© 1992 Oxford University Press

International Immunology, Vol. 4, No. 9, pp. 1031 - 1040

The gene organization of the human (37 subunit, the common p1 subunit of the leukocyte integrins HML-1 and LPAM-1 Wei-Meng Jiang, David Jenkins, Qian Yuan, Euphemia Leung, K. H. Andy Choo1, James D. Watson, and Geoffrey W. Krissansen

Key words: cell adhesion, evolution, genomic DNA, LeuCAM

Abstract The integrin j37 subunit associates with two alternative a subunits termed aHML-1 and a 4 to give the lymphocyte activation and homing receptors HML-1 and LPAM-1. Overlapping genomic clones that encoded the human j37 subunit gene were isolated from cosmid and phage X libraries. The coding portion of the gene spanned - 1 0 kb and was composed of 14 exons. Exon 1 (123 bp) encoded 5' untranslated sequences; exon 2 (204 bp) encoded the initiation codon, signal peptide, and 50 amino acid residues of the N-terminus of the mature protein; 7 exons (exons 3 - 9 ) , ranging in size from 90 bp to 242 bp, encoded most of the extracellular domain proximal to the four cysteine-rich repeats; the region corresponding to the (33 arginine-glycine-aspartic acid (RGD)-binding domain was divided amongst exons 4 and 5; the four cysteine-rich repeats were encoded by 3 exons (exons 1 0 - 1 2 ) with intron insertion into the first and third repeats; exon 13 (209 bp) provided a spacer between the cysteine-rich domains and the transmembrane domain; exon 14 (161 bp) encoded the transmembrane domain and exactly half of the cytoplasmic domain; the remainder of the cytoplasmic domain and most of the 3' untranslated region was contained in the largest 313 bp exon 15. Comparison of integrin /3 subunit genes revealed that the gene organization of j37 was almost identical to that of /32, but had diverged from that of /33. Amplification of integrin DNAs directly from genomic DNA, using PCR primers based on /? subunit consensus sequences corresponding to the 0 3 RGD-binding domain, yielded partial gene sequences for the /33, /3S, and /36 subunits only. Inspection of the amplified sequences revealed that, as for 0 3 , the regions in 0S and /36 corresponding to the /33 RGD-binding domain lacked the intron present in /37, 0,, and /32, which divides this region in j32 into two subdomains that contribute to subunit assembly. This study provides genetic evidence for at least two major branches to the integrin 0 subunit evolutionary tree, with 0 7 , /32, and probably 0, in one branch, and the cytoadhesin (33 and probably also /35 and /36 in the other.

Introduction The integrin gene family comprises a group of a/3 heterodimeric adhesion receptors involved in cell - cell and cell - matrix interactions (reviewed in 1 and 2). The distinct a and /3 gene superfamilies are not related, whereas the various members within each family are significantly homologous (a subunits are 1 7 - 63% similar, /? subunits 2 8 - 5 5 % similar). There are at least three human integrin subfamilies formed from groups of a and 0 subunits that assemble to give functionally homologous cell

surface glycoproteins. Each family is designated by a distinct 0 subunit, 0, (very late antigen family), fi2 [lymphocyte functionassociated antigens (LFA) family], and /33 (cytoadhesin family). The cDNAs encoding 11 distinct a subunits and eight separate /3 subunits have been cloned and sequenced. Complete genomic organizations for the coding regions of the p150,95a (3), GPIIb (4), and /32 subunit (5) genes have been determined. Partial genomic clones have been isolated and sequenced, encoding

Correspondence to: G. W. Krissansen Transmitting editor: R. Scollay

Received 13 March 1992, accepted 1 June 1992

Downloaded from http://intimm.oxfordjournals.org/ at University of California, San Fransisco on March 25, 2015

Department of Molecular Medicine, School of Medicine, The University of Auckland, Auckland, New Zealand 1 The Murdoch Institute for Research into Birth Defects, Royal Children's Hospital, Melbourne, Australia

1032

Organization of the integrin /37 subunit gene

Northern and Southern analysis of PCR-amplified 0 7 sequences of a small panel of cell types indicated that expression of the /37 subunit was probably leukocyte-restricted, suggesting it may be an immunoregulatory molecule like its closest family member, /32 (8). We subsequently reported (12) that an Nterminal region of'13 contiguous amino acids deduced from the cDNA encoding the mouse /37 homologue was identical with the N-terminus of a 120,000 Mr subunit that is part of an antigen, termed M290, found highly expressed on mouse intraepithelial lymphocytes (IEL) (13). This unexpected result focused two previously unconnected areas of research and suggested that integrins may have a special role to play in defence of the gut mucosa. We and others (14,15) subsequently demonstrated that the 0 7 subunit forms part of the human IEL antigen, HML-1, which had been suspected to be the human homologue of the M290 antigen. A panel of antibodies directed against the HML-1 antigen are proving useful for diagnosing certain T and B cell lymphomas and leukaemias, including Hairy Cell leukaemia (16 -18). A recent report by Kilshaw and Murant (19) has shown that @7 is present in mouse on most lymph node lymphocytes in association with a 4 , rather than aM290. a"/37 was shown to be identical to LPAM-1 (a4j3p), the Peyer's patch lymphocyte homing receptor (19,20). The 07 gene was mapped to band q13.13 of human chromosome 12 immediately distal to a fragile site, FRA12A (21, 22). This is notable because the a 5 gene of the FNR (a5/3,) also maps to this region at 12q11 - 1 3 (23). This paper describes the isolation and characterization of the genomic clones that encode the /37 subunit gene. The coding portion of the gene spans - 1 0 kb and is composed of 14 exons. Comparison of the genomic organization of the /37 subunit gene with the /32 gene and partial clones encoding the /3)t /33, and /36 subunits shows that the j32 and /37 genes share a very close evolutionary relationship, and provides evidence for the existence of at least two major branches to the evolutionary tree of integrin 0 subunits.

Methods Isolation of 07 subunit genomic clones The two human genomic libraries screened were a Charon 4A library constructed from human genomic DNA partially digested with EcoRI (obtained from the American Type Culture Collection, Rockville, MD, ATCC no. 37385, depositor A. Bank) and a cosmid library constructed in the cosmid vector pCV001 from DNA partially digested with Mbo\ (24). The Charon 4A library was plated on Escherichia coli strain LE392 and screened with a 5' 0.39 kb Pst\-Pst\ fragment (nucleotides 83-471) of the full length cDNA of human clone 1 (/37). To select for cosmid clones containing 5' sequences, the pCV001 library was screened with the extreme 5' 90 bp EcoRI -Pst\ fragment of clone 1 (/37). The libraries were replica plated onto Gene-Screen Plus filters (Du Pont, Boston, MA) and screened according to standard procedures (25) with probes labeled by random hexanucleotide priming (26). The pre-hybridizations and hybridizations were performed at 42°C in solutions containing 50% (v/v) formamide and 5 x SSC. Filters were given stringent washes with 0.1 x SSC, 0.1% SDS at 50°C. Positive clones were detected by autoradiography at - 70°C using Kodak X-AR5 X-ray film with Du Pont intensifying screens (Cronex Lightning-Plus) and purified by two additional rounds of screening.

° ~ .s 23-

lij cu X

6-

2.3-

0.5-

Fig. 1. Estimate of the size of the 0 7 gene by Southern analysis. Human genomic DNA (20 ng) was digested with EcoRI, Pst\, or H/ndlll and the fragments were separated by electrophoresis in 0.6% agarose gel and transferred to Gene Screen Plus. Integrin 0 7 exonic regions were detected by hybridization with the 32P-labelled full length insert of clone 1 lfi7). Size markers in kb are indicated in the left margin.

Downloaded from http://intimm.oxfordjournals.org/ at University of California, San Fransisco on March 25, 2015

the /?, (6) and /33 (6,7) subunits. These studies provided genetic evidence that integrin 0 subunits have a common evolutionary origin. We first described a novel integrin /3 subunit, termed /37, having used the PCR technique to amplify integrin /3 subunitrelated sequences from phytohaemagglutinin-activated peripheral blood lymphocytes (8,9). The cDNA for the /37 subunit encoded a 798 amino acid residue protein consisting of a typical signal peptide, a predominantly hydrophilic 708 amino acid residue domain with eight /V-glycosylation sites, a transmembrane domain, and an intracellular domain of 52 amino acids. The /37 sequence shows similarity to known /3,, /32, /33, /34, /35, and |36 sequences of 45, 49, 40, 33, 39, and 39%, respectively. Four cysteine-rich homologous repeat sequences were found in /37 that were homologous to the corresponding repeats in other integrin /3 subunits and to domain III of the laminin B chains. Part of the laminin cysteine-rich repeats are homologous to epidermal growth-factor-like repeat sequences that characterize a family of epidermal growth factor (EGF)-related proteins (10). Recently Erie et al. (11) revealed an in-frame alternatively spliced variant of 0 7 expressed at very low levels that had been generated by removal of exons that encode most of the cysteinerich domains.

Organization of the integrin 0T subunit gene Isolation of a 06 subunit genomic clone Amplification of integrin sequences directly from Jurkat cell genomic DNA, using PCR primers directed to conserved sequences in the integrin 0,, j32, and j33 subunits corresponding to the j33 subunit arginine-glycine-aspartic acid (RGD)binding domain (8,21), yielded partial gene sequences for the 0 3 , j35, and j36 subunits. The 32P-labelled PCR fragment encoding the /36 subunit was used to screen 4 x 105 recombinants of a human foetal liver genomic library in Charon 4A, constructed from a partial Haelll -Alu\ digest [obtained from the American Type Culture Collection, Rockville, MD, ATCC No. 37333 (27)], as described above. A single clone (H.A./Jg) was obtained (21).

Human genomic DNA (20 /*g) was digested with different restriction enzymes and electrophoresed in 0.6% agarose gel. The DNA was acid depurinated and alkali denatured in the gel, and transferred to Gene-Screen Plus nylon membranes. Filters were hybridized with a 32P-labelled full length /37 cDNA, or various fragments of this cDNA, according to the manufacturer's instructions. Filters were washed stringently with 0.1 x SSC, 0 . 1 % SDS at 50°C. Subcloning of restriction fragments and sequence determination Restriction enzyme fragments of the genomic clones were subcloned into pUC18 and M13 and sequenced by the dideoxy chain termination method using dATP-5'-[a-35S]thiophosphate (New England Nuclear Corp., Boston, MA). Pst\ proved to be an extremely useful enzyme since Pst\ sites were almost uniformly distributed amongst the /37 subunit exons. Certain fragments subcloned into M13 were sequenced using an ABI automated DNA sequenator. The entire transcribed portion of the 07 gene, previously defined by the /37 cDNA, was identified and

sequenced, except for a 25 bp exonic sequence at the extreme 5'-end of the /37 cDNA. Exon - intron boundaries were assigned by comparing the genomic and cDNA sequences, and according to the GT/AG rule for splicing. The determined DNA sequence has been submitted to GenBank databank.

Results and discussion Southern hybridization analysis of human genomic DNA We have previously described the isolation and sequencing of the cDNA clone 1 (j37), which encoded the human integrin /37 subunit (8,9). In order to obtain an approximate size for the 0 7 gene, human genomic DNA was digested with restriction enzymes and examined by Southern hybridization analysis (Fig. 1). Hybridization of EcoRI digested DNA with a full length 2.8 kb j37 cDNA revealed hybridizing bands of 14, 5, 3.5, and 0.7 kb. A 90 bp cDNA fragment encoding the extreme 5' untranslated region hybridized to the 14 kb fragment only (data not shown), suggesting that most of the coding region of the gene was contained within 10 kb. Isolation and sequencing of the /37 subunit gene A single phage X clone was obtained by screening 800,000 plaques from the Charon 4A genomic library. Restriction enzyme mapping, Southern hybridization with different fragments of clone 1 ()37), and DNA sequencing revealed that the Charon 4A clone contained the entire 3' untranslated region and most of the coding region, but did not contain exons encoding the first 67 amino acids of the /37 protein (Fig. 2). To isolate the missing portion of the j37 subunit gene, 500,000 colonies of a genomic library in a cosmid vector, pCV001, was screened with the most 5' 90 bp fragment of the /37 subunit cDNA. Three overlapping cosmid clones from 16 positive colonies were further characterized.

Fig. 2. Partial restriction map of the human 0 7 subunit gene. Phage X and cosmid clones used to characterize the 0 7 subunit gene are shown below the restriction enyzme map. The restriction sites shown and the sizes of the genomic fragments were determined by endonuclease digestion and Southern blot analysis. Restriction enzymes used were Alu\ (A), BamH\ (B), EcoRI (Rl), Psrl (P), Haelll (H), and Sst\ (S). Exons are represented by the solid boxed regions.

Downloaded from http://intimm.oxfordjournals.org/ at University of California, San Fransisco on March 25, 2015

Southern analysis of human chromosomal DNA

1033

1034

Organization of the integrin j37 subunit gene

exon 1

exon 9

cactttgcaflATGTGGAAACTAAAGCCCAGAGAGAAAGTCTGACTTGC CCCACAGCCAGTGAGTGACTGCAGCAGCACCAGAATCTGGTCTGTTTCCTGTTTGGCT CTTCTACCACTACGGCTTGGGATCTCGataaatagaa

cttcccataaAGCCTGTCTTCCACCGTGACCCTTGAACACTCTTCACT (371)S L S S T V T L E H S S L CCCTCCTGGGGTCCACATTTCTTACGAATCCCAGTGTGAGGGTCCTGAGAAGAGGGAG P P G V H I S Y E S Q C E G P E K R E GGTAAGGCTGAGGATCGAGGACAGTGCAACCACGTCCGAATCAACCAGACGal.gagag G K A E D R G Q C N H V R I N Q T(419) cca

exon 2 cctattacaaGGCATGGTGGCTTTGCCAATGGTCCTTGTTTTGCTGCT (-17)M V A L P M V L V L L L GGTCCTGAGCAGAGGTGAGAGTGAATTGGACGCCAAGATCCCATCCACAGGGGATGCC V L S R G E S E L D A K I P S T G D A ACAGAATGGCGGAATCCTCACCTGTCCATGCTGGGGTCCTGCCAGCCAGCCCCCTCCT T E W R N P H L S M L G S C Q P A P 3 GCCAGAAGTGCATCCTCTCACACCCCAGCTGTGCATGGTGCAAGCAACTGgtaaagat C Q K C I L S H P S C A W C K Q L(50) gg

exon 3

exon 4 caccccctaflGGGAGCCCCAGCAGCTCCAGGTCCGCTTCCTTCGTGC (118)G E P Q Q L Q V R F L R A TGAGGGATACCCGGTGGACCTGTACTACCTTATGGACCTGAGCTACTCCATGAAGGAC E G Y P V D L Y Y L M D L S Y S M K D GACCTGGAACGCGTGCGCCAGCTCGGGCACGCTCTGCTGGTCCGGCTGCAGGAAGTCA D L E R V R Q L G H A L L V R L Q E V CCCATTCTGTGCGCATTGatgagccgag T H S V R 1(174)

exon 5 cctctcccasGTTTTGGTTCCTTTGTGGACAAAACGGTGCTGCCCTTT (17S)G F G S F V D K T V L P F GTGAGCACAGTACCCTCCAAACTGCGCCACCCCTGCCCCACCCGGCTGGAGCGCTGCC V S T V P S K L R H P C P T R L E R C AGTCACCATTCAGCTTTCACCATGTGCTGTCCCTGACGGGGGACGCACAAGCCTTCGA Q S P F S F H H V L S L T G D A Q A F E GCGGGAGGTGGGGCGCCAGAGTGTGTCCGGCAATCTGGACTCGCCTGAAGGTGGCTTC R E V G R Q S V S G N L D S P E G G F GATGCCATTCTGCAGGCTGCACTCTGCCAGatgaggaggt D A I L Q A A L C Q