Myocyte Enhancer Factor-2 and Serum Response Factor Binding Elements Regulate Fast Myosin Heavy Chain Transcription in Vivo*

Adult fast muscle fibers express distinct myosin heavy chains (MyHC) in differing proportions, but the mechanisms underlying their differential expression remain undefined. We used a variety of in vitro and in vivo approaches to explore the contribution of transcriptional regulation to adult fast MyHC expression. Here we show that 800–1000 bp of a sequence upstream of the three mouse adult fast MyHC genes (Ia, IIb, and IId/x) are sufficient to drive muscle-specific and fiber-specific expression in vivo. We show that the upstream promoter region of the gene most abundantly expressed in mouse skeletal muscles, IIb MyHC, retains binding activity and transcriptional activation for three positive transcription factors, the serum response factor, Oct-1, and myocyte enhancer factor-2, whereas the other two genes (IIa and IId/x) have nucleotide substitutions in these sites that reduce binding and transcriptional activation. Finally, we demonstrate that regions upstream of 300 bp modulate the effects of these elements. Together, these data demonstrate that the quantitative differences in MyHC expression in mouse skeletal muscle have evolved at least in part through the elimination of positive-acting transcription factor binding sites.

The primary function of skeletal muscle is to produce contractile force via the formation and maintenance of a highly specialized contractile apparatus. The myosin heavy chain (MyHC) 1 protein contains both the ␣-helical rod domain necessary for thick filament formation and the ATPase domain necessary for converting chemical energy into mechanical force (1). Four different isoforms of MyHC, one slow and three fast, are typically expressed in adult skeletal muscle, each encoded by a separate gene. These isoforms are expressed in four distinct fiber types, types I, IIa, IId/x, and IIb (2), and have different force and velocity properties that greatly affect the functional capabilities of the muscle fibers. Each muscle is composed of a unique combination of these four fiber types depending on its particular function, and differences in MyHC expression between muscles are thought to have profound effects on the speed, force, and duration of muscle contraction.
The formation of different types of muscle fibers depends on differential expression of a suite of contractile and metabolic proteins specialized for each fiber type (3). The myogenic regulatory factor family of transcriptional activators, which includes myf-5, MyoD, myogenin, and myogenic regulatory factor-4 (4,5), and the myocyte enhancer factor-2 (MEF-2) family of transcriptional factors (6) are believed to underlie the expression of most if not all muscle-specific genes. Other muscleenriched transcriptional factors, including the serum response factor (SRF) (7), myocyte nuclear factor (8), and MEF-3 (9), have been shown to contribute to muscle-specific gene expression, either alone or by interacting with ubiquitously expressed factors such as nuclear factor-1 (NF1), Sp-1, and Oct-1 (10 -12).
In addition, the calcineurin/NFAT (nuclear factor of activated T cells) pathway appears to be involved in the slowspecific expression of the myoglobin and troponin I slow genes (13)(14)(15)(16)(17). Calcineurin is a phosphatase that dephosphorylates the NFAT family of transcription factors, allowing them to enter the nucleus, bind to specific sequences in slow-specific genes, and activate their expression (13). MEF-2 is also activated by calcineurin as well as by calcium-calmodulin kinase and may activate transcription of some slow fiber-specific genes (18,19). Finally, studies on the IIb MyHC promoter have identified a proximal MyoD binding E-box that is necessary for fast muscle-specific expression of the IIb gene in vivo (20). However, little is currently known about the factors that regulate fiberspecific expression of type I versus IIa versus IId/x versus IIb MyHC within individual muscle fibers.
We reported previously the isolation of the upstream promoter regions of the three adult fast skeletal MyHC genes and their activity in C 2 C 12 myotubes (21). These studies demonstrated that ϳ1 kb of the upstream promoter sequence was sufficient to confer muscle-specific, differentiation-sensitive expression in vitro. But although tissue culture systems are excellent models for studying muscle differentiation and for screening the quantitative effects of individual promoter mutations on gene expression, cultured myotubes cannot adequately mimic the complexity of skeletal muscle in vivo. Thus in vivo analyses are necessary to determine the role of specific regulatory elements in regulating MyHC gene expression in fully mature muscle fibers. In addition, fiber-specific gene expression cannot be modeled in vitro, because myotubes in culture do not mature to the point where fiber type specialization occurs.
The purpose of the present study was to evaluate the activity and specificity of the three adult fast MyHC promoters within skeletal muscle in vivo and to identify elements involved in their * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: MyHC, myosin heavy chain; CMV, cytomegalovirus; GFP, green fluorescent protein; eGFP, enhanced GFP; MEF, myocyte enhancer factor; MNF, myocyte nuclear factor; NF1, nuclear factor-1; SRE, serum response element; SRF, serum response factor; TA, tibialis anterior; TKp100, 100-bp thymidine kinase promoter. differential expression in different muscle and fiber types in vivo. We show that ϳ1 kb of the upstream promoter sequence of the adult fast MyHC genes is sufficient to confer muscle-specific and fiber-specific expression in vivo. Moreover, we determined the quantitative role of specific cis-regulatory domains on expression of the adult fast MyHC promoters in vivo.

EXPERIMENTAL PROCEDURES
DNA Constructs-Isolation of the three adult fast MyHC promoters from murine genomic DNA has been described previously (21). Briefly, ϳ0.8 -1 kb (1000 bp for IIa, 990 bp for IId/x, and 780 bp for IIb) was cloned into the firefly luciferase reporter plasmid VR1255 (Vical, San Diego, CA). A reporter plasmid containing a Renilla luciferase gene driven by a cytomegalovirus promoter (pRL-CMV; Promega, Madison, WI) was used as an injection control for all in vivo injections. We also created reporter constructs containing an enhanced green fluorescent protein gene (eGFP) by excising the firefly luciferase reporter gene from the promoter-VR1255 constructs using the SalI and BamHI sites and generating eGFP by high fidelity PCR with these sites added to the ends. All mutagenesis constructs were cloned by PCR using Turbo Pfu high fidelity polymerase (Stratagene) and standard techniques. All clones were checked by sequencing at the DNA Sequencing Core of the University of Colorado to ensure that they contained the appropriate deletions or mutations.
Cell Culture and Transfection-C 2 C 12 myoblasts were plated and cultured as described previously (21,22). Promoter-reporter plasmids were transfected into proliferating myoblasts using Lipofectamine 2000 (Invitrogen) per the manufacturer's instructions. Two days after transfection, myoblasts were differentiated into myotubes by adding media plus 2% horse serum to the plates. Myoblasts were then lysed and analyzed for luciferase expression using a firefly luciferase kit (Promega, Madison, WI). Values for both wild type and mutagenesis constructs were normalized to the average value for the wild type constructs. All experiments were replicated 3-6 times with 4 -6 wells per experiment and averaged.
Intramuscular Plasmid DNA Injections-Plasmid DNA injections were carried out using standard techniques (20). Plasmid DNA was purified by cesium chloride gradient centrifugation; 25 g of DNA was injected per muscle in 25 l of total solution for the tibialis anterior (TA), and 100 g of DNA in 100 l of total solution was injected into the calf. For all studies a control plasmid was co-injected (1.25 g of pRL-CMV) to control for injection efficiency. Seven days after injection, muscles were surgically removed and homogenized in lysis buffer (25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 4 mM EGTA, and 1 mM dithiothreitol), and the lysate was centrifuged for 30 min at 3000 ϫ g. The supernatant was removed and stored at Ϫ20°C until use. Dual luciferase assays were carried out using a commercially available kit (Promega, Madison, WI). For all experiments, firefly luciferase values were divided by the Renilla value to normalize for differences in transfection efficiency.
GFP Analysis-For constructs containing a GFP reporter, muscles were excised 7 days after injection, frozen in isopentane cooled in liquid nitrogen, and stored at Ϫ70°C until use. Muscles were sectioned and stained with anti-MyHC antibodies as described previously (24). The primary antibodies used were MHCs (Novocastra, Newcastle-on-Tyne, England) specific for type I MyHC, SC-71 specific for IIa MyHC, BF-F3 specific for IIb MyHC, and 6H1 specific for IId/x MyHC (23). Sections were incubated in a secondary antibody, namely goat anti-mouse IgG or IgM tetramethylrhodamine isothiocyanate (TRITC) conjugate for MyHC and goat anti-rabbit TRITC for laminin at 1:50 dilution for 1 h at room temperature. Sections were again washed in phosphate-buffered saline. After staining, the GFP-positive fibers were found by using the original video images as a guide, and their reaction with the various antibodies was scored.
Binding Assays-Nuclear extracts were isolated from 10 -20 15-cm dishes of cultured non-muscle L cells, C 2 C 12 myoblasts, and C 2 C 12 myotubes as described previously (22). Double-stranded oligonucleotides were labeled with [␣-32 P]dCTP, and unlabeled radionucleotides were removed by spin column purification. Binding reactions were carried out using 100 Ci of labeled probe, 10 g of nuclear extract Equal amounts of each promoter plasmid were injected into either the heart or the TA muscle of adult mice, and activity was evaluated using a luminometer 7 days after injection. B, Western blot analysis of MyHC isoform expression from the TA muscles of eight different adult mice. C-F, relative luciferase activity of the three adult fast MyHC promoter constructs following injection into the TA (C), gastrocnemius (Gastroc.) (D), plantaris (Plant.) (E), and soleus (Sol.) (F) muscles of adult mice. For all animals, MyHC promoter-driven firefly luciferase values were normalized to Renilla luciferase levels driven by the CMV promoter. Luciferase data in panels A and C are expressed as mean Ϯ S.E. for 3-5 animals per construct. *, significantly different from heart; p Ͻ 0.05. protein, and 1 g of poly(dI-dC) in binding buffer (100 mM Tris pH 7.5, 500 mM NaCl, 10 mM EDTA, 10 mM dithiothreitol, and 12.5% Ficoll) for 30 min at room temperature. All antibodies used for supershifting assays (Oct-1, MEF-2, SRF, NF1, and lamin A/C) were purchased from Santa Cruz Biotechnology.

RESULTS
The Adult Fast MyHC Promoters Are Skeletal Muscle-specific-We sought to determine whether the upstream promoter sequences of the three adult fast MyHC genes were sufficient to restrict reporter activity to skeletal muscle in vivo. Activity of the adult skeletal fast MyHC promoters was considerably greater when injected into the TA muscle than when injected into the heart. Absolute luciferase values for the IIa, IId/x, and IIb MyHC promoters were 5-, 1600-, and 3200-fold greater, respectively, in skeletal muscle versus heart (Fig. 1). The level of IIa MyHC promoter activity was much lower in the TA than that of IId/x or IIb, presumably because Ͻ10% of the fibers in the TA muscle express IIa MyHC (24). These data suggest that ϳ 1 kb of an upstream promoter sequence is sufficient to confer expression of the three adult fast MyHC genes in skeletal but not cardiac muscle in vivo.
The adult fast MyHC promoters also show muscle typespecific activity in skeletal muscle in vivo. In the TA muscle, IIb MyHC accounts for ϳ70% of the total MyHC protein, whereas IId/x and IIa MyHC account for ϳ20 and 10%, respectively (Fig.  1B). Quantification of the activity of the IIa, IId/x, and IIb promoters following injection into the TA muscle revealed a similar pattern, with IIb Ͼ IId/x Ͼ IIa (Fig. 1C). Injection into muscles containing different percentages of the three adult fast MyHCs further revealed that MyHC promoter activity mimicked endogenous MyHC protein expression. For example, the gastrocnemius, which has a fiber percentage similar to that of the TA, showed high levels of IIb promoter activity (Fig. 1D), whereas the soleus, which expresses ϳ50% type IIa, had the highest IIa promoter activity (Fig. 1F). The plantaris, which is composed of a mixture of primarily IIb and IId/x fibers, had comparable IIb and IId/x MyHC promoter activities (Fig. 1E). Thus, the three adult skeletal fast MyHC promoters were sufficient to confer muscle type-specific expression in vivo.
Fiber-specific Promoter Activity-To test whether the adult fast MyHC promoter regions were sufficient to drive muscle fiber-specific expression, we created constructs in which the adult skeletal fast promoters were linked to an eGFP reporter gene. These were injected into mouse TA muscles, and the relationship between GFP expression and endogenous MyHC protein expression was examined using immunohistochemistry. Expression of the IId/x MyHC promoter was fiber-specific; of 21 IId/x-GFP-positive fibers, 100% (21/21) were found to express IId/x MyHC (Table I and Fig. 2). Similarly, the IIb MyHC promoter was also fiber-specific; of 15 IIb-GFP-positive fibers, 100% (15/15) were positively stained for an antibody specific for IIb MyHC (Table I). In contrast, a positive control plasmid containing a CMV upstream enhancer driving GFP did not show fiber specificity and was expressed in all fiber types (Table I and Fig. 2).
Initially, we were unable to obtain any GFP-positive fibers following injection of the 1000-bp IIa-GFP construct into untreated control mouse TA; this was probably due to the small number of IIa-expressing fibers ordinarily present in the adult mouse TA muscle. We therefore injected the soleus muscle, which has more IIa-expressing fibers, and pretreated muscles with the myotoxin Marcaine, which has been shown to increase the uptake of plasmid DNA (25). When the IIa MyHC promoter construct was injected into Marcaine-treated calf muscles, Ͼ90% of the 1000-bp IIa-GFP fibers (42/46) were positive for IIa MyHC protein by immunohistochemistry (Table I). Thus, 800 -1000 bp of the three adult fast MyHC promoters was sufficient to drive fiber-specific expression in vivo.
The AT2 Element and Promoter Activity in Vitro-Based on the observation that these sequences were sufficient to direct fiber-specific expression, we sought to identify regulatory elements within the IIa, IId/x, and IIb MyHC promoter regions that are involved in their differential expression. For example, as shown above, Ͼ70% of the MyHC expressed in the TA muscle is the IIb isoform, suggesting that the IIb MyHC gene is more transcriptionally active in this muscle than the IId/x or IIa genes. We therefore searched for potential transcriptional activator binding sites present in the IIb but not in the IIa or IId/x MyHC genes. We focused on two elements in the proximal promoter regions of the IIb MyHC gene, an AT-rich region called AT2 (13) and a CArG element.
The AT2 region has been shown previously to bind to both the muscle-enriched transcription factor MEF-2 and the ubiquitously expressed transcription factor Oct-1 (13). In the IIa MyHC promoter, a TT to CC substitution has occurred in this Representative cross-sections from adult mouse TA muscle following injection with GFP reporter plasmids driven by either the IId/x, IIb, or CMV promoters followed by immunohistochemistry for specific MyHC isoforms. GFP-positive fibers have been numbered for easier identification between the GFP images (top row) and the tetramethylrhodamine isothiocyanate (TRITC)-labeled immunostaining (bottom row). For both the IId/x promoter and the IIb promoter, GFP-positive fibers were always positive for the corresponding MyHC isoform despite the presence of negative fibers nearby; three IIb-GFP-positive fibers and two IId/x-GFP-positive fibers are numbered in the respective panels for illustration. For CMV-GFP, no fiber specificity was evident; the representative section was immunostained with antibodies to both IIa (red) or IIb (green) to better illustrate this observation. The three CMV-GFP-positive fibers numbered in the right hand panel showed IIb immunostaining (fiber 1), IIa immunostaining (fiber 2), and immunostaining of neither (fiber 3). Note that the IIb images were taken at a slightly higher magnification than the IId/x or CMV images. region during the divergent evolution of these genes (Fig. 3A). In addition, both the IIa and the IId/x MyHC genes contain an AT to GA substitution in the region immediately flanking the MEF-2 binding site (Fig. 3A). Mutagenesis of the IIa AT2 element so that it was identical to the IId/x AT2 motif resulted in a significant increase in IIa MyHC promoter activity in C 2 C 12 myotubes compared with the wild type IIa MyHC construct (Fig. 3B). Further mutagenesis of the IIa MyHC AT2 region so that this element was then identical to that of the IIb AT2 resulted in a further increase in IIa MyHC promoter activity to ϳ150% that of the wild type IIa MyHC promoter (Fig. 3B). Conversely, mutation of the IId/x MyHC AT2 to the sequence of the IIa AT2 resulted in a significant 50% decrease in activity of this promoter, whereas mutation to the IIb AT2 sequence resulted in a significant 50% increase in activity of the IId/x promoter (Fig. 3C). Finally, mutagenesis of the IIb AT2 so that it was identical to the IIa AT2 element decreased IIb activity significantly (Fig. 3D). Thus, the nucleotide substitutions in the IIa and IId/x MyHC AT2 elements relative to the IIb promoter result in significant reductions in promoter activity relative to the IIb MyHC promoter sequence such that IIb Ͼ IId/x Ͼ IIa.
Electrophoretic Mobility Shift Assay Analysis of the AT2 Region-To complement these promoter activity experiments, we tested the three different AT2 sequences for protein binding with nuclear extracts from C 2 C 12 myotubes. Several complexes formed with the IIb AT2 oligonucleotide (Fig. 4, lane 10). The addition of an antibody to Oct-1 resulted in elimination of the second slowest migrating complex and the formation of a band shift (Fig. 4, lane 11), and the addition of an antibody to MEF-2 resulted in elimination of the topmost, slowest migrating complex and the formation of another supershift (Fig. 4, lane 12). These data confirm that the IIb AT2 element can bind both Oct-1 and MEF-2 as described previously (13). In contrast, the IIa AT2 showed little protein binding (Fig. 4, lanes 1-4). The faint, slowly migrating complex in the IIa AT2 lanes ran at the same position as the Oct-1 complex and was slightly decreased by the addition of an Oct-1 antibody (Fig. 4, lane 3). No complex was observed at the MEF-2 position, and the addition of an MEF-2 antibody did not produce a band shift (Fig. 4, lane 4). Thus, MEF-2 binding is eliminated, and Oct-1 binding is greatly reduced in the IIa AT2 element as compared with the IIb AT2.
The IId/x AT2 oligonucleotide did bind to both Oct-1 and MEF-2, but the binding was consistently less strong for both factors than that of the IIb AT2 (Fig. 4, lanes 5-8). We therefore created an oligonucleotide that contained the TG sequence found in the IIb AT2 flanking region in place of the GA found in the endogenous IId/x AT2 and found that this restored Oct-1 and MEF-2 binding as compared with the endogenous IId/x AT2 sequence (compare lane 6 with lane 14 in Fig. 4). In summary, the nucleotide substitutions in the IIa and IId/x AT2 regions result in impairments in Oct-1 and MEF-2 binding compared with the IIb AT2 sequence.
The CArG Element and Promoter Activity-A second potential activator element in the IIb promoter is the CArG box at approximately Ϫ80 to Ϫ120 bp upstream from the transcription start site. CArG boxes (CC(A/T) 6 GG) have been shown to bind SRF, a potent transcriptional activator (26). In the mouse IIb gene this element consists of an SRF consensus binding sequence flanked by two TTGCCN consensus sequences for the NF1/CCAAT binding factor (Fig. 5A). However, both the IIa and the IId/x MyHC promoters contain nucleotide substitutions within the SRF consensus core.
Mutation of the IIb SRF consensus site to that of the IIa or IId/x MyHC sequence significantly decreased activity of the IIb MyHC promoter compared with the wild type IIb construct, with the IId/x mutations having a larger decrement in IIb activity than the IIa mutations (Fig. 5B). Conversely, mutation of the IId/x CArG-like element to that of either the IIa or IIb sequences resulted in a significant increase in IId/x promoter activity (Fig. 5C). The IId/x MyHC promoter also contains a second CArG-like sequence immediately downstream of the first CArG-like box that differs from the SRF binding consensus sequence by a single nucleotide (CCAAATATGT, with the differing nucleotide underlined); mutation of this CArG2 sequence to further inhibit SRF binding did not significantly alter IId/x MyHC promoter activity (Fig. 5C). Together, these data suggest that the IIb CArG element acts as an activator, and substitutions in the IIa and IId/x promoters appear to eliminate this activator site. Moreover, the fact that the addition of either the IIa or the IIb CArG sequence to the IId/x MyHC promoter results in an increase in promoter activity suggests that the IId/x CArG-like element acts as a repressor.
We then tested the role of this IId/x CArG-like putative repressor element on IId/x MyHC promoter activity in TA muscle in vivo. For this study, we used a construct in which the entire CArG-like element was deleted, because this produced a more robust effect on IId/x MyHC promoter activity in C 2 C 12 myotubes in vitro (21). As shown in Fig. 5D, activity of the IId/x CArG-deleted construct was ϳ2-fold higher than that of the wild type IId/x MyHC promoter construct when injected into mouse TA. This demonstrates that the IId/x CArG-like element behaves similarly in cultured myotubes and in fully differentiated skeletal muscle fibers in vivo and further supports the hypothesis that the substitutions in the IId/x CArG box have created a repressor binding site.
Sensor Constructs Reveal That the IId/x CArG-like Element Is a Repressor-To test whether the IId/x CArG-like element acts as a repressor in a heterologous context, we created "sensor" constructs in which four tandem copies of either the IIb CArG element or IId/x CArG-like element were placed upstream of a minimal 100-bp thymidine kinase promoter (TKp100) driving luciferase expression (Fig. 6A). As expected, placing four copies of the IIb CArG element upstream of TKp100 increased activity by 1.5-fold (Fig. 6B). In contrast, adding four copies of the IId/x CArG-like element upstream of TKp100 significantly decreased promoter activity (Fig. 6B). Finally, a sensor construct consisting of five tandem copies of the c-fos SRE driving luciferase increased activity by ϳ5-fold (Fig. 6B). Together, these data suggest that the IIb CArG FIG. 5. The IIa and IId/x CArG-like box mutations reduce activity. A, the sequence of the relevant CArG-like elements for IIa, IId/x, and IIb and the IId/x CArG2 sequence and its mutation. The putative binding sites for NF1/CCAAT binding factor and SRF are boxed. Nucleotides differing from the IIb CArG element sequence are underlined. For the IId/x CArG2, the single nucleotide difference between this CArG-like element and a consensus SRF binding site is shown in boldfaced type, and the mutation is underlined. B, mutation of the IIb CArG sequence to that of the IIa or IId/x results in a significant decrease in IIb promoter activity. C, mutation of the IId/x CArG-like sequence to that of the IIa or IIb MyHC gene significantly increases IId/x promoter activity, whereas mutation of the IId/x CArG2 site has no significant effect. D, injection of the IId/x MyHC promoter construct and a IId/x MyHC construct in which the CArG-like element has been deleted into mouse TA muscle. All results are the mean Ϯ S.E. of 3-6 experiments each, 4 -6 wells per experiment. *, significantly different from wild type construct; p Ͻ 0.05. element is a relatively weak activator, whereas the IId/x CArGlike element acts as a repressor both in the context of the MyHC promoters as well as in a heterologous promoter context.
The IId/x CArG-like Element Binds a Protein That Is Not SRF-We then examined the protein binding pattern of the IIa, IId/x, and IIb CArG-like elements. As a positive control, we first examined the binding pattern of an oligonucleotide containing the c-fos consensus SRF binding sequence. This oligonucleotide showed at least four distinct binding activities, namely a slow migrating doublet and two faster migrating activities (Fig. 7A, lane 2). The topmost, slowest migrating complex was supershifted by addition of an antibody to SRF (Fig. 7A, lane 3). The IIb CArG element also showed four distinct binding activities, although these were typically less strong and less distinct than those observed in the c-fos SRE construct (Fig. 7B, lane 10). Again, the slowest migrating complex in the IIb CArG reactions appeared to be SRF, based on supershifting with an anti-SRF antibody (Fig. 7B, lane 11).
The middle complex seen in the IIb CArG reactions was also observed with the IIa and IId/x CArG-like elements and was not supershifted with an anti-SRF antibody (Fig. 7B, lanes 2  and 6). However, the IId/x CArG-like element also demonstrated a slower migrating activity (asterisk in Fig. 7B) that was also not supershifted by the addition of an anti-SRF antibody (Fig. 7B, lane 7) or an antibody to the repressor YY1, 2 which often binds to CArG elements, although it was almost completely eliminated by the addition of a 50-fold excess cold IId/x CArG-like element probe (Fig. 7B, lane 8). This binding activity was not observed in reactions with the IIa CArG-like element (Fig. 7B, lane 2). We tested several other nuclear extracts to see whether they contained proteins that would produce binding activities similar to that of myotube nuclear extracts with the IId/x CArG-like element. Interestingly, the complex was present in all extracts tested, including myoblasts, myotubes, neonatal skeletal muscle, non-muscle fibroblastic L cells, and liver. 2 Finally, the IId/x CArG2 element also appeared to bind to SRF, but the binding was extremely weak (Fig. 7C).
Combined AT2 and CArG-like Element Mutations and MyHC Promoter Activity-The data presented above show that nucleotide substitutions in the AT2 and CArG-like elements of the IIa and IId/x MyHC promoters eliminate activator sites found in the IIb MyHC promoter and that mutation of these elements individually affects MyHC promoter activity. We therefore tested the effects of mutations to both of these elements on MyHC promoter activity in C 2 C 12 myotubes in vitro and in mouse TA muscle in vivo. Mutation of the IIa MyHC promoter so that both the AT2 and CArG-like elements were identical to the IIb sequences resulted in a significant increase in promoter activity compared with that of the wild type constructs in C 2 C 12 myotubes and a 2-fold increase in mouse TA muscle, whereas changing both the IIb AT2 and CArG elements to the IIa sequences for these elements resulted in a large decrease in IIb promoter activity to ϳ10% of wild type in C 2 C 12 myotubes and in TA muscle (Fig. 8A). Our hypothesis from these experiments was that these two elements in the IIb promoter region contribute to the fiber specificity of this gene. However, when the constructs containing the IIa MyHC promoter driving GFP expression were mutated such that the AT2 and CArG elements were identical to those of the IIb MyHC promoter, we did not observe promiscuous expression of GFP in IIb muscle fibers in the IIb-fiber-enriched TA muscle. 2 Upstream Regions Modulate the Effects of the AT2 and CArG Elements-Finally, we wished to determine whether the effects of these elements were modified by upstream elements. We therefore created deletion constructs for all three adult fast MyHC promoters and injected them into mouse TA muscle. Deletion to either 300 or 150 bp, which still contain both the AT2 and CArG elements, resulted in a significant decrease in activity FIG. 6. The IIb CArG is an activator, and the IId/x CArG-like element is a repressor. A, schematic of the sensor constructs used in these experiments. The TKp100 parent construct consists of 100 bp of the thymidine kinase promoter linked to luciferase (luc), whereas the IIb CArG and the IId/x CArG-like element (CLE) were linked in four tandem copies upstream of this minimal heterologous promoter. The multiple cloning site (MCS)-luciferase and SRE-luciferase constructs were purchased from Stratagene and contain a multiple cloning site or five tandem copies of the c-fos SRE upstream of a minimal TATA box promoter. B, relative luciferase activities of the various sensor constructs in C 2 C 12 myotubes reported as fold of the parent construct. Each construct is reported as the mean Ϯ S.E. from four experiments each. *, significantly different from parent construct; p Ͻ 0.05. of all three adult fast MyHC promoters and eliminated the differential expression of these promoters in vivo (Fig. 9A). This result suggests that elements upstream of 300 bp are necessary to produce high level and differential expression in vivo.
We further tested the role of the upstream regions of the IIa and IIb MyHC promoters by creating domain swapped constructs in which the distal 500 bp was excised and either religated to the proximal 300 bp or swapped onto the proximal 300-bp sequence of the other gene. Removal of the distal 500 bp of the IIa gene and replacing it with the distal 500 bp of the IIb MyHC promoter resulted in a significant decrease in activity compared to that of the IIa gene with its own distal region religated onto its proximal region (Fig. 9B). Conversely, removing the IIb distal region and replacing it with the IIa distal region resulted in a nearly 3-fold increase in IIb MyHC promoter activity (Fig. 9B). These results suggest that the IIa distal region contains activator elements and/or that the IIb distal region contains inhibitor elements. To confirm this notion, heterologous constructs were created in which the IIa 500-bp distal region was ligated onto a minimal TKp100 promoter; addition of the IIa distal region resulted in a 20-fold increase over the TKp100 promoter alone (Fig. 9B). DISCUSSION In adult skeletal muscle, different isoforms of fast MyHC are expressed in distinct fast muscle fiber types. A greater understanding of the factors regulating expression of the fast MyHC isoforms at different levels in different muscles would provide insights into how these different skeletal muscle fiber types are established and maintained. Although considerable progress has been made in elucidating the molecular genetics underlying muscle-specific and slow versus fast gene expression, little is currently known regarding the molecular mechanisms governing gene expression within the fast subfamily. We recently reported the characterization of the upstream promoter sequences of the three adult skeletal fast MyHC genes in C 2 C 12 myotubes (21). This study identified several regulatory elements involved in differential expression of the fast MyHC genes in vitro. The present study is the first to compare directly the sequences and activities of all three adult fast promoters in skeletal muscle in vivo.
In the present study ϳ1 kb of the upstream promoter regions of the three adult fast MyHC genes was sufficient to confer the following: 1) skeletal muscle-specific expression in vivo, because activity was high in skeletal muscle but minimal in heart (Fig.  1A); 2) muscle type-specific expression, as each promoter was differentially active depending on the level of endogenous MyHC isoform expression in different muscles (Fig. 1, B-F); and 3) fiber type-specific expression, as ϳ800 -1000 bp of the promoter linked to GFP was sufficient to restrict expression to the individual fiber types ( Fig. 2 and Table I). Thus, most or all of the cis-regulatory elements necessary for this hierarchy of different regulatory decisions, i.e. skeletal versus cardiac, slow versus fast muscle, and IIa versus IId/x versus IIb, are contained within the first 800 -1000 bp upstream of the transcription start site, although this does not preclude regions up and downstream of this from contributing to and/or modifying MyHC gene expression.
We used sequence comparison to identify elements within this 1000-bp region that may be involved in the expression of the adult fast MyHC genes in adult muscle in vivo. We focused on two elements in the proximal promoter region, the AT2 and the CArG-like elements, because these had been demonstrated previously to act as activators of the IIb MyHC promoter (12,27). The IIb AT2 element binds both Oct-1 and MEF-2, but binding is greatly reduced for the IIa and IId/x AT2 elements (Fig. 4), resulting in decreased activity of the IIa and IId/x MyHC promoters (Fig. 3). Similarly, the IIb MyHC promoter contains a CArG box or consensus SRF binding sequence (CCAAAAATGG), but substitutions in the IIa and IId/x promoters have resulted in elimination of SRF binding (Fig. 7), which, in turn, decreases promoter activation (Fig. 5).
Experiments with sensor constructs demonstrated that the IId/x CArG-like element acts as a repressor in a heterologous context (Fig. 6). This IId/x CArG-like element binds a protein that is not antigenically related to SRF (Fig. 7) but is found in both muscle (Fig. 7) and non-muscle nuclear extracts. 2 Analyses using transcription factor binding motif recognition software such as MatInspector have failed to identify a likely candidate for this protein; we are currently attempting to identify this putative repressor. Nevertheless, these data support the hypothesis that the substitutions that occurred in the IId/x MyHC CArG-like element during its evolution from the ancestral MyHC gene have resulted not just in the elimination of an activator site but also in the creation of a repressor site.
The results of the present study support a role for the AT2 and CArG-like elements in the differential expression of the three adult fast MyHC genes. However, because all of the mutagenesis experiments were done using the ϳ1000-bp upstream promoter regions for all three MyHC genes, the possibility remains that other, as yet unidentified elements within this region may modulate the effects of these elements. Indeed, deletion to Ϫ300 bp resulted in a significant decrease in activity of all three MyHC promoters to a level comparable with one B, binding pattern of the IIa, IId/x, and IIb CArG-like elements. Lanes 1, 5, and 9, probe alone; lanes 2, 6, and 10, probe plus C 2 C 12 myotube nuclear extract; lanes 3, 7, and 11, probe plus C 2 C 12 myotube nuclear extract plus polyclonal anti-SRF antibody; lanes 4, 8, and 12, probe plus C 2 C 12 myotube nuclear extract plus 50-fold excess cold probe. C, binding pattern of the IId/x CArG2 element. Lane 1, probe alone; lane 2, probe plus C 2 C 12 myotube nuclear extract; lane 3, probe plus C 2 C 12 myotube nuclear extract plus anti-SRF antibody; lane 4, probe plus C 2 C 12 myotube nuclear extract plus 50-fold excess cold probe. The position of SRF is indicated by an arrow; the position of the SRF supershift (SS) is indicated by an arrowhead. The IId/x CArG-like element binding activity is indicated by an asterisk.
another both in vitro (21) and in vivo (Fig. 9A), strongly suggesting that these elements within the proximal 300 bp interact with element(s) upstream of 300 bp to establish differential expression of the adult fast MyHC genes.
One puzzling aspect of the present data is the role for MEF-2 and SRF in regulating quantitative expression of the three adult fast MyHC genes in vivo. Quantification of SRF levels (28) or MEF-2 sensor activity (18) in adult mice has suggested that expression and/or activity of these factors is low in adult fast skeletal muscle. How is it that these two factors, which are enriched in slow muscles such as the soleus, are involved in the regulation of expression of the fast MyHC genes, particularly IIb MyHC, which is rarely expressed in the soleus under normal conditions? First, there is some evidence that levels of these transcription factors are associated with IIb MyHC levels; dy/dy dystrophic mice, for example, have greatly reduced muscle MEF-2 and SRF concentrations (29) as well as decreased IIb MyHC levels and increased type I MyHC levels (30,31). Second, it may be that MEF-2 and SRF, which are typically expressed in much higher concentrations in developing muscle, are necessary for the initiation of IIb MyHC expression and that maintenance of fiber specificity requires much lower levels of these proteins in adult muscle. Third, both of these transcription factors associate with a number of other coactivators (32,33), and these cofactors may be differentially expressed in different fiber types. Moreover, both the MEF-2 and SRF gene are extensively alternatively spliced (26,34), and different isoforms may be expressed in a fiber-specific manner. Finally, our own data are consistent with the idea that additional elements upstream of the AT2 and CArG elements are also involved in modulating the effect of these elements (Fig. 9). Thus, MEF-2 and SRF appear to be critical for the quantitative expression of the IIb MyHC gene, but other transcription factors and their binding elements are necessary to restrict fast and slow MyHC expression to the appropriate muscle fiber types.
Our data suggest that the AT2 and CArG elements play a central role in the diversification of MyHC isoform expression across different fiber types. Interestingly, comparison of these sites across species reveals that these changes have not been conserved. For example, the IIa AT2 sequence differs in a single nucleotide in pigs from that in mouse or human (5Ј-TCAAATTATCCATAGGAGA-3Ј for mouse/human and 5Ј-TCAAATTATCCATAAGAGA-3Ј for pig, with the differing nucleotide underlined). The CArG-like element in the IIa MyHC promoter also differs between pig, mouse, and human (5Ј-CCAAGAGGGT-3Ј for mouse, 5Ј-CCAAGAGGTA-3Ј for human, and 5Ј-CCAAGACTTT-3Ј for pig, with the differing nucleotides underlined). Moreover, the consensus AT2 and CArG binding sites found in the mouse IIb MyHC promoter are not conserved in the human promoter and may indeed underlie some of the differences in IIb MyHC expression between mouse and human. 3 Together, these data suggest that these two elements may represent "hot spots" wherein mutations have accumulated that dramatically affect orthologous and paralogous expression of the adult fast MyHC cluster.
In summary, we have demonstrated that ϳ1 kb of the upstream promoter region of the adult fast MyHC genes is sufficient to confer muscle-specific, muscle type-specific, and fiberspecific activity in vivo. Moreover, we characterized the role of specific cis-elements in the three adult fast MyHC promoters. These studies provide a foundation for the analysis of two questions pertinent to muscle biology: 1) identification of the element(s) responsible for establishing and maintaining fiberspecific gene expression; and 2) elucidating the transcriptional signaling pathways that allow fibers to adapt to changes in neuromuscular activation.