Actinin-associated LIM Protein: Identification of a Domain Interaction between PDZ and Spectrin-like Repeat Motifs

RNA was isolated using the guanidine isothiocyanate/CsCl method, and mRNA was selected with oligo dT sepharose. For degenerate reverse transcriptase-PCR (RT-PCR), rat skeletal muscle mRNA was reverse transcribed with RTth polymerase using random hexamer primers. PCR, using a 54°C annealing temperature, was then performed using the following primer pair: GGGGGATCCGGXGGXCTXGGXHT and GCGGAATTCAGDARXATRTCXCC. Amplified products were resolved on agarose gels. Bands of 120–160 bp were isolated and subcloned into pBluescript vectors (Stratagene, La Jolla, CA) for sequence analysis. Before transformation, ligation products were digested with BglII to remove syntrophin cDNAs. Clones encoding ALP were isolated from a rat skeletal muscle cDNA library (catalog no. 937510; Stratagene) by plaque hybridization with the ³²P-labeled DNA fragment obtained above by RT-PCR. One clone of 1,586 nucleotides was sequenced on both strands and contained an open reading frame (ORF) of 362 amino acids. The initiation site for translation was assigned to the first methionine codon in the ORF at nucleotide 100. This ATG occurs in a consensus Kozak sequence for initiation (Kozak, 1991) and is preceded by an in-frame stop codon. A polyadenylation sequence occurs ∼400 bp downstream of the stop codon.

For Northern blotting, RNA was separated on a formaldehyde agarose gel and transferred to a nylon membrane. A random-primed ³²P probe was generated using the 160-bp ALP cDNA obtained from RT-PCR as a template. The filter was washed at high stringency, 63°C, 0.1% sodium chloride/sodium citrate and 0.1% SDS, and was exposed to x-ray film for 2 h at −70°C.

The nucleotides encoding amino acids 1–128 of ALP were amplified by PCR and subcloned into the GAL-4 DNA–binding domain plasmid pGBT9 (Clontech Laboratories, Palo Alto, CA). This construct was cotransformed into yeast strain HF7c with a library of human skeletal muscle cDNAs fused to the GAL-4 activation domain (Clontech). The transformation mixture was plated onto a synthetic dextrose plate lacking tryptophan, leucine, and histidine. Growth was monitored during a 5-d incubation at 30°C, and color was measured by a β-galactosidase colorimetric filter assay (Fields and Song, 1989). Interacting clones were rescued, retransformed to confirm interaction, and sequenced. Deletions of interacting clone 9-5 were generated by digestion with XcmI (9-5X), NarI (9-5N), or BglI (9-5B). Site-directed mutagenesis of ALP L78K was performed by PCR and confirmed by sequencing.

A glutathione S-transferase (GST) fusion protein encoding amino acids 1–207 of ALP was amplified by PCR and subcloned into pGEX2T (Pharmacia Biotech, Piscataway, NJ). The fusion protein was expressed and purified from Escherichia coli as described (Brenman et al., 1995). Rabbits were immunized with the GST–ALP fusion protein emulsified in complete and incomplete Freund's adjuvant. Serum bleeds were evaluated by ELISA. For affinity purification, ALP antiserum was applied to a column of GST– ALP fusion protein immobilized on Reacti-Gel (Pierce Chemical Co., Rockford, IL). The column was washed with 20 vol of buffer containing 500 mM NaCl and 10 mM Tris, pH 7.5. ALP antibody was eluted with 10 bed vol of 100 mM glycine, pH 2.5. Monoclonal antibodies to syntrophin (Peters et al., 1994), nNOS (Transduction Laboratories, Lexington, KY), and α-actinin (Sigma Immunochemicals, St. Louis, MO) were also used.

For immunofluorescent staining, skeletal muscle samples were flash frozen in liquid nitrogen–cooled isobutane, sectioned on a cryostat (10 μm), and melted directly onto glass slides. Sections were then postfixed in 2% paraformaldehyde in PBS. Tissues were blocked in PBS containing 1% normal goat serum. mAbs to α-actinin (1:100) and polyclonal ALP antibody were applied to sections overnight at 4°C. For indirect immunofluorescence, secondary goat anti–mouse FITC or donkey anti–rabbit Cy-3–conjugated antibodies were used according to the specifications of the manufacturer (Jackson ImmunoResearch Laboratories, Bar Harbor, ME).

For Western blotting, protein extracts were resolved by SDS-PAGE and transferred to polyvinyldifluoride membranes. Primary antibodies were diluted in block solution containing 3% BSA, and 0.1% Tween 20 in TBS, and were incubated with membranes overnight at 4°C. Labeled bands were visualized using enhanced chemiluminescence (Amersham, Arlington Heights, IL).

C2 myogenic cell line was grown on gelatin-coated dishes in Ham's F-10 nutrient mixture plus 0.8 mM extra CaCl₂ supplemented with 15% horse serum (Life Technologies, Inc., Bethesda, MD) and 2.5 ng/ml basic fibroblast growth factor (Promega, Madison, WI). Myoblasts were fused in media containing DME (GIBCO BRL, Gaithersburg, MD) supplemented with 2% horse serum. Protein was extracted from cultured cells in buffer containing 25 mM Tris-HCl, pH 7.4, containing 1% Triton X-100, 1 mM EDTA, and 100 μM PMSF. RNA was purified from cultured cells as described above.

Skeletal muscle tissue was homogenized in 10 vol of cold water containing 1 mM PMSF and centrifuged at 15,000 g for 10 min. To extract cytoskeletal proteins, the pellet was treated with 10 vol of buffer containing 2 mM Tris-HCl, pH 9, and 1 mM EGTA at 37°C for 30 min with gentle agitation. After centrifugation at 15,000 g for 30 min, the supernatant was titrated to pH 7.5. For “pull-down” assays, the solubilized tissue samples were incubated with control or GST fusion proteins linked to glutathione Sepharose beads for 1 h. Beads were washed three times with buffer containing 0.5% Triton X-100 and 350 mM NaCl, and proteins were eluted with SDS loading buffer. The GST–nNOS fusion protein was purified as described previously (Brenman et al., 1995).

For immunoprecipitation, polyclonal antibodies (1 μg) to ALP or preimmune serum were added to 0.5-ml aliquots of solubilized skeletal muscle extract, and samples were incubated on ice for 1 h. Protein A Sepharose (50 μl) was used to precipitate antibodies. Protein A pellets were washed three times with buffer containing 100 mM NaCl and 1% Triton X-100. Immunoprecipitated proteins were denatured with loading buffer and resolved by SDS-PAGE.

In situ hybridization used ³⁵S-labeled RNA probes exactly as described (Sassoon and Rosenthal, 1993). Sense and antisense probes to ALP (full length) were synthesized from a pBluescript vector using T3 and T7 polymerases.

Two P1 clones corresponding to human ALP (Genome Systems, St. Louis, MO) were used to determine the location of ALP on human chromosomes by fluorescence in situ hybridization (FISH). The hybridized signal was detected by antidigoxigenin conjugated with FITC, as described (Sakamoto et al., 1995; Stokke et al., 1995).

Fine mapping of ALP was performed by PCR amplification of human hamster somatic cell hybrids and radiation-derived hybrids containing all or portions of human chromosome 4q35. The somatic cell hybrids included HHW986, containing intact human chromosome 4 (Carlock et al., 1986), HHW986, retaining only 4q35 translocated to a derivative 5p, and HHW1372, in which only the telomeric region of 4q35 (distal to D4S187) is retained on a derivative X (Bodrug et al., 1990). The radiation hybrids were derived from HHW416 and retain varying fragments of 4q35 (Winokur et al., 1993). Negative controls included a human lymphoblastoid cell line GM7057 (NIGMS) and a Chinese hamster fibroblast cell line UCW 104.

Hybrids were screened by PCR for the presence of the ALP gene. The oligonucleotides (sense: GGGAGCTGTACTGCGAAA) and (antisense: CGATTGTTTTCTCGTGTA) amplify a 150-bp product within the 3′ untranslated region of ALP. Amplifications with a 51°C annealing temperature were done in 25 μl containing 250 ng genomic DNA, 50 μl each oligonucleotide, 1.25 mM each dNTP, 16.6 mM ammonium sulfate, 67 mM Tris-HCl, pH 8, 6.7 mM MgCl₂, 10 mM β-mercaptoethanol, and 2 U Taq polymerase.

To identify novel PDZ proteins expressed in skeletal muscle, we performed RT-PCR with degenerate primers that amplify amino acids GLGF (sense) and GDXIL (antisense; see Materials and Methods for experimental details). These sequences are conserved among many PDZ proteins. A library of these cDNA fragments was constructed and processed to remove the syntrophins leading to the identification of novel cDNAs encoding PDZ sequences. Northern analysis was then used to identify gene products that were enriched in skeletal muscle. The mRNA for one of the products, SK-2 (hereafter called ALP), migrated at 1.6 kb and occurred at high levels in skeletal muscle tissue, at low levels in the heart, and was undetectable in other tissues examined (Fig. 1,A). Human ALP showed a similar tissue-specific expression pattern (Fig. 1,B). To better understand mechanisms that regulate ALP during muscle development, we evaluated expression in myogenic cultures. ALP expression was dramatically induced after myotube fusion in culture (Fig. 1 C).

In situ hybridization of embryonic day (E15) mouse embryo with antisense ALP probe demonstrated the highest levels of ALP expression in developing skeletal muscle and heart (Fig. 1,D). ALP mRNA also occurred in embryonic intestine. No specific hybridization was observed using sense RNA control (Fig. 1 E).

cDNA clones encoding the full 1.6-kb mRNA were obtained from a rat skeletal muscle cDNA library. The mRNA contains a single ORF of 1,086 bp encoding a protein of 39 kD. Sequence analysis shows that ALP has a homology at the NH₂ terminus to PDZ domains. Alignment of ALP with other PDZ domains is shown (Fig. 2 A). The presence of histidine at amino acid 62 suggests that the PDZ domain may be in group I (Brenman and Bredt, 1997; Songyang et al., 1997).

The central region of ALP showed no homology to any other cloned gene while the COOH terminus encodes a LIM domain. Though ALP has not previously been reported, other proteins with a similar domain structure have been described. A database homology search with BLAST indicated that ALP shares high homology to a number of newly identified transcripts including CLP-36, RIL, and enigma (Fig. 2 B). CLP-36 was identified as a cDNA whose expression in the heart is downregulated by hypoxia (Wang et al., 1995). RIL, short for reversion-induced LIM protein, is downregulated in H-ras–transformed cells (Kiess et al., 1995). Enigma was identified as an insulin receptor–interacting protein (Wu et al., 1996). These investigators, however, did not recognize the homology of the NH₂-terminal regions of CLP-36, RIL, or enigma with the PDZ domain. The PDZ domain of ALP shares 55, 48, and 45% amino acid identity with the PDZ domains of CLP36, RIL, and enigma, respectively. The LIM domain of ALP shares even stronger homology (67% identity) with CLP36 and RIL. While ALP, CLP36, and RIL all only have one LIM domain, enigma has three LIM domains. The sequence homology indicates that ALP, CLP36, and RIL constitute a new family of proteins containing an NH₂-terminal PDZ domain and a COOH-terminal LIM domain.

Our analysis of the expressed sequence tag (EST) database showed that overlapping cDNAs corresponding to human ALP have been deposited. The human ALP is 91% identical to the rat sequence. We noted that EST clones from human heart libraries were consistently different in the central region from those in human skeletal muscle libraries (Fig. 2 C). Exons encoding the central 112 amino acids of skeletal muscle ALP are likely to be spliced out in the heart and replaced by exons encoding 64 different amino acids. To confirm this differential expression, we amplified the region that was unique to heart transcripts and reprobed the Northern blot. As expected, we found heart-specific expression of this region of ALP (data not shown). We therefore define two subtypes ALP_SK and ALP_H for the alternative transcripts that occur in skeletal muscle and heart, respectively.

Previous studies have shown that PDZ domains participate in protein–protein interactions. To determine potential targets for the PDZ domain of ALP, we used the yeast two-hybrid system. We screened 10⁶ clones from an adult skeletal muscle library (Clontech) and obtained 120 positive clones, 35 of which were recovered and analyzed. All positive clones encoded fragments of α-actinin-2, a muscle- specific cytoskeletal protein that contains an NH₂-terminal actin-binding domain, four central spectrin-like repeats, and a COOH-terminal region homologous to calcium-binding EF hands (Beggs et al., 1992). ALP interacts only with the spectrin-like repeat region of α-actinin-2, and all interacting clones encode spectrin repeat three (Fig. 3). However, a deletion construct (9-5N) containing repeat three did not interact with ALP, indicating that repeat three is necessary but not alone sufficient for binding.

Interaction of a PDZ domain with spectrin-like repeats is unprecedented. We therefore asked whether this interaction was specific. We found that the PDZ domains of nNOS, α1-syntrophin, and the three PDZ domains of PSD-95 (Brenman et al., 1996) did not interact with α-actinin-2 in the yeast two-hybrid system. We previously identified a point mutation that universally disrupts all known types of PDZ interactions (Christopherson, K., W. Lim, and D.S. Bredt, manuscript submitted for publication). This change corresponds to a mutation found in INAD (Shieh and Niemeyer, 1995), a PDZ protein in Drosophila that binds to the transient receptor potential (TRP) calcium channel (Shieh and Zhu, 1996). This mutation corresponds to position Leu78 in ALP. Mutating Leu78 of ALP to Lys abolished the interaction with α-actinin-2 (Fig. 3), demonstrating specificity of the PDZ domain interaction with α-actinin-2.

To further confirm this interaction, we expressed a bacterial fusion protein linking GST to the PDZ domain of ALP. We found that this fusion protein specifically retained α-actinin-2 from solubilized skeletal muscle cytoskeleton (Fig. 4) but did not retain α1-syntrophin. By contrast, an analogous column containing the PDZ domain of nNOS “pulled-down” α1-syntrophin but not α-actinin-2.

To determine whether the interaction with α-actinin-2 is physiologically important in localizing ALP to specific cytoskeletal domains, we developed an affinity-purified antiserum (see Materials and Methods). We evaluated the specificity of the serum by Western blot analysis of crude tissue extracts. As expected, the antiserum recognized a prominent band of 39 kD in skeletal muscle and a much less intense band of 35 kD in the heart (Fig. 5,A). No immunoreactive bands were noted in the spleen, kidney, brain, or liver. Western blotting of myogenic cell extracts showed that ALP is absent from myoblasts but is induced within 3 d after myotube fusion (Fig. 5 B).

To determine whether ALP and α-actinin-2 occur together in a protein complex in skeletal muscle, we performed immunoprecipitation studies (Fig. 4,B). We found that the antiserum to ALP specifically coimmunoprecipitated α-actinin-2 from solubilized cytoskeletal extracts from skeletal muscle. By contrast, neither nNOS nor syntrophin, cytoskeletal components of the dystrophin complex, were coimmunoprecipitated with ALP. We next compared the cellular distribution of ALP with α-actinin-2 in skeletal muscle. Immunofluorescent staining of longitudinal sections of adult skeletal muscle showed that ALP colocalized with α-actinin-2 on the Z lines (Fig. 5 C). No ALP immunoreactivity was found at the sarcolemma, nucleus, or other structures of the myofiber.

To determine whether ALP might map close to any known genetic mutations that are associated with muscle disease, we determined the chromosomal location of human ALP. We performed FISH with two independent digoxigenin-labeled genomic ALP probes on normal human metaphase chromosomes. Hybridization with these probes resulted in specific labeling of only chromosome 4. The location of the probes was determined by digital image microscopy after FISH and was localized by the fractional length from the p terminus (FLpter) as described previously (Sakamoto et al., 1995; Stokke et al., 1995). Both clones mapped to the most telomeric region of chromosome 4, at 4q34-qter with FLpter values of 0.962 ± 0.004 and 0.959 ± 0.005.

Interestingly, FSHD, the most common autosomal dominant muscular dystrophy, maps to the telomeric region of chromosome 4, at 4q35. Since FSHD is associated with deletions of a subtelomeric repeat sequence (van Deutekom et al., 1993), the localization of ALP relative to the 4q telomere was of interest. We therefore used somatic cell and radiation hybrid panels to map ALP within chromosomal band 4q35. PCR analysis of the somatic cell hybrids revealed that ALP maps within the proximal portion of this band (Fig. 6,B). ALP is not present in HHW1372, which contains only the telomeric 2 Mb of chromosome 4q distal to the locus D4S187. Analysis of the radiation hybrid panel DNAs, which contain independent fragments of 4q35, localizes ALP to the interval between D4S171 and the Factor XI gene (Fig. 6 C). Thus, the approximate distance of ALP from the telomere is 7–10 Mb.

The primary finding in this study is the identification of a functional interaction between a PDZ domain and the spectrin-like repeats of α-actinin-2. This association targets a novel LIM protein, ALP, to the actinin-rich Z lines of skeletal muscle fibers. PDZ domains are recently recognized protein–protein interaction motifs that are implicated in protein association with the cytoskeleton (Marfatia et al., 1996) and in signal transduction (Brenman and Bredt, 1997; Sheng, 1996). Previous studies demonstrated that the two PDZ proteins in skeletal muscle, nNOS and the syntrophins, are constituents of the dystrophin complex (Adams et al., 1993; Brenman et al., 1995). Our work here shows that the PDZ protein ALP does not associate with the dystrophin complex, but instead binds to α-actinin-2, which is in the dystrophin superfamily of cytoskeletal proteins.

Interaction with the spectrin-like repeat represents a new mode of binding for a PDZ domain. Previous work has shown that PDZ domains of the postsynaptic density protein, PSD-95, bind to certain glutamate receptors and K⁺ channels in the brain that terminate with a consensus of E-T/S-X-V/I (Cohen et al., 1996; Kim et al., 1995; Kornau et al., 1995). These interactions appear to anchor ion channels to synaptic sites in neurons. Interaction with specific COOH-terminal peptides may be a general property of PDZ domains, and two recent studies demonstrate that distinct PDZ domains, bind to specific COOH-terminal peptide sequences (Songyang et al., 1997; Stricker et al., 1997). Certain PDZ domains can also associate with each other in a homotypic-type interaction (Brenman et al., 1996). The PDZ domain of nNOS binds to the second PDZ domain of PSD-95 in the brain and to the PDZ domain of α1-syntrophin in skeletal muscle.

The binding interface between the PDZ domain of ALP and the spectrin-like repeats of α-actinin-2 represents a third mode for protein interactions mediated by PDZ domains. We suspect that this type of interaction is not unique to ALP and may explain cytoskeletal interactions of diverse PDZ proteins. Z-1 protein of epithelial tight junctions contains three PDZ domains and associates with spectrin in the cell cortex (Willott et al., 1993). Electron micrographic studies indicate that ZO-1 forms a complex with the central rodlike repeat domains of spectrin. It is not yet clear whether the PDZ domains of ZO-1 mediate this interaction. A complex ternary interaction between the spectrin-like repeats of dystrophin and the PDZ domains of nNOS/syntrophin may occur at the skeletal muscle sarcolemma (Chao et al., 1996). Thus, in vitro assays demonstrate that nNOS binds directly to syntrophin, but not to dystrophin. However, the nNOS/syntrophin interaction in skeletal muscle requires that certain spectrin-like repeats of dystrophin be intact. Also, nNOS is selectively absent from skeletal muscle sarcolemma in patients with Becker muscular dystrophy who have mutations within the spectrin-like repeats of dystrophin (Chao et al., 1996).

We find that ALP expression is normal in Duchenne and Becker muscular dystrophies (Xia, H., and D.S. Bredt, unpublished data). On the other hand, certain inherited muscular dystrophies result from mutations in cytoskeletal proteins that do not interact with the dystrophin complex (Hoffman et al., 1995). Plectin, a cytoskeleton–membrane anchorage protein of hemidesmosomes, links intermediate filaments to the sarcolemma and also occurs at the Z lines in skeletal muscle (Wiche et al., 1983). Mutations in plectin do not effect the dystrophin complex, but they cause an autosomal recessive muscular dystrophy associated with skin blistering (Smith et al., 1996). It will be important to assess ALP expression in a variety of inherited muscular dystrophies to determine whether it may play a role in any of these diseases.

Our chromosomal mapping studies indicate that ALP occurs on human chromosome 4q 35. Interestingly, the location is ∼7–10 Mb from the subtelomeric region that is mutated in FSHD, an autosomal dominant disease (Wijmenga et al., 1992). The specific genetic defect in FSHD disease appears to be a deletion of heterochromatin (Lyle et al., 1995; Winokur et al., 1994). It is not clear how this mutation results in muscular dystrophy. It is postulated that the telomeric mutation mediates a “position effect” that alters the expression of a nearby muscle-specific gene (Altherr et al., 1995). Genes separated by genomic distances >2 Mb from heterochromatin have been reported to be affected by position effect variegation in Drosophila (Bedell et al., 1996). Therefore, ALP should be considered a candidate gene for FSHD. In preliminary studies, we have not detected consistent changes in ALP expression in muscle biopsies from FSHD tissues. However, the muscle samples from FSHD patients analyzed for ALP expression may not have been from the critically affected muscle groups or from appropriate developmental stages.

What might be the normal function of ALP? Determining the role of the LIM motif in ALP remains a critical question. LIM motifs were first identified in protein products from three different genes, lin11 (Freyd et al., 1990), isl1 (Karlsson et al., 1990), and mec3 (Way and Chalfie, 1988), which all contain two LIM domains in association with a homeodomain DNA binding motif. These transcription factor LIM proteins participate in cell fate determination. Many distinct classes of LIM proteins have now been identified that do not have a homeodomain but still participate in cell fate determination (Sanchez-Garcia and Rabbitts, 1994). At the biochemical level, LIM motifs are implicated in protein–protein interactions. LIM motifs have been shown to interact with basic helix-loop-helix transcription factors (Wadman et al., 1994), protein kinase C (Kuroda et al., 1996), receptor tyrosine kinase tight turn structure (Wu et al., 1996), and other LIM motifs (Schmeichel and Beckerle, 1994). In striated muscle, the LIM-only protein, MLP, occurs at the Z lines and is an essential regulator of myogenic differentiation (Arber et al., 1994). Targeted disruption of MLP results in a disorganization of the myocyte cytoarchitecture (Arber et al., 1997). It will be interesting to determine whether the LIM domain of ALP interacts with MLP or other LIM proteins at the Z lines. To decisively determine the role of ALP in cellular physiology and muscular dystrophy, it will be important to delete the function of ALP by knockout and dominant negative approaches.

The authors thank Jay Brenman for performing in situ hybridization and Rick Topinka for excellent technical assistance.

This work was supported by grants (to D.S. Bredt) from the National Institutes of Health, the National Science Foundation, the Council for Tobacco Research, the Searle Scholars Program, and the Lucille P. Markey Charitable Trust, as well as by grants (to D.S. Bredt, S.T. Winokur, and M.R. Altherr) from the Muscular Dystrophy Association.

ALP

actinin-associated LIM protein

EST

expressed sequence tag

FISH

fluorescence in situ hybridization

FSHD

fascioscapulohumoral muscular dystrophy

GST

glutathione S-transferase

INAD

inactivation no afterpotential D

nNOS

neuronal nitric oxide synthase

ORF

open reading frame

RT-PCR

reverse transcription

ZO

zona occludens