Two mammalian UNC-45 isoforms are related to distinct cytoskeletal and muscle-specific functions

Previous studies have shown that the UNC-45 protein of C. elegans is required for normal thick filament assembly, binds Hsp90 and the myosin head, and shows molecular chaperone activity. We report here that mice and humans each have two genes that are located on different chromosomes, encode distinct UNC-45-like protein isoforms, and are expressed either in multiple tissues or only in cardiac and skeletal muscles. Their expression is regulated during muscle differentiation in vitro, with the striated muscle isoform mRNA appearing during myoblast fusion. Antisense experiments in C2C12 skeletal myogenic cells demonstrate that decreasing the general cell isoform mRNA reduces proliferation and fusion, while decreasing the striated muscle isoform mRNA affects fusion and sarcomere organization. These results suggest that the general cell UNC-45 isoform may have primarily cytoskeletal functions and that the striated muscle UNC-45 isoform may be restricted to roles in muscle-specific differentiation.

, however cardiac (sarcomeric) myosin heads are not (J. R. Sellers and H. L. Sweeney, personal communication). These results suggest that nonmuscle eukaryotic cells contain factors necessary for folding the non-muscle myosin motors (see Hutagalung et al., 2002). Previous research suggests that molecular chaperones assist the folding of muscle myosin heads. The chaperonin containing TCP-1 (CCT, where TCP-1 is t complex polypeptide 1) (Kubota et al., 1995) associates with and enhances the folding of nascent skeletal muscle heavy meromyosin in reticulocyte lysates (Srikakulam and Winkelmann, 1999). In addition, cultured muscle (C2C12) cells but not epithelial cells contain factors that permit the proper folding of recombinant muscle myosin subfragment 1 (Chow et al., 2002). The C. elegans protein UNC-45 has been shown by genetic experiments to be necessary for thick filament assembly (Epstein and Thomson, 1974;Venolia and Waterston, 1990) and by biochemical experiments to be a molecular chaperone with activity for the myosin head (Barral et al., 2002).
C. elegans UNC-45 protein has an apparent molecular mass of 107 kDa. It contains three amino-terminal tetratricopeptide repeats (TPR), a ~400 residue central region and a ~400 residue UNC-45/Cro1/She4p (UCS) domain (Barral et al., 1998;Venolia et al., 1999). The UNC-45 TPR domain binds the molecular chaperone Hsp90 in a stoichiometric manner (Barral et al., 2002). The remainder of the UNC-45 protein binds the myosin head and has chaperone activity on it (Barral et al., 2002). A temperature-sensitive mutation (unc-45 e286) in the UCS domain of UNC-45 reduced myosin accumulation and led to disordered assembly of the two myosin isoforms in body-Previous studies have shown that the UNC-45 protein of C. elegans is required for normal thick filament assembly, binds Hsp90 and the myosin head, and shows molecular chaperone activity. We report here that mice and humans each have two genes that are located on different chromosomes, encode distinct UNC-45-like protein isoforms, and are expressed either in multiple tissues or only in cardiac and skeletal muscles. Their expression is regulated during muscle differentiation in vitro, with the striated muscle isoform mRNA appearing during myoblast fusion. Antisense experiments in C2C12 skeletal myogenic cells demonstrate that decreasing the general cell isoform mRNA reduces proliferation and fusion, while decreasing the striated muscle isoform mRNA affects fusion and sarcomere organization. These results suggest that the general cell UNC-45 isoform may have primarily cytoskeletal functions and that the striated muscle UNC-45 isoform may be restricted to roles in muscle-specific differentiation.
wall muscle thick filaments (Barral et al., 1998). The resulting thick filaments were unstable upon isolation. UNC-45 protein was localized to the A-bands of body-wall muscle by immunostaining, and to all muscle types of C. elegans by a GFP-reporter (Venolia et al., 1999;Ao and Pilgrim, 2000). UNC-45 also has a role in non-muscle cells, as shown by its immunodetection in cleavage furrows and its interaction with non-muscle myosin II and unconventional myosin V in two hybrid studies (W. Ao and D. Pilgrim, personal communication).
Yeast and other fungal proteins containing an UCS-domain were identified in several distinct mutant screens and show functional linkage to the cytoskeleton. CRO1 of Podospora anserina is necessary for multiple processes in cell division (Berteaux-Lecellier et al., 1998). She4p of S. cerevisiae is required for a normal actin-based cytoskeleton, endocytosis, and a myosin-V-based molecular transport process (Jansen et al., 1996;Wendland et al., 1996;Beach and Bloom, 2001). A more recently identified UCS-containing protein, S. pombe Rng3p, is crucial for cell shape, normal actin cytoskeleton, and contractile ring assembly (Balasubramanian et al., 1998). It is essential for assembly of the myosin II-containing progenitors of the contractile ring (Wong et al., 2002). Widespread defects in the cytoskeleton are found in null mutants of all three fungal proteins.
The present investigation was aimed at identifying mammalian UNC-45 gene products and gaining insight into their functions. While there is a single C. elegans unc-45 gene, we found that humans and mice express two UNC-45-like gene products, one being present in all organs and the other highly expressed only in striated muscle. The expression of the general cell isoform decreased while the striated muscle isoform increased during muscle differentiation in vitro. Antisense experiments with the murine UNC-45 isoforms in the skeletal myogenic cell line C2C12 suggest that they have distinct roles.
Radiation hybrid chromosomal mapping of the human general cell UNC-45 locus PCR with total human genomic DNA and primers 5′-CCA AGG CTC ATG CAC ACG CTA CCT ATT GTG G-3′ and 5′-TGA TAG CAG GCT CAC AGC TGG TGA GGC TGC-3′ amplified only the 325 bp of the 3′ UTR of human general cell UNC-45. This primer pair was utilized at Research Genetics (Huntsville, AL) for dual mapping on the Stanford and G3 human/hamster hybrid panels (Cox et al., 1990).

Northern blotting
Total RNA was isolated from mouse tissues or C2C12 myogenic cells following Chomcznski and Sacci (Chomcznski and Sacci, 1987) or with TRIzol (Life Technologies, Rockville, MD), or was purchased (ovary, whole embryo) (Ambion, Austin, TX). Methods for the RNA gels and northern blotting were modified from Sambrook et al. (Sambrook et al., 1989). RNA molecular weight standards ranging from 0.24 to 9.46 kb (Life Technologies) were used to generate a linear plot of migration distance. Ten µg of each sample was separated on a 1.5% agarose-formaldehyde gel. Blots were pre-hybridized and hybridized in UltraHyb solution (Ambion) at 65°C. Two different general cell UNC-45 probes were used, one was 955 bp of cDNA from the UCS region, the product of primer set 5′-CTC GGC ATT GGT CAA TTG CAC CAA CAG C-3′ and 5′-GGA TCT CCA GGA CCT CAC TCT CCA TCA GGG-3′, while the other was 550 bp, half of which is 3′ UTR and was the product of primer set 5′-CTC ACC TCC ATG CGG CCA CAC-3′ and 5′-GAT GCT CCC AGC ATG TGA GGA TGC-3′. The striated muscle UNC-45 probe was 755 bp of cDNA, 350 of which was 3′ UTR, and was the product of primers 5′-TAC GGC AGG CAG CCA CCG AAT GCA TGT G-3′ and 5′-CAG TCT ACA GCC CGT TAT CTG GCC TGC-3′. Twenty-five ng of each probe was labeled with [ 32 P]-dCTP (Amersham, Piscataway, NJ) by the random nonomer method to specific activities of 5×10 8 -1×10 9 cpm/mg, and used at 1×10 6 cpm/ml. Blots were washed at room temperature with 2× sodium citrate/sodium chloride [SSC (see Sambrook et al., 1989)], 0.1% sodium dodecyl sulfate (SDS) for 2 minutes, and then at 63°C with three 15 minutes washes with 0.1× SSC, 0.1% SDS and 1 wash with 0.1× SSC, 0.5% SDS. Internal standards were obtained by re-labeling with a probe specific for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Ambion) or, in the case of multiple tissues constitutively expressing varying levels of GAPDH mRNA, with an 18 S RNA probe. Blots were exposed to Kodak BioMax MS film with BioMax intensifying screens for 3 days (general cell UNC-45), 11 hours (striated muscle UNC-45) or 2 hours (18S RNA and GAPDH) at -80°C. Films were digitized by scanning, and Photoshop (Abobe Systems, Mountain View, CA) was used for sizing and labeling. Absolute and relative amounts of general cell and striated muscle UNC-45 mRNA were obtained from films with UN-SCAN-IT software (version 5.1, Silk Scientific, Orem, UT) providing total pixels per band, and resulting data were analyzed using the Excel program (Microsoft, Redmond, WA).

In situ hybridization
The mouse general cell and striated muscle UNC-45 cDNA templates were identical to those used for northern blotting. Single-stranded sense and anti-sense digoxygenin-labeled RNA probes were generated as directed by the manufacturer (Roche Molecular Biochemicals, Indianapolis, IN). Whole-mount in situ hybridization was performed on mouse embryos of different ages as described in Conlon and Rossant (Conlon and Rossant, 1992). Labeled embryos were placed on a layer of agarose and photographed through a dissecting microscope with Kodak Ektachrome 160 T slide film. Slides were digitized via a Sprint Scan scanner (Polaroid, Wayland, MA) and images were adjusted for scaling and size with the Photoshop program.
C2C12 mouse skeletal myogenic cell culture Mouse C2C12 skeletal muscle cells were obtained from the American Type Culture Collection (ATCC CRL-1772) (Yaffe and Saxel, 1977;Blau et al., 1983;Silberstein et al., 1986). Cells were expanded and passaged 3 times, and frozen aliquots were used for antisense experiments. Cells were propagated in a humidified incubator at 10% CO2 in growth medium consisting of Dulbecco's Minimal Essential Medium containing 4.5 mg glucose/L, 110 mg sodium pyruvate/L and supplemented with 10% fetal bovine serum (FBS), 0.05 mg/ml gentamicin and a 1:400 dilution of Fungizone (all components from Invitrogen, Carlsbad, CA). Differentiation was induced by medium containing 2% horse serum instead of 10% FBS. Phase-contrast images were taken with a 10× lens on Kodak TriX Pan film at ASA 400. Negative film images were digitized by scanning and imported into Photoshop for contrast adjustment and cropping.

Antisense oligonucleotide experiments
Phosphorothioate 21-mer oligonucleotides (Sigma-Genosys, Woodlands, TX) designed to anneal at the start codon were used for the suppression of the mouse general cell UNC-45 (5′-GCC ACT CAC AGT CAT CAC GAA-3′) and striated muscle UNC-45 (5′-TTC AGC CTC TGC CAT AGT CTT-3′). The control oligonucleotide had a base composition similar to the UNC-45 oligonucleotides but randomized (avoiding tandem repeats) (5′-TAA GCA CTA GGA CAC CTC CAC-3′). Second-generation chimeric 18-mer oligonucleotides (trademark of Oligos Etc. Inc, Wilsonville, OR) were also used: general cell UNC-45 antisense: 5′-CGC ATT TGA ACA GCT CGT-3′, a control oligonucleotide that reversed the order of the previous bases (5′-TGC TCG ACA AGT TTA CGC-3′), and striated muscle UNC-45 antisense: 5′CCA TGA GGC TGC AGA TTC-3′. Oligonucleotides were resuspended in 10 mM Tris, pH 8.5 and lyophilized in aliquots. They were added daily, starting 1 hour after plating, to between 1.25 and 5 µM. Proliferating cells were treated for up to 5 days, while myotube-forming cells were treated for up to 9 days. Cell number was assessed by three methods: counting adherent cells in defined areas, counting trypsinized cells resuspended in trypan blue stain, and by the CyQUANT proliferation assay (Molecular Probes, Eugene, OR). For the latter, 96 well culture plates were seeded with 2,000 cells/well. Eight wells or triplicates of eight wells were used for each 3-day treatment. Frozen cells were lysed with the kit's lysis buffer supplemented with NaCl to 180 mM and EDTA to 1 mM, and 2 Kunitz units of RNase A/ml. After 1 hour at room temperature, one volume of GR dye solution was added to a final concentration of 2× dye, and the fluorescence resulting from GR dye binding to DNA (Jones et al., 2001) was read on a CytoFluor plate reader at 485 nm excitation and 530 nm emission, with 3 reads per scan and a gain of 50. A standard curve showing a linear relationship was obtained using 2, 4, 8, 12, and 16×10 3 C2C12 cells. Data was tabulated and averages±standard deviations were calculated for the general cell or striated muscle antisense data separately paired with the control data using the Excel program. The Student's t-test was applied using the two-tailed, equal variance parameters.
Immunoblotting C2C12 myotubes cultures treated for 8 days plus or minus 2.5 µM antisense oligonucleotides were rinsed several times with Hank's balanced salt solution. Cells within one well of a six-cell plate were lysed with 0.2 ml of 150 mM NaCl, 20 mM Tris pH 7.4, 2 mM EDTA, 2 mM adenosine triphosphate, 5 mM dithiothreitol, 1% Triton X-100, and Complete protease inhibitor cocktail (Boehringer Mannheim, Indianapolis, IN) to 5× concentration. SDS was then added to 1%. Proteins separated on 7.5% polyacrylamide gels were transferred to Immobilon-NC filters in 20% methanol in Laemmli running buffer, at 80 volts for 15 hours in the cold. Use of Kaleidoscope molecular weight markers (BioRad, Richmond, CA) and subsequent Commassie Blue-staining of the gels confirmed transfer. Blots were blocked with 1% nonfat dry milk in 150 mM NaCl, 50 mM Tris pH 7.6, 0.05% Tween-20, reacted with 2 µg/ml MF20 monoclonal anti-skeletal muscle myosin antibody (Bader et al., 1982), and counter-stained with 1:1000 dilution of horse radish peroxidase-labeled anti-mouse IgG. SuperSignal enhanced chemiluminescence (Pierce, Rockford, IL) and exposure to X-ray film detected labeling.

Results
Vertebrates have two UNC-45 isoforms that are products of separate genes cDNA sequences encoding human and mouse UNC-45 proteins were identified using BLAST (Altschul et al., 1990;Altschul et al., 1997;Gish and States, 1993;Madden et al., 1996) searches of EST databases with the C. elegans UNC-45 amino acid sequence as a query. Two different isoforms, a general cell (GC) and a striated muscle type (SM), of mouse and human UNC-45 exist, as shown by PCR bridging reactions that utilized fragments of mouse or human UNC-45-like coding sequences (Fig. 1). The names of these isoforms are based on the differential expression of their mRNAs.
Like C. elegans UNC-45, the predicted sequences for the mammalian proteins consist of three distinct regions: an amino-terminal triple TPR motif, a unique central region, and a C-terminal UCS domain. Although the two mammalian UNC-45 isoform proteins predicted from full-length cDNA sequences are each 31-32% identical and 53-54% similar to C. elegans UNC-45, they are quite different from one another. Overall, the mammalian GC and SM UNC-45 isoforms are only 55-56% identical and 74% similar in amino acid sequence. The most similar isoforms are 94-95% identical and 96-98% similar between mouse and human. Blocks of up to 10 consecutive identical and up to 25 conserved amino acids are present throughout the isoform sequences.
The GC and UNC-45 isoforms are predicted to have molecular weights of 103.450 and 103.41×10 3 . Structural prediction programs indicate that the pI of the GC UNC-45 protein would be 6.0, close to that of the C. elegans protein. The SM UNC-45 protein is predicted to have a more basic pI of 8.2. The GC and SM UNC-45 proteins are each predicted to consist of about 3:1 α-helical: random coil content with negligible amounts of beta sheet [ (Rost, 1996) www.embl-heidelberg.de/predictprotein].
The homologous human and mouse cDNAs for the ubiquitously expressed general cell UNC-45 (see below) that we cloned are represented in the GenBank database as human SMAP1 (direct submission, acc. no. BAB20273.1) and a human colon adenocarcinoma cDNA (direct submission, acc. no. AAH06214) and as a mouse cDNA from an induced mammary tumor cDNA (direct submission, acc. no. AAH04717). There is no full-length cDNA in the GenBank database for either the homologous human or mouse striated muscle UNC-45 isoform; (acc. nos. requested). The human SM UNC-45 gene product predicted in the NCBI database (acc. no. XP_091530; gene LOC146862, encoded by acc. no. XM_091530.1) is incorrect. It contains three extra exons encoding 122 additional residues when compared to the SM UNC-45 protein deduced from multiple PCR products.
Radiation hybrid mapping placed the human general cell gene locus on chromosome 15q25-26, correlating well with the human genome project placement of the gene at that location between 96,331,340-96,352,778 bp. The predicted mRNA XM_038413 and protein XP_038413 are in agreement with our cloned cDNA and the appropriate BAC sequences. No human disease loci have yet been mapped near this region.
residues plus 330 bases of 3′ UTR. The loci of the human and mouse GC unc-45 are syntenic. The mouse coding sequence is within acc. no. XM_124930, on NCBI contig NW_000327. The human SM UNC-45 gene is entirely contained in a bacterial artificial chromosome (acc. no. AC022916) and mapped to chromosome 17q11, between 33,872 and 33,834 Kb. The SM UNC-45 gene spans 38 kb, and consists of 19 exons. By synteny, the mouse SM UNC-45 gene would be on mouse chromosome 11 at locus 47.5 cM, and is in fact included in a mouse chromosome 11 BAC (acc. no. AL603745). The mouse gene is similar in size to the human gene, 34-kb. No human disease loci have yet been mapped to the SM UNC-45 locus.
Both the human and mouse GC UNC-45 genes are over 10 kb smaller than the SM UNC-45 genes, due to longer introns since total exonic sequences are nearly identical in length. Both GC UNC-45 genes contain one exon more than the SM UNC-45 genes, which encodes the extreme amino terminus. The intron/exon boundaries of the respective mouse and human genes for an isoform are identical, although the introns are of different sizes. The exons of the GC and SM UNC-45 genes are also identical within a species (Fig. 1). We assembled and aligned predicted UCS domains from the NCBI database (www.ncbi.nlm.nih.gov) (Fig. 2). The 17 full length UCS domains are from the species already mentioned and in addition from Neurospora crassa, Aspergillus fumigatus, a second genus of nematode, the mosquito

Fig. 2. Alignment of UCS domains from fungal homologues and animal UNC-45 proteins. Only one UCS protein is present in fungi and
invertebrates. Two UNC-45 isoforms, SM (striated muscle) and GC (general cell) are present in the bony fish Fugu rubripies, suggesting that the gene divergence is ancient. Note that the fungal proteins have extra residues in five places, compared with the animal proteins. Identity is indicated by black boxes, similarity by grey boxes. There is a clear distinction between invertebrate and vertebrate proteins. Blue arrowheads indicate completely conserved residues, and red arrowheads indicate 90-95% identity. Asterisks indicate conserved residues corresponding to C. elegans temperature-sensitive UNC-45 (Barral et al., 1998) and S. pombe Rng3p mutations (Wong et al., 2000). Ano. gam., Anopheles gambiae (coding sequence compiled from acc. no. AAAB01008844.1); Dros. melan., Drosophila melanogaster (acc. no. AAK93568); C. elegans, Caenorhabditis elegans [(Venolia et al., 1999) Fig. 2). These additions and deletions may represent specializations in function. The presence of two UNC-45 isoforms in the pufferfish gives rise to the notion that the second UNC-45 gene arose sometime during the chordate radiation [see accompanying Commentary ]. Pair-wise comparisons of identity showed that one of the Fugu UNC-45 isoforms is 72% identical and 84% similar to the human and mouse SM UNC-45. The other Fugu UNC-45 isoform is 64% identical and 78% similar to the human and mouse GC UNC-45 isoform.
The two murine UNC-45 isoform genes are differentially expressed in adult tissues The two murine UNC-45 mRNAs are differentially expressed in the adult. GC UNC-45 mRNA was detected in uterus, large intestine, kidney, spleen, lung, brain, liver and ovary using gene-specific labeling of duplicate northern blots containing total RNA from various adult organs (Fig. 3A). The mRNA was relatively less abundant in cardiac and skeletal muscle than in the non-striated muscle tissues. The GC UNC-45 mRNA was also found in whole 12 day mouse embryos. The demonstration of GC UNC-45 mRNA in every tissue examined here confirms the multi-organ expression pattern demonstrated by EST database searches which showed other GC UNC-45 expressing cells or organs include skin, bone marrow, T-cells, urinary bladder, mammary gland, optic nerve, various parts of the eye, germ cells, testis, prostate, pancreas, parathyroid gland and placenta. In addition, over a dozen tumors of various celltypes express GC UNC-45, some to a higher than normal level as suggested by SAGE (serial analysis of gene expression) analysis [see UniGene Cluster Hs.26110 Homo sapiens in the NCBI database (www.ncbi.nlm.nih.gov)].
The SM UNC-45 mRNA is abundant in skeletal muscle and the heart, both of which consist predominantly of striated muscle fibers (Fig. 3B). The same size mRNA was also present in whole embryo samples. However, SM UNC-45 mRNA was not detected in uterus and large intestine, which are rich in smooth muscle cells nor in non-muscle organs such as kidney, liver and ovary. The source of the minor SM reaction in the lung sample is unknown. A TBLASTN query (Altschul et al., 1997) of the EST database reveals that the partial SM UNC-45 sequences are present in cDNA libraries derived mostly from heart and segments of embryo containing the developing heart, such as embryonic body between the diaphragm region and the neck, as well as tissues containing skeletal muscle such as limbs and total head tissue.
The two unc-45 genes are differentially expressed during embryogenesis To determine the location of early expression of the GC and SM UNC-45 RNA species, in situ hybridization was performed on whole mouse embryos using sense controls and anti-sense RNA probes. The SM UNC-45 gene was strongly expressed in the heart at 8.75 days when it is already beating (Fig. 4A), and was not expressed in other organs above the background level seen in a sense control embryo (Fig. 4B). The GC UNC-45 gene was expressed at high levels in all tissues of an 8-day embryo, the earliest stage embryo obtainable, whereas by 9.75 days the most robust expression was in regions of intense development such as the branchial arches and the forelimb bud (data not shown).
Journal of Cell Science 115 (21) Fig. 3. The two murine UNC-45 mRNAs are differentially expressed. Duplicate northern blots of total RNA from various adult mouse organs and from 12 day embryos were labeled with isoformspecific probes (SM, skeletal muscle; He, heart; Ut, uterus; In, large intestine; Ki, kidney; Sp, spleen; Lu, lung; Br, brain; Li, liver; Ov, ovary; Em, embryo). The GC UNC-45 isoform was detected in all adult organs examined, and the SM UNC-45 isoform was detected in organs consisting mainly of striated muscle. A small amount of GC UNC-45 mRNA was also present in striated muscle tissues. The minor SM UNC-45 band in the lung sample is of unknown origin. Labeling of 18 S RNA indicated comparable loading. GC UNC-45 and SM UNC-45 mRNAs are differentially expressed during muscle differentiation in vitro Having determined that the mRNAs for the two isoforms of mammalian UNC-45 are differentially expressed during development and in adult tissues, we addressed when during muscle differentiation the striated muscle isoform is expressed. To assess the relative expression of the two UNC-45 isoform mRNAs, total RNA was isolated from C2C12 cells proliferating in growth medium, and from three stages of muscle differentiation induced by changing confluent cultures to differentiation medium, i.e. fusing myoblasts present at 2 days, young myotubes present at 3.5 days, and older myotubes present after 6 days of differentiation, some of which twitched (Fig. 5A). Total RNA from mouse skeletal muscle and uterus were used as respective positive controls for SM UNC-45 and GC UNC-45 mRNA.
As demonstrated in Fig. 5B, only GC UNC-45 mRNA was expressed in the proliferating C2C12 myoblasts. GC UNC-45 mRNA expression was at the highest level in these cells relative to the differentiating cultures at any stage, as shown by the total pixel number/band normalized to the internal standard of GAPDH (Fig. 5C). SM UNC-45 mRNA was first expressed when aligned myogenic cells were actively fusing (Fig. 5B). This population of fusing myoblasts and the earliest, narrow diameter myotubes expressed GC UNC-45 mRNA at 73% of the level in proliferating cells. Young myotubes in the process of assembling and remodeling myofibrils had the highest relative expression of SM UNC-45, 1.5 times greater than that in fusing myoblasts. The level of GC UNC-45 expression continued to decline during myotube maturation, from 25% in young myotubes to 14% in older myotubes, relative to that in proliferating cells. SM UNC-45 mRNA expression decreased in older myotubes to about half of the maximum found in younger myotubes.

GC UNC-45 functions in cell proliferation
Antisense experiments were performed to test whether the GC and SM UNC-45 isoforms exhibited different functions. The differential tissue expression of the two isoforms suggested that the GC UNC-45 may have a role in cytoskeletal functions and the SM UNC-45 may have a more specialized role in sarcomere assembly and function. Therefore the antisense treatments of C2C12 cells focused on cell proliferation and muscle differentiation. For studies of the effects of suppression of UNC-45 isoform mRNA in proliferation, C2C12 cells were plated at low density and antisense oligonucleotides were added daily to 2.5 µM for three days. This concentration permitted only partial mRNA suppression. However, higher concentrations of oligonucleotides had obvious toxic effects, evidenced by cell death in control cultures.
The extent of suppression of UNC-45 mRNA expression by treatment with antisense oligonucleotides was determined by northern blotting. Three days of treatment of proliferating C2C12 cultures with 2.5 µm GC UNC-45 antisense oligonucleotides reduced GC UNC-45 mRNA expression to half the amount in the control or SM UNC-45 antisense treated cultures (Fig. 6A). Cell proliferation was judged from total DNA as determined by fluorescence (Jones et al., 2001). The GC UNC-45 antisense oligonucleotides suppressed cell proliferation to 68-75% of values obtained by treatment with the control reverse GC oligonucleotide within 3 days of treatment (Fig.   6B,C). This reduction is significant (P=0.00008). Treatment with the SM UNC-45 antisense oligonucleotides did not significantly effect the proliferation rate (P=0.36). Cell viability was not affected by 2.5 µM oligonucleotide, as assessed by all cultures having less than 1% of trypan blue-stained cells.  to half the control levels (Fig. 7A). SM UNC-45 mRNA is expressed at control levels in the GC antisense-treated cultures, reflecting their limited differentiation (Fig. 7A).
For quantitative analysis of differences in myotube and sarcomere formation in C2C12 cultures treated with antisense oligonucleotides, over 600 cells from each treatment were scored for the number of nuclei per cell, expression of muscle-specific sarcomeric α-actinin as viewed in low magnification, and the pattern of staining of sarcomeric α-actinin viewed in high magnification. GC UNC-45 antisense treatment severely reduced myoblast fusion so that the positively stained cells contained predominantly one to four nuclei (Fig. 7B,D). Treatment with the SM UNC-45 antisense oligonucleotide reduced fusion but to a lesser extent. These cells were able to fuse to form multinuclear myotubes containing over a dozen nuclei (Fig. 7B,D). Reduction of SM UNC-45 mRNA affected sarcomere formation more directly, so that about half the myotubes had only small submembraneous structures with striated α-actinin while the interior of the myotubes was largely unstriated (Fig. 7C).
Considering the paucity and instability of thick filaments in C. elegans unc-45 temperature-sensitive mutants (Barral et al., 1998), we addressed whether reduction of SM UNC-45 had an effect on the total amount of skeletal myosin heavy chain in the C2C12 myotube cultures. Skeletal muscle myosin heavy chain (MHC) was specifically detected by immunoblotting using MF 20 antibody against equivalent amounts of total protein from myotube cultures treated 8 days plus or minus negative control or antisense oligonucleotides directed against GC or SM UNC-45. Reduction of SM UNC-45 mRNA had no significant effect on the amount of skeletal MHC in the myotube cultures compared to untreated and negative control antisense-treated cultures (Fig. 8). GC UNC-45 antisense-treated cultures had less than half as much skeletal MHC as the other samples, consistent with the suppression of myotube formation (Fig. 8). In these cultures skeletal MHC was derived from the small myotubes containing two to four nuclei and the rare larger myotubes, which were positively stained by the MF20 antibody, in contrast to the larger multi-nucleated myotubes in controls (see Fig. 7B). The effects of antisense suppression upon sarcomere formation therefore do not appear to be related to changes in the amounts of MHC or other major proteins.

Discussion
UNC-45 proteins throughout the animal kingdom have the same three-region structure of the canonical C. elegans protein, with an amino-terminal TPR domain consisting of three tandem TPR motifs, a central region that so far is unique to UNC-45 proteins and a C-terminal UCS domain. In contrast to flies and worms, there are two isoforms of UNC-45 in vertebrates, ranging from the pufferfish to humans. The mammalian UNC-45 isoforms have the same predicted molecular weight, but differ in predicted pI. The encoding genes show distinct chromosomal location and are differentially expressed in both embryonic and adult tissues. The general cell isoform appears to be ubiquitously expressed, while the striated muscle isoform was found only in cardiac and skeletal muscle.
The C2C12 skeletal myogenic cell line allows one to examine general cytoskeletal functions in the proliferative stage as well as muscle-specific functions during differentiation. Consistent with its widespread expression in organs, GC UNC-45 mRNA is expressed in proliferating, nondifferentiated myoblasts. This expression level is greater than any stage of muscle differentiation. By the time more robust myotubes have developed, the levels of GC UNC-45 mRNA are only 14% of those in proliferating cells. Reducing the GC UNC-45 mRNA to about a half-normal level by antisense treatment decreased C2C12 cell proliferation to 68-75% of control values. The GC UNC-45 antisense treatment also inhibited myoblast fusion. In contrast to the GC isoform, SM UNC-45 mRNA was not expressed until myogenic cells started fusing, with the highest levels of expression in young myotube cultures. Assembly and remodeling of myofibrils is highest Journal of Cell Science 115 (21) during young myotube formation, consistent with SM UNC-45 being involved in the process of thick filament assembly and sarcomere formation. Unlike the GC UNC-45 antisensetreated cells, cells treated with antisense SM UNC-45 oligonucleotides were able to fuse and form myotubes. However, half of these myotubes lacked striated myofibrils, as identified by staining for sarcomeric α-actinin.
The SM to GC UNC-45 mRNA expression ratio in older myotube cultures was 2.8. This ratio contrasts with the 18 to 1 ratio of SM to GC UNC-45 mRNA in adult mouse skeletal muscle. This difference is most likely a consequence of only partial differentiation in cell cultures versus in vivo. The basis for less SM UNC-45 mRNA in older as compared to younger  myotube cultures is unknown. However, protein half-life may vary and in addition the relationship between protein and mRNA expression need not be proportionate. This relationship can be determined for the UNC-45 isoforms when specific antibodies become available. Further work is needed to determine whether both GC and SM UNC-45 are present in striated muscle fibers. The antisense results correlate with the mRNA expression patterns indicating that GC UNC-45 has a role in processes involving the cytoskeleton and SM UNC-45 has a more muscle-specific function in sarcomere assembly. We cannot rule out the possibility of functional overlap between the two isoforms because suppression of either isoform had an affect on cell fusion and sarcomere formation. To date, it is not known what myosins are involved in the process of myocyte fusion. Based on their mRNA expression and the different roles of the two UNC-45 isoforms in C2C12 myogenesis, these two isoforms most likely have separate and distinct activities, possibly mediating functions that involve different classes of myosin. However, either isoform may function in myosin folding, assembly, and/or contractile activity.
Previous studies have shown that proteins in the UCS domain family are required for a variety of myosin-and actin-based processes utilizing both conventional and unconventional myosins. She4p is required for mRNA transport involving an unconventional myosin type V in S. cerevisiae (Beach and Bloom, 2001;Jansen et al., 1996). Temperature-sensitive mutations in the essential gene rng3, a member of the UCS family, block the assembly of the actomyosin ring during cytokinesis, and are synthetically lethal with mutations in the cytoskeletal myosin II gene (Balasubramanian et al., 1998). C. elegans UNC-45 has been shown to directly bind the head of muscle myosin II (Barral et al., 2002), and to interact with cytoskeletal type II and unconventional type V myosins through two-hybrid analysis (W. Ao and D. Pilgrim, personal communication). In addition, C. elegans UNC-45 was recently shown to be a myosin-targeted chaperone since it prevents the thermally induced aggregation of the myosin head (Barral et al., 2002). This finding, in conjunction with UNC-45 binding the chaperone Hsp90, suggests that UNC-45 might influence thick filament assembly through a role in myosin folding. The other UCS domain proteins may function likewise for different myosin substrates. In this regard, it is significant that wild-type S. pombe Rng3p is sequestered only by myo2-E1 myosin II, which has a mutation in the myosin head and leads to defective contractile ring assembly (Wong et al., 2000).
Based on their mRNA expression and the different effects of suppression of the two UNC-45 isoforms in C2C12 muscle differentiation, we propose that they are involved in distinct functions. The GC UNC-45 isoform appears to have more of a general function, possibly being involved in cell division, whereas the SM isoform is related to striated muscle differentiation including myofibril formation. The functions of the two isoforms are not necessarily independent of one another, because the cytoskeleton is needed for formation and maintenance of myofibrils. Lack of cytoskeletal proteins such as vinculin (Barstead and Waterston, 1991), talin (Moulder et al., 1996), desmin (Li et al., 1997;Milner et al., 1996), certain isoforms of integrin (Gettner et al., 1995;Moorthy et al., 2000;Volk et al., 1990) and spectrin (Hammarlunda et al., 2000;Norman and Moerman, 2002) as well as the extracellular matrix protein perlecan (Rogalski et al., 1995) leads to defects in myofibril formation and maintenance. Therefore both UNC-45 isoforms may be necessary for functional sarcomeres.
Our results are consistent with a model (Fig. 9) in which the two UNC-45 isoforms have separate, but possibly overlapping functions in striated muscle differentiation. In this model, GC UNC-45 would be involved in cell proliferation and cytoskeletal maintenance of myofibrils once they have formed. SM UNC-45 would function in the development of sarcomeres. These two proteins may interact with different myosins in their respective functions.