ABSTRACT
We recently reported the cloning and sequencing of the7 integrin chain and its regulated expression during the development of skeletal muscle (Song et al. (1992) J. Cell Biol. 117, 643-657). The 7 chain is expressed during the development of the myogenic lineage and on adult muscle fibers and this suggests that it participates in multiple and diverse functions at different times during muscle development. One interesting portion of this isoform is its cytoplasmic domain; comprised of 77 amino acids it is the largest in the alpha chains thus reported. In these experiments we begin to study the potential functions of the 7 cytoplasmic domain by ana-lyzing homologies between the rat and human sequences, by immunologic studies using an anti-cyto-plasmic domain antiserum, and by identifying two alternate forms. In keeping with the nomenclature used to describe the 3 and 6 alternate cytoplasmic domains, we refer to the originally reported species as 7B and the two additional forms as 7A and 7C. These three cytoplasmic domains likely arise as a consequence of alternate splicing. A splice site at the junctions of the transmembrane and cytoplasmic domains is used to gen-erate the 3, 6 and 7 A and B forms. The 7A form RNA contains an additional 113 nucleotides compared to the B form, and a common coding region in the A and B form RNAs is used in alternate reading frames. Part of the coding region of 7B appears to be used as the 3-untranslated region of the 7A form. The 7C mRNA is 595 nucleotides smaller than the 7B RNA and part of the 3-untranslated region of the 7B isoform is used as coding sequence in 7C. There is developmental specificity in expression of these alternate mRNAs: 7A and 7C transcripts are found upon terminal myogenic differentiation whereas 7B is present earlier in replicating cells and diminishes upon differentiation. We suggest this selective expression of the 7 cytoplasmic domains underlies the diversity in function of the 7 1 integrin at different stages of muscle development.
Immunochemical analyses indicate that the 7B cytoplasmic domain undergoes a change in conformation in response to binding laminin or upon crosslinking initiated with antibody reactive with the integrin extracel-lular domain. Crosslinking also promotes association of the integrin with the cell cytoskeleton. Analysis of the amino acid sequence of the 7B cytoplasmic domain reveals several motifs that may relate to the function of this protein. Two regions in the 7B cytoplasmic domain have homology and similar apposition to those in the catalytic phosphotransfer domain and the ATP-binding site of serine/threonine protein kinases. There is also a sequence of 15 amino acids in the 7B cytoplasmic domain that is homologous to that in many receptor-like protein tyrosine phosphatases. Although this cytoplasmic domain may be too small to have catalytic properties, it may effect the localization or regulation of these enzymes, or other proteins that interact with them. There is also a potential actin-binding sequence and a unique three-fold DXHPX repeat towards the carboxyl end of the 7B cytoplasmic domain. Clearly, the 7B cytoplasmic domain contains a rich potential for par-ticipating in the transduction of signals initiated outside the cell. This diversity in features, conformational changes, and forms of the 7 cytoplasmic domains likely underlie its diverse functions on skeletal muscle.
INTRODUCTION
The integrins are a diverse family of heterodimeric cell sur-face integral membrane proteins that mediate the interactions of cells with each other, with extracellular matrix proteins, and either directly or indirectly, with additional molecules in their environment. The interactions of integrins with extracellular matrix proteins or with counter receptors
on other cells play significant roles in cell adhesion, migration and differentiation. The presence of integrins on most cells and their diversity of functions have made them a focus of study of many laboratories interested in cell, tissue and embryonic development, motility, cell structure and proliferation, clotting, and signal transduction, as well as specific pathologies associated with these processes, including metastasis and thrombosis (for reviews see, Albelda and Buck, 1990; Hemler, 1990; Springer, 1990; Hynes, 1992; Juliano and Haskill, 1993). The functional diversity exhib-ited by the integrins is largely due to the heterogeneity in integrin structure that is a consequence of the associations of alpha and beta chains that comprise the heterodimer. At least eight beta chains and fourteen alpha chains have been reported and the variety of heterodimers derived from these is believed to underlie the diversity in ligand binding and tissue specificity (Hynes, 1992). Preferred pairs of alpha and beta chains are expressed and associated in different cell types and more than one integrin is often expressed on individual cells.
Both alpha and beta integrin chains have a single long extracellular domain, a hydrophobic transmembrane domain, and (with exception of β4) a relatively short cytoplasmic domain. The capacity to bind ligand outside the cell is a function of both proteins. The sequences of the cytoplasmic domains of the different alpha and beta chain isoforms are quite divergent from one another: this is especially true among the alpha chains. Alternate forms of cytoplasmic domains that arise from alternate splicing have been found in the β1, β3, β4, α3 and α6 integrin cytoplas-mic domains (van Kuppevelt et al., 1989; Altruda et al., 1990; Tamura et al., 1990, 1991; Cooper et al., 1991). Presumably, these diversities in the cytoplasmic domains also contribute to the varied capacities of the integrins to mediate the transduction of signals that initiate from extracellu-lar interactions into the cell as well as signals that arise within the cell and are directed outward via activation of the integrins. The association of integrins with the cell cytoskeleton and the formation of adhesion plaques have been the most widely studied aspects of cytoplasmic domain functions and the beta chains seem to have a pre-eminent role. Some cytoplasmic proteins that mediate these interactions, for example vinculin and talin, have been identified (Springer and Paradiso, 1981; Horwitz et al., 1986; Otey et al., 1990) and phosphorylation of specific residues in the cytoplasmic domain appears to have a regulatory role (Hirst et al., 1986; Dahl and Grabel, 1989; Otey et al., 1990; Reska et al., 1992). Additional interactions of both alpha and beta chain cytoplasmic domains likely mediate integrin functions and these may be as diverse as the cytoplasmic domains themselves.
We recently reported the cloning and sequencing of the α7 integrin chain and its regulated expression during the development of skeletal muscle (Song et al., 1992; George-Weinstein et al., 1993). The α7 integrin also appears on some cells derived from the neural crest, including melanoma cells (Kramer et al., 1991), dorsal root ganglia and PC12 cells (S. J. Kaufman and M. George-Weinstein, unpublished data). The α7 integrin is expressed on replicating secondary myoblasts (Kaufman and Foster, 1988; Kaufman et al., 1991; George-Weinstein et al., 1993). When these cells are grown on laminin, they undergo a change in shape, become more mobile and maintain their proliferation (Foster et al., 1987; Ocalan et al., 1988; Goodman et al., 1989). As the α7β1 integrin is the sole functional laminin-binding integrin on these cells (von der Mark et al., 1991), we believe it underlies these behaviors and also func-tions to localize these cells at the laminin-rich sites of sec-ondary fiber formation (George-Weinstein et al., 1993). Upon terminal differentiation there is an increase in the expression of α7 on myotubes and this persists in adult fibers. α7 localizes between adult fibers and the surrounding matrix (Song et al., 1992), and at myotendinous junctions where it likely serves to tether the fibers at its ends (M. George-Weinstein and S. J. Kaufman, unpublished results). The function of α7 at these sites in adult muscle appears to be quite different from that on mobile, replicating myoblasts. These observations suggest that α7 has multiple and diverse functions, and this raises the question, ‘How are the requirements for these diverse functions resolved?’
One interesting characteristic of the α7B isoform is its large cytoplasmic domain comprised of 77 amino acids, the largest of the alpha chains thus reported. In the experiments reported here we begin to study the potential functions of this cytoplasmic domain by analyzing homologies between the rat and human sequences, by immunologic studies using an anti-cytoplasmic domain antiserum, and by identifying two alternate forms of the α7 cytoplasmic domain. One of these alternate forms, α7A, is identical in amino acid sequence to that recently reported by Collo et al. (1993). The results of our experiments demonstrate multiple con-served and unique features of the α7 cytoplasmic domain and conformational changes. This diversity in features, con-formational states, and alternate forms, likely underlie the diversity in function of the α7 integrin chain on skeletal muscle.
MATERIALS AND METHODS
Isolation of a human 7 cDNA clone
A human fetal muscle λgt11cDNA library, kindly provided by Dr George Dickson, was screened by plaque filter hybridization using clone 05A rat α7 cDNA (Song et al., 1992), labeled by the random priming method (Oligolabeling kit; Pharmacia) with [α-32P]dCTP (Amersham; 3000 Ci/mmol). Hybridizations were performed at 65°C as described (Song et al., 1992), positive plaques were iso-lated, and recombinant phage DNA was purified from small scale plate lysates (Silhavy et al., 1984). cDNA fragments were sub-cloned in the EcoRI sites of pBluescript SK− and sequenced by the dideoxy chain termination method (Sanger et al., 1977).
Production of anti-7 cytoplasmic domain antiserum
A 27 amino acid peptide, NH2-CEDRQQFKEEKTG-TIQRSNWGNSQWEG, was synthesized using t-BOC chemistry and an Applied Biosystems model 430A synthesizer, at the University of Illinois Biotechnology Center, and purified using a Vydak C-18 reverse phase chromatography column, eluted with a 0-60% gradient of 0.1% trifluoroacetic acid (TFA) and 0.1% TFA/70% acetonitrile. This peptide (excluding the N-terminal cysteine) is within the α7B integrin cytoplasmic domain. The peptide was further purified using a Bio-Rad 10 DG exclusion column.
6.7 mg of peptide in 300 μl phosphate buffered saline (PBS), was added to 10 mg of maleimide-activated keyhole limpet hemo-cyanin (KLH) (Pierce) in 10 ml PBS and incubated at room tem-perature for 2 hours. Then 100 μl of 100 mM cysteine was added for 15 minutes and the protein was dialyzed against 500 ml PBS overnight at 4°C. New Zealand White rabbits were immunized at two intramuscular and subcutaneous sites with approximately 1 mg of conjugated protein emulsified in an equal volume of Ribi adjuvant (Ribi Immunochemical). Each rabbit was immunized 7 times over five months. The serum was delipified with 0.25% sodium dextran sulfate and 1% CaCl2, centrifuged at 10,000 g for 10 minutes, and the antibody was precipitated with 50% ammonium sulfate and dialyzed extensively against PBS. Immunoblot analyses using the peptide coupled to bovine serum albumin (BSA) demonstrated the specificity of the respective sera for the immunizing peptide.
Immunofluorescence
L8E63 myogenic cells were grown in Dulbecco’s modified Eagle’s medium (DME) supplemented with 10% horse serum (Gibco), on 12 mm glass coverslips coated with 0.1% gelatin as described (Kaufman and Parks, 1977). Cultures of cells from the thighs of newborn rat hindlimbs were prepared and grown as indi-cated (Foster et al., 1987). Cells crosslinked prior to staining with anti-cytoplasmic domain antibody (anti-CD) were first reacted either with anti-α7 integrin or anti-laminin 2E8 (Engvall et al., 1986; Developmental Hybridoma Bank) monoclonal antibodies followed by fluorescein-conjugated donkey antimouse immunoglobulin (Jackson Immunoresearch). These cells were then fixed with 95% ethanol or treated with 0.25% Triton X-100 for 10 minutes. Staining was then done with anti-CD antibody (a 1:200 dilution of a 50% ammonium sulfate cut), followed by rho-damine-conjugated donkey anti-rabbit immunoglobulin (Jackson Immunoresearch). After staining, the cells previously extracted with Triton X-100 were fixed with 95% ethanol. Live cells were reacted for 5 minutes with laminin diluted to 1, 10 or 100 μg/ml in Dulbecco’s PBS (DPBS) containing 0.1% gelatin. These cells were immediately fixed with 95% ethanol and stained with anti-CD antibody. To demonstrate specificity, the anti-CD antibody was blocked with α7 cytoplasmic domain peptide by incubating equal volumes of antibody and 200 μg/ml peptide for 30 minutes. The antibody was then diluted to the appropriate concentration for staining. To determine the effect of temperature on promoting accessibility of the cytoplasmic domain, individual coverslips in dishes were maintained at the specified temperatures for 5 min-utes. The cells were immediately fixed with 95% ethanol at room temperature. The coverslips were mounted in glycerol/PBS (9/1, v/v), pH 8.5, containing 10 mM p-phenylenediamine (Eastman), sealed with Flo-texx (Fisher), and examined with a Zeiss pho-tomicroscope III equipped with epi-illumination optics and an HBO 100 W mercury lamp.
Immunoblotting
Cell lysates were made from cells rinsed three times with DPBS, then with DPBS containing 1 mM phenylmethylsulfonyl fluoride (PMSF). The cells were then scraped with a rubber policeman and pelleted by centrifugation. The pellets were resuspended in PBS containing 1 mM PMSF and sonicated three times, for 5 seconds using a Branson 200 Sonifier, set at output 5 and 50% duty cycle. The lysates were then electrophoresed in 8% SDS-polyacrylamide gels at 40 mA. The gels were equilibrated in 25 mM Tris, 200 mM glycine, pH 8.8, and 20% methanol and transferred onto nitro-cellulose at 100 V, for 60 minutes. The blots were rinsed at room temperature for 60 minutes in TSTB (5 mM Tris, 75 mM NaCl,
and 0.5% Tween-20, pH 7.5) containing 2% gelatin. The nitro-cellulose was then rinsed 3 times, 10 minutes each, in TSTB con-taining 0.5% gelatin and then reacted with the appropriate anti-bodies diluted in this same buffer. After rinsing, the blots were reacted with alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin (Jackson Immunoresearch) for 60 minutes, washed, and developed with nitro blue tetrazolium (NBT, 55 mg/ml in 70% N, N-dimethylformamide) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP, 50 mg/ml in 100% N, N-dimethylfor-mamide) in 10 ml 150 mM NaCl, 5 mM EDTA, and 100 mM Tris-HCl, pH 9.5. The blots were then rinsed with TSTB, then with 20 mM Tris, pH 2.9, 1 mM EDTA, and allowed to air dry. Prestained molecular mass markers were included in each gel.
Deglycosylation of 7 integrin
L8E63 myotubes, on three 100 mm dishes, were collected using a rubber policeman, pelleted and extracted twice, for 30 minutes, at 4°C, with 100 μl of extraction buffer (200 mM octyl-β-D-glu-copyranoside, 1 mM PMSF and 100 mM Tris-HCl, pH 7.4). For deglycosylation, SDS was added to a final concentration of 1%, the extract was then boiled for 2 minutes, then adjusted to 20 mM sodium phosphate buffer, pH 7.2, 10 mM sodium azide, 50 mM EDTA and 0.5% NP-40, and boiled for 2 minutes. The denatured cell extract was then incubated with peptide-N-glycosidase F (PNGase, 20 mU/50 μl cell lysate, Boehringer Mannheim) at 37°C for 18 hours and analyzed in immunoblots.
Reverse transcriptase-polymerase chain reaction (RT-PCR)
RT-PCR was performed as described (Song et al., 1992). Single-stranded cDNA was synthesized using 100 ng poly(A)+ RNA puri-fied from L8E63 myoblasts or myotubes, or 1 μg total RNA extracted from L8E63 cells, fu-1 cells or newborn myoblasts grown as indicated (Foster et al., 1987), 15 units of AMV reverse transcriptase (Promega) and 0.2 μg of oligo (dT) primer. The single-stranded cDNAs were amplified in PCR buffer (50 mM KCl, 1.5 mM MgCl2, and 10 mM Tris-HCl, pH 8.4) containing 1.0 μM of 17mer sense primer (5′-AGCCGTGCTTCATGTCT-3′) and 1.0 μM 17mer antisense primer (3′-GGCTGGTACAGCA-CACT-5′) for amplification of the mRNAs encoding α7A and α7B cytoplasmic domains, or, 1.0 μM of 20mer antisense primer (3′-TGGTACAGCACACTTGAAGA-5′) for amplification of the α7C mRNA, 250 μM each of dATP, dCTP, dTTP and dGTP, and 1.25 units/50 μl reaction mixture of Taq polymerase (Promega). The samples were cycled 30 times at 94°C for 1 minute, 55°C for 1.5 minutes, and 72°C for 1.5 minutes. The PCR products were separ-ated on 0.8% agarose gels, the fragments were cut out, purified using GeneClean II (Bio101) and cloned using the ddT-tailed vector system described by Holton and Graham (1990). Then 5 μg of pBluescript SK− was digested with EcoRV and was tailed in 25 mM Tris-HCl, pH 6.6, 200 mM potassium cacodylate, 250 μg/ml BSA, 1.5 mM CoCl2 and 10 μM ddTTP (Pharmacia) using 50 units of terminal transferase (Promega), for 1 hour at 37°C. ddT-tailed plasmid was purified with phenol/chloroform, followed by ethanol precipitation, and ligated with the purified PCR frag-ment using T4 DNA ligase (BRL). XL-1 cells were transformed with the ligation product, plated on X-gal and white colonies were selected.
Computer analyses of protein sequences
Protein sequence homology searches were carried out using the BLAST programs developed by National Center for Biotechnol-ogy Information at the National Library of Medicine (Altschul et al., 1990). Protein secondary structure analyses were performed with the algorithms of Chou and Fasman (1974), Garnier et al. (1978), and Hopp and Woods (1981), using MacVector 3.0 (IBI, New Haven, CT).
RESULTS
Alternate forms of the 7 cytoplasmic domain
Reverse-transcriptase PCR analysis, using paired primers specific for the cytoplasmic domain of the α7 integrin, revealed two products. This suggested that two α7 isoforms were expressed during the differentiation of the L8E63 myogenic cell line (Fig. 1). In contrast with the 899 bp product detected throughout the course of myogenesis in vitro, a 113 nucleotide larger fragment was generated from poly(A)+ RNA prepared from cells in cultures that had undergone differentiation. The 1012 bp fragment was cloned into pBluescript KS− and its nucleotide and inferred amino acid sequences were determined. The larger PCR fragment represents an mRNA that encodes an alternate 58 amino acid cytoplasmic domain. In keeping with the nomenclature previously used for isoforms of the α3 and α6 cytoplasmic domains, and the homologies of the respec-tive A and B isoforms, this species has been referred to as α7A and the original form as α7B. An alternate isoform of the mouse α7 cytoplasmic domain was recently cloned and sequenced by Collo et al. (1993). The sequence of the rat α7A cytoplasmic domain (Fig. 2A) differs at five nucleotides from that of the mouse, and there is a single residue difference in the 58 amino acid sequence. A common coding region in the A and B form RNAs is used in alternate reading frames, and part of the coding region of α7B is in the 3′-untranslated region of the α7A RNA. Two splice sites appear to be located 5′ to the GFFKR coding regions of the A and B form RNAs.
An additional alternate isoform of the α7 cytoplasmic domain was detected in a lambda UniZap rat muscle cDNA library screened with a 293 nucleotide PstI fragment produced from the 5′-end of the 05B α7 cDNA clone. This fragment encodes a 97 amino acid region in the extracellular domain that is unique to the α7 integrin. Forty-five clones were identified and inserted into the pBluescript SK− plasmid by in vivo excision and transformation into XL-1 blue cells. Plasmid from these cells was pooled into nine groups, each representing five positive clones. The DNA sequences bounded by the 17mer sense and antisense primers (Fig. 2B) were amplified using Taq polymerase. Two fragments were generated, one 899 bp, representing the amplified fragment of the α7 cDNA previously reported (Song et al., 1992), and a second, 304 bp fragment. Indi-vidual clones in the positive groups were analyzed by PCR amplification and two clones containing the shorter frag-ment were identified (Fig. 3A) and sequenced from their 3′-ends. The 5′-ends of these sequences, which encode the membrane spanning region and extend into the extracellu-lar domain, were identical with the sequence of α7B (Fig. 2B). A deletion of 595 nucleotides in α7B accounts for the shorter sequence, which we refer to as α7C. A new open reading frame is generated as a consequence of this dele-tion and results in the alternate protein sequence in the cyto-plasmic domain that begins with a switch from the GFFKR sequence in α7B to GFFKC in α7C. Part of the 3′-untrans-lated region present in the α7B RNA is used to encode the α7C form and a new termination codon, TAA (Fig. 2).
In contrast with the 58 amino acids in the α7A form and 77 amino acids in the α7B cytoplasmic domain, the α7C cytoplasmic domain contains 18 amino acids. Potential sites of phosphorylation in the three α7 cytoplasmic domains were determined using reported consensus sequences (Aitken, 1990). As indicated in Fig. 2C, each cytoplasmic domain has a single site of potential tyrosine phosphorylation and this appears to be the sole potential phosphorylation site in the α7B cytoplasmic domain. The α7A and α7C domains each have a serine residue that may be phosphorylated by calcium-, calmodulin-dependent protein kinase II: this same serine in α7A may be phosphorylated by cGMP-dependent protein kinase. An additional serine and threonine residue in the α7A cytoplasmic domain may be a substrate for phosphorylation by protein kinase C.
Expression of the7cytoplasmic domains is developmentally regulated
The differentiation of primary cultures of myogenic cells in vitro more closely resembles in vivo development and these cells were used to confirm that the expression of the α7 cytoplasmic domain isoforms was developmentally regulated. RNA was isolated from cultures of differentiating newborn skeletal myoblasts at various times after these cells were placed in culture. The medium in the cultures was switched on day 2 to one that promotes differentiation. As seen in the results of RT-PCR analysis, expression of α7A accompanies myogenic differentiation whereas α7B is present earlier, in replicating myoblasts, and diminishes upon differentiation (Fig. 4). fu-1 cells, a developmentally defective line derived from L8E63 myoblasts, which have lost their normal control of proliferation, do not differentiate and are transformed and tumorigenic (Kaufman and Parks, 1977). Immunofluorescence (Kaufman et al., 1985) and northern analyses demonstrate that these cells express reduced α7 integrin (Song et al., 1992) and as shown here using RT-PCR, fu-1 cells express the α7B but not the α7A integrin chain (Fig. 4). This supports the conclusion that the shift in expression of α7B to α7A is part of the developmental process that accompanies the differentiation of skeletal muscle.
To detect the mRNA that encodes the α7C cytoplasmic domain RT-PCR was performed using an antisense primer that is specific to this form (Fig. 2A) and poly(A)+ RNA purified from L8E63 cells. The fragment produced by PCR was ddT-tailed, cloned into the pBluescript SK− plasmid and sequenced. This nucleotide sequence was identical to that obtained from the cDNA library, con-firming the expression of this alternate form of the α7 protein. RT-PCR detected the α7C mRNA in poly(A) + RNA prepared from myotubes but not myoblasts (Fig. 3B). This isoform of α7 mRNA appears to be present in relatively low amounts as it was detected only after two rounds of PCR.
These three forms of the α7 chain likely originate by alternate RNA splicing. The structures of the mRNAs that are consistent with the procedure used to generate these cDNAs are indicated in Fig. 2. We suggest that part of the diversity in function of the α7 integrin at different stages of muscle development is due to the developmentally reg-ulated expression of the alternate cytoplasmic domains. Additional studies on the α7B cytoplasmic domain were undertaken to further define its function in replicating myoblasts.
Comparison of the human and rat 7B cytoplasmic domains
A 1.9 kb clone of the 3′-end of α7 cDNA was identified in a cDNA library prepared from human fetal muscle, isolated, and sequenced. The nucleotide coding regions and inferred amino acid sequences of the rat and human α7B cytoplas-mic domains were determined to be 84% identical (Fig. 5). Two regions in the α7B integrin cytoplasmic domain are highly conserved between the human and rat sequences: these are the 32 amino acids directly on the carboxyl side of the GFFKR sequence (97% indentity between species) and the sequence of 21 amino acids that begins 4 amino acids from the carboxyl terminus and contains a DXHPX repeat (discussed below). All four threonine residues are maintained at the identical positions in the cytoplasmic domains. These identities in the human and rat α7B cyto-plasmic domains suggest that this integrin alpha chain has the same function in different species, and that the highly conserved portions likely play a significant role in mediat-ing this function.
Structural motifs in the 7B cytoplasmic domain
α-Helix
Computer analysis (Chou and Fasman, 1974; Garnier et al., 1978) of the α7B cytoplasmic domain sequence suggests an α-helical conformation comprised of approximately 25 amino acids (Fig. 5).
Serine/threonine protein kinase homology regions
The sequence, AVKIL(P)R, is present in the α-helical region of the α7B cytoplasmic domain. This same sequence is con-served in subdomain II, the catalytic site of serine/threonine protein kinases. Site-directed mutagenesis of the lysine residue (K) within this sequence results in the loss of protein kinase enzymatic activity (Hanks et al., 1986). The ATP-binding domain of these protein kinases have the consensus sequence LGXGXXGXV. The rat α7B sequence, LGXXGXXVXV, located towards the carboxyl end of the cytoplasmic domain, shares this homology. The human sequence, LGXXGXXGXG, also shares this homology.
DXHPX repeats in the α7 cytoplasmic domain
The sequence Asp/X/His/Pro/X, where X is a hydrophilic residue, is repeated three times in the α7B cytoplasmic domain. The most carboxyl repeat is within the potential ATP-binding site cited above. These repeats, and the three amino acids that separate them, are conserved 95% between the human and rat α7B integrins. The DXHPX repeat in the α7B cytoplasmic domain is unique amongst integrin alpha chains, moreover, it was not found in any sequence in cur-rent protein or nucleic acid data banks.
Receptor-like protein tyrosine phosphatase homology region
The α−helical region in the cytoplasmic domain also con-tains a sequence of 15 amino acids, TVPQYHAVK-ILREDR, that is 80% homologous with a region in the family of receptor-like protein tyrosine phosphatases (Gebbink et al., 1991). As noted above, the AVKIL motif within this sequence is also common and essential to the enzymatic activity of protein kinases. A sequence with 92% homology to PQYHAVKILREDR is also present in the cytoplasmic domains of the human α3B and α6B integrins, however, the lysine residue essential to protein kinase activity has been replaced by arginine in α3B and α6B whereas it is maintained in the human and rat α7B (Fig. 6).
Putative actin-binding site
A potential actin-binding site, QQFKEEK, that closely resembles the actin-binding site of villin (QQNLKKEK), is also found within the α-helix region. This sequence in villin is also within an α-helical domain, and mutation to QQN-LKEEK, which resembles α7 integrin even more closely, still has 80% actin-binding capacity (Friederich et al., 1992). The other two lysine residues are essential for actin-binding.
Structural changes in the 7B integrin cytoplasmic domain
A 26 amino acid segment of the cytoplasmic domain of the rat α7B chain (Fig. 5) was synthesized with an additional cysteine residue at the amino terminus. This portion of the cytoplasmic domain was chosen because its sequence is unique from that of other reported alpha chains and its potential high antigenicity (Hopp and Woods, 1981). The antiserum was evaluated for reactivity with α7 integrin by immunofluorescence and immunoblot analyses.
The major polypeptides in unreduced extracts of L8E63 myoblasts and myotubes that were reactive after elec-trophoresis in 8% polyacrylamide gels had mobilities corresponding to approximately 121,000 and 70,000 Da. These two bands represent the intact α7 chain and a product of one of two cleavage sites in the molecule. Upon reduction, the predominant band reactive with this antiserum migrated at approximately 38,000 Da (Fig. 7A). This peptide represents the carboxyl terminal peptide that can originate either from a single proteolytic cleavage of the intact α7B chain at the RRRRE site or from the 70,000 Da peptide (Fig. 7C). A small amount of the 121,000 Da protein persisted, indicating that all the α7B in the extract had not undergone proteolytic cleavage. The molecular mass of the 38,000 Da peptide deduced from its amino acid sequence is 26,000 Da and glycosylation at three potential sites in this peptide would decrease its mobility to approximately 38,000 Da (de Curtis et al., 1991; Parham et al., 1977). To demonstrate this, the integrin was treated with endoglycosidase F to remove carbohydrate. As a result, the mobility of the 38,000 Dapeptide changed to 26,000, confirming that one or more of the putative glycosylation sites inferred from amino acid sequence analysis (Song et al., 1992) is in fact used (Fig. 7B,C). The mobilities of the 121,000 and 70,000 Da polypeptides are also influenced by deglycosylation of the protein (Fig. 7B). Reactivity of the protein with the anti-cytoplasmic domain antiserum was inhibited by preincuba-tion of the antibody with the peptide, confirming the inden-tity of the polypeptides reactive with the antibody (Fig. 7A). In contrast with the results obtained by immunoblots, the results of immunofluorescence analyses of myoblasts and myotubes with this antiserum were negative or weak. In those experiments, the cells were treated either with 95% ethanol or 2% p-formaldehyde, both of which are routinely used to permeabilize these cells and render cytoplasmic pro-teins accessible to antibody. This failure of the cytoplasmic domain to react with the antiserum was apparently due to inaccessibility of the epitopes on the cytoplasmic domain, since immunoblots indicate that the α7 chain is present sub-sequent to fixation.
Previous experiments demonstrated that reactivity of the α7 chain with primary and secondary antibodies promotes the association of the integrin with the cell cytoskeleton, as noted by colocalization with actin filaments and a shift from being extractable in detergents such as Triton X-100 to becoming part of the detergent insoluble cytoskeletal net-work (Kaufman et al., 1985; Lowrey and Kaufman, 1989). The association of α7 with the cytoskeleton upon crosslink-ing is rapid and dependent on bivalent secondary antibody. This association of α7 with the cytoskeleton and change in its extractability suggested that upon crosslinking of the extracellular domain, the cytoplasmic domain may undergo a change in conformation or association with other proteins, and this might result in its accessibility to antibody. As seen in Fig. 8, reaction of α7 integrin on live cells with mono-clonal antibodies specific for the extracellular domain, fol-lowed by secondary antibody to crosslink the complexes, resulted in reactivity of the cytoplasmic domain with the antiserum. Antibodies reactive with other molecules in the membrane (A5, I3, H58 and H73; Kaufman and Foster, 1985) had no effect on accessibility of the α7B cytoplasmic domain. Specific binding of the anti-cytoplasmic domain antibody was inhibited by prior incubation with the immunizing peptide. In these experiments the cells were first reacted with the monoclonal antibody followed by secondary donkey antimouse IgG. The cells were then either fixed using 95% ethanol or extracted with 0.25% Triton X-100 and then fixed. In both cases, staining with the anti-serum was dependent on prior reactivity of cells with the primary and secondary antibodies, demonstrating that the accessibility of the cytoplasmic domain and shift in the capacity to extract α7 with detergent were commensurate and dependent on crosslinking by the antibodies.
In addition to crosslinking α7 in the membrane with anti-body reactive with the α7 chain, laminin promoted both the accessibility of the cytoplasmic domain to the anti-cyto-plasmic domain antibody and the shift in α7 extractability by detergent. The capacity of laminin to promote these changes was dose dependent and an excess of the extra-cellular matrix protein was suboptimal (Table 1) as would be expected if laminin too must bridge at least two α7 molecules. Thus crosslinking this integrin, either with antibod-ies or its ligand, promotes its association with the cell cytoskeleton and an alteration in the cytoplasmic domain. Monoclonal antibody reactive with laminin could also pro-mote this change to a limited extent. Exposure of the cells to 45°C or 56°C for 10 minutes, followed by fixation with 95% ethanol also rendered the cytoplasmic domain accessible to the antiserum, further indicating an unmasking of this region of the cytoplasmic domain by physical perturbations (Fig. 8; Table 1).
DISCUSSION
The α7β1 integrin on skeletal myoblasts is a laminin-binding protein. First identified with monoclonal antibody H36 (Kaufman et al., 1985), expression of the α7 integrin has been used as a marker of the development of the skeletal myogenic lineages (Kaufman et al., 1991; George-Wein-stein et al., 1993). The expression of α7 during the development of the primary and secondary muscle fiber lineages differs and this is believed related to the diverse functions and requirement of these cells at different stages of development (George-Weinstein et al., 1993). During the development of primary muscle, α7 is first seen on fibers, after terminal differentiation is initiated. This corresponds to the time these fibers envelop themselves in a laminin-rich extra-cellular matrix. It is within this lamina that secondary fiber formation subsequently takes place. In contrast with primary fiber formation, during the development of secondary fibers α7 is expressed on precursor cells. The α7β1 integrin may then serve to localize these precursor cells at the sites of secondary fiber formation and support the expansion of this population of cells. This is consistent with the findings that laminin selectively promotes a change in the shape and mobility of secondary myoblasts and maintains them proliferating (Foster et al., 1987; Ocalan et al., 1988). As α7β1 is the sole functional laminin-binding integrin on skeletal myoblasts (von der Mark et al., 1991), we suggest that it has a significant role in mediating these behaviors of myoblasts in a laminin-rich environment.
Upon terminal differentiation, secondary myoblasts cease replicating, decrease their mobility, align and fuse to form multinulceate fibers. Immunofluorescence, immunoblots, and northern analyses demonstrate that an increase in the expression of α7 integrin accompanies and is dependent on this stage of differentiation (Kaufman and Foster 1985; Kaufman et al., 1985; Song et al., 1992). Expression of the alternate A and C forms of the α7 chain cytoplasmic domain takes place at this stage in development, as does the switch in expression of numerous other isoforms of muscle proteins, such as creatine kinase, NCAM, actin, and myosin.
Newly formed fibers associate laterally into bundles and distally with tendon. The α7 integrin is localized along the membrane of mature muscle fibers (Song et al., 1992) and at myotendinous junctions (George-Weinstein and Kauf-man, unpublished results), thereby participating in the distal and lateral cohesion of muscle fibers needed for the directed generation of force and movement. The roles of the α7 inte-grin at these sites in functional muscle appear quite differ-ent from that in earlier myogenic development. We suggest that at least part of the functional diversity of the α7 chain results from the use of alternate cytoplasmic domains.
The cytoplasmic domains of the 14 known alpha inte-grins have common as well as distinct features that undoubtedly underlie the function of these proteins. At the most amino terminal end of the cytoplasmic domains the alpha chains have a common sequence, GFFKR. This region may be essential for the stable association of α and β chains (Solowska et al., 1991). Some divergences in this region are in the GFFDR sequence of the chick α8 integrin (Bossy et al., 1991), the GFFNR sequence of the Drosophila PS2 chain (Bogaert et al., 1987), the DFFKP sequence of human α3B (Takada et al., 1991) and as reported here, the GFFRR of α7A and GFFKC of the α7C form. The α3B, α6B and α7B cytoplasmic domains have the common amino acid sequence, P-YHAV-I--E-R, which is highly homologous to a region in receptor-like protein tyrosine phosphatases. Whatever function this motif may serve is likely consistent in all these proteins. The α3A, α6A and α7A cytoplasmic domains differ from the respective B forms in this region, but they are similar to each other, indicating that the functional changes between the A and B forms of α3, α6 and α7 may be conserved. The remaining portions of these alpha chain isoforms and all other inte-grin α chain cytoplasmic domains are generally quite diverse from one another. Since the unique sequences of the cytoplasmic domains of the same integrin isoforms are highly conserved between species, it is reasonable to conclude that they are functionally significant. As demonstrated here, there is 84% identity in the human and rat α7B cyto-plasmic domains.
The potential roles of the alpha chain cytoplasmic domains are now coming under experimental scrutiny. Deletion or replacement of an alpha chain cytoplasmic domain can alter the binding affinity of the heterodimer to its ligand (O’Toole et al., 1991b) and alter its capacity to promote collagen gel contraction (Chan et al., 1992). These results imply that the cytoplasmic domain may confer specific functions to the integrins and perhaps participate in signal transduction, both into the cell and to the extra-cellular ligand-binding regions of the molecule. The expression of integrin alpha chain isoforms with different cytoplasmic domains at different stages of development (Cooper et al., 1991 and Tamura et al., 1991) further suggests that isoforms with different cytoplasmic domains may mediate specific functions during development. The developmentally regulated expression of multiple α7 cyto-plasmic domains supports this.
The mechanism of formation of the α3, α6, and α7 A and B alternate cytoplasmic domains appears to be similar. The B forms originate from a splice site located 5′ to a second GFFKX coding region, that is downstream from the A form GFFXX coding regions (Cooper et al., 1991; Hogervorst et al., 1991; Tamura et al., 1991). In the case of α7A and α7B the 3′-splice site is followed by a common coding region in the A and B forms that is used in alternate reading frames. In contrast, the α3 and α6 3′-splice sites are in the 3′-untranslated regions and there is no common coding sequence in these A and B cytoplasmic domains. α7C originates from an alternate splice that is common to a single GFFKX coding region and the 3′-untranslated region of α7B. Identical splice site junctions, AAGTGTG, are found in α3B, α6B and α7C, at the junctions of the transmembrane and cytoplasmic domains. The greatest homology between the α7B integrin and other α chain isoforms is with α6B: there is 47% identity in a 1,047 amino acid overlap, 70% identity in the transmembrane domains and 34% identity (and 77% homology) in a 56 amino acid overlap in the cytoplasmic domains. This suggests a common ancestry between the genes that encode these proteins, although the α7B cytoplasmic domain appears to have evolved greater functional diversity.
Homologous phosphotransfer and ATP-binding sequences in the α7B cytoplasmic domain suggest that it may have serine/threonine kinase activity, however, the cat-alytic activity of these kinases is localized in domains that are significantly larger than the 77 amino acids of the α7B cytoplasmic domain. Alternatively, these motifs and the phosphatase homology sequence in the α7B cytoplasmic domain, may modulate the activity or localization of the respective enzymes, or other proteins that bind to them, at the inner periphery of the cell. The maintenance of the receptor-like protein tyrosine phosphatase homology region in α3B, α6B and α7B suggests that whatever function it serves is conserved in the integrins that contain any of these alpha chains. In contrast, only the α7B sequence contains the crucial lysine residue essential to serine/threonine pro-tein kinase activity and both the phosphotransfer and ATP-binding motifs. It remains to be determined if α7B has enzy-matic activity, perhaps as a subunit of a larger complex, or if it has other regulatory functions.
The ATP-binding motif in the α7B cytoplasmic domain contains and is adjacent to two additional DXHPX repeats. This threefold repeat is not evident in other sequences in current data banks. The 21 amino acid stretch that contains these repeats is highly hydrophobic and is 90% identical in human and rat.
Although we have not demonstrated that any of these motifs are functional within the α7B cytoplasmic domain, they do present a rich potential for participating in the transduction of signals initiated outside the cell. Several interesting observations suggest that integrins are intimately involved in signal transductions in which phosphorylation and association with actin are certain to have key roles. The different potential tyrosine and serine/threonine phosphorylation sites in the three α7 cytoplasmic domains could underlie diverse mechanisms of signal transduction at different stages of muscle development. The localization of integrin-regulated kinase activity (pp125FAK) has recently been demonstrated at focal adhesions (Schaller et al., 1992), and tyrosine phosphorylation of pp125FAK appears to be activated by integrin and dependent on actin filaments (Lipfert et al., 1992). A single tyrosine residue in a protein kinase phosphorylation consensus site is present in all three α7 cytoplasmic domains. Tyrosine phosphorylation of other cytoplasmic proteins also appears to be initiated through integrins and this may be enhanced in transformed cells. Phosphorylation of tyrosine in the integrin β1 chain cytoplasmic domain in cells expressing pp60v-src leads to decreased association of the integrin with talin inside the cell and decreased association with extracellular matrix (Tapley et al., 1989; Horvath et al., 1990). As a consequence, these transformed cells become rounded. Modul-ation of the interactions of cells and extracellular matrix also take place during normal processes such as cell migra-tion and cell division and these interactions too may be mediated by tyrosine phosphorylation (Hynes, 1992). The intimate association of α7 and β1 and the potential phos-photransfer activity, or kinase- and phosphatase-modulat-ing capacities of the α7B cytoplasmic domain, suggest that regulation of cell adhesion, mobility, shape and prolifera-tion of myoblasts may take place through this integrin com-plex. In contrast with myogenic cells that exhibit normal expression of the α7 integrin and differentiate, mutants that are deficient in α7 have often lost their ability to control replication, are transformed and tumorigenic (Kaufman et al., 1985; Foster and Kaufman, 1985) and do not express α7A. Association of extracellular matrix proteins with these cells is also altered. Thus both expression of α7B and the switch to α7A and α7C appear to be important to myogenic differentiation. Immunologic reagents specific to the α7A and α7C forms will be used to confirm their expression and localization.
Immunolocalization of α7 in the cell membrane and crosslinking α7 integrin with antibodies reactive with its extracellular domain renders the α7B cytoplasmic domain accessible to an antiserum raised against an immunogenic portion of its sequence. It should be noted that this crosslinking does not merely serve to aggregate these recep-tors and render them more readily detectable. The α7 chain can be detected by immunofluorescence on cells processed at 4°C or fixed with 1% p-formaldehyde to prevent changes in its native distribution (Kaufman et al., 1985; Lowrey and Kaufman, 1989). Of more physiologic significance, incu-bation of cells with laminin also promotes unmasking within the cytoplasmic domain and this suggests that occu-pancy of the receptor in vivo also modulates such changes in the protein. It is highly likely that the capacity of laminin to promote the association between integrin and the cell cytoskeleton is significant to the changes in mobility, shape and proliferative state of these cells growing in a laminin environment. The actin-binding motif in the α7B cytoplas-mic domain is similar to that in villin (Friederich et al., 1992) and may be directly involved in this association. The change in reactivity of the cytoplasmic domain with anti-body promoted by antibody or ligand may result from alter-ing the association of the α7 cytoplasmic domain with other proteins, and/or, from a conformational change initiated in the extracellular portion of the molecule. In either case, modulation of the structure of the integrin cytoplasmic domain initiated outside the cell likely results in the phys-iologic responses of these cells. Similarly, activation of the platelet αIIbβ3 integrin by thrombin or collagen, or by anti-bodies reactive with the receptor, leads to a change in con-formation in the β chain that also results in its accessibil-ity to an antibody (Shatill et al., 1985; Gulino et al., 1990; Kouns et al., 1990; Andrieux et al., 1991; O’Toole et al., 1991a). Fab′ fragments of secondary antibodies are inef-fective at promoting association of α7 with the cytoskele-ton, indicating that bridging at least two α7β1 heterodimers is necessary to alter its cytoplasmic domain and direct its interaction with the cell cytoskeleton (Lowrey and Kauf-man, 1989). This suggests that some signals are transduced into the cell by a mechanism in which integrin crosslink-ing and association with the cytoskeleton are important fac-tors. As discussed, the α7B cytoplasmic domain does con-tain a rich potential for participating in the transduction of signals initiated outside the cell. Further defining the roles of the motifs in the α7B cytoplasmic domain and the diver-sity in functions that arise from use of the alternate cyto-plasmic domains is of great interest to understanding the significance of extracellular matrix and integrins in the development of skeletal muscle, and the mechanisms and molecules involved in matrix induced signal transduction.
ACKNOWLEDGEMENTS
We thank Mr Maojian Gu for his skillful assistance with the endoglycosidase experiments and Dr George Dickson for gener-ously providing the human fetal muscle cDNA library. We also thank Dr Vito Quaranta for allowing us to discuss results from his laboratory prior to their publication. This work was supported by National Institutes of Health grant GM28842. The α7A, α7B and α7C nucleotide sequence data are available from the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under acces-sion numbers X74293, X74295 and X74294, respectively.