![]() | ![]() |
Formats:
|
||||||||||||||||||||||||
Copyright © 2004 Rumfelt et al; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL. Diversity and repertoire of IgW and IgM VH families in the newborn nurse shark 1Department of Microbiology and Immunology, University of Miami School of Medicine, Miami, FL 33101 USA 2Department of Immunology, University of Toronto, Sunnybrook & Women's College Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5 Canada 3Department of Microbiology and Immunology, University of Maryland School of Medicine, 655 W. Baltimore St., Rm. BRB-13009, Baltimore, MD 21201 USA Corresponding author.Lynn L Rumfelt: Lynn.Rumfelt/at/sw.ca; Rebecca L Lohr: Rebecca_Lohr/at/yahoo.com; Helen Dooley: HDooley/at/som.umaryland.edu; Martin F Flajnik: MFlajnik/at/som.umaryland.edu Received January 27, 2004; Accepted May 6, 2004. This article has been cited by other articles in PMC.Abstract Background Adult cartilaginous fish express three immunoglobulin (Ig) isotypes, IgM, IgNAR and IgW. Newborn nurse sharks, Ginglymostoma cirratum, produce 19S (multimeric) IgM and monomeric/dimeric IgM1gj, a germline-joined, IgM-related VH, and very low amounts of 7S (monomeric) IgM and IgNAR proteins. Newborn IgNAR VH mRNAs are diverse in the complementarity-determining region 3 (CDR3) with non-templated nucleotide (N-region) addition, which suggests that, unlike in many other vertebrates, terminal deoxynucleotidyl transferase (TdT) expressed at birth is functional. IgW is present in the lungfish, a bony fish sharing a common ancestor with sharks 460 million years ago, implying that the IgW VH family is as old as the IgM VH family. This nurse shark study examined the IgM and IgW VH repertoire from birth through adult life, and analyzed the phylogenetic relationships of these gene families. Results IgM and IgW VH cDNA clones isolated from newborn nurse shark primary and secondary lymphoid tissues had highly diverse and unique CDR3 with N-region addition and VDJ gene rearrangement, implicating functional TdT and RAG gene activity. Despite the clear presence of N-region additions, newborn CDR3 were significantly shorter than those of adults. The IgM clones are all included in a conventional VH family that can be classified into five discrete groups, none of which is orthologous to IgM VH genes in other elasmobranchs. In addition, a novel divergent VH family was orthologous to a published monotypic VH horn shark family. IgW VH genes have diverged sufficiently to form three families. IgM and IgW VH serine codons using the potential somatic hypermutation hotspot sequence occur mainly in VH framework 1 (FR1) and CDR1. Phylogenetic analysis of cartilaginous fish and lungfish IgM and IgW demonstrated they form two major ancient gene groups; furthermore, these VH genes generally diversify (duplicate and diverge) within a species. Conclusion As in ratfish, sandbar and horn sharks, most nurse shark IgM VH genes are from one family with multiple, heterogeneous loci. Their IgW VH genes have diversified, forming at least three families. The neonatal shark Ig VH CDR3 repertoire, diversified via N-region addition, is shorter than the adult VDJ junction, suggesting one means of postnatal repertoire diversification is expression of longer CDR3 junctions. Background The major components of adaptive immunity, including Ig, T cell receptors, and MHC class I and class II, have been identified in the ancient jawed vertebrate taxon cartilaginous fish (Chondrichthyes), which includes sharks, skates, rays (elasmobranchs), and ratfish (holocephali) (Figure (Figure1)1
In all elasmobranchs examined, secretory IgW transcripts are expressed in two forms, a full-length, long form and a truncated, short form that is probably derived by alternative splicing [19,20]. Recent molecular characterization of Ig H chains also identified both IgW forms in the African lungfish, Protopterus aethiopicus, a lobe-finned bony fish, which phylogenetic studies suggest are closely related to land vertebrates (tetrapods) (refer to Figure Figure1)1 To further our understanding of Ig class expression and regulation during ontogeny, we herein describe IgM and IgW VH family expression in the newborn nurse shark, Ginglymostoma cirratum. Results Newborn IgM is expressed from multiple loci of one VH family Developmental regulation of IgM expression was examined by screening two cDNA libraries constructed from newborn nurse shark epigonal organ (primary lymphoid tissue) and spleen (secondary lymphoid tissue). Libraries were probed with a canonical IgM VH and partial CH1 probe under high and low stringency hybridization conditions [5]. Deduced amino acid (AA) sequences of the leader, VH, and partial CH1 for 25 positive clones were compared to the predicted AA sequence of the single previously published adult nurse shark IgM (Genbank accession #M92851 [26]) and aligned in Clustal W (Figure (Figure2)2
Excluding the positions of VDJ rearrangement (CDR3), VH members within a group are quite similar in FR and CDR1-2, suggesting that they are expressed from one locus or several highly related loci. Indeed, microheterogeneity of 5'UT sequences of the 24 clones identified expression of 15 highly related loci from the five groups (data not shown). We estimated from previous Southern blotting analysis that there were ~15–25 IgM loci in the nurse shark [5]. Thus, our library screening results would suggest that most IgM loci are expressed at birth, which is consistent with our previous immunohistochemistry study demonstrating that the great majority of splenic B cells in newborn nurse sharks are IgM+ [12]. If this estimate of gene number is confirmed in future studies of nurse shark VH genes, then this elasmobranch species has far fewer IgM genes as compared to the published estimates of 100 IgM loci for the horn shark and 50 loci for the skate [1,28]. The IgM clones were analyzed for the conserved residues necessary to maintain the Ig structural fold using the ImMunoGeneTics (IMGT) system [29-31]. Representatives of the most divergent groups I and VI were compared to the originally isolated nurse shark IgM M92851 VH (Figure (Figure3)3
The CDR3 sequences of each clone were unique in length and sequence, suggesting gene rearrangement and N-region addition via RAG1/2 and TdT action, respectively. This was further investigated by comparison of CDR3 nucleotide sequences to a representative genomic horn shark IgM locus (X13447, [33]) in Table 2. The DH gene sequence that exactly matched the horn shark sequence (and thus could be encoded by a DH segment) is underlined (at least 4 bases corresponding to the D segment), and putative N-region additions are indicated by italicized G and C nucleotides [37]. At least four of 22 newborn clones (18%) putatively utilize both DH1 and DH2 genes, as do most adult VH clones. The majority of newborn VH may rearrange only the DH1 gene, or more likely, based upon the RSS sequences of the rearranging segments in most shark clusters (V-23, 12-D1-22, 12-D2-12, 23-J) there may instead be extensive trimming of the DH2 segment during rearrangement so that it cannot be recognized in the cDNA clones. There is a GC bias in the non-D-encoding CDR3 sequences for ~75% of the clones implicating TdT activity, similar to previous findings for neonatal IgNAR VH [14]. In summary, newborn IgM is expressed from multiple heterogeneous loci of one VH family that have conserved the Ig fold structure, and these expressed VH genes are diverse in CDR3, indicative of TdT and RAG activity at birth.
cDNA clone 36E in group VI is an unrearranged VH containing intronic sequence. The deduced sequence of 36E has only 38–50% AA identity with the conventional VH, and thus forms a separate VH family (Table 1) [31]. This VH family contains the residues important for the Ig fold, with the exception of an Arg replacing the typical Gly at position 49 (Figure (Figure3).3 IgW VHs are diverse and consist of multiple families expressed at birth The ontogeny of IgW expression in neonate sharks was investigated by screening nurse shark pup spleen and epigonal organ cDNA libraries with an IgW VH and partial CH1 probe [5]. Clones were negatively selected for IgM VH+ cross-hybridizing clones and positively selected for strong, intermediate, and weak hybridization intensity signals. Their deduced AA sequences were aligned to the published nurse shark IgW VH (accession #U51450 [43]) in Clustal W and shown to form six groups (Figure (Figure4).4
In the first constant domain (CH1) of conventional Ig H chains a Cys in the A strand is typically present that forms a disulfide bond with light (L) chains; this position is Gly in group II clones and Arg in group I clone 8S (Figure (Figure4)4 IgW VH (#U51450 [43]) structure from IMGT [46] was compared to representative members of the divergent groups III and V (refer to Figure Figure3).3 The potential somatic hypermutation motif AGY is biased to FR1 and CDR1 The V region of secretory IgNAR in the adult nurse shark has a high frequency of mutation in adult sharks, yet this region is seldom mutated in young pups suggesting that the newborn immune system requires further maturation to enable an effective immune responsive environment [14,48]. Previously the Ser codon motif AGY (Y=C/T) had been identified as a heavily targeted site of IgNAR somatic hypermutation [48]. Therefore, we analyzed the newborn IgM and IgW VH FR and CDR for the AGY potential hypermutation motif and Ser codon TCN (N=C/G/T/A), which tends not to be somatically mutated (Figure (Figure5).5
Newborn IgW and IgM CDR3 are shorter than in adults Newborn mouse, human, and frog VH CDR3s are less diverse and shorter in length due to non-random VH gene usage and lack of TdT activity [16,17,49-51]. In newborn elasmobranchs TdT is expressed and functional early in shark ontogeny, as shown in previous studies of TdT expression and implied by the diversity of the newborn CDR3 repertoire in the IgM and IgW classes (Figures (Figures2,2
IgM and IgW classes were present in the ancestors of all jawed vertebrates 1The identification of IgW in the lungfish, close relatives of land vertebrates, substantially changed the phylogenetic distribution and evolution of this Ig class, since these lobe-finned bony fish shared a common ancestor with cartilaginous fish more than 460 MYA (refer to Figure Figure1)1
Discussion We show that the newborn nurse shark expresses five IgM VH genes from one conventional VH family, as shown previously for the adult horn shark, sandbar shark, and ratfish [33]. Additionally, newborn nurse sharks express at least two other VH families (36E and IgM1gj [5]). Newborn IgM VH CDR3 are diverse with N-nucleotide additions, which shows that TdT is active in embryonic life, as previously documented for newborn IgNAR CDR3 [5,14]. The majority of IgM loci are expressed at birth. Among the five VH groups these loci differ mainly in the CDR1 and CDR2, implying that diversity-enhancing selection pressures are active in germline CDR as previously reported for other vertebrates [53]. The IgM VH CDR3 length is significantly shorter in neonates, showing that a developmental program may control the CDR3 length. One explanation for this finding may be that there is expression of different VH clusters in adults that have longer DH and JH gene segments, as is seen in the mouse [56]; this possibility seems unlikely as germline diversity seems similar in neonates and adults (although the adult nurse shark repertoire must be studied in more detail to prove this point). Secondly, TdT expression in pro/pre B cells in adults may be higher than in young animals [57]. Finally, TdT activity may be modified (e.g. decreased N-region addition or increased exonuclease activity – our data suggest that DH2 segments may be extensively trimmed in neonatal junctions) by expression of splice variants or developmental regulation of DNA-dependent protein kinase, a known modulator of TdT activity [58,59]. Newborn IgW VH expression is from at least three rather divergent VH families [43]. CDR3 lengths of IgW are also significantly shorter than that of adults, as seen for IgM, implying that a developmental program selects for increased CDR3 length as the animals mature. Potential somatic hypermutation Ser AGY motifs are preferentially encoded in the FR1 and CDR1, whilst the untargeted TCN motif is present in FR3. No Ser residues at all are encoded in FR2. This tendency to target mutations predominantly in the CDR is expected, but is paradoxical in FR1 of IgM VH. There are generally fewer mutations in FR1 in Ig from all vertebrates so far examined. This suggests that despite the presence of hypermutation targeting motifs in this region, there is a higher order of control over the mutational mechanism that inhibits the targeting to FR1 AGY hotspots. The IgM and IgW VH phylogenetic analysis suggest that both VH families were present in the cartilaginous and bony fish ancestral pool. A more thorough understanding of the evolution of these VH families will require identification of IgW in the cartilaginous fish class holocephali and in bony fish groups such as the chondrostean, holosteans and teleosts. Furthermore, isolation of all of the germline genes is required to examine relative expression levels of the various genes, and the level of diversity gained by somatic hypermutation after antigenic stimulation of B cells. Conclusion Most IgM loci expressed at birth in nurse shark are from one heterogeneous highly-related VH family which differs in the CDR1-2. This repertoire is increased in its diversity via N-region addition in CDR3. In spite of the N-region addition the newborn VDJ junction is significantly shorter than the adult, most likely due to extensive trimming of the DH2 segment during rearrangement. In contrast, nurse shark IgW VH genes have diverged to form at least three VH families expressed at birth. Methods Animals Nurse shark pups were delivered by Caesarian section from a gestating female shark near term as described [60]. cDNA library construction and screening Tissues were dissected from nurse shark pups and total RNA was isolated as described [5]. The newborn pup spleen and epigonal cDNA libraries were constructed as described previously [5]. Libraries were plated and screened with canonical nurse shark IgM VH probe under both high and low stringency conditions as described [5,61]. Canonical nurse shark IgW VH and partial CH1 probe was amplified by PCR from plasmid DNA containing cloned cDNA insert using specific primers and labeled as described [5,43]. Clones were selected and isolated based on hybridization signal intensity of strong, intermediate and weak with more than 60 clones analyzed for each Ig class. Alignment and phylogenetic analysis IgW and IgM cDNA clones were translated into amino acid sequences using the EXPASY translate tool [62] and aligned in ClustalW v1.8 [63] for alignment analysis or ClustalX v 1.8 for phylogenetic tree analysis [27,55]. Phylogenetic tree analysis was performed using the VH regions from FR1-FR3 excluding CDR1-2. Amino acid sequences were aligned in ClustalX v1.8 using the multiple alignment parameter which does pairwise alignments in the Gonnet series protein weight matrix under default conditions of 10.00 gap opening, 0.20 gap extension, and 30% delay divergent sequences. A neighbor-joining (NJ) tree in PHYLIP output was drawn using a dendrogram as the guide and the reliability of branching order was determined by 1000 replications (bootstrap analysis) [52,55]. The NJ phylogram tree was drawn in Treeview v1.6.6 and rooted using mouse VL kappa as outgroup [54]. The NJ tree was labeled in Canvas v9.0 (ACD Deneba Software, Miami, FL, USA). Genbank and Swiss-Prot and TrEMBL accession numbers used for phylogenetic analysis are as follows: M. musculus kappa LC 29725591; P. aethiopicus IgW AF437727 clone 28; P. aethiopicus IgM AF437734 clone 76; P. aethiopicus IgM AF437724 clone 27; R. erinacea IgM S10387; R. erinacea IgM S12838; R. erinacea IgW S12839; R. erinacea IgW REU08009; C. plumbeus IgW 1117935; C. plumbeus IgW1255130; C. plumbeus IgW 1255132; H. colliei IgM AAC12920; H. colliei IgM 2653745; H. colliei IgM 2653755; H. colliei IgM 2653743; H. francisci IgM 64003; H. francisci IgM 64005; H. francisci IgM Z11776; H. francisci IgW C6-26m13f (clone W26) P83907; H. francisci IgM clone 14-1 AY612427; C. plumbeus IgM clones 12, 19, 26, 27, 35 [44]; G. cirratum IgW U51450; G. cirratum IgM1gj AF327520; G. cirratum IgM M92851; G. cirratum IgM 57S AY609270; G. cirratum IgM 30E AY609260; G. cirratum IgM 36E AY609263; G. cirratum IgM 72S AY609272; G. cirratum IgM 24S AY609256; G. cirratum IgW14S AY609229-AY609230; G. cirratum IgW 20E AY609231-AY609232; G. cirratum IgW 25E AY531553-AY531554; G. cirratum IgW 99S AY609242-AY609243; G. cirratum IgW 130S AY609246; R. productus IgM clone 23A AY612424-AY612425; R. productus IgM 3-1 AY612426. Adult and newborn CDR3 length comparisons Adult nurse shark IgM VH cDNA Genbank accession numbers AY608337-AY608404 (clones Mary M2-Mary M34, Jesus M3-M47, and Joseph M1-M27). Newborn nurse shark IgM cDNA clones1E, 2E, 2S, 6E, 14E, 15S, 17S, 21S, 22E, 24S, 26E, 27S, 29S, 30E, 35S, 36E, 46S, 47S, 49S, 50S, 54S, 55E, 57S, 72S, 73S (Genbank accession numbers AY609247-AY609274). Newborn nurse shark IgW cDNA clones 1E, 5E, 8S, 9E, 14S, 20E, 25E, 27S, 37E, 40S, 76S, 90S, 92S, 99S, 105S, 114S, 130S (Genbank accession numbers AY609225-AY609246, AY531553-AY531554, AY524297). Sandbar shark and horn shark IgM accession numbers listed in the phylogenetic tree analysis were used for determination of CDR3 length. Human (Hs) preterm neonate and adult IgM VH cDNA sequences were obtained from Zemlin et al [64] and Xenopus laevis (Xl) 5–48 day larvae and adult IgM VH cDNA sequences published in Schwager et al [51] and Du Pasquier et al [49]. Authors' contributions LR dissected shark tissue, isolated RNA, constructed and screened cDNA libraries, isolated cDNA clones, performed phylogenetic and statistical analyses, wrote the manuscript draft and made the figures. BL screened nurse shark PBL (Y) cDNA library with IgW VH probe and isolated, analyzed and aligned positive clones. Those results have been reported in a paper currently in submission. HD provided 64 IgM VH cDNA sequences from three immunized adult nurse sharks for the statistical study of adult IgM CDR3 lengths. MF delivered the shark pups by Caesarian section, and participated in the bleeding and dissection of the sharks; he also coordinated the study, participated in the analysis of results, and the manuscript draft. All authors read and approved the final manuscript. Acknowledgements This study was supported by NIH grant RR06603 (MFF). References
|
PubMed related articles
Your browsing activity is empty. Activity recording is turned off. |
|||||||||||||||||||||||
Annu Rev Immunol. 1999; 17():109-47.
[Annu Rev Immunol. 1999]Curr Top Microbiol Immunol. 2000; 248():249-70.
[Curr Top Microbiol Immunol. 2000]J Immunol. 2000 Jul 1; 165(1):306-12.
[J Immunol. 2000]Nature. 1986 Apr 10-16; 320(6062):546-9.
[Nature. 1986]Comp Biochem Physiol. 1969 Oct 15; 31(2):365-71.
[Comp Biochem Physiol. 1969]Nucleic Acids Res. 1990 Nov 11; 18(21):6369-76.
[Nucleic Acids Res. 1990]Immunogenetics. 1999 Jan; 49(1):56-67.
[Immunogenetics. 1999]Annu Rev Immunol. 1999; 17():109-47.
[Annu Rev Immunol. 1999]Proc Natl Acad Sci U S A. 2003 Mar 4; 100(5):2501-6.
[Proc Natl Acad Sci U S A. 2003]Nature. 2002 Aug 15; 418(6899):767-70.
[Nature. 2002]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Mol Immunol. 1992 Sep; 29(9):1157-8.
[Mol Immunol. 1992]Nucleic Acids Res. 1994 Nov 11; 22(22):4673-80.
[Nucleic Acids Res. 1994]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Scand J Immunol. 2002 Aug; 56(2):130-48.
[Scand J Immunol. 2002]Annu Rev Immunol. 1999; 17():109-47.
[Annu Rev Immunol. 1999]Nucleic Acids Res. 1990 Feb 25; 18(4):1015-20.
[Nucleic Acids Res. 1990]Immunol Today. 1997 Nov; 18(11):509.
[Immunol Today. 1997]EMBO J. 1988 Nov; 7(11):3413-22.
[EMBO J. 1988]Int Immunol. 1996 Jun; 8(6):847-54.
[Int Immunol. 1996]Immunol Today. 1994 Aug; 15(8):367-73.
[Immunol Today. 1994]EMBO J. 1988 Nov; 7(11):3413-22.
[EMBO J. 1988]J Exp Med. 1990 Nov 1; 172(5):1377-90.
[J Exp Med. 1990]Immunogenetics. 2002 Oct; 54(7):501-12.
[Immunogenetics. 2002]J Mol Biol. 1990 Oct 5; 215(3):403-10.
[J Mol Biol. 1990]J Exp Med. 1993 Sep 1; 178(3):815-24.
[J Exp Med. 1993]Proc Natl Acad Sci U S A. 1996 Apr 16; 93(8):3289-93.
[Proc Natl Acad Sci U S A. 1996]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Immunol Cell Biol. 1996 Aug; 74(4):357-64.
[Immunol Cell Biol. 1996]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Int Immunol. 1994 Nov; 6(11):1661-70.
[Int Immunol. 1994]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Mol Immunol. 1992 Sep; 29(9):1157-8.
[Mol Immunol. 1992]EMBO J. 1988 Nov; 7(11):3413-22.
[EMBO J. 1988]Immunogenetics. 1998; 47(3):234-45.
[Immunogenetics. 1998]Immunogenetics. 2002 Oct; 54(7):501-12.
[Immunogenetics. 2002]Proc Natl Acad Sci U S A. 1998 Nov 24; 95(24):14343-8.
[Proc Natl Acad Sci U S A. 1998]J Immunol. 1998 Feb 1; 160(3):1233-9.
[J Immunol. 1998]Int Rev Immunol. 1997; 15(3-4):207-41.
[Int Rev Immunol. 1997]Immunol Res. 2002; 26(1-3):265-78.
[Immunol Res. 2002]Immunol Rev. 2000 Jun; 175():201-13.
[Immunol Rev. 2000]Proc Natl Acad Sci U S A. 1988 Apr; 85(7):2245-9.
[Proc Natl Acad Sci U S A. 1988]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Proc Natl Acad Sci U S A. 2003 Mar 4; 100(5):2501-6.
[Proc Natl Acad Sci U S A. 2003]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Mol Biol Evol. 1987 Jul; 4(4):406-25.
[Mol Biol Evol. 1987]Nucleic Acids Res. 1997 Dec 15; 25(24):4876-82.
[Nucleic Acids Res. 1997]EMBO J. 1988 Nov; 7(11):3413-22.
[EMBO J. 1988]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]Immunogenetics. 2002 Oct; 54(7):501-12.
[Immunogenetics. 2002]Mol Biol Evol. 1989 Sep; 6(5):447-59.
[Mol Biol Evol. 1989]J Immunol. 1991 Mar 15; 146(6):1996-2004.
[J Immunol. 1991]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Proc Natl Acad Sci U S A. 2000 Apr 25; 97(9):4712-7.
[Proc Natl Acad Sci U S A. 2000]Proc Natl Acad Sci U S A. 2001 Feb 13; 98(4):1775-80.
[Proc Natl Acad Sci U S A. 2001]J Immunol. 1997 Dec 15; 159(12):6097-104.
[J Immunol. 1997]Eur J Immunol. 1996 May; 26(5):1123-9.
[Eur J Immunol. 1996]Nucleic Acids Res. 1994 Nov 11; 22(22):4673-80.
[Nucleic Acids Res. 1994]Nucleic Acids Res. 1997 Dec 15; 25(24):4876-82.
[Nucleic Acids Res. 1997]Mol Biol Evol. 1987 Jul; 4(4):406-25.
[Mol Biol Evol. 1987]Comput Appl Biosci. 1996 Aug; 12(4):357-8.
[Comput Appl Biosci. 1996]Immunol Cell Biol. 1996 Aug; 74(4):357-64.
[Immunol Cell Biol. 1996]Blood. 2001 Mar 1; 97(5):1511-3.
[Blood. 2001]Proc Natl Acad Sci U S A. 1988 Apr; 85(7):2245-9.
[Proc Natl Acad Sci U S A. 1988]Immunol Rev. 2000 Jun; 175():201-13.
[Immunol Rev. 2000]