| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Centro di Ricerca Interdipartimentale per le Biotecnologie Innovative (CRIBI) Biotechnology Center, University of Padova, Padua, Italy
2 Department of Anatomy and Physiology, University of Padova, Padua, Italy
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
denervation; cDNA microarray; mitochondria; myosin
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
It is well known that neural stimulation is one of the most important determinants of gene expression in skeletal muscles (for a review, see Refs. 19 and 54) and this makes it very interesting to study how gene expression changes when neural control is removed. In slow postural muscles such as soleus, the low-frequency neural discharge is the essential factor in determining and maintaining the phenotype, whereas in fast muscles, the role of neural control is more controversial (54). Actually, fast muscle fibers receive 
100-fold less neural stimuli than slow muscle fibers do (25).
Denervation causes atrophy and change in fiber phenotype in fast as well as in slow muscles. The molecular mechanisms responsible for these changes are only partially understood; it is, however, clear that atrophy in response to denervation is different in slow and fast muscles. In denervated rat soleus, both slow and fast fibers undergo a very rapid atrophy (17), which is associated with a marked transformation of the muscle toward a fast phenotype (28, 41). In denervated extensor digitorum longus (EDL) and tibialis anterior (TA) muscles of adult rats, atrophy is not less severe, but it affects specifically fast fibers, whereas slow fibers seem more resistant (9, 17). In the same muscles, a progressive shift of myosin heavy chain (MyHC) isoform expression along the transition pathway 2B 
2X 2A 1 has been reported (28). In a recent study on rat EDL denervation (21), we have observed early changes in twitch time course that were accompanied by changes in sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA), parvalbumin, and calsequestrin expression. Electron microscopy has shown in rat fast muscle a proliferation of the junctional structures (triads and diads) that occurs a few days after denervation (58, 63). Thus, despite the limited amount of nervous stimuli received, fast muscles also respond to denervation, and the excitation-contraction coupling mechanism seems to be more sensitive than the contractile mechanisms to removal of neural control. Mitochondria and energy production systems are also affected by denervation in fast as well as in slow muscles (43, 72, 79). Evidence is, however, still fragmentary, as each study has focused on a single specific functional or structural aspect.
The development of the adaptive response requires several days or even weeks, and this makes necessary the identification of the precise time frame in which gene expression is studied. For instance, it is known that denervation of fast muscles produces a rapid upregulation of myogenic regulatory factors (MRFs) (6, 69), whereas the gradual transition in fiber type occurs more slowly (74). It is also important to take into account that data on denervation are mostly derived from studies on young adult rat muscles, and results obtained in other species (e.g., rabbit) suggest that the denervation-induced changes may be species, muscle, or developmental stage specific (15).
Many genes differentially expressed in denervated muscle have been identified so far (6, 64). However, only one high-throughput study on gene expression in denervation has been published and is focused on comparing gene expression in denervated neonatal rat muscles (16 days after denervation) with gene expression in space flight atrophy (51). A more complete mapping of the expression changes in an appropriate time window might provide information relevant to understanding the mechanisms leading to atrophy and fiber phenotype change after denervation in muscles.
In this study, we extended our previous work (21) on early changes induced by denervation in fast muscles. By means of real-time quantitative PCR (RQ-PCR) and cDNA microarray expression profiling, we followed changes in the expression of 2,000 genes in murine TA for 2 wk after surgical denervation. In particular, we focused on genes implicated in contractile and metabolic functions. A few essential genes were further analyzed at the protein level, and the possible functional relevance was also tested with physiological experiments. The results provided the first general timeline of changes in gene expression after denervation in a fast muscle and revealed that removal of neural control determines coordinated and sequential changes in gene expression.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
General design of microarray experiments.
Microarray experiments were carried out with cDNA targets from whole TA muscles, which were directly labeled during reverse transcription to preserve abundance relationships among messenger RNAs. Total RNA from denervated and contralateral TA was extracted from five mice for each time point. RNA from each muscle was prepared separately, and competitive hybridizations were carried out between denervated and contralateral samples of the same animal. To correct for biases due to the different incorporation rate and emission signal intensity of each fluorophore, we used reversed labeling of denervated and contralateral samples in each experiment, thus carrying out at least two separate hybridizations for each time point and for each animal. For each time point, the best experiments were selected for analysis. Criteria for selection included low background noise and absence of artifacts. The use of contralateral leg muscles is widely accepted in denervation studies. There is, however, evidence of effects of peripheral nerve section on contralateral motoneurons with possible influence on muscles (reviewed by Koltzenburg et al., Ref. 35). To evaluate whether contralateral muscles would undergo significant changes in gene expression during the time course of the experiment, competitive hybridization experiments were performed between day 1 and day 14 contralateral TA samples. Only 23 spots (each probe is represented on the array by 4 spots) and only 1 probe showed variations greater than two times.
Extraction of RNA.
Muscles were removed, weighed, cut into pieces, and frozen in liquid nitrogen immediately after each animal was killed. Frozen tissues were dispersed in 5 vol of Trizol (Invitrogen, Carlsbad, CA) and homogenized with an ultra-turrax-T8 blender (IKA-Werke, Staufen, Germany). Total RNA was purified according to the standard protocol. We checked for the absence of significant degradation by running an aliquot of RNA in an Agilent Bioanalyzer 2100, using the RNA 6000 LabChip kit (Agilent Technologies, Palo Alto, CA).
Microarray fabrication.
The tissue-specific microarrays used for this work (Mouse MuscleArray 1.0) consisted of 9,216 spots representing 2,061 cDNA probes, 1,402 isolated from cultured C2C12 myoblasts and 659 from adult hindlimb skeletal muscles. About six thousand muscle-specific expressed sequence tags (ESTs) were produced at CRIBI from specially designed cDNA libraries, following a strategy already applied to study human muscles (12, 38, 39). This strategy resulted in a nonredundant collection of cDNA clones. Before cloning, cDNAs were fragmented by sonication, and only the very 3'-end of the mRNAs was selected (38). This distinctive feature of our cDNA probes has two main advantages: the cDNA inserts are uniform in size (300–700 bp) and can be amplified with high efficiency; furthermore, hybridization to targets is more specific. All cDNAs considered in this work were sequence verified. Details on microarray printing and postprocessing have been deposited at the Gene Expression Omnibus (GEO) database (platform accession no. GPL1523).
Microarray hybridization.
Total RNA for each sample (8 µg) was reverse transcribed with SuperScript II (Invitrogen) and oligo-dT(21) as primer in the presence of either Cy3-dCTP or Cy5-dCTP (Amersham Biosciences, Uppsala, Sweden). Labeled cDNAs were further purified with a GenElute PCR Clean-Up kit (Sigma). Target cDNAs destined for paired analysis (i.e., denervated vs. contralateral) were mixed and precipitated with ethanol and salts. Competitive hybridizations were carried out in a dual slide chamber (HybChamber; GeneMachines, San Carlos, CA). Pellets of purified, labeled cDNAs were dissolved in 40 µl of hybridization buffer (Northern Max, Ambion) and applied on the microarrays, covered with a 22 x 22-mm glass coverslip. Hybridization proceeded overnight at 42°C by submersion of the chamber in a high-precision water bath (W28; Grant, Cambridge, UK). At the end of hybridization, slides were dived for 5 min sequentially in buffer A (1x SSC-0.1% SDS) at 65°C, in buffer B (0.1x SSC-0.1% SDS) at room temperature, and finally in 0.1x SSC.
Analysis of expression data.
Digital images were generated in a GSI Lumonics LITE dual confocal laser scanner (ScanArray Microarray Analysis software) and processed with QuantArray Analysis software (GSI Lumonics, Ottawa, Canada). Intensity values were processed at the SNOMAD web site (http://pevsnerlab.kennedykrieger.org/snomadinput.html). Normalized values were calculated for each spot and converted to a logarithmic scale. Final values correspond to log2 ratio of the normalized intensities; positive numbers correspond to RNA overexpressed in denervated muscles. The microarray data have been deposited in the GEO database (series accession no. GSE1893). To identify differentially expressed genes, we performed comparison statistical tests implemented in significance analysis of microarray (SAM) (68). SAM analyses were performed at the level of cDNA probes, thus grouping replicate data of the same probe (4 spots on the array). The following criteria were adopted to filter the row data. 1) After visual inspection of MA scatter plots generated by the SNOMAD software, spots with a weak hybridization signal (normalized meanlogint values < –5, which corresponded approximately to absolute intensities <700 in at least 1 channel) were considered not reliable. 2) Empty spots and cDNA probes with a defective signal were not analyzed. The multiclass SAM analysis was performed on a selected set of 2,056 cDNA probes as follows: four different classes have been set, one for each time point, and genes differentially expressed across all classes (namely, genes differentially expressed in at least 1 class) were selected with a false discovery rate (FDR) <1%. On the other hand, a one-class SAM analysis allowed selection of genes up- and downregulated at each single time point. Results of the two tests were consistent. Probes associated with higher FDRs are shown in Supplemental Materials (see below).
Real-time quantitative PCR.
Real-time quantitative PCR (RQ-PCR) based on the SYBR Green chemistry (Applied Biosystems, Foster City, CA) was carried out as described in a previous study (12). Total RNA (pooled individuals) from each time point was reverse transcribed with SuperScript II and oligo-dT(21). Diluted cDNA was amplified in 10 µl of PCR reactions in a GeneAmp 9600 thermocycler, coupled with a GeneAmp 5700 Sequence Detection System (Applied Biosystems). Gene-specific primers were selected with Primer 3 software; sequences of distinct exons were chosen to avoid amplifying contaminant genomic DNA. For each time point, we amplified samples and controls (denervated and contralateral TA) from multiple serial dilutions of the cDNA input. Differences in gene expression were evaluated by a relative quantification method, as described by Pfaffl (55). Values were normalized to the expression of the ß2-microglobulin internal reference, whose abundance did not change under our experimental conditions. Normalized ratios were converted in logarithmic scale, and standard deviation was calculated according to the mathematical methodologies described by Marino et al. (42).
Supplemental data.
The cDNA sequences, as well as more information about the Mouse MuscleArrays, are available as Supplemental Materials (available at the Physiological Genomics web site).1
All information relative to this study is available as Supplemental Materials: 1) RQ-PCR data; 2) a full list of differentially expressed genes detected by cDNA microarray analysis; 3) Supplemental Fig. S8 (nuclear genes with mitochondrial functions); 4) additional tables showing expression data of genes with biological functions relevant to the present study; these genes have been grouped according to Gene Ontology categories, and we noted that several muscle-specific genes have incomplete annotation.
In the Supplemental Tables, genes have been ranked according to their FDR: the lower the FDR associated, the higher the significance of the expression data. We identify five different classes: extremely significant genes (*****) with a probability of being a false positive <1%; moderately significant genes (****) with probability <5% or (***) <10%; and low significant genes (**) with a probability <15% or (*) <20%. Only genes of the first class (*****) are shown (see Fig. 3 and Table 2).
Gel electrophoresis and Western blot.
Small muscle fragments from control and denervated TA [3 for each time point, also used for succinate dehydrogenase (SDH) staining and single-fiber dissection] were shattered with a ceramic pestle in liquid nitrogen and dissolved at a concentration of 2 mg/ml in SDS-PAGE solubilization buffer (62.5 mM Tris, pH 6.8, 2.3% SDS, 5% 2-mercaptoethanol, 10% glycerol). MyHC isoform composition was determined by SDS-PAGE on a 4% stacking gel and on an 8% separating gel. Electrophoresis was run for 42 h at low temperature (70 V constant). After the gel run, the protein bands were revealed by Coomassie brilliant blue or silver staining, as specified. MyHC isoform percentage composition was determined by densitometry with a Bio-Rad Imaging Densitometer (GS-670). For the analysis of low-molecular-weight proteins, 12% electrophoresis gels with 4% stacking gels were used. The electrophoresis was run at 50 V constant in the stacking gel and 100 V constant in the separating gel. Proteins were transferred from gels to nitrocellulose membranes and identified by immunostaining. Transfer was obtained by applying a current of 300 mA constant for 2 h. Nitrocellulose sheets were reacted first with primary antibodies directed against the 13-kDa component of complex III (polyclonal antiserum, kindly provided by R. Bisson, Univ. of Padova), troponin I (T8-G, kindly provided by S. Schiaffino, Univ. of Padova), troponin T (RV-C2, kindly provided by S. Schiaffino), and myoglobin (FL-154, Santa Cruz) and then with a peroxidase-conjugated secondary antibody. The bands were visualized by an enhanced chemiluminescent method (ECL-Plus, Amersham Bioscience).
SDH.
SDH activity was determined by the method of Nachlas et al. (49) on transversal TA cryosections.
Physiological analysis on TA single fibers.
Single fibers were manually dissected from the superficial part of freshly excised TA muscles under a stereomicroscope (x10–60 magnification) while immersed in skinning solution. At the end of the dissection, fibers were bathed for 1 h in a skinning solution containing 1% Triton X-100 (Sigma) to ensure complete membrane solubilization. Segments 1–2 mm in length were then cut from the fibers, and light aluminum clips were applied at both ends.
Skinning, relaxing, preactivating, and activating solutions employed for mechanical experiments with single fibers were prepared as previously described (16). The pH of all solutions was adjusted to 7.0 at the temperature at which solutions were used (20°C). Protease inhibitors (E64, 10 µM; leupeptin, 40 µM) were added to all solutions.
Once the clips were applied, the fiber segments were transferred to the experimental setup and mounted in a drop of relaxing solution between the force transducer (AME-801 SensorOne, Sausalito, CA) and the electromagnetic puller (SI, Heidelberg, Germany) equipped with a displacement transducer. The setup and the recording and analysis system has been described in previous studies (66). After being mounted in the setup, the fiber segment could be rapidly moved between six drops containing different solutions with progressively decreasing pCa, from relaxing (pCa = 9) to activating solution (pCa = 4.8).
The fiber segments mounted in relaxing solution were stretched by 20%, reaching a final average sarcomere length of 2.7 µm. Then, the fiber segments were transferred into the subsequent solutions with decreasing pCa. The pCa-tension curve was obtained by measuring tension development at six progressively increasing calcium concentrations. Tension was normalized to tension developed at the highest calcium concentration (pCa = 4.8) and plotted vs. the pCa values. The resulting sigmoid curve was fitted by the Hill equation: relative tension = 100/[1 + 10(pCa – pK) x n]. The parameters pK, corresponding to the pCa value at which relative tension is 50% of the maximum, and n, corresponding to the Hill slope of the curve, were determined with a least-squares fitting procedure. Myosin composition was assessed by single-fiber gel electrophoresis (66).
Physiological analysis on intact EDL.
Intact EDL muscles were dissected from the denervated and contralateral legs of seven mice for each time point (7 and 14 days after denervation). Isometric properties were tested in vitro at 37°C. Details of the setup and the registration system have been given previously (47). Muscle responses were recorded via an AT-MIO 16 AD card, and data were analyzed by a virtual instrument created with the LabView computer program (National Instruments, Austin, TX). Twitch parameters, tension (Pt), contraction time (CT), and half-relaxation time (HRT), were measured. Force-frequency curves were determined by stimulating EDL muscles at 30-60-100-120-130 Hz for 500 ms. Force recorded at each stimulation frequency was normalized to the value at 130 Hz. Fatigue was induced with a 6-min-long protocol consisting of short tetani at low frequency (60 Hz for 300 ms) delivered every 3 s. Force recorded during the fatigue protocol was normalized to the initial value.
Statistical analysis.
Data are expressed as means ± SE unless stated differently. Nonparametric Mann-Whitney test was used for paired comparison between denervated and contralateral legs. Variance analysis (ANOVA) was used for comparison between control and denervated muscles at various times. Pearson correlation coefficients were calculated for comparison between variations in gene expression studied with RQ-PCR and with microarray. A probability <0.05 was considered statistically significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
3%, see Ref. 10). After resection of the sciatic nerve, there was a progressive and considerable atrophy of the denervated TA muscles, reaching 34% of the contralateral muscles after 14 days (Table 1). The cross-sectional area of the single fibers containing MyHC 2B isolated for mechanical studies showed an even more dramatic reduction (see Fig. 4C).
|
|
|
MRF expression.
Because expression of myogenin in adult fast muscles is strongly regulated by neural activity (30, 37), we tested by RQ-PCR the expression of all MRFs. Myogenin was in fact the most strongly induced MRF on denervation; mRNAs for MyoD and MRF4 were also upregulated, although not as strongly as myogenin, while expression of Myf5 did not change (Fig. 2B).
Microarray and differentially expressed genes.
To extend our analyses to other genes relevant for muscle functions, we took advantage of the microarray technology. The Mouse MuscleArrays used in this study are muscle-specific microarrays that contain a total of 2,061 cDNA probes, mostly representing mRNAs expressed at high or intermediate levels in skeletal muscle (see METHODS). Due to the composition of the probe set, this microarray provided a reliable description of genes coding for proteins involved in contractile and metabolic functions, whereas less information was available about regulatory genes controlling transcription, cellular growth, proliferation, and signaling cascades (see DISCUSSION). Total RNAs from denervated and contralateral TA muscles of the same animal were directly labeled during reverse transcription, and the cDNA target was subjected to competitive hybridizations.
After processing normalized data with a multiclass comparison statistical test, we found extremely significant expression changes (i.e., with FDR <1%; see METHODS) in 83 cDNAs, which were assigned to 71 unique gene products (Fig. 3). The genes relevant for muscle contractile and metabolic functions are listed in Table 2, whereas a full list of all differentially expressed genes can be found in the Supplemental Materials (see METHODS). To confirm the general validity of the microarray results, the expression of several differentially expressed genes was tested by RQ-PCR, with the finding of a good correlation with microarray data (Table 3).
|
|
|
To complete the analysis of the changes affecting the contractile mechanism, the expression of the Ca-calmodulin-dependent kinase Mlck, which is responsible for regulatory myosin light chain phosphorylation, was assessed by RQ-PCR and found strongly reduced after 7 days (Supplemental Materials). Calmodulin expression was also generally reduced (Supplemental Materials).
To assess whether the changes in myofibrillar gene expression were accompanied by functional changes, the contractile performance of single skinned fibers was analyzed. Despite the dramatic reduction of cross-sectional area (see Fig. 4C), tension (force/cross-sectional area) developed during maximal activation by fast fibers containing MyHC 2B was not reduced (Fig. 4A). The response to activator calcium was, however, decreased, as indicated by the shift to the right of the pCa-tension curve (Fig. 4B). No changes in troponin I and T expression at the protein level were found (see Fig. 4D, III, for troponin I; data not shown for troponin T). Therefore, the reduced calcium sensitivity might be explained by a lower phosphorylation of myosin light chain in view of the decreased expression of calmodulin and myosin light chain kinase.
|
|
B-crystallin and cysteine and glycine-rich protein 3 (Csrp3 or muscle LIM protein), which are generally more expressed in slow fibers, showed upregulation starting from day 3, whereas myozenin 1 (alias FATZ-1, calsarcin-2), a protein typical of fast fibers, showed downregulation at all time points (see Fig. 3 and Table 2).
Energy production.
Reduced transcription of genes involved in ATP synthesis or glucose utilization is a common feature of the rapid atrophy induced by fasting or systemic diseases (40). Ten differentially expressed genes involved in energy production are listed in Table 2, and all were downregulated after denervation. Five genes (Atp5b, mt-Co1, mt-Cytb, mt-Nd2, mt-Nd5) encode for proteins of the inner mitochondrial membrane (4 of them are mitochondrial encoded), taking part in electron transport and/or ATP synthesis, and three genes (Mdh2, Cs, Aco2) encode for enzymes participating in the tricarboxylic acid (TCA) cycle. The mRNAs of several glycolytic enzymes were reduced, with the muscle-specific isoforms of phosphofructokinase (Pfkm) and enolase (Eno3) reaching significant levels (Supplemental Materials). Adenylate kinase and mitochondrial creatine kinase (Ckmt2) were downregulated as well (Table 2). Interestingly, Pfkm, enolase, and adenylate kinase (Ak1) localize to the M band, where myomesin is also localized, pointing to the M band as a specific target for denervation atrophy. Myoglobin gene expression also showed significant changes, decreasing at 3 days and increasing at 14 days after denervation. A clear decrease was detected also at the protein level (Fig. 4D, II).
Mitochondrial enzymes.
In mammals, genes on the H-strand of the mitochondrial genome are transcribed as a polycistronic precursor molecule that is later processed to give individual mRNAs for 12 different polypeptides. Because all genes in the mitochondrial genome were represented on the array (in most cases with >1 probe), we checked whether the probes that did not reach statistical significance according to the most severe criteria (FDR <1%) also exhibited the same expression profile observed for the four genes described above (mt-Co1, mt-Cytb, mt-Nd2, mt-Nd5). The mRNAs of all 12 mitochondrial products shared a common expression profile, being progressively reduced up to 7 days after denervation (Fig. 6). At 14 days, a partial recovery of mitochondrial gene expression occurred. To explain this, we must take into account that variations of mitochondrial mRNAs might reflect not only changes in gene transcription but also the overall density of mitochondria in the muscle fibers. Of note, SDH staining revealed that denervation induced an altered intracellular distribution of mitochondria. In particular, subsarcolemmal mitochondria were reduced compared with controls, while the SDH activity became higher in the intermyofibrillar region (Fig. 7A). A similar pattern of mitochondria location had been observed also in muscles of spaceflight rats (51). Using a well-characterized antibody that reacts only with the 13-kDa component of the complex III proteins (7), the proteins of the mitochondrial fraction were examined by Western blot. A reduction of complex III proteins was apparent during the first week but did not reach statistical significance (Fig. 7B).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
2,000 genes was monitored during 2 wk after denervation. Denervation has distinct and, in some cases, opposite effects on slow and fast muscles. Available evidence (see Refs. 19, 54) suggests that the neural stimulation is essential to induce the slow type of gene expression, whereas it is controversial how much it is relevant to determine gene expression in a fast muscle. The present study was focused on the changes induced by the removal of nerve stimulation in fast muscles, and the TA of the mouse was chosen as a model. Before the current study, the response of murine muscles to denervation was not widely studied. Our analysis of the changes at the protein level based on electrophoresis and Western blot, plus the determination of some functional parameters, accompanied the gene expression profile obtained with cDNA microarrays, and RQ-PCR provided novel information about gene expression in response to denervation. The results show that, after the first 2 wk of TA denervation, a large number of genes changed their expression in a coordinated way, while the changes at the protein level and in the functional properties were limited. Before a discussion of the expression data, some important features of our approach need to be pointed out. First, total RNA was extracted from whole muscle homogenates, and therefore we could not discriminate among changes occurring in specific cell types. This problem is common in expression studies on muscle tissue (24) and cannot be overcome until single-fiber analyses are feasible. Second, the cDNA microarrays (Mouse MuscleArray) chosen for the expression analysis are derived from muscle cDNA libraries, and therefore the probe set is particularly reliable for assessing changes of transcripts typical of muscle cells. Thus the emerging view of muscle plasticity is centered on contractile and metabolic functions, whereas less information can be extracted about other important classes of proteins, like components of the proteasome subunits, membrane channels implicated in electrical activity, or families of transcription factors (see Supplemental Materials). The peculiar features of the cDNA probes on Mouse MuscleArray (see METHODS) make hybridizations more specific compared with arrays using full-length cDNAs. In fact, we found a good correlation between results of microarrays and RQ-PCR for all the transcripts tested (Table 3). To aid in the comparison of our data with other global gene expression studies, the microarray data have been deposited in the GEO database (series accession no. GSE1893).
Within the time window explored, denervated TA showed clear signs of a transition from fast to slow phenotype involving several myofibrillar proteins. Combined microarray and RQ-PCR analysis indicated a decrease in MyHC 2B mRNA and an increase in MyHC 2A mRNA, in agreement with previous results of ribonuclease protection assays in TA muscles of rats (28). In substantial agreement with changes in gene expression, a significant increase in MyHC 2A protein expression and little variation of MyHC 2B protein were detected 2 wk after denervation. It is not surprising that changes at the protein level are less pronounced than changes at the mRNA level, particularly in view of the long half-life (14.7 days) of MyHC 2B (65). A generally accepted scheme of reversible fiber type transitions assumes that fast-to-slow transforming muscles undergo a gradual replacement of the fast isoform MyHC 2B by MyHC 2A through the intermediate MyHC 2X (53, 60). The RQ-PCR results indicating overexpression of the slow MyHC isoform should be interpreted with caution, as the expression of MyHC 1 is highly dependent on the neural discharge [see Spangenburg and Booth (61) for a review]. In the rat, a substantial increase in MyHC 1 expression has been observed only 5 wk after denervation (46). Slow fibers in the deep part of TA are more resistant to atrophy than fast fibers in the superficial part (5, 9, 17). Taking into account that, in the denervated soleus, type 1 fibers undergo a massive atrophy (17), one might speculate that slow fibers in fast muscles such as TA have a different sensitivity to neural stimulation compared with slow fibers in soleus muscle, because they are committed during early myogenesis in a nerve-independent way (50), and for this reason they remain relatively insensitive to neural influences. The greater resistance to atrophy of slow fibers might explain why genes typical of slow fibers appear overexpressed without the need to assume a change in gene expression in fast fibers. Aside from MyHC isoforms, the expression of other genes coding for myofibrillar proteins differentially expressed in fast and slow fibers also suggests a fast-to-slow transition: for example, troponin I slow is overexpressed, whereas MyBP-C fast is downregulated.
The fast muscle phenotype is not only determined by the expression of fast isoforms of myofibrillar proteins but also by specific expression of genes implicated in cytosolic Ca2+ regulation. The transition toward a slower muscle phenotype is evident also at this level, as parvalbumin and SERCA1 expression decreases. Among genes coding for proteins involved in excitation-contraction coupling, the gene coding for the 
1-subunit of dihydropyridine receptor (DHPR) showed a significantly increased expression (see Supplemental Materials), whereas mRNA coding for ryanodine receptor (RyR) or other triad components did not vary significantly. The lack of variation of RyR expression is in contrast to the expectations based on electron microscope observations of a rapid doubling of the contact areas between T tubules and terminal cisternae in denervated muscles (58, 63). These ultrastructural findings contrast with the lack of changes in RyR expression (52) in EDL after long-term denervation (25–50 days). A recent paper by Radzyukevich and Heiny (56) shows an increase of mRNA of DHPR in various models of disuse, including denervation of fast muscle. Other studies (57), however, do not report any variations in DHPR expression. A possible explanation of the ultrastructural findings is that triad morphological changes represent a reorganization in response to the early reduction of myofibrillar mass that likely results in only minor transcriptional changes. The changes in gene expression are, on the whole, suggestive of slower kinetics of intracellular calcium, which is consistent with the prolonged time to peak and the leftward shift of the force-frequency curve observed in denervated murine EDL (Fig. 5), a finding that is in agreement with the our previous observations in rat denervated EDL (21). The choice to analyze contraction kinetics in vitro in EDL and not in TA was due to the great thickness and the lack of a proximal tendon of the TA, which make impossible a reliable analysis in vitro. Also, a determination of contractile performance of TA in vivo is made difficult when the nerve supply is interrupted and direct electrical stimulation of the muscle is needed. Because of these difficulties, twitch time parameters, force-frequency curve, and fatigue curve were determined in EDL muscles of the same animals used for RNA preparation. EDL can be considered a good model of TA because it is very similar to TA in its function (they are both foot dorsal flexor muscles without much of a postural role), innervation (they are both innervated by peroneus nerve), and fiber type composition: MyHC isoform analysis shows that predominant isoforms were 2X and 2B in control TA with 35 and 65%, respectively (see Fig. 1), and in control EDL with 16 and 81%, respectively.
A dramatic reduction in mitochondrial enzyme activities in denervated muscles has been reported in previous studies (26) (72). Our microarray data also show changes consistent with the view that denervation has pronounced effects on mitochondrial function and point to the transcriptional nature of those effects. In the first 3 days after denervation, all genes encoded by the mitochondrial DNA and several nuclear-encoded mitochondrial genes involved in energy production were markedly downregulated. However, during the progression of atrophy, we observed a tendency to rescue mitochondrial gene expression. After performing a cluster analysis of all genes present on the Mouse MuscleArrays with proved mitochondrial localization, we selected 50 genes that exhibited similar expression profiles even if they did not reached statistical significance (Supplemental Fig. S8). Among those genes, soluble enzymes (particularly of the TCA cycle) and membrane complexes involved in energy production through oxidative phosphorylation were particularly represented. Interestingly, some molecular components of the permeability transition pore complex, like voltage-dependent anion channel (VDAC) or adenine nucleotide translocase (ANT), were also identified by these criteria.
Distinct gene expression programs may be activated during the progression of atrophy to achieve a coordinated variation of mitochondrial functions. We noted that nuclear-encoded genes are mostly underexpressed 3 days after denervation, while the mitochondrial-encoded genes reach the lowest expression levels after 1 wk (Table 2). Analysis by computational approaches of the promoter region of the selected genes is underway, to identify putative transcription factor-binding sites and modules responsible for a common gene regulation. The pathway based on peroxisome proliferator-activated receptor G co-activator, or PGC-1, appears a likely candidate. Actually, it is known that Ca2+-calmodulin-dependent kinases (CaMK) are implicated in the regulation of mitochondrial gene expression of skeletal muscle through the expression of the master regulator of mitochondrial biogenesis, PGC-1 (76). PGC-1 causes induction of nuclear respiratory factor-1 (NRF-1) and NRF-2 gene expression and co-activates the transcriptional function of NRF-1 on the promoter of the mitochondrial transcription factor TFAM (77). A role for calmodulin and CaMK in activating expression of transcription factor NRF-1 has been demonstrated also in cultured cardiomyocytes (78). Our data show that all calmodulin genes are downregulated after denervation, and cytosolic Ca2+ levels are known to be low in denervated muscles (27).
Although the major focus of the study was on genes related to energy production and contractile responses, the present results provide some information on transcriptional mechanisms responsible for denervation atrophy. In agreement with previous studies (6), we found that denervation of TA induced the rapid overexpression of muscle-specific F-box (atrogin-1/MAFbx) and RING (MuRF-1) proteins that cause accelerated proteolysis through the ubiquitin-proteasome system. Recent data point either to Forkhead box O (FOXO) (59) or to NF-
B (31) as transcriptional regulator for ubiquitin ligases in atrophying muscles. Gene-silencing studies have further shown that distinct components of the NF-B family are recruited during atrophy of fast or slow fibers (31). Interestingly, the overexpression of atrogin-1/MAFbx and MuRF-1 peaked at day 3 and thereafter declined, suggesting that other signaling pathways and other proteolytic mechanisms might become predominant at later times, in general agreement with previous observations on disuse-induced atrophy (62).
The transcription factors of the MRF family are major regulators of the muscle phenotype, although their role in adult muscle has not yet been investigated in detail. Our RQ-PCR data on denervated TA are in good agreement with mRNA expression levels of myogenin and MyoD mRNA in the mouse TA muscle (32, 69). It is well known that MRF expression differs among various hindlimb muscles, correlating with their respective differences in fiber type composition (69); old animals have fewer satellite cells and spontaneous muscle fiber denervation, and this may in part account for differential expression levels among muscle types (37). The strong increase of myogenin mRNA after denervation is well documented (Ref. 37 and references therein), and immunohistochemical studies showed that myogenin protein is localized in nuclei of both adult fibers and satellite cells (32). The expression of MRF in satellite cells may be related with their activation, which is known to occur during early phases after denervation (see Ref. 8 for a recent review). Apparently, the overexpression of two potent inhibitors of cell proliferation, Gadd45a and Cdkn1a (or p21), also reported by Caiozzo et al. (11) in denervated laryngeal muscles, might regulate but not inhibit satellite cell activation.
TA denervation was followed by upregulation of genes like Csrp3 (see Fig. 3) and Ankrd2 (RQ-PCR data not shown), which, in accordance with previous studies (2, 67), are involved in transition toward slow phenotype. These proteins are normally expressed in slow muscles (13, 75) and can mediate protein-protein interactions through LIM or ankyrin domains, respectively (48, 70). Their possible role, however, remains controversial because of their dual localization. While in the nucleus, the CRP3/MLP protein may influence the myogenic activities of the MRFs through a direct physical interaction, as demonstrated for MyoD and myogenin (36). On the other hand, CRP3/MLP can bind several cytoskeletal proteins (14), and these interactions could be important for the rearrangement of the Z-disc during redifferentiation of adult muscle fibers (73). More recently, a multiple role has been suggested for the ANKRD2 protein, based on its ability to bind Z-disc proteins (e.g., telethonin) and to act as gene regulator (34). TA denervation also caused upregulation of B-crystallin, which is known to be more expressed in slow than in fast muscles (13). This small heat shock protein displays chaperone-like properties and might be involved in the remodeling of myofibrillar structures (18). On the whole, the upregulation in denervated TA of genes expressed at nearly undetectable levels in fast muscles supports the view that a real transformation of fast fibers in slow fibers is on the way.
In conclusion, the present results show that denervation of murine TA is followed by marked changes in gene expression that appear to be coordinated in the direction of a fast-to-slow transformation and of a reduced metabolic activity, particularly affecting mitochondrial and aerobic-oxidative metabolism. Our findings that removal of nerve supply has a great impact on gene expression in fast muscles were unexpected and open important questions concerning the signals and their intracellular mediation. Electrical activity, neurotrophic factors, and load might be the signals relevant to explain transcriptional changes.
1) The membrane electrical activity is an important regulator of the activity of Ca2+-dependent transcription factors via calcium/calmodulin-regulated enzymes such as calcineurin (phosphatase 2B), CaMK, and myosin light chain kinases. In particular, activated calcineurin promotes nuclear translocation of nuclear factor of activated T cells (NFAT), which in turn controls the expression of slow fiber type genes (44). Calcineurin activity is not only influenced by intracellular calcium but also by a family of calcineurin-interacting proteins called filamin-actinin-telothonin-binding protein of the Z disc (FATZ)/calsarcins (14). Gene silencing of the slow isoform of FATZ (calsarcin-1) resulted in constitutively enhanced calcineurin signaling and an excess of type 1 fibers in skeletal muscles (20). Similarly, the marked downregulation of the Myoz1 gene (see Fig. 3 and Table 2), encoding the fast isoform of FATZ (calsarcin-2), might eventually lead to increased calcineurin activity in denervated TA muscles. Marked slow-to-fast transformation occurs after denervation in the soleus muscle (28), whereas only partial fast-to-slow transitions occur in the denervated TA. The huge difference in the amount of neural stimulation delivered to a slow compared with a fast muscle (100-fold, see Ref. 25) might account for this discrepancy. Interestingly, there are also some common effects, for example the upregulation of MyHC 2A in both denervated slow muscles (28) and fast muscles (this study). Common to fast and slow muscles is the appearance of a new type of electrical activity: fibrillation appears in denervated rat TA after 54–55 h (45) and might cause an increase in electrical activity above that experienced by fast muscle fibers in physiological conditions. Among fast MyHC isoforms, the MyHC 2A promoter was shown to be by far the most responsive to intracellular calcium (1).
2) Neurotrophic factors might contribute to the control of transcription. The neuromuscular junction is a potential source of neurotrophic factors (for a review, see Ref. 23). Actually, recent studies have shown clear differences between the lack of nerve-muscle contact and the lack of the mere nerve electrical activity, suggesting a role for the ciliary neurotrophic factor to blunt muscle atrophy (32).
3) The load experienced by the leg muscles is dramatically reduced after sciatic nerve interruption. Both flexors and extensors of the foot were paralyzed, and the load applied by tonic and phasic activity of ankle extensors on TA disappeared. In this respect, it is paradoxical to note that several genes typical of slow muscles found upregulated in the present study (e.g., Ankrd2, Cryab, Csrp3) were induced in the TA muscle even after a single bout of eccentric contraction (3). It is tempting to speculate that those genes are implicated in a load-related intracellular signaling that would cause remodeling of myofibrillar structures, although the molecular mechanisms by which muscle cells are sensing such mechanical stimuli remain elusive at present. It must be underlined, however, that reduction of mechanical load is known to shift gene expression toward the fast and not the slow phenotype (29).
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: C. Reggiani, Dept. of Anatomy and Physiology, Univ. of Padova, Via Marzolo 3, 35131 Padua, Italy (e-mail: carlo.reggiani{at}unipd.it).
* A. Raffaello and P. Laveder contributed equally to this work. 
1 The Supplemental Material for this article is available online at http://physiolgenomics.physiology.org/cgi/content/full/00051.2005/DC1.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
oi-Vrani T, and Jerkovi R. Muscle fiber type composition and morphometric properties of denervated rat extensor digitorum longus muscle. Croat Med J 41: 294–297, 2000.[ISI][Medline]
