γ1- and γ2-Syntrophins, Two Novel Dystrophin-binding Proteins Localized in Neuronal Cells*

Dystrophin is the scaffold of a protein complex, disrupted in inherited muscular dystrophies. At the last 3′ terminus of the gene, a protein domain is encoded, where syntrophins are tightly bound. These are a family of cytoplasmic peripheral membrane proteins. Three genes have been described encoding one acidic (α1) and two basic (β1 and β2) proteins of ∼57–60 kDa. Here, we describe the characterization of two novel putative members of the syntrophin family, named γ1- and γ2-syntrophins. The human γ1-syntrophin gene is composed of 19 exons and encodes a brain-specific protein of 517 amino acids. The human γ2-syntrophin gene is composed of at least 17 exons, and its transcript is expressed in brain and, to a lesser degree, in other tissues. We mapped the γ1-syntrophin gene to human chromosome 8q11 and the γ2-syntrophin gene to chromosome 2p25. Yeast two-hybrid experiments and pull-down studies showed that both proteins can bind the C-terminal region of dystrophin and related proteins. We raised antibodies against these proteins and recognized expression in both rat and human central neurons, coincident with RNA in situ hybridization of adjacent sections. Our present findings suggest a differentiated role of a modified dystrophin-associated complex in the central nervous system.

Since the identification of dystrophin, the product of the Duchenne muscular dystrophy gene at Xp21, molecular genetics has moved quickly (1,2). The deficiency of dystrophin in Duchenne muscular dystrophy (DMD) 1 and its first animal model, the mdx mouse, leads to a dramatic reduction in a group of previously unknown proteins identified as the dystrophinassociated protein complex.
The extreme C terminus of dystrophin, which lies beneath the muscle membrane, is associated directly with a group of cytoplasmic peripheral membrane proteins known as syntrophins (29 -34). A similar interaction has been demonstrated with utrophin, the autosomally encoded dystrophin-related protein (35), and with some dystrobrevin isoforms (36 -38). The three known syntrophin isoforms, ␣1, ␤1, and ␤2, are encoded by distinct genes with specific expression (33). ␣1-Syntrophin is most abundant in skeletal muscle, where it is located close to the sarcolemma together with ␤1-syntrophin. In contrast, ␤2syntrophin is largely concentrated at the neuromuscular junction, but is barely detectable at the sarcolemma. Syntrophins bind directly to the C-terminal domain of dystrophin, in the region encoded by exons 74 and 75 (34, 39 -41). This region is contained in almost all mini-dystrophin transcripts starting from alternative distal promoters (42,43). These shorter dystrophins, called apodystrophins or Dp, have nearly ubiquitous expression. Similar binding has been ascertained for a homologous region in utrophin and ␣and ␤-dystrobrevins (36,38,44,45). In DMD, the sarcolemmal syntrophins are lost, whereas ␤2-syntrophin remains localized at the neuromuscular junction.
Each syntrophin has a characteristic domain organization in mammals (human, mouse, rabbit) as well as in the genetically distant Torpedo californica. Two pleckstrin homology (PH) domains and one PDZ domain are constantly present (46,47), with the first PH domain split into two regions (PH1a and PH1b) by the PDZ domain and its flanking regions. In addition, another region has been recognized in the C-terminal 57 amino acids and termed the syntrophin unique (SU) domain (32,33). There are indications that this domain may be directly involved in dystrophin binding. Here, we report the identification, through expressed sequence tag (EST) data base searching and cDNA library screening, of two novel human genes belonging to the syntrophin family.

EXPERIMENTAL PROCEDURES
Isolation of Human ␥1and ␥2-Syntrophin cDNAs-Approximately 500,000 plaques from a human fetal brain cDNA library in the gt10 vector (CLONTECH) and the NT2 neuronal precursor cell cDNA library in the Uni-ZAP XR vector (Stratagene) were screened using EST clones 49263 and c-1gb01 (IMAGE Consortium) according to standard procedures (48). Sixteen positive clones (eight for each probe) were isolated. The open reading frames of ␥1and ␥2-syntrophins were sequenced in at least two independent clones.
Northern Blotting-Human total RNA and rat poly(A) RNA Northern blots were purchased from CLONTECH and Origene, respectively, and hybridized according to standard procedures (48). The 701-and 432-bp rat cDNA fragments, used as probes for ␥1and ␥2-syntrophins, were obtained by RT-PCR using human-specific primer pairs. PCR Conditions-All PCR amplifications were performed in a PTC-100 thermal cycler (MJ Research, Inc.) as described previously (18).
Computational Analysis-Multiple sequence alignment of proteins was realized using the CLUSTAL W1.7 program with default parameters (49). The alignment data were also utilized to obtain a phylogenetic tree of proteins. A prevision of secondary structure was obtained using the PHDsec algorithm (50). To search domains or functional sites in the sequences, we scanned PROFILE and PROSITE data bases using the ISREC web site.
The GST fusion proteins used in rabbit polyclonal antibody production (GST-␥1-syntrophin and GST-␥2-syntrophin) were purified using a continuous elution SDS-polyacrylamide gel electrophoresis apparatus (Model 491 Prep Cell, Bio-Rad), digested with thrombin protease (Amersham Pharmacia Biotech), and reloaded on the apparatus to separate GST from peptide. Protein concentration was determined (Bio-Rad protein assay).
For GST pull-down assay, cDNAs encoding dystrophin (amino acids 3194 -3685), ␣-dystrobrevin (amino acids 422-564), and ␤-dystrobrevin (amino acids 444 -606) were cloned in frame with GST into the pGEX-2TK vector and introduced into E. coli JM109 cells. Twenty-ml overnight cultures were diluted 1:10 with LB medium supplemented with 1 mM 2-mercaptoethanol and 20 mM glucose. After induction with 0.1 mM isopropyl-␤-D-thiogalactopyranoside, cells were washed in PBS and resuspended in a 0.02 initial volume with 50 mM glucose, 25 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 10 mg/ml lysozyme. After incubation for 10 min at room temperature (all following steps were carried out at 4°C or on ice), an equal volume of 2ϫ lysis buffer (final concentration: 50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitors) was added, and cells were fractured with repeated cycles of freezing-thawing. The protein extract was clarified by centrifugation at 20,000 ϫ g for 30 min and incubated with 100 l of a 50% slurry of glutathione-agarose CL-4B (Fluka) for 2 h with shaking. After extensive washing in PBS supplemented with 1 mM phenylmethylsulfonyl fluoride and protease inhibitors, GST-dystrophin/dystrobrevin bead-bound fusion proteins was utilized in assay.
The pGBT9-DYS, pGBT9-DTNA, and pGBT9-DTNB plasmids were cotransformed in the YRG-2 yeast strain, containing two Gal4-inducible reporter genes, HIS3 and lacZ, with pGAD-G1SYN, pGAD-G1SYN⌬18, pGAD-G2SYN, or pGAD-A1SYN according to the specifications of the manufacturer (CLONTECH). The double transformants were plated onto selective plates lacking tryptophan, leucine, and histidine to show the activation of the Gal4-inducible HIS3 reporter gene through protein-protein interaction. Positive clones were also replica-plated on selective medium and tested for ␤-galactosidase activity using the colony lift assay following the directions supplied with the kit obtained from CLONTECH to confirm the interaction through the activation of the second Gal4-inducible reporter gene, lacZ.
Rat Tissues-Adult Wistar rats (3 months old, 250 -300 g) were killed under deep anesthesia (0.1 mg/g sodium barbital), and their brains and spinal cords were quickly dissected out and frozen immedi-ately on dry ice. Cryostat sections 10 m thick were collected onto Superfrost Plus slides (Erie Scientific, Portsmouth, NH) and stored at Ϫ80°C until processed.
Riboprobe Preparation-The rat ␥1and ␥2-syntrophin cDNA fragments obtained by RT-PCR were subcloned into the EcoRV site of the pBluescript SK vector. The 701-and 432-nucleotide cRNA probes were transcribed in both the antisense and sense orientations from the properly linearized plasmids, using 20 units of T3 or T7 RNA polymerase (Promega), 30 M 35 S-UTP (1300 Ci/mmol), and 10 mM each unlabeled ATP, CTP, and GTP.
In Situ Hybridization-In situ hybridization was performed on 10-m tissue sections from fresh frozen, non-perfused rat brain and spinal cord as described (51,52) with minor modifications. Briefly, after fixation in 4% paraformaldehyde, sections were treated with 0.5% acetic anhydride in 100 mM triethanolamine (pH 8.0) and prehybridized for 1 h at 55°C in prehybridization buffer (50% formamide, 750 mM NaCl, 50 mM sodium phosphate buffer (pH 7.0), 10 mM EDTA, 200 g/ml heparin, 5ϫ Denhardt's solution, and 0.5 mg/ml purified tRNA) containing 10% dextran sulfate and 35 S-RNA probe (2 ϫ 10 8 cpm/ml). After hybridization, sections were treated with 20 g/ml RNase, washed at high stringency for 1 h at 60°C, and agitated overnight in 1ϫ SSC. The dehydrated sections were coated with NBT-2 emulsion (Eastman Kodak Co.) and exposed for 2-5 weeks. The autoradiograms were developed, lightly counterstained with methylene blue or hematoxylin and eosin, and examined by dark-and bright-field microscopy. A sense strand probe labeled to the equivalent specific radioactivity of the antisense probe was used on adjacent sections in all experiments to check for background hybridization.
Immunohistochemistry-Sections were dried at room temperature, fixed in cold acetone, and pretreated with 0.3% H 2 O 2 in PBS to quench the endogenous peroxidase activity; rinsed in PBS; and incubated with 10% normal goat serum and 0.2% Triton X-100 for 60 min to mask nonspecific adsorption sites. Sections were then incubated for 1 h at room temperature with the anti-␥1-and anti-␥2-syntrophin rabbit polyclonal antibodies (diluted 1:200 in PBS). Omission of the primary antibodies or their replacement by preimmune sera was used for control experiments. After several rinses in PBS, the sections were incubated with biotinylated goat anti-rabbit IgG, washed in PBS, and then incubated with the ABC complex according to the manufacturer's instructions (Vectastain ABC, Vector Labs, Inc.). Peroxidase staining was obtained by incubating the sections in 0.075% 3,3Ј-diaminobenzidine and 0.002% H 2 O 2 in 50 mM Tris buffer (pH 7.6) for 10 min. In control experiments, when primary antibodies were omitted or replaced by nonimmune sera, the immunoreaction did not take place.
Immunofluorescence of Human Muscle-Unfixed cryostat sections (5 m thick) were cut from diagnostic muscle biopsies, air-dried, fixed in cold acetone, preincubated with 10% normal goat serum in PBS for 30 min, and incubated for 1 h at room temperature with a 1:50 dilution of anti-␥1-or anti-␥2-syntrophin antibody in PBS containing 1% bovine serum albumin. After several rinses in PBS, indirect immunofluorescence was visualized using biotinylated secondary antibodies (1:40) and fluorescein isothiocyanate-labeled streptavidin (1:250) (Amersham Pharmacia Biotech). Sections were covered with a glycerol mount and examined with an Olympus photomicroscope equipped with epifluorescence.

Isolation and Sequence Analysis of Syntrophin cDNAs-
The peptide sequences of human and mouse ␣1-syntrophin (Gen-Bank TM /EBI Data Bank accession numbers U40571 and U00677, respectively), ␤1-syntrophin (L31529 and U89997, respectively), and ␤2-syntrophin (U40572 and U00678, respectively) were matched using the TBLASTn algorithm with electronic data base sequences, including the subset of ESTs. Two distinct EST groups were retrieved. Each belong to novel cDNAs, having significant amino acid similarity to the three human syntrophins. The former sequence was found in human ESTs H16675, R13432, and R25045, and the latter in human EST Z43606. Clones 49263 and c-1gb01, for the first and second sequences, respectively, derived from an infant brain cDNA library, provided by IMAGE Consortium. These clones were sequenced and used as probes for further screening of human fetal brain and NT2 neuronal precursor cell cDNA libraries. Altogether, 16 clones (eight for each probe) were isolated to confirm and complete the sequence of both cDNAs.
The former cDNA, initially named syn4, is 1898 bp long and includes an open reading frame of 1554 nucleotides (Gen-Bank TM /EBI accession number AJ003030). It encodes a protein of 517 amino acids with a predicted molecular mass of 57,932 Da and a calculated isoelectric point of 6.24. The latter cDNA, initially named syn5, is 1938 bp long and contains an open reading frame of 1620 nucleotides (GenBank TM /EBI accession number AJ003029). It encodes a protein of 539 amino acids with a predicted molecular mass of 60,066 Da and a calculated isoelectric point of 7.59. Some differences have been observed for a few syn4 and syn5 clones. These are likely due to alternative splicing. For both cDNAs, the sequence flanking the first ATG is in accordance with the Kozak consensus for translational start sites (53). In agreement with other investigators, we have renamed syn4 and syn5 as ␥1-syntrophin and ␥2syntrophin, respectively.
Peptide similarity (40 -44%) to the human syntrophins spanned the entire open reading frames. Similarity to a Caenorhabditis elegans protein (U49829) is higher (46 -52%). The ␥1and ␥2-syntrophins share a 73% amino acid similarity, thus suggesting that the two proteins are much more closely related and probably derived from a common single syntrophin precursor.
Computational Analysis of Protein Sequence-The syntrophin sequences share distinctive conserved motifs: two PH domains and one PDZ domain, with the first PH domain split into two regions (PH1a and PH1b) by the PDZ domain and its flanking regions (32,33). In addition, another region has been recognized in the C-terminal 57 amino acids and termed the SU domain (32,33). There are indications that this domain is involved in dystrophin binding.
We have produced a multiple sequence alignment of all syntrophins using the CLUSTAL W1.7 program (49) (Fig. 1). The ␥1and ␥2-syntrophins are very similar to the other syntrophins in the PDZ domain. The PH domain is not easy to recognize since it includes a series of relatively poorly conserved peptides interspersed with less conserved linker sequences. These may vary from a few amino acids to 100 or more, often containing other functional domains. Nevertheless, the C-terminal 15 or so amino acids contain only an invariant Trp residue, and the six N-terminal residues from this amino acid frequently include two or more negatively charged residues as well as glutamic acid (46). The alignment shows that both the C-terminal Trp residues of the PH1 and PH2 domains are present in ␥1and ␥2-syntrophins (Fig. 1), suggesting that both domains can be conserved.
The SU domain is less homologous, with only a weak similarity to the other three syntrophins. The proposed secondary structure of the SU domain is composed of three to five ␤-sheets separated by as many turns (33). Using the PHDsec algorithm (50), the predicted secondary structure of ␥1and ␥2-syntrophins is mainly arranged into an ␣-helix.
We analyzed ␥1and ␥2-syntrophins using the PROFILES-CAN program. A PDZ domain was identified in both proteins (residues 57-140 for ␥1-syntrophin and residues 73-156 for ␥2-syntrophin) and only one PH domain, corresponding to the PH2 domain of the other syntrophins (residues 283-390 for ␥1-syntrophin and residues 296 -421 for ␥2-syntrophin) (Fig.  1). No other domain was found. In addition, we have also identified, in both proteins, an ATP/GTP-binding site motif A (P-loop) (54) (residues 440 -448 for ␥1-syntrophin and residues 471-479 for ␥2-syntrophin). This motif is a glycine-rich region that typically forms a loop between a ␤-strand and an ␣-helix. This loop interacts with one of the phosphate groups of the nucleotide. Many potential phosphorylation sites have also been identified with no clear relationship to the other syntrophins.
Phylogenetic analysis of all syntrophins establishes a common origin of the syntrophins with a early separation into two groups: the first including ␣1-, ␤1-, and ␤2-syntrophins, and the second including ␥1and ␥2-syntrophins. The two C. elegans syntrophins are final confirmation.
Tissue Distribution of ␥1and ␥2-Syntrophin Expression-The expression of ␥1and ␥2-syntrophins in human and rat adult tissues was assayed by Northern blotting, using fragments of the respective cDNAs as probes.
Expression of ␥1-Syntrophin Is Brain-specific in Humans and Rats-In man, only an ϳ7.0-kb transcript was observed using a 426-bp cDNA probe encoding the last 126 amino acids (Fig. 2a). Conversely, the rat cDNA probe (701 bp), corresponding to amino acids 29 -290 of human ␥1-syntrophin, hybridized with three equally abundant transcripts of ϳ2.6, 3.4, and 7.5 kb (Fig. 2b). The different probes may reflect the different expression pattern. The mRNAs are much longer than the coding sequence, and a long 3Ј-UTR and/or 5Ј-UTR is probably present. The expression of further ␥1-syntrophin isoforms most likely originates by alternative splicing. For example, exon 18 is in-frame spliced out in clone 49263, and so is the fourth exon in another ␥1-syntrophin cDNA clone. Other alternatively spliced products were identified by RT-PCR (data not shown).
Conversely, ␥2-syntrophin has a broader but weaker expression. A clearer picture of ␥2-syntrophin comes from rat Northern blots (Fig. 2c). In addition to brain with two transcripts of ϳ2.1 and 2.3 kb and a third very weakly hybridizing transcript of 2.5 kb, a mRNA of 2.1 kb was also observed in testis. Weak signals at 0.8, 2.2, and 2.5 kb were also present in kidney and lung, as well as a weaker 1.6-kb transcript in heart. These different transcripts are probably related to isoforms of this gene. Evidence for this hypothesis comes from cDNA library screening and RT-PCR of human brain cDNA. In particular, we have characterized two isoforms. In the first, exons 3-5, corresponding to the first half of the PDZ domain, are in-frame spliced out; and in exon 9, a putative 5Ј-donor splicing site at bp 693 leads to in-frame deletion of the last 27 bp of this exon, corresponding to amino acids 830 -839. Interestingly, the consensus of the protein kinase C phosphorylation site can be recognized in this fragment. In the second isoform, exons 3-6 are equally in-frame spliced out, and the PDZ domain is nearly entirely eliminated.
Genomic Structure and Chromosomal Mapping-Genomic library screening and long-range and vectorette PCR were used  (Table I). The first two exons contain the 5Ј-UTR, and the first methionine is in the third exon. The ␥2-syntrophin gene (SNTG2) has at least 17 exons (Table II). The first exon includes the first methionine. For both genes, the termination signal in the last exon is followed by a 3Ј-UTR colinear with the genomic sequence. The exon structure and splicing sites are conserved in the coding sequences between ␥1and ␥2-syntrophins (Tables I and II), suggesting a common origin. Two methods were employed for chromosomal mapping. For ␥1-syntrophin, PCR primers, designed from genomic sequences around exons 1 and 19, were used to screen a yeast artificial chromosome library and a radiation hybrid panel (Genebridge 4, Research Genetics). One yeast artificial chromosome clone (856d11) containing the markers AFMB353XD9, D8S589, and D8S1652 (chromosome 8q11) was identified. Radiation hybrid mapping confirmed the chromosomal location at 8q11 between D8S1622 and AFMB353XD9 (4.7 centirays from AFMB353XD9). ␥2-Syntrophin was mapped to chromosome 2 at p25 between D2S323 and D2S330 (1.7 centirays from D2S323) by radiation hybrid using PCR primers flanking exon 12.
In Vivo and in Vitro Assays of Binding with Dystrophin-The ␣1-, ␤1-, and ␤2-syntrophins have been shown to bind dystrophin, utrophin, and dystrobrevins in vitro (34,35,39,40,55). To verify whether ␥1and ␥2-syntrophins can also bind dystrophin or related proteins, assays of protein-protein interaction were used. In addition, we performed a GST pull-down assay. The C terminus of dystrophin or ␣or ␤-dystrobrevin was fused to glutathione S-transferase protein in the pGEX-2TK plasmid, expressed in E. coli cells, and purified on glutathione-agarose beads. The full-length coding sequences of ␥1and ␥2-syntrophins were then cloned, as sense and antisense, into the pCT plasmid and transfected into COS-7 cells. The protein extracts were incubated with the bead-bound fusion proteins and, after extensive washing, analyzed by Western blotting.
The two-hybrid system is a yeast-based genetic assay to detect in vivo protein-protein interaction (56,57). In the assay, one protein is fused with the DNA-binding domain, and the other with the transcription activation domain of GAL4. Should an interaction occur, the resulting dimer induces reporter gene activation (HIS3 and lacZ). To confirm this interaction in vivo, the dystrophin C terminus (residues 3194 -3685) corresponding to exons 66 -79 and regions of ␣-dystrobrevin (residues 422-564) and ␤-dystrobrevin (residues 444 -606), all with the syntrophin-binding site, were fused to the DNA-binding domain of GAL4 in the pGBT9 plasmid (pGBT9-DYS, pGBT9-DTNA, and pGBT9-DTNB, respectively), whereas ␥1syntrophin (residues 19 -517) and ␥2-syntrophin (residues 1-539) were fused to the activating domain of GAL4 in the pGAD plasmid (pGAD-G1SYN and pGAD-G2SYN, respectively). In addition, human ␣1-syntrophin (residues 173-505), fused to the activating domain of GAL4 in the pGAD plasmid (pGAD-A1SYN), was used as a positive control of interaction with dystrophin and related proteins. The pGBT9-DYS, pGBT9-DTNA, and pGBT9-DTNB plasmids were cotrans-TABLE II Genomic organization of ␥2-syntrophin For each exon, the sequence at exon-intron boundaries is shown, as well as the length and the corresponding region or domain. The numbers 0, 1, and 2 indicate the splicing phase when the interruption falls before the first, second, or third base of a codon. The ␥1and ␥2-syntrophins interacted with dystrophin in vivo (Fig. 4, a and b). For ␥1-syntrophin, we also used an isoform without exon 18 (pGAD-G1SYN⌬18) that includes the ATP/GTP-binding site. This isoform did not interact with dystrophin, confirming that the dystrophin-binding site is located at the C terminus, presumptively in the last 80 amino acids. ␣-Dystrobrevin interacted with ␥2-syntrophin, but only weakly with ␥1-syntrophin (Fig. 4c). Conversely, both syntrophins bound to ␤-dystrobrevin (Fig. 4d). These findings were confirmed by testing for ␤-galactosidase activity (data not shown).
In Situ Hybridization in Rat Central Nervous System Tissues-␥1-and ␥2-syntrophin mRNAs were detected by in situ hybridization in all rat central nervous system regions examined; and in particular, they were highly expressed by neuronal cells. The expression of syntrophin transcripts was mainly restricted to cells with neuronal morphology. Hybridization grains for the ␥1and ␥2-syntrophin mRNAs were localized in the perikaryon and proximal portion of the neuronal processes. Strong hybridization signals were localized in the hippocampus, neuron-rich dentate granule cells, and pyramidal cell layers (Fig. 5). Intense labeling was observed in neurons of the cerebral (parietal and frontal) cortex (Fig. 5G). ␥1and ␥2syntrophin mRNAs were also expressed in the cerebellar cortex, deep cerebellar nuclei, thalamus, and basal ganglia (data not shown). A very strong mRNA signal was present within neurons of both anterior and posterior horns of the spinal cord, whereas a lower signal could be detected in the white matter (Fig. 5, C and E). No specific signal over the background was detectable after hybridization of adjacent sections with the sense strand probe (data not shown). This confirmed the specificity of hybridization. Moreover, RNA studies were concordant with the distribution of the ␥1and ␥2-syntrophin immunoreactivities. Using polyclonal antibodies raised against ␥1and ␥2-syntrophins, we observed reactivities in the same cell populations containing abundant levels of the corresponding mRNAs (Fig. 5, B, D, F, and H), thus confirming that ␥1and ␥2-syntrophin gene products are indeed highly expressed in the central nervous system. The widespread, albeit uneven, distribution of ␥1and ␥2-syntrophin transcripts and proteins throughout different cerebral and spinal areas suggests that ␥1and ␥2-syntrophin genes play an important housekeeping role in neurons. Immunofluorescence of Human Muscle-No signal was found with anti-␥1-syntrophin antibody (Fig. 6A). ␥2-Syntrophin showed a strong sarcolemmal immunoreactivity in all muscle fibers (Fig. 6B). In DMD patients with absence of dystrophin, ␥2-syntrophin was absent or severely reduced (Fig. 6C); In contrast, a normal plasmalemmal signal was observed in patients with neurogenic atrophy (Fig. 6D). Upon immunofluorescence, ␥2-syntrophin was not selectively localized at the neuromuscular junctions and was present on the membrane of cardiomyocytes in two biopsy specimens (data not shown). DISCUSSION Originally identified in postsynaptic membranes of T. californica (29), syntrophins are intracellular peripheral membrane proteins of ϳ58 kDa that, in man and mouse, exist in three highly conserved but distinct isoforms (␣1, ␤1, and ␤2) encoded by different genes (30,31,33). The name "syntrophin" was introduced to indicate that this group of proteins accompanies dystrophin. Further studies have demonstrated that syntrophins also accompany all the other proteins of the dystrophin family such as utrophin (dystrophin-related protein 1) (35), dystrophin-related protein 2, and dystrobrevins (␣ and ␤) (37,38). Binding is mediated by amino acid sequences that are homologous to the cysteine-rich domain and C-terminal region of dystrophin encoded by exon 74 (34,39,40).
The role of syntrophins and their requirement for muscle and nerve cell function are still obscure. To date, no human disease has been associated with a syntrophin gene mutation. Likewise, no gross histological changes in the skeletal muscle of ␣1-syntrophin knockout mouse (the first syntrophin-deficient animal model) have been reported (58). This may be due to a redundancy of the genes of the syntrophin family. The lack of function of one gene could be replaced by other members of the family, coexpressed in the same tissues. In man, the ␣1-syntrophin transcript is predominantly expressed in skeletal and cardiac muscle, whereas the ␤1and ␤2-syntrophin transcripts are expressed in a wide variety of tissues (31,33,37). In addition, histochemical studies with specific antibodies revealed that the three syntrophins are all present at the neuromuscular junction. ␣1-Syntrophin is also found at the sarcolemma, whereas ␤1-syntrophin occurs at the sarcolemma of fast twitch muscle fibers (37). We indicate here that the syntrophin gene family should include at least two other members, which are less related to the primary sequences of known syntrophins, but retain the property of dystrophin binding. ␥1and ␥2-syntrophins should be considered a separate entity because of their relatedness (73% similarity), the common C. elegans ancestor gene, and the genomic organization. In particular, these two genes are split into 17-19 exons by long introns at corresponding positions that are different from the exon-intron boundaries found in the other syntrophins. In addition, we observed a complex pattern of alternatively spliced products, with the presence, at least for ␥2-syntrophin, of both translated and nonfunctional transcripts.
Expression data indicate that ␥1-syntrophin is restricted to neurons. ␥2-Syntrophin has a broader expression and a more complex pattern of splicing and is presumably included in the dystrophin-associated complex beneath the muscle membrane at the sarcolemma. In fact, some DMD patients show a secondary reduction of ␥2-syntrophin protein expression.
The ability of ␥1and ␥2-syntrophins to bind dystrophin and dystrobrevins (␣ and ␤) has been confirmed in vivo and in vitro. Binding is mediated by the last 80 amino acids, a region with weak similarity to the SU domain, the putative dystrophinbinding site. The homologous cysteine-rich domain and C-terminal region of dystrophin and related proteins include the syntrophin-binding site and several potential binding domains (59,60). In particular, a coiled-coil motif, flanking the syntrophin-binding site, links dystrophin to the same motif in the C terminus of dystrobrevin. Therefore, dystrophin interacts with dystrophin-glycoprotein complexes via the cysteine-rich domain and heterodimerizes with dystrobrevins via the coiledcoil motif (61). Together, dystrophin and dystrobrevin can recruit two syntrophins. This model seems to be confirmed by the observation that dystrophin complexes are highly enriched in ␣1and ␤1-syntrophins, whereas utrophin complexes contain mostly ␤1and ␤2-syntrophins (37). Different pairings are possible between the syntrophins: ␥1and ␥2-syntrophins increase the possible combinations in brain.
DMD and Becker muscular dystrophy patients with point mutations or deletions localized in the 3Ј-region of the dystrophin gene in addition to the progressive muscle wasting show a higher incidence of mental retardation, with learning disorders and speech difficulties (62). In contrast, mutations localized in other parts of the gene are usually not associated with mental retardation. There is no explanation for this clinical observation because dystrophin is absent or severely reduced in DMD patients, regardless of where the primary nonsense mutation occurs. The presence of additional promoters located in certain dystrophin gene introns could still generate functional minidystrophin 3Ј-transcripts (apodystrophins), unless the mutation does not directly involve the 3Ј-exons. The latter part of the dystrophin gene encodes the region endowed with the dystroglycan-binding site (exon 65) and the syntrophin-binding site (exons 73-74).
In the brain, ␥1and ␥2-syntrophins as well as ␤-dystrobrevin can bind dystrophin isoforms Dp71 and Dp140 (63). In common with dystrophin and ␤-dystrobrevin, ␥1and ␥2-syntrophins are found in the cortex and hippocampal formation. These data provide evidence that the composition of the dystrophin-associated protein complex in the brain differs from that in muscle (64). A mouse transgene overexpressing apodystrophin-1/Dp71 (exons 63-79) in dystrophin-deficient animals (mdx mice) could not restore the normal muscle phenotype (65,66). This suggests that this dystrophin fragment alone has no influence on the muscle disease progression. It is possible that the presence of an intact C-terminal fragment could be important for the associated mental disorders. The apodystrophindystroglycan complex in the central nervous system can bind the ␥1and ␥2-syntrophins. Further studies are needed to determine the role of these novel syntrophins in neuron signaling processes and whether their concomitant lack affects learning.