Slow and fast fiber isoform gene expression is systematically altered in skeletal muscle of the Sox6 mutant, p100H

Authors

  • Nobuko Hagiwara,

    Corresponding author
    1. University of California, Davis, Division of Cardiovascular Medicine, Rowe Program in Genetics, Davis, California
    • Division of Cardiovascular Medicine/Rowe Program in Human Genetics Tupper Hall Room 4446 University of California, Davis One Shields Avenue, Davis, CA 95616
    Search for more papers by this author
  • Betty Ma,

    1. University of California, Davis, Division of Cardiovascular Medicine, Rowe Program in Genetics, Davis, California
    Search for more papers by this author
  • Alice Ly

    1. University of California, Davis, Division of Cardiovascular Medicine, Rowe Program in Genetics, Davis, California
    Search for more papers by this author

Abstract

We have previously demonstrated that p100H mutant mice, which lack a functional Sox6 gene, exhibit skeletal and cardiac muscle degeneration and develop cardiac conduction abnormalities soon after birth. To understand the role of Sox6 in skeletal muscle development, we identified muscle-specific genes differentially expressed between wild-type and p100H mutant skeletal muscles and investigated their temporal expression in the mutant muscle. We found that, in the mutant skeletal muscle, slow fiber and cardiac isoform genes are expressed at significantly higher levels, whereas fast fiber isoform genes are expressed at significantly lower levels than wild-type. Onset of this aberrant fiber type-specific gene expression in the mutant coincides with the beginning of the secondary myotube formation, at embryonic day 15–16 in mice. Together with our earlier report, demonstrating early postnatal muscle defects in the Sox6 null-p100H mutant, the present results suggest that Sox6 likely plays an important role in muscle development. Developmental Dynamics 234:301–311, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

During vertebrate development, the Sox (Sry related HMG box) family of transcription factors governs differentiation of various types of tissues (Wegner, 1999; Kamachi et al., 2000). Mutations of the SOX genes are known to cause severe hereditary diseases, such as Campomelic dysplasia (SOX9; Wagner et al., 1994; McDowall et al., 1999) and Waardenburg-Hirschsprung disease (SOX10; Pingault et al., 1998), suggesting their critical roles in mammalian development. Since the identification of Sry, the original Sox family member (Sinclair et al., 1990), close to 30 Sox genes have been identified in vertebrates (Wegner, 1999; Schepers et al., 2002). Sox genes contain the HMG box, a highly conserved DNA binding domain in the family. Although the HMG boxes of Sox genes bind to the same heptamer sequence in vitro (Wegner, 1999), each individual Sox gene regulates a distinct set of genes during the differentiation of different cell types. It is now thought that Sox proteins achieve their specificity as part of multiprotein complexes unique to different cell types (Kamachi et al., 2000; Wilson and Koopman, 2002).

As with many members of the Sox family, Sox6 is expressed in multiple cell types and is implicated in the regulation of more than one gene. The Sox6 gene was first isolated from an adult testis cDNA library and was also shown to be highly expressed in the developing central nervous system (Connor et al., 1995). Previously, we have reported that Sox6 mRNA is expressed in a wide variety of adult tissues in both mice and humans, with the most abundant expression in skeletal muscle (Hagiwara et al., 2000; Cohen-Barak et al., 2001). Sox6 was first implicated in bone development, where it functions as a transcriptional activator of the Collagen 2a1 gene along with Sox5 and Sox9 (Lefebvre et al., 1998). The importance of the Sox5 and Sox6 genes in bone development was later confirmed when severe bone defects were observed in the Sox5/Sox6 double knockout mouse (Smits et al., 2001). Reflecting its expression in multiple tissues, Sox6 has also been implicated in the regulation of FGF3 expression in the developing inner ear (Murakami et al., 2001), mRNA splicing (Ohe et al., 2002), and neuronal differentiation in culture (Hamada-Kanazawa et al., 2004). However, despite its abundant expression in skeletal muscle (Hagiwara et al., 2000; Cohen-Barak et al., 2001), the role of Sox6 in muscle development has not been determined.

We reported previously that Sox6 is critical for normal muscle development by documenting the myopathy and cardiac conduction abnormalities in the Sox6 null mutation p100H (Hagiwara et al., 2000). In the p100H mutant, a chromosomal inversion disrupts the Sox6 gene along with the pink-eyed dilution (p) gene, which solely regulates pigmentation. The distal break point of the inversion causes truncation of the Sox6 gene, resulting in the loss of the HMG DNA binding domain (Hagiwara et al., 2000). We reported that the Sox6 null p100H mouse exhibited stunted growth and died within 2 weeks after birth, which was also reported in Sox6 knockout mice (Smits et al., 2001). In addition, we have found degeneration in both skeletal and cardiac muscles and cardiac conduction abnormalities in the Sox6 null p100H mouse. Based on these observations, we hypothesized that Sox6 plays an important role in muscle development.

In this report, we demonstrate that p100H mutant skeletal muscles show a significant increase in mRNA expression of slow skeletal and cardiac muscle isoform genes and a significant decrease in the expression of fast isoform genes. The differential expression of these genes in the mutant skeletal muscle begins around embryonic day (E) 15.5, a stage coinciding with the beginning of secondary myotube formation during embryonic muscle development in mice (Kelly and Zacks, 1969; Ontell and Kozeka, 1984; Stockdale, 1992). Because of the systematic change in fiber type-specific gene expression observed in the Sox6 null-p100H mutant during critical stages of embryonic muscle development, it is likely that Sox6 plays an important role in terminal differentiation and/or maturation of skeletal muscles.

RESULTS

Expression of Fiber Type-Specific Genes Is Systematically Altered in the p100H Mutant Skeletal Muscle

Microarray analyses were performed to identify differentially expressed genes between p100H mutant and wild-type skeletal muscle at E18.5. Genes showing more than a 1.5-fold difference with a P value smaller than 0.05 were categorized by their biological functions and further analyzed. From this analysis, we have found clear trends in muscle-specific genes whose expression is altered in the mutant skeletal muscle. The expression of slow skeletal isoform and cardiac isoform genes is significantly increased in the mutant, whereas the expression of fast skeletal isoform genes is significantly decreased in the mutant (Table 1). The microarray results were confirmed by Northern hybridization. As shown in Figure 1, skeletal muscle RNA from the p100H homozygote showed significantly higher expression of slow skeletal isoform (myosin light chain 2v [MLC2v], sarcoplasmic reticulum Ca2+-ATPase 2 [SERCA2], troponin I slow [TnI slow], myosin heavy chain β [MHCβ], troponin C slow [TnC slow]), and cardiac isoform genes (Troponin T2, cardiac [Tnnt2]), and lower expression of fast skeletal isoform genes (TnI fast, Calsequestrin 1).

Table 1. List of Muscle Structural Genes Showing Significant Changes in the p100H Mutant Skeletal Musclea
Classification and gene nameGenBank acc. no.Fold changeFiber typeParametric P value
  • a

    The table represents the muscle genes that showed greater than 1.5-fold changes with statistical significance by class comparison (p100H/p100H vs. wild type). − indicate that those genes show decreased expression in the mutant skeletal muscle. The two genes in italics were deleted twice by two independent oligomer probe sets. GenBank accession numbers listed in the table correspond to those listed by Affymetrix. Fold changes and p-values were calculated by BRB Array Tools.

HIGHER in the p100H mutant    
Slow-twitch muscle isoforms    
 Myosin light chain, MLC2vM916022.01Slow/cardiac0.00187
 ATPase, Ca2+ transporting, SERCA2AF0299822.42Slow/cardiac0.000617
 Tropomyosin 3, gammaU045412.55Slow0.00101
 Troponin 1, skeletal, slow 1AJ2428742.68Slow0.000405
 Troponin T1, skeletal, slowAV2134312.74Slow0.00293
 Myosin alkali light chain, MLCIvX129723.09Slow/cardiac8.90E-06
 myh 7, (MHCβ)AJ2233623.33Slow/cardiac0.00144
 Troponin T1, skeletal, slowAJ1317113.67Slow2.70E-05
 Troponin C, cardiac/slow skeletalM297934.39Slow/cardiac0.000189
Cardiac muscle isoforms    
 Myosin light chain, MLCIaA16488501.56Cardiac0.00172
 Myosin light chain, MLCIaM194362.09Cardiac0.0116
 Troponin T2, cardiacL476002.86Cardiac5.91E-05
Others    
 Myosin binding protein HU682671.60 0.00595
 LOWER in the p100H mutant    
Fast-twitch muscle isoforms    
 Calsequestrin 1U93291−2.34Fast0.000175
 Myozenin 1/calsarcin-2AA733946−1.58Fast0.0399
 ParvalbuminX59382−1.50Fast0.02
 Troponin 1, skeletal, fast 2J04992−1.50Fast0.0127
 Troponin T3, skeletal, fastL48989−1.50Fast0.0259
Figure 1.

Northern hybridization of representative genes of Table 1. Each lane contains 5 μg of total RNA isolated from skeletal muscle of p100H homozygote and wild-type term embryos (embryonic day [E] 18.5). The blot was hybridized with 32P-α-dCTP–labeled cDNA fragments of the selected genes that were amplified by polymerase chain reaction. Control hybridization with β-actin was performed to confirm equal loading of RNA.

Postnatal expression of fiber type-specific genes was also examined using RNA isolated from P10 mice, prompted by the observations that, in postnatal life, fast skeletal isoform gene expression dominates that of slow skeletal isoform genes in most skeletal muscles (Weydert et al., 1987; Narusawa et al., 1987; Condon et al., 1990a; Lu et al., 1999; Calvo et al., 2001). The mRNA expression of the MLC2v, MHCβ, TnI slow (slow skeletal isoforms), and TnI fast (fast skeletal isoform) genes was analyzed by semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) using total RNA from pooled skeletal muscles of the body wall or diaphragm (which predominantly consists of fast twitch fibers; Fournier and Lewis, 2000). As shown in Figure 2, in both pooled skeletal muscles and diaphragm, slow skeletal isoform gene expression (MLC2v, MHCβ, TnI slow) is significantly higher in the mutant than in wild-type; on the other hand, fast skeletal isoform gene expression (TnI fast) is considerably lower in the mutant. The altered fiber type-specific gene expression in the mutant muscle, therefore, is not limited to the fetal stage in the Sox6 mutant.

Figure 2.

Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analysis confirming postnatal differential expression of fiber type-specific genes. Total RNA was prepared from P10 mouse muscle. A,B: Serial dilutions (10−1, 10−2, 10−3) of cDNA from pooled skeletal muscle (A) and diaphragm (B) were amplified for 30 cycles, except for MHCβ. For MHCβ, serial dilutions of 100, 10−1, 10−2 of cDNA were used as template and amplified for 35 cycles. RT(−) is a negative control without reverse transcriptase. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as ubiquitously expressed positive control to confirm the cDNA input level between the mutant (p100H/p100H) and wild-type.

Deviation of Fiber Type-Specific Gene mRNA Expression in the p100H Mutant Begins at Embryonic Day 15.5

The temporal expression of slow and fast isoform genes was examined in the Sox6 mutant and wild-type mice. To determine when the mRNA expression of slow and fast skeletal isoform genes in the mutant muscle starts to deviate from wild-type, TaqMan real-time PCR was performed. In all experiments, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used as a standard to normalize the fiber type-specific gene expression. To normalize gene expression between p100H mutant and wild-type mice, the comparative cycle threshold (Ct) was calculated (see Experimental Procedures section). In Figure 3A,B, the relative expression levels (mutant to wild-type) of TnC slow, TnI slow, and TnI fast are presented. At E15.5, both slow skeletal isoform genes (TnC and TnI slow) begin to show an increase in the p100H mutant skeletal muscle. Conversely, TnI fast shows a significant decrease in the mutant, down to half the level of wild-type by E18.5 (Fig. 3A). At the term embryo stage (E18.5), the expression of TnC slow in the mutant is six times higher than in wild-type. This trend continues through early postnatal stages and by postnatal day (P) 9, TnC expression in the mutant is ∼70-fold higher than wild-type (Fig. 3B). In wild-type, on the other hand, the exact opposite is seen during embryonic skeletal muscle development; fast skeletal isoform gene expression starts to dominate slow skeletal and cardiac isoform-specific gene expression beginning in mid-gestation (Calvo et al., 2001; Wang et al., 2001). This transition to fast fibers occurring in wild-type muscle is depicted in Figure 3C, in which TnI fast expression becomes greater than TnI slow expression after E15.5. In the mutant muscle, however, TnI fast expression never exceeds the expression of TnI slow, even at E18.5. Another group of thin filament-associating proteins troponin T (Tnnt1 slow, Tnnt2 cardiac, and Tnnt3 fast) also exhibited the same pattern of expression in the mutant muscle. As shown in Figure 4A, beginning at E15.5, mRNA expression of Tnnt1 (slow skeletal) and Tnnt2 (cardiac) continually increases in the p100H mutant, whereas Tnnt3 (fast skeletal) decreases. In postnatal stages, differences in their expression between wild-type and the mutant keep widening (Fig. 4B). In the tongue, which is a pure fast muscle organ and whose myogenesis is completed earlier than other muscles (Dalrymple et al., 1999; Yamane et al., 2003), the reduction of Tnnt3 (fast) expression in the mutant is even more evident (Fig. 4C). Altogether, differential mRNA expression of the fast and slow fiber isoforms of the troponin genes in p100H embryonic skeletal mutant muscle becomes apparent at E15.5 and is sustained through postnatal stages with an even higher magnitude of difference.

Figure 3.

Relative mRNA expression levels (p100H mutant to wild-type) of troponin (Tn) C, TnI slow, and TnI fast genes. A: Embryonic stages. B: Late embryonic to early postnatal stages. C: Comparison of temporal expression of TnI fast and TnI slow in wild-type and the mutant. A,B: Gray lines indicate the ratio = 1.0, meaning that the expression level in the mutant (p100H/p100H) equals the wild-type. Samples were taken from developmental stages embryonic day (E) 10.5, 12.5, 14.5, 15.5, 16.5, 18.5, postnatal day (P) 1, 3, and 9. Both slow fiber-specific genes show a significant increase in the mutant, whereas the TnI fast gene shows a significant decrease in the mutant compared with wild-type. C: Ratio of expression of TnI fast to TnI slow at each developmental stage was calculated and plotted for the wild-type and the mutant. In the wild-type, expression of TnI fast becomes dominant as muscle matures, which is indicated by a steady increase in the ratios of TnI fast over TnI slow. Contrary to this finding, in the mutant even at E18.5, TnI fast expression is still slightly below TnI slow expression. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 4.

Relative mRNA expression levels (p100H mutant to wild-type) of the troponin T genes (Tnnt1, 2, and 3). A: Embryonic stages. B: late embryonic to early postnatal stages. C: Comparison of Tnnt3 (fast) expression between tongue and body muscles. A,B: Gray lines in A and B indicate the ratio = 1.0, meaning that the expression level in the mutant (p100H/p100H) equals the wild-type. Some points are shifted horizontally to make them visible. Total RNA was prepared from developmental stages embryonic day (E) 9.5, 10.5, 12.5, 14.5, 15.5, 16.5, 18.5, postnatal day (P) 1, 3, and 9. Tnnt1 (slow) and Tnnt2 (cardiac) show a significantly higher expression in the mutant muscles, whereas Tnnt3 (fast) is significantly reduced in the mutant. C: Relative expression levels of the Tnnt3 gene in the tongue (E16.5, E18.5, and P6) and pooled body muscles are plotted to compare the rate of decrease in two types of muscles, mixed muscles and almost pure fast muscles. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

MHC mRNA Expression in the p100H Mutant Skeletal Muscle Reproduces the Differential Expression Seen in the Troponin Genes

Because the classification of skeletal muscle fiber types is operationally defined by the expression of MHC isoforms (e.g., reviewed by Pette and Staron, 2000), their temporal expression in p100H mutant skeletal muscle was also examined. During normal skeletal muscle development, MHCemb (embryonic fast) and MHCβ (slow) are among the first isoforms detected in the embryonic skeletal muscle, followed by MHCneo (neonatal fast) (Condon et al., 1990a; Lyons et al., 1990). MHCneo eventually is replaced in mature skeletal muscle by the adult fast MHC isoforms, MHCIIb, MHCIId/x, and MHCIIa, depending on the physiological nature of the fiber; MHCIIb being the most glycolytic and MHCIIa being the least glycolytic (Gunning and Hardeman, 1991; Hamalainen and Pette, 1993).

We focused on the mRNA expression of the MHC isoforms during the developmental stages beginning at E15.5 to see whether the fast/slow isoform gene expression pattern observed in the troponin genes was occurring also in the MHC genes. Figure 5A summarizes the mRNA expression of MHCemb and MHCβ. MHCemb shows no significant difference between the p100H mutant and wild-type, likely reflecting its neutrality within the fiber types where it is expressed, whereas the expression of MHCβ, a slow isoform, continually increases in the mutant, reaching more than a magnitude difference in the postnatal stages (Fig. 5A). The mRNA expression of three fast fiber MHC isoforms (MHCneo, MHCIIa, and MHCIIb) clearly shows reduced expression in p100H mutant skeletal muscle compared with wild-type, progressing with developmental stages (Fig. 5B). MHCneo expression in the mutant slightly increases postnatally (Fig. 5B, P3–P9). This finding is likely to be a reflection of the fact that, in wild-type, MHCneo is replaced by adult MHC isoforms, whereas in the mutant, that transition is delayed. Taken together, in the p100H mutant skeletal muscle, fiber type-specific gene expression is systematically altered; slow skeletal and cardiac isoform gene expression is significantly increased, whereas fast skeletal isoform gene expression is decreased compared with wild-type.

Figure 5.

Comparison of temporal expression of myosin heavy chain (MHC) mRNA between the p100H mutant and wild-type skeletal muscles. A: MHCβ (slow) and MHCemb (embryonic) isoforms. B: MHCneo (neonatal fast), MHCIIa (adult fast), and MHCIIb (adult fast). A: Relative expression levels (mutant to wild-type) of the two MHC isoform genes are calculated and expressed as a bar graph. Developmental stages of the samples used are embryonic day (E) 15.5, 16.5, 18.5, postnatal day (P) 1, 3, 5, and 9. Gray lines correspond to the expression ratio of 1.0, which indicates an equal expression level between the mutant and wild-type. Although MHCemb is expressed at equivalent levels in the mutant and wild-type, MHCβ shows a substantially higher expression in the mutant skeletal muscles. B: Expression of three fast isoform MHCs (neonatal, IIa, and IIb) is compared between the mutant and wild-type. Developmental stages examined are as same as A. The mRNA expression of MHCIIa at E15.5 and E16.5 was too low to obtain reproducible data; therefore, the results are shown from E18.5. Expression of all three MHC genes shows a significant decrease in the mutant skeletal muscle. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Histological Examination of p100H Mutant Skeletal Muscle and Sox6 Expression in Slow and Fast Muscle Groups

To examine the differentiation of mutant skeletal muscle at the cellular level, muscle cell morphology as well as the expression of TnI fast mRNA in E18.5 hindlimb muscles were compared between wild-type and the mutant. In E18.5 wild-type muscle, primary and secondary myotubes are readily distinguishable by their morphological differences. In addition, by E18.5 the hindlimb muscles, except for the soleus, show high mRNA expression levels of the TnI fast gene (Calvo et al., 2001). If the mutant muscle has a defect in the formation of either the primary or secondary myotubes and/or has a decrease in the expression of TnI fast at the cellular level, we reasoned that those defects in the mutant muscle would be most easily detected at E18.5. First, the morphology of the hindlimb muscle cross-section of term embryos (E18.5) was examined using hematoxylin and eosin (H&E) staining. At this stage, the gross morphology of the mutant muscle is normal (Fig. 6). Also, as shown in Figure 6, both primary (large, donut-shaped fiber) and secondary (small fiber) myotubes are observed in the mutant skeletal muscle in comparable numbers. Three fields in gastrocnemius of the wild-type and the mutant sections were examined for the number of primary and secondary myotubes, and there appears to be no defect in the formation of both myotubes in the mutant (ratio of secondary/primary: wild-type 1.20+0.038; p100H 1.47+0.099; mean + SE). Second, to analyze fiber type-specific gene expression in mutant skeletal muscle at the cellular level, we examined the mRNA expression of TnI fast by in situ hybridization. Reflecting the previous real-time PCR result showing a significant reduction of TnI fast at the tissue level in the mutant (Fig. 3A), the intensities of the signals, as well as the numbers of TnI fast positive cells in the mutant are significantly decreased compared with wild-type (Fig. 7). The reduction of the TnI fast mRNA affected both muscle fiber types, fast and slow, because all extensor digitorum longus (primarily fast), gastrocnemius (mixed), and soleus (primarily slow) showed a significantly lower TnI fast signals compared with their wild-type counterpart (Fig. 7C, E,G compared with D,F,H, respectively).

Figure 6.

Morphological comparison of primary and secondary myotubes between the p100H mutant and wild-type muscle. Cross-sections of hindlimbs of the mutant and wild-type were stained by H&E and observed under brightfield illumination (×200 magnification). An arrowhead indicates an example of the primary myotubes, which are donut-shaped and larger than the secondary myotubes, whose example is indicated by arrows. The secondary myotubes are more densely stained and are formed associated with the primary myotubes. The number and morphology of both types of myotubes appear normal in the mutant muscle.

Figure 7.

Expression of troponin (Tn) I fast mRNA in p100H mutant and wild-type hindlimb muscles (in situ hybridization). A,C,E,G: Sox6 mutant (p100H/p100H). B,D,F,H: Wild-type. Cross-sections of E18.5 hindlimb were hybridized with antisense TnI fast riboprobe. TnI fast mRNA expression is visualized in blue. A and B are observed at a low magnification (×40); C–H are individual muscle groups indicated in A and B observed in a higher magnification (×200). EDL, extensor digitorum longus (predominantly fast fibers); MG, medial gastrocnemius (mixed with fast and slow fibers); Sol, soleus (predominantly slow fibers); T, tibia; F, fibula. The number of TnI fast positive cells in the mutant skeletal muscle is significantly reduced overall.

A higher level of slow skeletal isoform gene expression in the mutant muscle and a lower level of fast skeletal isoform gene expression could be regulated by differential expression levels of the Sox6 gene itself in slow and fast fiber types. To test this hypothesis, Sox6 expression in the three adult muscle groups was examined at both the mRNA and protein levels. Diaphragm, gastrocnemius, and soleus were chosen as predominantly fast, mixed, and predominantly slow muscles, respectively (Voytik et al., 1993; Fournier and Lewis, 2000). As shown in Figure 8A, the Sox6 mRNA expression was not differentially expressed among these muscles. Sox6 protein expression was also examined by Western blot in these three muscle groups as well as in the testis, which was used as a positive control (Connor et al., 1995). As shown in Figure 8B, Sox6 protein expression was also comparable among these three muscles. These results suggest that regulatory mechanisms other than differential expression of Sox6 between the slow and fast fibers may be involved in the regulation of fiber type-specific gene expression.

Figure 8.

Sox6 expression in three different muscle groups. A: Northern hybridization. B: Western blot. A: Total RNA was isolated from adult (∼2 months old) gastrocnemius, diaphragm, and soleus muscles, and 10 μg was applied to each lane. The blot was first hybridized with the Sox6 cDNA and then stripped and hybridized with cytoplasmic β-actin to confirm an equal loading. B: Muscle and testis crude extracts were prepared, and 80 μg of protein was separated in each lane. The blot was first incubated with anti-Sox6 rabbit polyclonal antibody. The same blot was stripped and incubated with anti–α-skeletal actin monoclonal antibody. Estimated molecular weight for the full-length Sox6 protein is 91.8 kDa (Connor et al., 1995). The lower band could be the product of one of the alternatively spliced Sox6 transcripts. Sox6 expression at both the mRNA and protein levels is comparable among the three muscles.

DISCUSSION

Sox6 mutations in mice result in early postnatal lethality (Hagiwara et al., 2000; Smits et al., 2001). We previously reported skeletal muscle degeneration, cardiomyopathy, and conduction abnormalities of the heart in the early postnatal Sox6 mutant p100H (Hagiwara et al., 2000). These mutant phenotypes led us to hypothesize that Sox6 is required for normal development and/or function of both skeletal muscle and the heart. It is known that Sox6 plays an important role in chondrogenesis (Smits et al., 2001; Uusitalo et al., 2001) and functions as a mediator of bonemorphogenetic protein 2–induced chondrocyte-specific gene expression (Fernandez-Lloris et al., 2003). However, Sox6's function in other tissues has remained unknown, despite its widespread expression in a variety of tissues, including muscle, testis, and the central nervous system (Connor et al., 1995; Hagiwara et al., 2000; Cohen-Barak et al., 2001).

In this study, we have characterized for the first time the developmental defects in the Sox6 mutant skeletal muscle using the Sox null mutant p100H (Hagiwara et al., 2000). In the absence of a functional Sox6 gene, the expression of slow, cardiac, and fast skeletal isoform genes, including the myosin heavy chain and troponin genes becomes distorted. Instead of the normal transition from slow isoforms to fast isoforms beginning at E15–E16 seen in wild-type muscle (Lu et al., 1999; Calvo et al., 2001; Wang et al., 2001), the expression of slow isoforms predominates over that of fast isoforms in the Sox6 mutant skeletal muscles. The aberrant predominance of the slow isoform gene expression in the mutant starts around E15.5, a stage coinciding with the appearance of secondary myotubes in mice (Ontell and Kozeka, 1984; Cossu et al., 1988; Cachaco et al., 2003). The formation of primary and secondary myotubes occurs at different developmental stages and exhibits differences in the expression of slow and fast skeletal isoform genes both in vivo and in vitro (Stockdale, 1992). In vivo, primary myotubes initially express MHCβ (slow isoform) along with MHCemb (embryonic fast isoform), although later most of them, except soleus, convert to fast fibers; whereas secondary myotubes initially express MHCneo (neonatal fast isoform) along with MHCemb and most of them stay as fast fibers (Condon et al., 1990a; Gunning and Hardeman, 1991; Cho et al., 1994; Dunglison et al., 1999). Limb myoblasts induced to form myotubes in culture replicate the initial in vivo phenotypes, suggesting there is a default phenotype for primary and secondary myotubes (Smith and Miller, 1992; Pin and Merrifield, 1993; Cho et al., 1994). It is interesting that, although both primary and secondary myotubes seem to be formed normally at the gross morphological level in the p100H mutant hindlimb muscles, the functional differentiation of myotubes, fast or slow, is abnormal in the mutant. Because the beginning of distorted expression of fiber-specific genes in the mutant muscle coincides with the formation of the secondary myotubes (E15–E16), Sox6 may be required for the differentiation of the functional characteristics of the secondary myotubes. Because of the impaired differentiation program, the mutant skeletal muscle may differentiate into slow fibers. In addition, it is also possible that the mutant primary myotubes, which are located in the future fast muscle groups fail to express fast isoform genes during the process of maturation. Alternatively, Sox6 may be required for induction of the fast isoform genes and/or suppression of the slow isoform genes in both primary and secondary myotubes. We currently are examining these possibilities using muscle cell cultures.

Uncovering the mechanisms of slow–fast fiber formation regulated by Sox6 is critical to understand the early stages of skeletal muscle differentiation. Because the Sox6 protein is not differentially expressed between slow and fast muscles in adults, regulatory mechanisms other than differential expression of Sox6 between the slow and fast fibers could be involved. For example, it has been reported that the Six1 protein, which activates fast isoform genes, is equally expressed in fast and slow fibers, although the protein is enriched in the nuclei of the fast muscle fibers (Grifone et al., 2004). Alternatively, Sox6 may exert its effect on regulation of fiber type-specific gene expression during the period of differentiation by means of posttranslational modification and/or differential expression but may be inactive in fully differentiated adult muscle. It should be also considered that Sox6 may act extrinsically because Sox6 expression is detected in many tissues, including the nervous system and cartilage (Conner et al., 1995; Lefebvre et al., 1998; Hagiwara et al., 2000). We are in the process of testing these alternative hypotheses.

In this report, we described roles of Sox6 in embryonic skeletal muscle development. The regulatory mechanisms for fiber type differentiation in embryonic skeletal muscle are not well known compared with those in adult skeletal muscle. In adult muscle, the most intensely investigated mechanism regulating transcription of fiber type-specific genes is modulation by the motoneuron activity (e.g., Chin et al., 1998; Murgia et al., 2000; Wu et al., 2000). In embryonic muscle, the initial fiber type specialization does not appear to be regulated by innervation (Condon et al., 1990b; Sheard and Duxon, 1996; Robson and Hughes, 1999), suggesting that signals different from those responsible in adult muscles regulate fiber type differentiation. In mice, it has also been shown that the ectopic expression of the transcription coactivator PGC-1α in fast fibers can convert fast fibers to slow fibers in conjunction with MEF2 (Lin et al., 2002). The expression of PGC-1α is not changed in the p100H mutant postnatal muscles (Hagiwara, unpublished data). Whether the fiber type-regulatory factors identified in adult muscle have similar regulatory effects on developing embryonic muscle or regulatory factors unique to embryonic muscle are necessary for embryonic muscle differentiation will require further investigation.

In summary, we have reported that the loss of Sox6 results in systematic changes of isoform-specific muscle gene expression. The present results show the first evidence of a possible role of Sox6 in the development of skeletal muscle. Taken together, the Sox6 mutant mouse will be a useful model for elucidating the still unclear regulatory mechanisms of mammalian skeletal muscle differentiation during embryogenesis.

EXPERIMENTAL PROCEDURES

Microarray Analysis

Skeletal muscles of the body wall and limbs were collected from term embryos (E18.5) of p100H (Sox6 mutant allele) and wild-type homozygotes (C57B6 background). Total RNA was prepared using Trizol (Invitrogen, Carlsbad, CA), then treated with RQ-DNase (Promega, Madison, WI) and further purified using RNeasy midi-kit (Qiagen, Valencia, CA). Synthesis of double-stranded cDNA, labeling of cRNA with biotin, and fragmentation of the cRNA were performed following the Affymetrix Expression Analysis Technical Manual (Affymetrix, Santa Clara, CA). Hybridization of the Affymetrix Murine Genome U74Av2 set (containing the ∼6,000 known genes in the Mouse UniGene data base and another ∼6,000 expressed sequence tag clusters) and scanning of the hybridized GeneChips were performed at the UC Davis Medical School Microarray core facility. To assess the reproducibility of the hybridization results, three hybridization experiments were performed for both the Sox6 mutant and wild-type. Raw scanning data were converted to a CHP file by the Microarray Suite 5.0 program (Affymetrix). Data were then normalized and analyzed by BRB-Array Tools (version 3.0) software program (http://linus.nci.nih.gov/BRB-ArrayTools.html, developed by Dr. Richard Simon and Amy Peng). Genes showing more than 1.5-fold differences (up or down) with a parametric P value less than 0.05 were selected for further analysis.

Northern Hybridization

The mRNA expression of representative genes identified by microarray analysis was also examined by Northern hybridization. Five micrograms of E18.5 skeletal muscle total RNA was separated on 1% agarose gel containing 2.2% formaldehyde, then transferred to a Biodine B nylon membrane (Pall Life Sciences, East Hills, NY). The blot was hybridized with a PCR-amplified cDNA fragment of the selected genes. Probes were labeled with 32P-α-dCTP using a random primer labeling kit (RediprimeII, Amersham Pharmacia Biotech, Piscataway, NJ). The membrane was hybridized in ExpressHyb (Clontech, Palo Alto, CA), washed following the manufacturer's instruction, and then exposed to an X-Omat AR film (Kodak) with an intensifying screen at −80°C.

Primer sequences and annealing temperatures of the selected genes were as follows: myosin heavy chain β (MHCβ), 5′-GCCAACACCAACCTGT-CCAAGTTC-3′ (forward) and 5′-TGCAAAGGCTCCAGGTCTGAGGGC-3′ (reverse), 64°C; cardiac Troponin T2 (Tnnt2), 5′-CGACCACCTGAATGAAGACC-3′ (forward), 5′-TTCTAGCTAAGCCAGCTCCC-3′ (reverse), 60°C; myosin light chain 2v (MLC2v), 5′-TGTTCCTCACGATGTTTGGG-3′ (forward), 5′-CTCAGTCCTTCTCTTCTCCG-3′ (reverse), 60°C; sarco(endo)plasmic reticulum Ca2+ transporting ATPase 2 (SERCA2), 5′-GCTCATTTTCCAGATCACACC-3′ (forward), 5′-CACATTTCCTCCACATCACAC-3′ (reverse), 58°C; troponin C (TnC slow skeletal/cardiac) 5′-AGGACGACAGCAAAGGGAAG-3′ (forward), 5′-GCCAAGGTTCAAGGACACAG-3′ (reverse), 60°C; troponin I slow (TnI slow), 5′-AGACTGGAGGAAGAATGTGG-3′ (forward), 5′-AGAAAGATAGGTGAGTGGGG-3′ (reverse) 58°C; TnI fast 5′-ACTACCTGTCAGAACACTGCC-3′ (forward), 5′-ACACCTTGTGCTTAGAGCCC-3′ (reverse) 61°C; calsequestrin 1, 5′-ATCCCAGACAAGCCCAACAG-3′ (forward), 5′-TCCTCCTCGTTATCCATCTCC-3′ (reverse), 60°C. β-actin mRNA expression was used as a loading control. The β-actin cDNA fragment was amplified using 5′-TC-ATGAAGTGTGACGTTGACATCC-3′ (forward) and 5′-GTAAAACGCAGCTCAGTAACAGTC-3′ (reverse) primers at the annealing temperature of 61°C. The identity of the PCR-amplified cDNA fragments was verified by sequencing.

Semiquantitative RT-PCR

A total of 5 μg of total RNA (skeletal muscle or diaphragm of P10 mice) was reverse-transcribed at 42°C with 100 pmol of random hexamers (Invitrogen) in the presence of 1× reaction buffer, 10 mM dithiothreitol, 1 mM dNTPs, 20 units of SUPERaseIn RNase inhibitor (Ambion, Austin, TX), and 200 units of SuperScript II (Invitrogen) in a 20-μl reaction mixture for 1.5 hr. cDNA was then diluted with water to generate 10-fold serial dilutions up to ×1,000. One microliter of each serial dilution was used for PCR amplification for 30 cycles with gene-specific primers in a 30-μl reaction mixture, except for MHCβ PCR. For MHCβ PCR, the original, 10×, and 100× diluted cDNA samples were used, and amplified for 35 cycles. A total of 5 μl of PCR reaction was then separated on a 2% agarose gel. Photography of the gel and quantification of the intensity of the ethidium bromide–stained bands were performed using the Fluorchem 8000 imaging system (Alpha Innotech, San Leandro, CA). GAPDH was used as a standard to compare the expression levels of the tested genes [primer sequence for GAPDH: 5′-ACCACAGTCCATGCCATCAC-3′ (forward), 5′-TCCACCACCCTGTTGCTGTA-3′ (reverse)].

Real-Time PCR

Total RNA was prepared from wild-type and the mutant embryos collected at different embryonic stages as indicated (E9.5–E18.5) and postnatal stages (P1, 3, 5, and 9). To prepare RNA containing skeletal muscle from E9.5 to E15.5 embryos, the heart was removed, and then RNA was prepared from the rest of the body. For E16.5 embryos, eviscerated bodies were used to prepare RNA. For older animals (E18.5 and early postnatal stages), skeletal muscle was collected from the whole body. Total RNA from tongue was prepared from tongues of E16.5, E18.5, and P6 mice. TaqMan real-time PCR was performed using Assays-on-Demands gene expression probes and TaqMan universal PCR mix, following the manufacturer's protocols (Applied Biosystems, Foster City, CA). Reactions were monitored by the ABI Prism 7900HT. In all experiments, GAPDH expression was used as a standard to normalize expression of muscle-specific genes in wild-type and the mutant. To compare the expression of muscle-specific genes in the mutant to wild-type, the comparative cycle threshold (Ct) method was used. In this method, the relative expression level between two samples is calculated as a ratio of the cycle numbers required to reach an arbitrary amplification value [formula = 2(-ddCt)]. A relative expression level greater than 1 means higher expression in the mutant, less than 1 means lower expression in the mutant, and equal to 1 means equivalent expression in the mutant and wild-type. In the real-time PCR graphs in the Results section, each data point represents two independent samples with three replicas when there are error bars, and one sample with three replicas when there are no error bars. Two independent samples were collected from different litters; therefore, their developmental stages may differ up to 16 hr.

Histology

Hindlimbs of E18.5 embryos were collected from both wild-type and the p100H of the same litters. Embryos of specific developmental stages were obtained from timed matings of heterozygous intercross, designated E0.5 at noon on the day when a vaginal plug was detected. Legs were fixed in 4% paraformaldehyde in PBS, washed in PBS, then dehydrated and embedded in paraffin. Cross-sections (10 μm) were processed for H&E staining.

In Situ Hybridization

In situ hybridization using sections was performed following the protocol reported by Etchevers et al. (2001). For a TnI fast riboprobe, a cDNA fragment was amplified by PCR using the primers listed above, then cloned into pCRII-TOPO (Invitrogen). Digoxigenin (DIG) -labeled antisense or sense probe RNA was transcribed using SP6 or T7 polymerase, respectively. Alkaline phosphatase conjugated anti-DIG antibody and nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) were used for detecting the hybridized probe. A probe concentration of 1 μg/ml was used for hybridization.

Western Blotting

Crude tissue lysate was prepared from diaphragm, gastrocnemius muscle, soleus muscle, and testis of 2-month-old wild-type mice. Tissues were homogenized in RIPA buffer (1% NP-40, 0.5% deoxycholate, and 0.1% sodium dodecyl sulfate (SDS) in PBS with 2 μg/ml of aprotinin, 2 μg/ml of leupeptin, 100 μg/ml of phenylmethyl sulfonyl fluoride, and 1 mM sodium orthovanadate), and supernatant was collected. A total of 80 μg of each protein sample was loaded on 8% SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. The blot was incubated with anti-Sox6 rabbit polyclonal antibody (Sigma, St. Louis, MO) at 1:800, and the signal was visualized by ECL system (Amersham Pharmacia Biotech). To estimate the amount of muscle protein loaded in each lane, the same blot was stripped and then incubated with anti-α skeletal actin monoclonal antibody (Sigma) at 1:2000.

Acknowledgements

We thank Drs. Ann Bonham, Anne Knowlton, Murray Brilliant, Orit Cohen-Barak, Ray Runyan, Everett Bandman, Sue Bodine, Richard Tucker, Michael Ferns, Paul Fitzgerald, and Mr. Adam Jenkins for helpful discussion; and Dr. Michael George for the help with microarray data analysis. We also thank Quan Nugyen and Michael Yeh for their technical assistance. This work has been supported by a UCD seed grant, a UCD faculty grant, AHA Beginning Grant in Aid, and a grant from the March of Dimes foundation (to N.H.).

Ancillary