The three human syntrophin genes are expressed in diverse tissues, have distinct chromosomal locations, and each bind to dystrophin and its relatives.

The syntrophins are a biochemically heterogeneous group of 58-kDa intracellular membrane-associated dystrophin-binding proteins. We have cloned and characterized human acidic (α1-) syntrophin and a second isoform of human basic (β2-) syntrophin. Comparison of the deduced amino acid structure of the three human isoforms of syntrophin (together with the previously reported human β1-syntrophin) demonstrates their overall similarity. The deduced amino acid sequences of human α1- and β2-syntrophin are nearly identical to their homologues in mouse, suggesting a strong functional conservation among the individual isoforms. Much like β1-syntrophin, human β2-syntrophin has multiple transcript classes and is expressed widely, although in a distinct pattern of relative abundance. In contrast, human α1-syntrophin is most abundant in heart and skeletal muscle, and less so in other tissues. Somatic cell hybrids and fluorescent in situ hybridization were both used to determine their chromosomal locations: β2-syntrophin to chromosome 16q22-23 and α1-syntrophin to chromosome 20q11.2. Finally, we used in vitro translated proteins in an immunoprecipitation assay to show that, like β1-syntrophin, both β2- and α1-syntrophin interact with peptides encoding the syntrophin-binding region of dystrophin, utrophin/dystrophin related protein, and the Torpedo 87K protein.

Dystrophin, the protein product of the Duchenne muscular dystrophy locus, is a large membrane-associated cytoskeletal protein (1). In order to understand the function of this protein in skeletal muscle, it is important to establish the molecular organization of dystrophin in the context of the membrane cytoskeleton. Dystrophin copurifies with a group of integral membrane glycoproteins and membrane-associated proteins called the dystrophin glycoprotein complex (2)(3)(4). A number of these proteins have been further defined by their primary sequence and their biochemical properties.
A 58-kDa cytoplasmic peripheral membrane protein was independently identified in the Torpedo electric organ, and shown to localize to the postsynaptic neuromuscular junction in mammals (5). This 58-kDa synaptic protein also copurifies with dystrophin and is now known as syntrophin (6 -9). Dystrophinassociated syntrophin isolated from rabbit skeletal muscle is heterogeneous; it appears as a triplet by one-dimensional SDSelectrophoresis, and when separated by two-dimensional gel electrophoresis, appears as two clusters of 58-kDa proteins with different isoelectric points (pI), one which is slightly acidic (␣, pI ϭ 6.4) and the other which is quite basic (␤, pI ϭ 9) (10). Phosphatase pretreatment of the isolated microsomes results in some signal consolidation (10), and phosphoamino acid analysis of syntrophin isolated from Torpedo electric organ shows that serine and tyrosine residues are phosphorylated (11).
The isolation of two distinct isoforms of syntrophin in mouse (9), and antibodies to a single cloned isoform of rabbit syntrophin (12), confirmed the biochemical evidence that there are at least two distinct genes. Based upon partial peptide sequences from purified rabbit muscle syntrophin, we independently isolated human ␤1-syntrophin cDNA which was also used to identify a distinct but related human muscle expressed sequence tag (EST), 1 EST25263 (13). The deduced amino acid sequence of this human EST fragment was nearly identical to a portion of mouse ␤2-syntrophin (9). From all the available sequences, we proposed that there are at least three syntrophin genes in the mammalian genome. From their predicted amino acid sequences and their calculated pI values, the acidic isoform was named ␣1-syntrophin, and the two basic isoforms ␤1-syntrophin and ␤2-syntrophin (see Table I).
The widely expressed C-terminal product of the DMD gene, dystrophin protein of 71 kDa (Dp71), the dystrophin related protein (DRP or utrophin), and the 87K relative of dystrophin also copurify with syntrophin when isolated by immunoaffinity techniques (6,8,14,15). The suggestion that dystrophin interacts with syntrophin via its C terminus was independently determined by blot overlay of dystrophin onto isolated syntrophin (16,17). Recombinantly produced ␤1-syntrophin interacts with a small region within the C terminus of dystrophin, revealing a strong binding site within exon 74 of dystrophin (18). This result was independently determined by using bacterially expressed or in vitro translated portions of dystrophin to overlay onto purified syntrophin bound to a solid support and to show that syntrophin may have yet another binding site in a more distal region on dystrophin (19,20). In addition, ␤1syntrophin was shown to also interact with the homologous * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U40571 and U40572.
ʈ To whom reprint requests should be addressed.
regions of utrophin/DRP and the Torpedo 87K protein (18). Despite their mRNA expression in a wide variety of tissues (9,12,13), the syntrophin isoforms appear to have a remarkable specificity in their submembranous localization in muscle. Isoform-specific antibodies discriminate the localization of ␣1syntrophin, which is expressed throughout the sarcolemma, from ␤2-syntrophin, which localizes specifically to the neuromuscular junction (NMJ) (21). The determinants of the isoform-specific localization are presumably due to structural differences between these isoforms, but these are yet to be determined.
Gibson and colleagues have noted that the syntrophins contain two pleckstrin homology (PH) domains (22), a small ϳ100residue domain originally found as an internally duplicated motif in pleckstrin, the major substrate of protein kinase C in erythrocytes (23,24). This domain has captured wider attention because it is also found in a number of other intracellular signaling and cytoskeletal proteins, such as ␤-spectrin, phospholipase C␥, the ␤-adrenergic receptor kinase, a number of GTPases, and GTPase-activating proteins, many of which are membrane-associated (22). The deduced three-dimensional structure of the N-terminal pleckstrin domain and the PH region of ␤-spectrin has been determined (25,26). A hydrophobic lip of the pleckstrin ␤-barrel has been shown to bind to phosphatidylinositol 4,5-bisphosphate, which may explain how many PH-containing proteins are associated with the membrane without containing classical membrane-anchoring groups (27).
Adams and colleagues (28) have noted the homology between a conserved region in the middle of the syntrophin genes and a number of other membrane-associated proteins, the PDZ domain, named for the Post-synaptic density protein-95 (PSD-95, (29)), the Drosophila discs large tumor suppresser protein (30), and the Zonula occludens-1 protein (ZO-1 (31)). Since this motif is shared among these intracellular peripheral membrane proteins, it may also be the basis by which the syntrophins interact with another component of the membrane or membrane cytoskeleton.
We report here the cloning and characterization of human ␤2-syntrophin and ␣1-syntrophin. By comparing the amino acid sequences of human ␤1-, ␤2-, and ␣1-syntrophin, we have identified the C-terminal 57 amino acids of syntrophin as a conserved syntrophin-unique domain. The mRNA of ␤2-syntrophin is expressed in a wide variety of tissues, whereas ␣1syntrophin is predominantly expressed in striated muscle. The human chromosomal sublocalization of ␤2-syntrophin is 16q23-24, and that of ␣1-syntrophin is 20q11.2. We have also verified the functional conservation of ␤1-, ␤2-, and ␣1-syntrophin in their ability to interact with dystrophin and its relatives in an in vitro binding assay.
For ␣1-syntrophin, the oligonucleotide pair 5Ј-TGG GAT CCA GGA CAT CAA GCA GAT TGG CT-3Ј and 5Ј-GTG AAT TCC CGT GCG CAG GGC AAA GGA GA-3Ј, amplified a reverse-transcribed cDNA template from human adult muscle, using 100 ng of template DNA and 100 pmol of each primer, with 30 cycles of the following conditions: 55°C for 1 min, 72°C for 10 s, and 94°C for 1 min. Amplified DNA was separated by 6% polyacrylamide gel electrophoresis, and the 250-bp fragment was visualized by staining with ethidium bromide (32).
cDNA Isolation and Sequencing-PCR probes were prepared incorporating [␣-32 P]dCTP to screen a gt10 cDNA library (13). In each round of screening, 10 6 phage were plated and replicated onto nitrocellulose filters and hybridized by standard methods (32). Positively hy-bridizing plaques were plaque-purified and the inserts subcloned into pBluescript II ϩ SK or KS.
cDNA clones of ␤2-syntrophin were obtained by screening a human adult brain library with the PCR probe that corresponded to EST25263 (above). The phage 19-1 was isolated and the insert subcloned into the EcoRI site of Bluescript II ϩ SK for sequencing. The DNA of this partial cDNA clone was then isolated and radiolabeled (OLB labeling kit, Boehringer Mannheim) to screen a human adult muscle cDNA library. The clones HAM1 and HAM12 were subcloned into Bluescript for sequencing.
cDNA clones of ␣1-syntrophin was obtained by screening a human adult heart (left ventricle) cDNA library (33) with a PCR probe prepared from the primers described above. Of the six resulting clones, all were subcloned into plasmid, shown to map similarly, and partially sequenced for verification. The clones LV31-1 and LV6-2 were sequenced entirely.
The entire sequence of both strands were obtained by the dideoxy nucleotide chain-termination method, with either Sequenase T7 polymerase (U. S. Biochemical Corp.) or an ABI automated sequencer with Taq DNA polymerase (Perkin-Elmer). Sequences were analyzed and aligned using the GCG software suite (Wisconsin University) in their default settings. Gaps determined pairwise identity scores, PileUp produced multiple alignments (Figs. 1-3), and ProteinStructure made local secondary structure predictions.
The mRNA Expression of the Syntrophins-A panel of poly(A) ϩ RNA from human tissues blotted onto positively charged membrane (Clontech) was hybridized to PCR probes using standard conditions of high stringency (32). For ␤2-syntrophin, the primer pair 5Ј-GGA AAA CAG ATT GAT AGA GCT ACA TTC-3Ј and 5Ј-CGA GAG CCT GTC CTG GTA GCA AAT-3Ј amplified a 401-bp fragment from the open reading frame of clone 19-1. In addition, the gel-purified insert of clone 19-1 was radiolabeled (OLB, Boehringer) and used in a separate hybridization of an identical blot. For ␣1-syntrophin, the primer pair described above in "PCR cloning" was used to amplify the 250-bp open reading frame region of ␣1-syntrophin from the subcloned PCR fragment.
Somatic Cell Hybrid Mapping of ␣1-Syntrophin-Based upon the sequence of the PCR product for ␣1-syntrophin, another pair of internal primers was designed to amplify the human sequence specifically, so that only the human sequence would be amplified from human-rodent somatic cell hybrids. The DNAs from the NIGMS monochromosomal hybrid collection (35,36) were amplified in pools as described by Beck (36) by PCR using the primers 5Ј-GAA AAG GAA CTG CTC CTC TAC TTG-3Ј and 5Ј-GAG TGG GGC AGT ACG GGC TGG CCG-3Ј. After a 5-min denaturation in 94°C, 30 PCR amplification cycles were performed in a 50-l reaction volume with 200 ng of genomic DNA, 50 pmol of each primer, and 2.5 units of Taq DNA polymerase. The annealing, elongation, and denaturation cycles were the same as for the PCR cloning of ␣1-syntrophin. The 78-bp reaction product was separated by 7% polyacrylamide gel electrophoresis and detected with ethidium bromide.
Immunoprecipitation of in Vitro Translated Syntrophin, Dystrophin, DRP, and Torpedo 87K Protein-The translation of the syntrophin binding site of dystrophin, DRP, and 87K protein, and their use in the co-precipitation of ␤1-syntrophin was described in detail previously (18). The dystrophin peptides C2979 and Dp71D110, the utrophin/DRP peptide TDR3, and the full-length 87K peptide T87 were translated from their respective expression vectors without the presence of [ 14 C]leucine; we also translated the exon 74 containing region of dystrophin or its homologous region of DRP or the Torpedo 87K protein as fusion proteins with the FLAG octapeptide (IBI/Kodak, New Haven, CT). We showed previously that a translated polypeptide that corresponds to the C-terminal two-thirds of ␤1-syntrophin, amino acids 204 to 538, coprecipitates with all three FLAG fusion proteins with the anti-FLAG monoclonal antibody M2 (IBI/Kodak). In the present experiments, the ␣1-syntrophin partial cDNA LV6-2 and the ␤2-syntrophin partial cDNA 19-1 were cloned into the EcoRI site of the in vitro transcription/translation vector pTR3 (18) and translated in the presence of [ 14 C]leucine. These constructs permit the production of truncated syntrophin polypeptides homologous to the truncated ␤1-syntrophin peptide used previously (18). The ␤2-syntrophin C-terminal expression construct (T␤2S-14) expressed amino acids 208 to 540, and the ␣1-syntrophin C-terminal expression construct (T␣6) expressed amino acids 157 to 505. Five to 10 microliters of the two translation reaction mixtures, unlabeled peptides (dystrophin, DRP, 87K) and radiolabeled syntrophin peptides, were combined in Tris-buffered saline (pH 8.0) with 0.1% Tween 20 (TBST), then incubated and immunoprecipitated with 10 l of the respective antibody: anti-dystrophin antibody d11, the anti-DRP antibody BH3, or anti-87K monoclonal antibody AA4.1 (kindly provided by Jonathan B. Cohen) and protein G-Sepharose (Sigma). FLAG fusion proteins were used to co-precipitate ␤2and ␣1-syntrophin as described for ␤1-syntrophin. After three 1-ml washes with TBST, the pellets were resuspended in 20 l of protein loading buffer, boiled, and separated by SDS-polyacrylamide electrophoresis, fixed, dried, and visualized by PhosphorImager autoradiography (18).

RESULTS
Cloning of Human ␤2-Syntrophin-Having previously amplified a PCR product corresponding to EST25263 (13), we were able to radiolabel this product and isolate full-length cDNA clones (see "Materials and Methods"). A brain-derived clone, 19-1, and two muscle clones, HAM1 and HAM12, were subcloned into plasmid and sequenced. Of these three overlapping clones, 19-1 and HAM12 are shorter, spanning the 3Ј-half of the open reading frame (Fig. 1A). The 5Ј end of HAM1 contains 223 bp of chimeric cDNA corresponding to a portion of the HHR23A protein (39), followed by 20 bp of 5Ј-untranslated region that is highly similar to the 5Ј-untranslated of mouse ␤2-syntrophin reported in Ref. 28.
The ␤2-syntrophin candidate HAM1 contains a single large open reading frame (Fig. 1A), which begins with an ATG start codon in a favorable context for the initiation of translation, and is flanked at the 3Ј end with a polyadenylation signal at the appropriate distance from a poly(A) tail (GenBank accession no. U40572). The ATG start codon is 45 nucleotides upstream of another in-frame ATG start codon which is the initiation codon in mouse ␤2-syntrophin (28). In the mouse ␤2-syntrophin gene, the codon corresponding to the first human ATG is ATC, but the former ATG codon in human is in an extremely favorable context for the initiation of translation (40). The deduced peptide is 540 amino acids in length, 58,000 in molecular weight, and has a pI of 9.4. Its amino acid sequence is 96% identical to its mouse homologue (Fig. 1B) (9,28).
Cloning of Human ␣1-Syntrophin-Based upon the published sequence of ␣1-syntrophin in mouse and rabbit, a pair of PCR primers (see "Materials and Methods") were designed to amplify a conserved region of the cDNA from a human muscle cDNA template. This PCR product was subcloned into plasmid, sequenced to confirm the specificity of the reaction, and then used as a template to screen a human left ventricle cDNA library. The six resulting clones yielded five partial cDNAs and a single full-length cDNA. The full-length cDNA, LV31-1, is 2136 bp long and encodes a single large open reading frame. The cDNA clone LV6-2 spans a 3Ј portion of the open reading frame ( Fig. 2A; GenBank accession no. U40571).
The deduced peptide is 505 amino acids in length, predicted to be a molecular mass of 54 kDa, and have a pI of 6.4 (Fig. 2B). The open reading frame begins with the first ATG start codon in the cDNA, which is in a favorable context for the initiation of translation, and is flanked at the 3Ј end with a polyadenylation signal at the appropriate distance from a poly(A) tail. At the amino acid level, this human isoform is 94% identical to the published mouse sequence and 93% identical to the published rabbit sequence. All three sequences contain homologous start codons. In comparison to rabbit and human, the mouse cDNA bears an internal deletion of 6 amino acids near its N terminus (GAPREQ). The 4-amino acid internal insertion in mouse (SSAH) is considered to represent a rare splicing event to a nearby splice acceptor (28).
Comparison of the Three Human Syntrophins-The deduced amino acid sequence of human ␤2-syntrophin is 57% identical to human ␤1-syntrophin. The human ␣1-syntrophin peptide sequence is 54 and 50% identical to human ␤1and ␤2-syntrophin, respectively. Our alignment of three human syntrophin isoforms indicates the two tandem pleckstrin homology domains as aligned in Ref. 22 and represents them schematically as well (Fig. 3, A and B). The first PH domain is split into two The EST25263 sequence was used as the basis for cloning 19-1 from a human adult brain cDNA library. The 19-1 clone was used as the probe to isolate HAM1 and HAM12 from a human adult muscle cDNA library. The region HHR is a chimeric sequence from an unrelated gene (see "Results"). B, interspecies comparison of the deduced amino acid sequence of the ␤2-syntrophins, between human and mouse. Includes data of the N terminus of mouse ␤2-syntrophin reported (28). A dash indicates identity to the residue at that position for the human homologue, and a space indicates gaps introduced by PileUp software to optimize the alignment (see "Materials and Methods"). The clone 19-1 contains an open reading frame from the amino acid under the symbol "Ͼ." regions (PH1a and PH1b) by a large region of variable sequences flanking a core of close homology among the three syntrophins. This core of high homology, the PDZ domain, is aligned as by Adams and colleagues (28) (Fig. 3A, black box), and is predicted to have mainly ␣-helical secondary structure by Chou-Fasman and Garnier-Osguthorpe-Robson analyses (see "Materials and Methods").
The C-terminal 57 amino acids (Fig. 3 in gray) also forms a region of strong homology among the three human syntrophins, but does not have homology to other characterized proteins. This 57-amino acid sequence has been labeled the syntrophinunique domain, and is predicted by Chou-Fasman and Garnier-Osguthorpe-Robson analysis to consist of from three to five strands of ␤-sheet separated by as many turns (see "Materials and Methods").
Tissue Expression of the Syntrophin mRNA-Hybridization to Northern blots of a series of human tissues reveals that each of the syntrophins are expressed distinctly (Fig. 4). We have previously reported the expression of ␤1-syntrophin (13), but show it here for comparison (Fig. 4A). The Northern analysis of ␤2-syntrophin shows that it also has a wide ranging pattern of expression, but that it is expressed in relatively low but even levels throughout all tissues (Fig. 4B). It also has at least three distinct transcript classes, of 10, 5, and 2 kilobase pairs. The diverse distribution of distinct transcript classes is also observed in the mRNA expression of mouse ␤2-syntrophin as well (9). Neither our PCR amplifications nor our cDNA clones reflect any differences in the open reading frame that would account for this size heterogeneity.
A similar hybridization with ␣1-syntrophin probe reveals a distinct pattern of expression from that of the ␤-syntrophins (Fig. 4C). A single 2.5-kilobase pair transcript is expressed in relatively high levels in both skeletal muscle and heart, with some low level expression in brain, pancreas, liver, kidney, and lung, and none detected in placenta.
Chromosomal Localization of Syntrophins-Using the same NIGMS somatic cell hybrids that were used to map ␤1and ␤2-syntrophin (13), we determined the chromosomal location of ␣1-syntrophin. Using a specific pair of oligonucleotides (see "Materials and Methods"), a 780-bp PCR product was amplified in pools of human-rodent somatic cell hybrids containing chromosome 20 (not shown). DNA from the cell lines was then used individually to further confirm that this PCR product uniquely amplifies from chromosome 20-derived cell lines, and not from cell lines containing other members of the pool or either of the other two syntrophin genes, chromosomes 8 or 16 (Fig. 5). The 78-bp product is amplified in human genomic DNA, but not in DNA isolated from rat or hamster cell lines. Furthermore, two independent somatic cell lines, each containing human chromosome 20, amplify the specific PCR product.
Human genomic clones of both ␤2-syntrophin and ␣1-syntrophin isolated from an EMBL3 human genomic library were used for FISH analysis to independently confirm the mapping panel results (see "Materials and Methods"). The ␤2-syntrophin signal localized to the region between 16q23 and 16q24 (Fig. 6A), and ␣1-syntrophin uniquely localized a signal to 20q11.2 (Fig. 6B). No secondary hybridization signals were consistently seen to suggest other closely related loci elsewhere in the genome.
The syntrophin proteins used in this assay consistently showed a low level rate of aggregation that is also seen in the control lanes (Fig. 7, A, lanes 4, 5, 9, and 10, B, lanes 4 and 8).
To address this issue, the exon 74 homologous regions of dystrophin, utrophin, and 87K protein were produced as FLAG fusion peptide as reported previously for coprecipitation with ␤1-syntrophin (18). All three proteins were able to coprecipitate both ␤2- (Fig. 7B, lanes 1-3) and ␣1-syntrophin ( Fig. 7B lanes  5-7). A background of nonspecific aggregation of ␤2and ␣1syntrophin was seen regardless of whether a specific antiserum or the anti-FLAG monoclonal antibody was used. This back-ground is variable from different experiments but is never higher than the specific coprecipitation reactions. DISCUSSION In this report we conclusively confirm our previous hypothesis that there are three distinct but homologous human syntrophin genes. Their biochemical and genetic characteristics are summarized in Table I. We have also shown that these homologous proteins are functionally conserved with respect to their in vitro binding properties to dystrophin, utrophin, and the 87K protein.
Comparisons of the three human syntrophins to each other, as well as those sequences available in mouse and rabbit, demonstrate that for a particular isoform of syntrophin there is a high degree of interspecies conservation, with 96% identity for ␤2-syntrophin and at least 93% for the three mammalian ␣1-syntrophins. In contrast, the three human syntrophins are less strongly conserved with respect to each other. The ␣1syntrophin is 54 and 50% identical to its ␤1and ␤2-syntrophin counterparts, respectively, and the ␤-syntrophins are only 57% identical to each other.
The syntrophins contain two tandem PH domains (Fig. 3A, in plain box) (22). The first PH domain is interrupted by a 162-to 182-amino acid region in which 80 amino acids are highly conserved among the three syntrophins (Fig. 3A, in black box). Adams and colleagues (28) have noted the homology between this conserved region and a number of other membrane-associated proteins, the PDZ domain. The PH and PDZ domains, either together or individually, may determine the specific membrane localization of syntrophin, either directly to a lipophilic membrane component (27) or via an integral membrane protein (41).
In comparing the three human syntrophins, we have found that the C-terminal 57 amino acids of the three proteins are highly homologous to each other (Fig. 3A, in gray). This region is predicted to contain as many as five strands of ␤-sheet, and because it appears to be a unique motif among other known proteins, we have called it the syntrophin-unique domain. It is possible that this 57-amino acid syntrophin-specific domain subserves syntrophin's specific interaction with dystrophin and its relatives. To this point, we have shown that the C-terminal two-thirds of the three translated syntrophins can coprecipitate with the exon 74 region of dystrophin and its relatives (Ref. 18 and Fig. 7). These polypeptides do not contain the domain PH1a and the PDZ region, but do contain PH1b, all of PH2, and the syntrophin-unique domain (see Fig. 3). Further functional analysis of syntrophin structure will be necessary.
Dystrophin and utrophin/DRP also have distinct localizations to the muscle membrane, with dystrophin distributed throughout the sarcolemma and utrophin/DRP found mainly at the neuromuscular junction (42). Since mouse ␣1-syntrophin is distributed throughout the membrane and ␤2-syntrophin is found at the neuromuscular junction (21), we hypothesized that FIG. 7. ␤2and ␣1-syntrophin interact with dystrophin and its relatives. A, translated peptides of dystrophin (dys), utrophin/DRP (drp), and the Torpedo 87K protein (87K) are used to coprecipitate ␤2-syntrophin (lanes 1-5) and ␣1-syntrophin (lanes 6 -10) peptides. The syntrophin peptides were also combined with Dp71⌬110 (18), which lacks the syntrophin binding region (lanes 4 and 9), or with d11 antibody alone (lanes 5 and 10). B, FLAG fusion proteins of the syntrophin binding domains of dystrophin (1 and 5), utrophin/DRP (2 and 6), and the Torpedo 87K protein (3 and 7) are used to coprecipitate translated partial cDNAs of ␤2-syntrophin and ␣1-syntrophin. In the control lane (4 and 8) an identical precipitation was performed in the absence of FLAG fusion protein. these two isoforms of syntrophin had unique binding properties to the respective dystrophin and utrophin/DRP proteins. The finding that all three syntrophins can each bind to dystrophin and its relatives in vitro falls short in providing some clue as to how either the syntrophins or dystrophins can localize to different specializations of the sarcolemma. The coprecipitation procedure used here (Fig. 7) does not quantitatively address this question. However, the differences among the three syntrophins, which are especially marked in the connecting loops to the PDZ domain (Fig. 3A), may reflect the specialization of these individual genes to a particular function, such as to interact with another protein, or as determinants of their distinct subcellular localization. At the level of the mRNA, the ␤-syntrophins share a common characteristic in that they give rise to a set of transcript classes (Fig. 4). In contrast to ␤1-syntrophin, whose five transcript classes are most abundant in liver, ␤2-syntrophin transcripts are more homogeneously expressed, most abundant in lung, and have three transcript classes (Fig. 4B). These results are similar to those found in mouse ␤2-syntrophin, but the relative abundance of ␤2-syntrophin in human brain is much lower than that observed in mouse (9). In the cDNA clones that we have isolated so far, we have not noticed any large differences in the sequences among the clones that can account for these alternative forms, nor do the other reported cDNAs from rabbit and mouse (9,12,28).
The ␣1-syntrophin transcript is expressed as a single-sized transcript of 2.5 kilobase pairs (Fig. 4C), and is strongly dominant in cardiac and skeletal muscle. This representation of ␣1-syntrophin mRNA expression shows somewhat more expression in extramuscular tissues than that reported in mouse and rabbit tissues previously (9,12), but may be attributable only to a higher sensitivity in detection.
The tissue distribution of the respective syntrophin mRNA also allow us to hypothesize what kinds of inherited disorders may be caused by defects of these genes. The sublocalization of ␤2-syntrophin to 16q23-24 (Fig. 6A), and its widespread tissue distribution suggests that a defect of this gene would have consequences in multiple organs. Because ␣1-syntrophin is so abundantly expressed in striated muscle, we would predict that a defect of this gene would be more inclined to result in a myopathic phenotype. The question of whether this 20q11.2encoded gene (Fig. 6A) is linked to any autosomal neuromuscular diseases is currently under investigation.