex3 F | GTT TGC TTA CGA AGG CCA AG |
Tw5bc ex3 R | CGT GAT CAG AGC TGC CAT G |
Tw5bc 3′UTR R | CTT AAC TTC TGA GCC ATC TCT CC |
Primer Name . | Sequence (5′ to 3′) . |
---|---|
Cyclophilin F | CCA TCG TGT CAT CAA GGA CTT CAT |
Cyclophilin R | TTG CCA TCC AGC CAG GAG GTC T |
GAPDH F | TGA AGG TCG GTG TGA ACG GAT TTG |
GAPDH R | GGC CTT CTC CAT GGT GGT GAA GAC |
β2m F | ATG GCT CGC TCG GTG ACC CTG |
β2m R | ATT GCT CAG CTA TCT AGG ATA |
Kb ex2 F | GAG CCC CGG TAC ATG GAA |
Kb ex3 R | CAG GTA GGC CCT GAG TCT |
Db ex2 F | GAG CCC CGG TAC ATC TCT |
Db ex3 R | CAG GTA GGC CTT GTA ATG |
Xho I/Q1b ex1 F | TGG CAG CTC GAG TGA CCC TGA CCA AAA CCG GA |
EcoR I/Q1b ex3 R | ACC CCA GAA TTC AGC CAG ACA ACT TCT GGA AG |
Q1b/Q2b 5′UTR/ex1 F | GCC TCA GAT GCC CTG TAT TCC |
Q1b/Q2b 3′UTR R | CTC AGT CTA CTC CAG GCA GCT GTC |
Q2b 5′UTR/ex1 F | TCA GAT GCC CTG TAT CCC AGA TGG |
Xho I/Q2b ex2 F | CCT GGC CTC GAG AGG AGC CCC GGT TCA TTA TC |
EcoR I/Q2b ex3 R | GTA ATC GAA TTC ATC GTA GGC AGA CTG CTC A |
Q2b 3′UTR R | CAC CAG AGT GTC ACC TTT ACA ATT C |
T3b 5′UTR/ex1 F | CTT CAG ATT TCC CTA ACA TGA GG |
T3b ex2 F | GTA CAT AGC TGT GGG CTA CC |
T3b ex3 R | CCA TCA TAG CCA TGC TGC TC |
T3b 3′UTR R | GAA GAA GTA ACA AGA CAT TGT CAG G |
T23b ex2 F | CAG AGT AAA CCT GAG GAC CC |
T23b ex3 R | AGG CCT CCT GAC AAT ACC CG |
H2-Blbc 5′UTR/ex1 F1 | CAG ATG CCC TGT ATT CCA AAT GG |
H2-Blbc 5′UTR/ex1 F2 | TAT TCC AAA TGG GGC AAT GGC GCA |
H2-Blbc int3 F | GCA GTC GGG TGC TCT TAC C |
H2-Blbc ex3 R1 | CGC TAC CAG ATC CGC CGC CA |
H2-Blbc ex3 R2 | CGC ACT CGC CCT CTA GGT AGA A |
H2-Blbc ex5 R | CTC TTC AAC ACA AAA GCC AC |
H2-Blbc 3′UTR R | ACT CCA CTA ATC AAC CCT CAG |
Tw5bc 5′UTR F | TCA GAC ATC CAG GAT CCC AG |
Tw5bc 5′UTR/ex1 F | ACA TCC AGG ATC CCA GAT GG |
Tw5bc ex3 F | GTT TGC TTA CGA AGG CCA AG |
Tw5bc ex3 R | CGT GAT CAG AGC TGC CAT G |
Tw5bc 3′UTR R | CTT AAC TTC TGA GCC ATC TCT CC |
Reverse transcription and PCR
Two micrograms of total mRNA was reverse transcribed to cDNA using an Omniscript RT kit (Qiagen) and oligo(dT) primers. Two microliters of this 20-μl cDNA reaction was used in each RT-PCR requiring primers in the 3′ untranslated region (UTR) and 5′ UTR of the transcript. One microliter of cDNA was used in all other RT-PCRs. Where genomic DNA was PCR amplified, 90 ng of DNA was used as the template. All PCR amplification was performed using either HotStarTaq DNA Polymerase (Qiagen) or Platinum Taq DNA Polymerase High Fidelity (Invitrogen) according to the manufacturer’s protocols. Primer pairs for all MHC genes used in RT-PCR spanned at least one intron, and PCRs controlling for genomic DNA contamination of RNA samples were performed using a 10-fold excess of total RNA as a template. All PCRs were performed at an annealing temperature of 55°C with a 90-s extension time. The number of cycles was adjusted individually for each experiment and is described in the accompanying figure legend.
Molecular cloning
PCR products were separated on an agarose gel, and individual bands were extracted using a QIAquick Gel Extraction Kit (Qiagen). This purified DNA was then integrated into the pCRII vector using the TA Cloning Kit (Invitrogen) according to the manufacturer’s protocols. Plasmid DNA was isolated using a QIAfilter Plasmid Midi Kit (Qiagen).
Sequence analysis
DNA sequences were generated using BigDye Terminator v3.0 Cycle Sequencing and 3100 and 3730 capillary analyzers (Applied Biosystems). Where PCR-amplified cDNA was sequenced, the primers used in the amplification were also used for sequencing. Where cloned PCR products were sequenced, M13 (−20) and M13 reverse primers were used. Sequence alignments were performed using the ClustalV function (19) in Lasergene v5 MegAlign software (DNASTAR). Amino acid predictions were performed with Lasergene v5 EditSeq software. Similarity searches were performed by blastn analysis (National Center for Biomedical Information). DNA sequences were submitted to GenBank (AY989821–AY989882).
Results
Intestinal tissues of C57BL/6J mice transcribe multiple class Ib MHC genes
One of the difficulties in studying the multigene family of class I MHC Ags is their close sequence homology and the inherent difficulty in distinguishing individual members by standard Northern blot hybridizations or Ab-based approaches. To overcome this limitation we relied on a sensitive RT-PCR assay to survey body-wide expression patterns of selected class I genes. A panel of highly specific primers, diagnostic for 18 sequenced or partially sequenced class I MHC genes from H2b and H2bc mice (Table I), was used to examine C57BL/6J tissues (Fig. 1 and data not shown). Transcription of class Ia (Kb and Db), β2-microglobulin (β2m), as well as the majority of the tested class Ib was detected in all sampled organs: the small and large intestine, stomach, thymus, spleen, liver, kidney, brain, heart, lung, testes, and ovaries/uterus. The ubiquitously transcribed class Ib included T23, encoding Qa-1 Ag (20, 21); Q6–Q9, encoding Qa-2 Ags (22, 23); Q4 (24); T22 (25); and M3/M4 (26). The transcripts of Q10 (27) were detected only in liver, whereas transcripts of M9 (28) were absent from all tested tissues (data not shown). Specificity of RT-PCR signals was confirmed for each gene by sizing and sequencing of amplified bands from small intestinal samples.
Class Ib MHC genes Q1 (29, 30), Q2 (29, 31), T3 (32, 33), and unexpectedly H2-Bl were found to be transcribed predominantly in the gastrointestinal tract (Fig. 1, top panel). Transcription of Q1, Q2, and T3-related genes was previously observed in gastrointestinal tissues (31, 32). In contrast, H2-Bl was thought to be silent in adult tissues and to express only in embryonic blastocyst and placenta, in a manner reminiscent of human HLA-G. The original study reporting absence of H2-Bl transcripts in adult tissues did not examine samples from the gastrointestinal tract, thus explaining the seeming discrepancy between our conclusions and those reached by Sipes et al. (11).
Single residue polymorphisms of H2-Bl alleles in inbred mouse strains
Sipes et al. (11) generated the complete sequence of H2-Bl gene from cosmids of 129/SvJ DNA and identified an open reading frame spanning six exons. The exon-intron organization of H2-Bl was found to be similar to other class I MHC genes, with individual exons encoding distinct protein domains. The three extracellular domains of H2-Bl protein, α1, α2, and α3, were most closely related to class Ia sequences, the transmembrane domain (TM) was class Ib-like, and the truncated cytoplasmic tail (CYT) domain was unique. Accordingly, the 129/SvJ H2-Bl molecule was predicted to have a structure similar to class Ia proteins. Partial sequencing of H2-Bl from C57BL/6J suggested also that this allele encodes a full-size molecule identical with the 129/SvJ protein. Although the gene was found to have null alleles (BALB/c and DBA/2), it appeared to be highly conserved in a subset of the analyzed inbred, H2-Bl-positive strains.
Because the pattern of substitutions in MHC exons is often indicative of functional adaptations (34, 35), we sought to identify the polymorphisms occurring within the entire coding region of H2-Bl and to extend the scope of the H2-Bl sequence analysis to additional strains/genetic backgrounds.
Initially we examined contiguous coding regions of H2-Bl alleles partially sequenced by Sipes et al. (11): from C57BL/6J, FVB/NJ, and, as a reference, 129/SvJ inbred strains. Small intestinal H2-Bl transcripts containing six exons were amplified with H2-Bl-specific primers derived from the 5′ and 3′ UTR of the gene (Table I). The RT-PCR conditions were optimized for each strain to correct for differences in the transcriptional levels of gut-synthesized, canonical H2-Bl mRNA. Full-size, ∼1200-bp cDNA products were subcloned and sequenced. To correct for errors in PCR amplification, multiple clones were sequenced from each reaction.
The DNA sequence and the conceptual translation of the region corresponding to the mature H2-Bl protein (exon 2 to exon 6) confirmed the original nucleotide and amino acid predictions for the 129/SvJ allele made by Sipes et al. (11) (Fig. 2; AY989383).
Canonically spliced H2-Bl from FVB/NJ mice differed from its 129/SvJ counterpart in 22 nucleotide positions (AY989854–55). Exon 2 contained four synonymous substitutions (GAT→GAC at Asp39, CCG→CCC at Pro50, GAA→GAG at Glu53, and GGC→GGT at Gly90) and two nonsynonymous substitutions (CGG→GGG, Arg14→Gly14, GCG→TCG, Ala49→Ser49). Exon 3 contained three synonymous polymorphisms (GAT→GAC at Asp137, GAG→GAA at Glu163, and CTC→CTT at Leu168) and one nonsynonymous substitution (ACC→GCC, Thr125→Ala125). In exon 4, five synonymous (GCA→GCC at Ala187, ACT→ACC at Thr225, AAG→AAA at Lys253, TAC→TAT at Tyr257, and CCC→CCA at Pro269) and three nonsynonymous polymorphisms (CAT→CGT, His191→Arg191, CAT→CGT, His260→Arg260, and TAC→CAC, Tyr262→His262) were detected. Finally, exon 5 contained one synonymous (TCC→TCA at Ser279) and two nonsynonymous substitutions (GTG→GCG, Val306→Ala306, and AGC→AGG, Ser310→Arg310) and exon 6 contained one nonsynonymous polymorphism (AAG→AGG, Lys323→Arg323).
Comparison of the intact canonical alleles from 129/SvJ and FVB/NJ strains revealed strong conservation of class I structure and peptide-binding residues in particular. The synonymous H2-Bl substitutions predominated over the nonsynonymous ones in the three extracellular domains (12 vs 6), whereas the reverse was true in the buried, TM and CYT, regions (1 vs 3). More significantly, all six nonsynonymous polymorphisms fell outside of the peptide-binding region defined by Bjorkman et al. (36) (Figs. 2 and 3). Five of the six nonsynonymous substitutions occurred in residues that are variable in murine and human class I MHC (37), and the one exception (in the conserved Ala125 position) did not interfere with any known structural properties of class I proteins. Furthermore, none of the nonsynonymous polymorphisms changed the β2m-interacting residues in the HLA-A2 model (36) or affected the CD8 binding site (38). Interestingly, although the majority of the synonymous substitutions occurred at random locations, two of them may have functional significance. One is in the Gly90 codon, at the junction of exon 2 and intron 2, which therefore may be responsible for splicing choices of H2-Bl mRNA in the parental strains (see below). The other is located at a codon corresponding to residue 163, predicted to face into the groove and toward the TCR (36). This position is highly variable in class Ia molecules, and its variation is known to alter recognition by CTL. Its conservation in H2-Bl may be indicative of functional selection for binding of invariant ligands, such as peptides and/or T/NK cell receptors.
In contrast with the 129/SvJ and FVB/NJ alleles, we found that C57BL/6J H2-Bl encodes a truncated protein (AY989821). The H2-Bl cDNAs from C57BL/6J differed from their 129/SvJ counterparts by five nucleotides in the exon 2 to exon 6 coding region (Fig. 2). These included four nucleotides in exon 3 (one synonymous substitution, CTA→CTG at Leu160, and three nonsynonymous substitutions, GGC→GCC [Gly162→Ala162], GAG→CAG [Glu173→Gln173], and AAT→AAG [Asn176→Lys176]) and one nonsynonymous substitution in exon 4 (CAG→TAG). The latter polymorphism converts Glu235 into a stop codon and was independently confirmed by sequencing of H2-Bl genomic DNA from a C57BL/6NCr mouse. Thus, if translated, H2-Bl from C57BL/6 mice would generate a truncated, secreted protein, lacking the cysteine bridge (Cys195:Cys246) in the β2m-binding domain. Because such a product is unlikely to perform classical Ag-presenting functions, the nonsynonymous polymorphism in the position corresponding to the putative peptide contact residue (Gly162→Ala162) appears to be of little predictive value.
In summary, our data indicate that two of the three sequenced cDNA alleles of H2-Bl (129/SvJ and FVB/NJ) encode canonical, membrane-bound class I proteins with highly conserved peptide binding sites. The third allele (C57BL/6J) is truncated and predicted to be secreted and lacking in class I protein features.
H2-Bl sequences in wild-derived mice
To gain further knowledge of the diversity of H2-Bl, we analyzed 5′ segments of this gene from eight H2-Bl-positive, wild-derived mouse strains of various Mus species and subspecies further characterized in a following section. The nucleotide compositions were established by direct sequencing of uncloned PCR products amplified from genomic DNA (Fig. 4; AY989849, AY989855, AY989858–66, and AY989868). All but one of the strains (M. musculus spp.; MPR/Pakistan) appeared homozygous at the H2-Bl locus (or had a null allele on the other chromosome). Interestingly, their polymorphism patterns showed that the wild alleles either are identical with FVB/NJ H2-Bl (BIK/Israel and SMZ/Morocco) or to 129/SvJ (BZO/Algeria) or are exact, or nearly exact, chimeric versions of these two alleles. DIK/Israel and DEB/Spain differed from 129/SvJ by one silent FVB/NJ-like substitution at the Asp137 codon. MGA/Georgia differed from the prototype 129/SvJ H2-Bl by one silent substitution at Leu141. The most divergent alleles (M. musculus spp.; BID/Iran and MPR/Pakistan) carried a more extensive “microchimerism” of 129/SvJ and FVB/NJ sequences. In addition, each of these alleles contained one unique (not found in the two major alleles) nonsynonymous substitution in the predicted mature protein sequences: Pro50→Gln50 (MPR/Pakistan) and Met103→Val103 (BID/Iran). Both these polymorphisms are located outside of the highly conserved class I MHC regions and the predicted peptide-binding residues of the groove (Fig. 3).
Strain-dependent alternative splicing of H2-Bl genes
During our experiments seeking to identify the canonical H2-Bl gut transcripts, we detected strain-dependent variation in length and in relative abundance of H2-Bl-specific cDNAs (Fig. 5; H2-Bl panel). The H2-Bl cDNA amplified from 129/SvJ with primers specific for the 5′ and 3′ UTR of H2-Bl migrated as two major bands in agarose gels (∼1200 and ∼900 bp). cDNA from FVB/NJ migrated as a single band (∼1200 bp), as did the C3H/HeJ cDNA (∼900 bp). Interestingly, the H2-Bl-positive C57BL/6J mice, like the control BALB/cJ H2-Bl-negative mice, typed negative in this semiquantitative assay, suggesting that the steady state gut H2-Bl mRNA levels in C57BL/6J strain are too low to be detected under the standard RT-PCR conditions used in this set of experiments (Fig. 5). Similar results were observed by Northern blot analysis (data not shown), thus confirming overall low transcript abundance of the C57BL/6J H2-Bl allele.
After changing experimental conditions to enhance detection of C57BL/6J mRNAs (Fig. 6), two major H2-Bl cDNA species were visualized by gel electrophoresis. The relative intensities of the upper (∼1200 bp) and the lower (∼900 bp) bands were reproducibly altered in C57BL/6J compared with 129/SvJ lanes (Fig. 6).
The H2-Bl RT-PCR products amplified from small intestinal RNA of each of the studied strains were eluted from an agarose gel, cloned, and sequenced (Figs. 5 and 6; AY989821–71, AY989854, and AY989856–57). Multiple clones were analyzed to correct for possible PCR errors. The ∼1200-bp cDNA bands from 129/SvJ and C57BL/6J mice contained two common species, H2-Bl.1 and H2-Bl.1a, present in an ∼1:1 ratio. The first corresponds to the 1186-bp canonical isoform encoding L, α1, α2, α3, TM, and CYT regions, and the second one is a differentially spliced, 1153-bp product, lacking the first 33 nucleotides of exon 2. The ∼900-bp bands in both strains also contained two common products: a 910-bp isoform, H2-Bl.2, lacking exon 3, and its 877-bp truncation variant, H2-Bl.2a, lacking 33 nucleotides in the 5′ end of exon 2. In addition, C57BL/6J gave rise to at least two other isoforms. The first of them, H2-Bl/B6.1, is generated by splicing at a cryptic site in exon 3, which leads to an alternative reading frame ending with a premature stop codon in the α2 domain. The second, H2-Bl/B6.2, skips exon 4. Because removal of this coding region deletes the stop codon, the H2-Bl/B6.2 transcript from the C57BL/6J allele may give rise to a protein structurally reminiscent of NKG2D ligands such as α3-domain-lacking, class I-related Rae-I polypeptides (39). H2-Bl.1, H2-Bl.2, and H2-Bl/B6.2 isoforms correspond in exon/exon organization to HLA-G isoforms designated as HLA-G1, HLA-G2, and HLA-G4, respectively (12, 40).
In many genes a premature stop codon leads to stop-codon-mediated mRNA degradation (41) and alternative splicing of the transcripts (42). We suggest that the nonsense mutation CAG→TAG in exon 4 of C57BL/6J H2-Bl is responsible for drastically reduced H2-Bl transcript levels and the synthesis of unique H2-Bl/B6.1 and H2-Bl/B6.2 isoforms detected in this strain. This polymorphism may also contribute to the generation of additional short H2-Bl splicing variants observed preferentially in this strain (Fig. 6).
The H2-Bl cDNAs from FVB/NJ (∼1200-bp band) contained the canonical, H2-Bl.1, and the truncated canonical, H2-Bl.1a, isoforms in a 1:1 ratio, but they lacked detectable H2-Bl.2/2a isoforms (data not shown). Because one of the polymorphisms between FVB/NJ and 129/SvJ alleles occurs at the junction of exon 2 and intron 2 (Fig. 2), it is likely that this silent substitution, CAG→TAG, is responsible for differential splicing of exon 3.
The exon/exon organization of the C3H/HeJ H2-Bl cDNAs corresponded with H2-Bl.2/2a isoforms (data not shown). The H2-Bl.1/1a isoforms were not detected in this strain. Because the sequences of the C3H/HeJ H2-Bl.2/2a were identical with the corresponding isoforms from 129/SvJ, the polymorphism regulating differential splicing of exon 3 in this allele may be located outside of the sequenced regions, possibly in intron 2 or 3 or in the spliced-out exon 3.
Expression profiles of H2-Bl transcripts are not influenced by genes located outside of the Mhc
To test whether non-MHC genes contribute to the observed low expression of H2-Bl mRNA in C57BL/6J mice, we took advantage of Kb−/− Db−/− mice (16) derived from the 129/SvJ strain and backcrossed for six generations to C57BL/6J. Presence of 129/SvJ loci in the Mhc was confirmed by showing that Kb−/− Db−/− gut transcripts contain sequences unique to Tw5, a gene related to H2-Bl and shown to exist in 129/SvJ but not in C57BL/6J MHC (13). The RT-PCR analysis of Tw5 transcripts in several tissues revealed that this gene, proposed to be a paralog of H2-Bl on the basis of sequence homology (13), has a restricted expression pattern, identical with the H2-Bl allele of Kb−/− Db−/− strain (Fig. 7). H2-Bl transcript levels in gut tissues from C57BL/6J mice were reproducibly reduced compared with Kb−/− Db−/− control, indicating that the background genes in C57BL/6J do not influence this phenotype. Other gut-restricted genes, Q1, Q2, and T3, identified in the initial survey of C57BL/6J, had similar expression profiles in the two strains. Minor, strain-dependent differences in the mRNA levels were detected in a few of the secondary organs (Q2 and T3 in thymus, Q1 in liver) suggesting that Q1, Q2, and T3 alleles, or their paralogs from C57BL/6J and Kb−/− Db−/−, may have slightly different tissue distributions in the two strains.
We also considered the possibility that alternative splicing of H2-Bl is imposed by the gut milieu and controlled by non-MHC genes expressed in this tissue. To address this possibility we tested transcriptional profiles of Kb and Db genes (data not shown) as well as gut-restricted class I genes Q1, Q2, T3, and Tw5 (Fig. 5). In each case, only a single species of class I cDNA, with a size predicted for the relevant canonical isoform, was amplified.
Taken together, the data support the conclusion that the alternative splicing pattern and the polymorphism in expression levels of H2-Bl transcripts are allele-specific and are not modulated by strain-dependent variation in non-MHC genes.
Comparison of gut-restricted class Ib proteins
We demonstrated here that two T region genes, H2-Bl and Tw5, are preferentially transcribed in the gut tissues of adult mice. Exon-intron organization of Tw5 was originally deduced from the genomic sequence of 129/SvJ mice (13). It predicted six translated exons for this gene, of which one, exon 6, was postulated to encode a 19-aa-long CYT. Our analysis of Tw5 cDNAs revealed that Tw5 transcripts contain eight exons, of which the last three encode a 48-aa-long CYT (AY989869). Interestingly, the comparison of amino acid sequences of Tw5 with other gut-restricted class Ib studied here showed that, in the C terminus, Tw5 is more related to Q1, Q2, and T3 than to its proposed paralog, H2-Bl (Fig. 2). Tw5, unlike H2-Bl, encodes a tyrosine residue (at position 230) implicated in the endocytic signaling in class Ia Ags (43). Tyrosine residues with potentially similar functions are also found in the CYT of Q2 and T3 (Fig. 2). Divergence in the predicted amino acid composition is also seen within exon 1 of Tw5 and H2-Bl (11, 13). The putative leader peptide of Tw5, unlike the leader peptide of H2-Bl, bears no substantial homology to the prototype class Ia leader, Qdm (44), and is not predicted to bind to Qa-1 or to interact with CD94/NKG2 receptors. The remainder of the Tw5 protein, particularly the α1, α2, and α3 domains, is closely related to H2-Bl (85.8% identity at the amino acid level). Unexpectedly, T region-encoded Tw5 is more closely related to Q region Q1 and Q2 in α1, α2, and α3 (83.2% and 80.7% identity, respectively) than to T region-encoded T3 (70.8%). Both H2-Bl and Tw5 are only distantly related to their proposed human homologue, HLA-G (68.2% and 72.3% identity, respectively), and share only one variable residue (Met5) that distinguishes these three proteins from all other gut-restricted class Ib studied here.
Finally, because H2-Bl is believed to represent a murine homologue of HLA-G, we investigated whether these two proteins share a common pattern of polymorphism. Fig. 3 shows that the nonsynonymous polymorphic substitutions in the predicted H2-Bl and HLA-G α1/α2 domains affect distinct structural regions, but in both cases they are located outside of the putative peptide-contact residues.
Distribution of gut-specific class Ib genes in wild-derived mouse strains
Because H2-Bl and Tw5 have null alleles among inbred strains (11, 13), we wondered whether these genes and Q1-, Q2-, and T3-like, gut-specific loci have been maintained in the genomes of wild mice under pathogen-imposed and breeding-dependent selections.
DNA from 32 specimens of mice from distinct locations in Europe, Africa, Asia, and the Pacific were typed by PCR for Q1-, Q2-, H2-Bl-, Tw5-, and T3-like genes with class Ib-specific primers (Table II). The identity of each type of gut-specific gene was confirmed by sequencing of PCR products amplified from SEB (Spain) M. spretus mouse DNA (AY989866, AY898970, AY989872, AY989879, and AY989881) and scoring for the presence of PCR bands with the predicted size in all other cases. We assume that negative typing is indicative of homozygous null alleles, though some genes that diverged in their sequences may have been missed due to the polymorphism in the sequences corresponding with primer locations. Only one of the gut-restricted genes, T3-like, was present in all studied mice, whereas other genes were detected with variable frequency. Q1-like genes were found in 25 mice, Q2-like in 17, H2-Bl-like in 24, and Tw5-like in 15 of 32 mice tested. In addition, one mouse scored as “ambiguously positive” for Q2 (see footnote to Table II), and three similar determinations were made for H2-Bl.
Mouse Type . | Strain Name . | Country of Origin . | MHC Class Ib Geneb . | . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | . | . | Q1 . | Q2 . | H2-Bl . | Tw5 . | T3 (TL) . | ||||
Inbred Strain | C57BL/6NCr | – | + | + | + | − | + | ||||
129/SvJ | – | + | + | + | + | + | |||||
FVB/NJ | – | + | + | + | + | + | |||||
Mus m. domesticus | BIK | Israel | + | + | + | + | + | ||||
BZO | Algeria | + | +/− | + | − | + | |||||
DIK | Israel | + | + | + | + | + | |||||
DJO | Italy | + | + | − | − | + | |||||
DMZ | Morocco | + | + | + | − | + | |||||
DOT | Tahiti | + | + | + | − | + | |||||
ZZMO | Tunisia | + | + | + | − | + | |||||
DDO | Denmark | + | − | + | − | + | |||||
DEB | Spain | + | + | + | + | + | |||||
DGA | Georgia | − | − | + | + | + | |||||
Mus m. musculus | MAM | Armenia | − | − | + | + | + | ||||
MGA | Georgia | + | + | + | − | + | |||||
MBK | Bulgaria | + | − | + | − | + | |||||
MBS | Bulgaria | + | − | + | + | + | |||||
MDH | Denmark | + | − | +/− | − | + | |||||
MPB | Poland | − | − | + | − | + | |||||
Mus m. castaneus | CIM | India | + | + | − | − | + | ||||
CTA | Taiwan | − | + | + | + | + | |||||
CTP | Thailand | + | + | +/− | − | + | |||||
Mus m. molossinus | MOL | Japan | − | − | + | + | + | ||||
Mus m. spp. | TEH | Iran | + | − | + | + | + | ||||
BID | Iran | − | − | + | − | + | |||||
MAC | Iran | + | − | + | + | + | |||||
KAK | Iran | + | + | + | + | + | |||||
DHA | India | + | + | − | + | + | |||||
MPR | Pakistan | + | + | + | + | + | |||||
Mus spretus | SFM | France | − | − | − | − | + | ||||
SMZ | Morocco | + | + | + | − | + | |||||
STF | Tunisia | + | + | − | + | + | |||||
SEB | Spain | + | + | + | + | + | |||||
Mus macedonicus | XBS | Bulgaria | + | − | +/− | − | + | ||||
Mus spicilegus | ZRU | Ukraine | + | − | + | − | + |
Mouse Type . | Strain Name . | Country of Origin . | MHC Class Ib Geneb . | . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | . | . | Q1 . | Q2 . | H2-Bl . | Tw5 . | T3 (TL) . | ||||
Inbred Strain | C57BL/6NCr | – | + | + | + | − | + | ||||
129/SvJ | – | + | + | + | + | + | |||||
FVB/NJ | – | + | + | + | + | + | |||||
Mus m. domesticus | BIK | Israel | + | + | + | + | + | ||||
BZO | Algeria | + | +/− | + | − | + | |||||
DIK | Israel | + | + | + | + | + | |||||
DJO | Italy | + | + | − | − | + | |||||
DMZ | Morocco | + | + | + | − | + | |||||
DOT | Tahiti | + | + | + | − | + | |||||
ZZMO | Tunisia | + | + | + | − | + | |||||
DDO | Denmark | + | − | + | − | + | |||||
DEB | Spain | + | + | + | + | + | |||||
DGA | Georgia | − | − | + | + | + | |||||
Mus m. musculus | MAM | Armenia | − | − | + | + | + | ||||
MGA | Georgia | + | + | + | − | + | |||||
MBK | Bulgaria | + | − | + | − | + | |||||
MBS | Bulgaria | + | − | + | + | + | |||||
MDH | Denmark | + | − | +/− | − | + | |||||
MPB | Poland | − | − | + | − | + | |||||
Mus m. castaneus | CIM | India | + | + | − | − | + | ||||
CTA | Taiwan | − | + | + | + | + | |||||
CTP | Thailand | + | + | +/− | − | + | |||||
Mus m. molossinus | MOL | Japan | − | − | + | + | + | ||||
Mus m. spp. | TEH | Iran | + | − | + | + | + | ||||
BID | Iran | − | − | + | − | + | |||||
MAC | Iran | + | − | + | + | + | |||||
KAK | Iran | + | + | + | + | + | |||||
DHA | India | + | + | − | + | + | |||||
MPR | Pakistan | + | + | + | + | + | |||||
Mus spretus | SFM | France | − | − | − | − | + | ||||
SMZ | Morocco | + | + | + | − | + | |||||
STF | Tunisia | + | + | − | + | + | |||||
SEB | Spain | + | + | + | + | + | |||||
Mus macedonicus | XBS | Bulgaria | + | − | +/− | − | + | ||||
Mus spicilegus | ZRU | Ukraine | + | − | + | − | + |
Genomic DNA from the indicated mice was tested by PCR with Platinum Taq over 30 cycles using the following primers: Xho I/Q1b ex1 F and EcoR I/Q1b ex3 R; Xho I/Q2b ex2 F and EcoR I/Q2b ex3 R; H2-BIbc 5′UTR/ex1 F2 and H2-Blbc ex3 R1; Tw5bc ex3 F and Tw5bc ex3 R; T3b ex2 F and T3b ex3 R. Primers are specific to C57BL/6J alleles (Q1, Q2, and T3) or 129/SvJ alleles (H2-Bl and Tw5) but cross-react with other alleles. Because modern inbred strains of mice are predominantly descendants of the Mus musculus subspecies, sequence confirmation of PCR products from all five genes was performed in the distantly related Mus spretus mouse, SEB (AY989866, AY989870, AY989872, AY989879, and AY989881).
Reactions yielding a band at the expected size were scored as positive, “+.” Those with a very faint band at the appropriate size were scored as ambiguously positive, “+/−.” Those with no visible band or a band of the incorrect size were scored as negative, “−.” Mice positive for a particular gene are considered to have that gene or a highly homologous gene.
Only one mouse scored negative for four (Q1, Q2, H2-Bl, and Tw5) of the gut-specific genes, and two mice lacked three of the gut-specific genes (Q1, Q2, and Tw5 in both cases). H2-Bl was absent in three cases along with its paralog, Tw5, and in two cases its loss was accompanied by the presence of all other gut-specific class Ib.
Judging from this data, it appears that gut-specific class Ib are inherited and maintained independently of each other, as there is no clear correlation between presence/absence of these genes and the genetic classification/geographical origin of the tested mice. Thus, we conclude that the gut-expressed class Ib may be selectively maintained in the environments in which they contribute to their host’s fitness.
Inheritance of H2-Bl alleles selected for reproductive success
The H2-Bl cDNA clones described by Sipes et al. (11) were derived from a library made from ∼2000 blastocysts of outbred ICR mice. All three independently identified clones were identical in their regions of overlap and corresponded with the 129/SvJ allele. Because ICR mice originated from a heterogeneous pool of mice encoding different MHC and were bred for fecundity (45), we wondered whether the 129/SvJ allele has been selectively passed on because it contributed to the reproductive success of the ICR stock.
To address this issue, nine ICR mice from at least two different litters were examined here. All expressed H2-Bl in gut tissues as judged by RT-PCR (data not shown). Full-size (5′ UTR to 3′ UTR) RT-PCR products from five of these nine ICR mice were analyzed by cloning/DNA sequencing (AY989842–47, AY989850–53, and AY989867). Two of the mice appeared homozygous for the 129/SvJ allele of H2-Bl (100% identity) and gave rise to the predicted alternatively spliced isoforms. Two other mice were consistent with homozygosity for the FVB/NJ allele and did not express the truncated, exon 3-deleted products (H2-Bl.2/2a). An alternative interpretation of the data is that the putative homozygotes carried a null allele on one of the sister chromosomes. The fifth mouse gave rise to a mixture of 129/SvJ and FVB/NJ cDNA products, suggesting that it is a heterozygote for the H2-Bl locus.
Thus, we conclude that the selection for fecundity in ICR mice did not eliminate either of the two major alleles of H2-Bl from the breeding pool.
Discussion
Recent advances in the understanding of the complex immunological roles played by the members of the extended family of class I and class I-like proteins revived our interest in class I genes encoded by the Q and T regions of the mouse Mhc. Specifically, we were interested in reexamining those MHC class Ib genes that were previously sequenced, or partially sequenced, but whose functions remain unknown. To address whether any of them represent candidates for NK/IEL receptor ligands, we surveyed their body-wide expression patterns, emphasizing tissues enriched for ECs, particularly the intestines.
Unexpectedly, in addition to the class Ib genes identified earlier as gut-restricted, Q1 (29, 30), Q2 (29, 31), and T3 (32), we discovered that H2-Bl, a gene so far detected only in blastocyst and placenta (11) and proposed to play a role in inducing NK cell tolerance at the maternal-fetal interface (14), is selectively expressed in the intestinal tissues enriched for intestinal ECs. We generated a series of complete, contiguous sequences of H2-Bl cDNAs corresponding with mature, canonical class Ib H2-Bl proteins from several inbred and outbred mice. In addition, we sequenced 5′ DNA sequences of H2-Bl genes from wild-derived mouse strains originating from different locations in Europe, Asia, Africa, and the Pacific. Two predominant alleles were identified among the studied mice, typified by H2-Bl from 129/SvJ and FVB/NJ strains. They differed from each other at 22 nt positions, 9 of which resulted in amino acid replacements. This degree of divergence between the two “prototype” H2-Bl alleles is larger than in most stringently conserved class Ib (23, 27), but it is similar to that observed in T23 and Q2 genes (21, 29). Although we favor the interpretation that the H2-Bl sequences from 129/SvJ and FVB/NJ correspond with true alleles, we cannot formally exclude the possibility that they were amplified from distinct genomic loci.
Past studies demonstrated that extensive polymorphism in the classical class I MHC genes predominantly affects the residues of the groove and correlates with the ability to bind peptides with different sequence motifs (46, 47). It is also thought that the polymorphism patterns represent, at least to some degree, a record of past pathogen-driven selections and are predictive of functionally important locations in the studied proteins (34, 35). The pattern of amino acid substitutions in the major H2-Bl alleles and in closely related sequences from wild-derived mice appears to be nonrandomly distributed in relation to the functionally important protein regions. The conceptual translation of each H2-Bl allele, with the exception of the C57BL/6J allele, is expected to give rise to a canonical protein with three extracellular domains, TM and CYT regions. Remarkably, the residues that are highly conserved between mouse and human class I proteins, as well as those that are universally present in class I chains throughout vertebrate evolution (46), remain invariant among different H2-Bl alleles. Because the conservation of these residues is thought to be critical for structural integrity of class I MHC, the allelic variants of H2-Bl proteins are likely to fold into class Ia-like structures.
The repertoire of nucleotide substitutions in the variants of the canonical H2-Bl outside of the class I “framework” residues also appears to be nonrandom. Even though the polymorphisms are spread throughout the coding regions (six in exon 2, four in exon 3, eight in exon 4, three in exon 5, and one in exon 6 for the two major alleles), none affect residues in the putative peptide-contact sites. Furthermore, none of the residues that are highly polymorphic in class Ia is substituted in H2-Bl variants. In one case, where nucleotide substitution occurred at such a position, at Glu163 (a residue that is predicted to face into the groove and toward the TCR), the change is synonymous. In general, the synonymous substitutions in H2-Bl alleles predominate over the nonsynonymous in the three extracellular domains, whereas the reverse correlation is observed in the TM and CYT regions. This suggests that selective conservation of the H2-Bl N terminus indeed might have occurred during evolutionary history of this gene in Mus species. Because the side chains involved in peptide binding in typical class I (Tyr7, Tyr59, Tyr159, and Tyr171 for the N terminus and Thr143, Tyr84, Lys146, and Trp147 for the C terminus (48)) are strictly conserved in the H2-Bl sequences analyzed here, it is most likely that this class Ib molecule binds peptides. Alternatively, the observed pattern of substitutions may be indicative of selection for binding of conserved, nonpeptidic short ligands or a necessity to interact with invariant receptors, such as specific members of the TCR, NKG2D, Ly49, or CD94/NKG2 families.
The most striking feature of the strain-dependent variation of H2-Bl alleles is the diversity of their alternatively spliced transcripts. We identified here at least four different repertoires of transcriptional isoforms generated in the gut of 129/SvJ, FVB/NJ, C3H/HeJ, and C57BL/6J mice. The 129/SvJ allele gives rise to at least four major isoforms: the canonical transcript (H2-Bl.1), its truncated variant lacking codons for 11 aa at the N terminus (H2-Bl.1a), α2-domain-lacking isoform (H2-Bl.2), and its N-terminal truncated variant (H2-Bl.2a). The FVB/NJ mice express only H2-Bl.1 and H2-Bl.1a, whereas C3H/HeJ express only detectable levels of H2-Bl.2 and H2-Bl.2a. Most interestingly of all, a polymorphic substitution of Glu235 into a premature stop codon not only converts canonical H2-Bl into a potentially secreted, truncated protein, but it also alters qualitatively and quantitatively its profile of isoforms. The relative proportions of H2-Bl.1/1a to H2-Bl.2/2a are reversed compared with 129/SvJ, new isoforms are synthesized, including an α1 and α2 domain-encoding, Rae I-like transcript (H2-Bl/B6.2), and the overall level of H2-Bl transcripts from this allele is severely depressed. We propose that the reduction in the steady state levels of H2-Bl mRNA in C57BL/6J and the change in the profile of alternative splicing are due to the presence of the stop codon because the nonsense-mediated effects on RNA degradation and splicing patterns have been reported in other systems (41, 42). The differential splicing between 129/SvJ and FVB/NJ alleles may also be attributed to a specific polymorphism; in this case one possible candidate is a synonymous substitution at the Gly90 codon near the junction of exon 2/intron 2.
To the best of our knowledge, this is the first reported case of a class I MHC gene in which polymorphism affects alternative splicing of exons encoding extracellular domains. Its direct consequence is the expression of variant sets of canonical and noncanonical products that may be shaped by balancing evolution (49). Interestingly, the truncated, noncanonical isoform found thus far only in H2-Bl, H2-Bl.1a, is theoretically compatible with class I-like structure, despite the predicted displacement/rearrangement of β strands (50). The H2-Bl isoform lacking the α2 domain, H2-Bl.2, resembles similar truncated polypeptides encoded by human and mouse class I MHC and non-MHC genes, including HLA-G (12), MICA/B (51), and MR1 (52). H2-Bl/B6.2, lacking the α3 domain, is in turn structurally similar to the HLA-G4 isoform (12) and several members of the mouse family of NKG2D ligands, such as Rae 1 (9, 39).
One feature common to all H2-Bl isoforms studied here is the presence of exon 1, encoding the leader peptide. The N-terminal leaders of class I MHC are cotranslationally cleaved before the mature protein leaves the endoplasmic reticulum. Many of these leaders bind to Qa-1 (or its human homologue, HLA-E). When the complexes are presented at the cell surface they engage inhibitory or activating NKG2 receptors on NK cells or T cells and either potentiate or down-regulate effector functions (44). It recently has been shown that H2-Bl.1 and H2-Bl.2 isoforms can provide leader peptides for binding to Qa-1 molecules in transfected RMA-S cells (14). The leader/Qa-1 complexes expressed on the surface of tumor cells were able to attenuate NK activity in vitro and in vivo (14). By extension, we propose that the polymorphism affecting the profiles of H2-Bl transcripts impacts effector functions of gut-localized cells expressing CD94/NKG2. In addition to NK cells, these might include IELs and other populations of immune cells in the lamina propria or the epithelial layer of the intestines.
It is of interest to note that Tw5, shown here to be gut-specific and earlier proposed to be a paralog of H2-Bl on the basis of sequence homology (13), does not give rise to alternatively spliced transcripts. This feature, as well as the differences in the length and the composition of the cytoplasmic domains, may be indicative of their distinct roles in the mucosal immune system.
H2-Bl was originally postulated to be the murine homologue of human HLA-G (11). Finding that H2-Bl is highly expressed in adult gut tissues may not necessarily contradict this hypothesis, because the intestinal milieu, like the maternal/fetal interface, is dominated by immunological tolerance. H2-Bl encodes leader peptides that provide protection from NK cytolysis (14) as does HLA-G (53), and both class I encode similar alternatively spliced isoforms that may perform homologous functions. In contrast, the two class Ib molecules do not share overlapping patterns of nonsynonymous polymorphism that would indicate similar evolutionary pressures. They also direct synthesis of a nonoverlapping set of truncated transcripts. It is thus possible that detection of H2-Bl in the blastocyst is related to genome-wide epigenetic reprogramming that allows expression of a multitude of otherwise tissue-specific genes at this stage of embryonic development (54). Indeed, we detected a CpG island (55) in the 5′ region of the H2-Bl gene (data not shown) which, upon demethylation, could give rise to transcription of H2-Bl in blastocysts.
Little is currently known about the roles of other class Ib proteins expressed selectively in the gut tissues. The best characterized member of this family in the mouse, TL, encoded by T3 and related genes, is thought to modulate cytokine production by IELs upon binding to CD8αα receptors (3). The Q1 and Q2 proteins have not yet been studied. Whereas Q1 gene is highly conserved among different mouse strains (29), Q2, like H2-Bl described here, has divergent alleles. Q2 sequences from H2b and H2k differ at 97 nucleotides in exons 2 and 3, resulting in 46 amino acid replacements. Strikingly, this diversity, unlike the diversity of H2-Bl alleles but similar to polymorphisms in classical class I Mhc or MICA/B genes (56), appears to result from selection for diversification of the peptide-binding site.
Given the distinctive structural properties of the gut-specific class Ib examined here, we wondered whether they are all essential for survival in the wild. Our results demonstrated that only T3-like genes are maintained in all genomes of wild-derived mouse strains, whereas other gut-specific class Ib genes have null loci and are inherited independently of each other. This diversity of class Ib functions in the gut is probably magnified by contributions of MHC products with a broad tissue distribution. We have shown here that at least 11 ubiquitously transcribed class I MHC are coexpressed in intestinal EC-enriched tissues, along with the gut-specific class Ib genes. Two of the widely expressed class I-like proteins, MR1 and Qa-1, were shown to have local effects in the gut and to regulate expression of unique, gastrointestinal T cell subsets (4, 6).
Taken together, our data suggest that a large number of class I MHC and class I-like genes are active in the tolerance-dominated, pathogen-reactive milieu of the gut. The complexity of their canonical product functions may be potentiated by expression of multiple alternatively spliced isoforms.
Acknowledgments
We are very much indebted to Drs. A. Orth and F. Bonhomme for the gift of the wild-derived mouse DNAs and to Drs. F. Bonhomme, J. Deisenhofer, L. Hooper, and E. Wakeland for helpful comments and discussions. We also thank J. McLean and V. Johnson for assistance in RNA generation and sequencing and Dr. E. Chiang and S. Shanmuganad for critical reading of this manuscript.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from Ellison Medical Foundation (ID SS 020401), the National Institutes of Health (2 RO1 19624), the National Science Foundation Graduate Research Fellowship, and the National Institute of Allergy and Infectious Diseases (Training Grant AI005284-27).
Abbreviations used in this paper: IEL, intraepithelial lymphocyte; EC, epithelial cell; UTR, untranslated region; β2m, β2-microglobulin; TM, transmembrane domain; CYT, cytoplasmic tail.