Corresponding author D. Vullhorst: Section of Molecular Neurobiology, NICHD, Bethesda, MD, USA. Email: email@example.com
Firing patterns typical of slow motor units activate genes for slow isoforms of contractile proteins, but it remains unclear if there is a distinct pathway for fast isoforms or if their expression simply occurs in the absence of slow activity. Here we first show that denervation in adult soleus and EDL muscles reverses the postnatal increase in expression of troponin I (TnI) isoforms, suggesting that high-level transcription of both genes in mature muscles is under neural control. We then use a combination of in vivo transfection, live muscle imaging and fluorescence quantification to investigate the role of patterned electrical activity in the transcriptional control of troponin I slow (TnIs) and fast (TnIf) regulatory sequences by directly stimulating denervated muscles with pattern that mimic fast and slow motor units. Rat soleus muscles were electroporated with green fluorescent protein (GFP) reporter constructs harbouring 2.7 and 2.1 kb of TnIs and TnIf regulatory sequences, respectively. One week later, electrodes were implanted and muscles stimulated for 12 days. The change in GFP fluorescence of individual muscle fibres before and after the stimulation was used as a measure for transcriptional responses to different patterns of action potentials. Our results indicate that the response of TnI promoter sequences to electrical stimulation is consistent with the regulation of the endogenous genes. The TnIf and TnIs enhancers were activated by matching fast and slow activity patterns, respectively. Removal of nerve-evoked activity by denervation, or stimulation with a mismatching pattern reduced transcriptional activity of both enhancers. These results strongly suggest that distinct signalling pathways couple both fast and slow patterns of activity to enhancers that regulate transcription from the fast and slow troponin I isoforms.
The presence of slow- and fast-twitch fibre types is a hallmark of adult vertebrate skeletal muscles. Their relative abundance within any particular muscle is tuned to match its functional demand as part of the motor system. Fibre types express distinct sets of contractile and metabolic proteins, as well as receptors and ion channels (Schiaffino & Reggiani, 1996). They are established during fetal development largely independent of environmental cues (Condon et al. 1990; Stockdale, 1997). Postnatally, however, skeletal muscles acquire an exquisite ability to respond to changing work demands by adjusting muscle mass and fibre type composition (Pette, 2002; Rennie et al. 2004). For example, paradigms for increased mechanical activity such as functional overload, resistance training and endurance exercise lead to an adaptive fast-to-slow shift in MHC expression (Baldwin & Haddad, 2001). Since the seminal studies of Buller and Eccles (Buller et al. 1960) it is widely accepted that maintenance and transformation of adult skeletal muscle fibre types are largely under neural control (Salmons & Henriksson, 1981; Pette & Vrbova, 1985; Gundersen, 1998). Motoneurones modulate gene expression of their muscle targets through signals encoded by the pattern of action potentials they evoke in the muscle fibres. Electrical stimulation with paradigms mimicking natural motoneurone activity can fully substitute for the nerve in maintaining fibre phenotypes or triggering adaptive fibre type transformations, depending on whether a matching or a mismatching pattern is applied (Eken & Gundersen, 1988; Windisch et al. 1998).
Fibre type-specific expression of slow and fast troponin I (TnI) isoforms in rodent hindlimb muscles is established during fetal development (Zhu et al. 1995; Calvo et al. 2001). Postnatally, high-level expression of the slow and fast TnI mRNAs depends on ‘slow’ and ‘fast’ innvervation, respectively (Calvo et al. 1996). In particular, the frequency of action potentials seems to be a critical determinant of fibre type-specific TnI expression. Thus, normal transcript levels of the slow TnI isoform (TnIs) can be maintained in denervated rat soleus muscles that are electrically stimulated with a 10-Hz pattern. In contrast, a 100-Hz pattern that delivers the same amount of action potentials is ineffective in maintaining TnIs mRNA levels but is sufficient to up-regulate expression of the fast TnI isoform (TnIf). Moreover, the effect of the phasic, fast stimulation differs dramatically from the effects of denervation (Calvo et al. 1996).
Transcriptional control elements governing fibre type-specific gene expression in transgenic mice have been identified for both fast and slow TnI isoforms (Yutzey et al. 1989; Banerjee-Basu & Buonanno, 1993; Nakayama et al. 1996): a 128 bp slow-specific upstream regulatory element, termed SURE, and a fast-specific 144 bp intronic element, FIRE, that share common cis-regulatory DNA motifs. We have recently shown that SURE comprises two functionally interacting components (Calvo et al. 2001). The upstream half confers pan-muscle expression, while the downstream half confines enhancer activity to slow fibres. However, it is at present not known whether SURE and FIRE enhancers represent only developmental control elements required for the establishment of fibre type-specific TnI expression, or if they also respond to patterned electrical activity in mature muscles.
In this study we utilize a new combination of in vivo transfection, electrical stimulation and live microscopic analysis of single cells to investigate the responsiveness of slow and fast TnI promoters/enhancers to distinct electrical activity patterns. We have taken advantage of the green fluorescent protein (GFP) reporter, driven by these enhancers, to monitor transcription in vivo. The activity patterns used for stimulation were designed to mimick important aspects of those found in natural slow and fast motor units of the rat, and have previously been shown to be efficient in transforming muscle contractile properties (Eken & Gundersen, 1988; Windisch et al. 1998). By imaging transfected rat soleus muscles and quantifying fluorescence of distinct fibres in the animal before the onset and upon completion of stimulation, we were able to calculate the change in reporter expression over time. We demonstrate that transcription from the 2.1 kb upstream/intronic fragment of the fast TnI gene, containing the FIRE element, is activated by fast activity compared to muscles that were either denervated, stimulated with a slow pattern or innervated by a slow nerve. In contrast, transcription from the 2.7 kb upstream fragment of the rat slow TnI gene, which includes the SURE, is activated by slow activity.
All surgical procedures were performed on 250–350 g male Wistar rats (6–8 weeks) anaesthetized with equithesin (5 μl g−1 body weight) delivered intraperitoneally. Stimulation experiments were approved by the Norwegian Animal Research Authority. Denervation experiments presented in Fig. 1 were approved by the NIH Office of Laboratory Animal Welfare. Animals were killed by CO2 inhalation and cervical dislocation.
Troponin I regulatory regions were subcloned into pEGFP (Clontech) containing the 3′-intron and the polyadenylation sequence of the mouse protamine 1 gene (Johnson et al. 1988) downstream of the GFP coding sequence (pEGFPmP1). A segment of the rat TnIs gene encompassing 2.7 kb of upstream sequence down to nucleotide position (nt) + 21 was excised from TrpI.CAT (Banerjee-Basu & Buonanno, 1993) as a BamHI–AvaI fragment, and inserted between the BamHI and SmaI sites of pEGFPmP1 after blunt-ending the AvaI site with Klenow polymerase. A 2.1 kb TnIf segment between nt −530 and nt +1604 of the quail gene was released from the parental plasmid TnICAT1 (Yutzey et al. 1989) by SalI and partial Sau3AI digestion, and inserted between the SalI and BamHI sites of pEGFPmP1. A plasmid expressing bacterial β-galactosidase under control of the CMV promoter was used for control transfections.
Soleus and EDL muscles from young rats (P0–P21) were used to determine the time course of postnatal expression of troponin I genes. For each time point, muscles from 7 to 14 pups were pooled. For experiments in adult rats, both legs were denervated by removing a ∼5 mm segment of the sciatic nerve high up in the thigh. Muscles were harvested 1–14 days later. Total RNA was extracted from normal and denervated rat soleus and EDL muscles using Trizol® (Life Technologies) or RNAwiz (Ambion). Ten micrograms of RNA was size fractionated by electrophoresis in 1.5% agarose–2.2 m formaldehyde gels, and electroblotted onto nylon membrane (Gene Screen, NEN). Blots were successively hybridized with 32P-labelled full-length probes for mouse fast and rat slow TnI transcripts (Koppe et al. 1989) and washed stringently at 65°C with 0.1 × SSC (1 × SSC = 150 mm NaCl, 15 mm sodium citrate pH 7.0). A 32P-labelled oligonucleotide encompassing the sequence between nt +207 and nt +247 of the mouse 28S ribosomal RNA was used for normalization. Signal intensities were directly quantified using a Phosphorimager (Amersham Biosciences).
In vivo transfection
DNA used for injections was purified by ion exchange chromatography (Qiagen). To facilitate subsequent imaging procedures, transfection conditions were adjusted to favour DNA uptake by superficial fibres by bathing the muscles in, rather than directly injecting them with, the DNA solution. In this way, 50 μl of a 0.5 μg μl−1 DNA solution in saline was applied on the soleus muscle surface. After 30 s the muscles were electroporated with five pulse trains each consisting of 1000 symmetrical 200-μs bipolar pulses with an amplitude of 50 V (∼150 V cm−1). Two 1-mm silver wires spaced at 3 mm and connected to a Pulsar 6bp-a/s stimulator (Frederick Haer & Co, Brunswick, ME, USA) were used as electrodes. Electrodes were placed on each side of the muscle such that the electrical field ran perpendicular to its long axis. The electrodes were moved stepwise along the muscle in between trains.
Electrical stimulation and denervation
Stimulations were performed essentially as previously described (Eken & Gundersen, 1988; Gorza et al. 1988). In brief, 1 week after transfection, soleus muscles of the right hindlimb were denervated by removing a 0.5–1 mm long segment of the sciatic nerve high up in the thigh. Then, the uninsulated distal ends of two Teflon coated steel electrodes (AS632, Cooner Sales, Chatsworth, CA, USA) were placed around the proximal and distal tips of the muscle. Their proximal ends were placed under the skin, exiting through the head via a 6-mm silicon tube fixed to the scalp with steel screws (Frederick Haer & Co) and dental cement (Swebond, Svedia Dental-Industri AB, Sweden). The electrodes were then connected to a rotating contact approximately 0.5 m above the animal's head to allow free movement in the holding cage, and finally to electrical stimulators. Stimulation commenced 24 h later. Two different stimulation patterns were used. A 15 Hz pulse train lasting 10 s and repeated every 30 s was used to simulate activity of slow type I motor units. The fast pattern was designed based on the frequency of motoneuronal discharges observed in type IIB motor units (Hennig & Lømo, 1985), and consisted of trains of 25 pulses at 150 Hz every 15 min. Muscles were stimulated for 12–13 days. Control contralateral soleus muscles were either innervated or denervated. Upon completion of an experiment, muscles were excised and assayed for acetylcholine supersensitivity to gauge the success of the stimulation procedure (Lømo & Rosenthal, 1972). Only stimulated muscles that generated less than 10 mN of additional force were included in the data sets.
Imaging and quantification
Muscles were exposed and placed under a Plan Fluorite 1 × Par focal objective (DFPLFL 1 × PF, Parfocal/PF series, Olympus) mounted on an Olympus SZX-RFL fluorescence microscope to give high quality colour images. A filter cube (SZF-FAD, Olympus) with a barrier filter (BA510IF) and an excitation filter (BP460-490) combination was used. Expression of enhanced GFP in transfected muscles was quantified in situ by taking digital pictures using a chilled 3CCD video colour camera with the gamma value set to 1 (Hamamatsu, C5810). By placing calibrated neutral filters (12.5%, 25%, 50% and 79% transmission) into the light path, pixel values up to 125 were found to fall within the linear range of the camera. All data acquisitions were done within this range. When the epi-illumination was projected onto a uniformly white surface, the variability of pixel values across the field was < 3%. To account for possible variability in camera sensitivity and the mercury light source over time, 15 μm InSpeck green (505/515) fluorescent calibration beads (Molecular Probes) were used as internal controls. For in vivo imaging, the lower hindlimb was placed under the microscope with the lateral side facing up. Surgical retractors were used to pull the gastrocnemius to the side and the soleus muscle was carefully positioned using a smoothened glass rod. The dorsal and lateral sides of the muscle were examined for GFP positive fibres. In preliminary control experiments, the calibration beads were implanted in the rat hindlimb and bead intensity was measured in situ before closing the wound, and then again after 2 weeks (corresponding to the typical duration of a stimulation experiment). After normalization against an external fluorescence standard (Leica Fluorescence Standard, Tamro Medical Laboratory AS, Oslo) that was introduced in the field immediately after recording from the beads, light emission was found to be unchanged, affirming that bead fluorescence was stable for 2 weeks in situ. The beads also proved to be resistant to bleaching by prolonged illumination (data not shown). Bead-to-bead variability was negligible. For stimulation experiments, the calibration beads were added at the onset of stimulation to areas that later were to be imaged. Fluorescence from both beads and fibres was measured at this time and again upon completion of the experiment. The positions of beads with respect to each other and relative to certain landmarks such as blood vessels was used to identify the same fibres in both image sets. Upon acquisition and normalization of the image, windows of approximately 300 pixels were placed over fluorescent fibres and used to determine the total fluorescence intensity within the selected areas. When comparing the same fibre before and after treatment, fluorescence intensities were measured from windows with the same number of pixels. The overall image light intensity was adjusted as to give the same pixel value from the relevant beads at both time points in order to compensate for illumination instability. Necessary adjustments, however, were always less than 3%. The corrected average pixel value for each fibre at the onset of stimulation was recorded as ‘GFP1’ and for the same fibre upon completion of stimulation 12–13 days later (‘GFP2’). The relative change in GFP expression in individual fibres was calculated as:
The relative change in TnI expression was calculated only for those fibres that could unambiguously be identified in both images. The number of quantified fibres per muscle varied between 1 and 12 (median/mean = 3/3), with a range of 17–47 fibres analysed for each treatment. In addition, the total number of fibres was recorded from image sets; only those sets in which all fluorescent fibres could unambiguously be identified were included in this analysis. Pairwise comparisons of the different treatments were tested by an one-way ANOVA post testing with Student's t test and Bonferroni's correction.
Nerve- and pattern-dependent expression of troponin I genes in the rat
To investigate the role of innervation in the regulation of TnI gene expression after birth, we followed the expression of both genes by Northern blot analysis in the rat during postnatal development and after denervation (Fig. 1). First, we determined the postnatal time course of TnIs expression in rat soleus and EDL muscles, which are predominantly composed of slow and fast fibres, respectively. As shown in Fig. 1, TnIs expression at postnatal day 0 (P0) is only about 2-fold higher in soleus than in EDL muscles and remains mostly unchanged for the first postnatal week. However, starting at P7, TnIs mRNA levels steadily increase in the soleus from an initial 26% to adult levels (set as 100%), while in the EDL they are down-regulated to very low levels. In mature muscles, TnIs transcripts are 50-fold more abundant in the soleus than in the EDL, corresponding to a ∼25-fold augmentation of fibre type specificity from the time of birth. The importance of innervation for TnIs expression is illustrated in Fig. 1B. Within 4 days following denervation, TnIs expression in adult soleus muscles decreases to levels similar to those found during the first postnatal week. Interestingly, residual expression in the soleus remains around 20% even upon prolonged periods of denervation (2 weeks), indicating the existence of a nerve-independent component in the regulation of TnIs gene transcription. Moreover, removal of the nerve does not result in the concomitant up-regulation of TnIs expression in the EDL, strongly suggesting that down-regulation of the TnIs gene in fast muscles is not mediated by a nerve-dependent repressive mechanism.
As shown in Fig. 1C, TnIf expression in the EDL steadily increases during postnatal development. Relative mRNA abundance starts from 35% at P0 and reaches adult levels after P21. Similar to the initial slow-specificity of TnIs, TnIf mRNA at birth is only about 2.5-fold more abundant in EDL than in soleus, but this ratio increases to ∼12.5 in the adult animal. Interestingly, TnIf expression modestly increases in soleus muscles up to P9, suggesting that processes driving TnIf expression during the first postnatal week are acting in both fast and slow muscles. Like TnIs, the postnatal increase of TnIf expression in the EDL is reversed in the adult within 4 days after denervation, although mRNA levels transiently return to ∼65% around day 9 before decreasing again (Fig. 1D). Taken together, high-level expression of both TnIs and TnIf genes in adult muscles depends on motoneurone innervation. TnI expression in adult denervated soleus and EDL muscles in contrast resembles a perinatal state that is established in the absence of distinct motoneuronal activity patterns.
In vivo transfection and imaging of troponin I reporter gene expression
To test the responsiveness of TnI regulatory sequences to patterned electrical activity in vivo, we developed a strategy that included transfection and live imaging of GFP fluorescence in muscle fibres. Most critically, we sought to record changes in reporter gene activity in the same muscle fibre before and after electrical stimulation. Figure 2A shows a flow diagram of the major steps involved. At the start of an experiment (day 0), soleus muscles were transfected with a GFP reporter plasmid harbouring regulatory sequences for either TnIs or TnIf (see below). A week later, the first picture was taken (‘GFP1’), and the muscle was denervated to eliminate endogenous motoneuronal activity. Electrodes were implanted and muscle stimulations commenced 24 h later lasting for a total of 12–13 days. Upon completion, the second picture was acquired (‘GFP2’). Because surface fibres are more favourable for imaging, we modified the standard transfection conditions (Mathiesen, 1999) such that muscles were bathed in DNA solution rather than directly injected prior to electroporation. Figure 2B shows cross-sections of solei transfected with a control β-galactosidase expression plasmid and stained with X-Gal. On the left, DNA was injected into muscles, while on the right it was applied exogenously. As evident from the staining pattern, exogenous application of DNA strongly favoured the uptake by peripheral fibres while injection resulted in the transfection of both deep and surface fibres.
TnI regulatory sequences used in this study are schematically depicted in Fig. 3A. TnIs2700 represents a 2.7 kb upstream fragment of the rat gene; TnIf2100 encompasses the sequence between −530 and −1604 of the corresponding quail gene. These fragments confer fibre type-specific reporter gene expression in transgenic mice (Hallauer et al. 1993) and include bona fide enhancers for TnIs (SURE) and TnIf (FIRE). In preliminary experiments using luciferase-based constructs, we found that transcriptional activities of these contiguous genomic sequences are ∼10-fold higher in transfected muscles than constructs containing minimal enhancer/promoter arrangements (data not shown), thus facilitating the acquisition and subsequent quantification of digital images. Figure 3B shows a representative image of a surface-transfected soleus muscle expressing GFP under the control of the TnIf2100 enhancer. Also visible in this image is fluorescence emitted from GFP calibration beads that were used to normalize image sets against each other prior to quantification. The beads also aided in identifying distinct fibres in both images as they maintained their position on the muscle surface during the treatment. Importantly, the transcriptional activity of the TnIf2100 enhancer was strong enough to allow detection of ‘baseline’ GFP expression in normal innervated soleus muscle fibres, thereby enabling us to study activity dependence of both TnI promoters in the same muscle.
Effect of slow and fast activity pattern on troponin I promoters
Using the strategy outlined above, we investigated the effects of distinct depolarization patterns on the transcriptional activity of the TnI promoters. Stimulation commenced 1 week after transfection to allow the muscle to recover from any possible effects of the electroporation procedure itself. A total of four conditions were tested and compared against each other: normal innervated, denervated, denervated/fast-stimulated, and denervated/slow-stimulated. In addition, in Table 1 data sets from the different conditions are compared against each other to evaluate statistical significance.
Table 1. Pairwise comparisons of transcriptional responses of TnI promoters between the four conditions used in this study
Mean difference (%)
Data sets were compared using Bonferroni's adjustment. †Data sets for TnIs2700/DEN and TnIs2700/SLOW treatments displayed a bimodal distribution (see Fig. 5B). Therefore, comparisons including these conditions were made separately for both the entire cohorts (normal numbers) and for the respective major subpopulations (italicized numbers; corresponding means in Fig. 5B indicated by filled bars).
INN versus DEN
INN versus FAST
INN versus SLOW
DEN versus FAST
DEN versus SLOW
FAST versus SLOW
INN versus DEN
INN versus FAST
INN versus SLOW
DEN versus FAST
DEN versus SLOW
FAST versus SLOW
Figure 4A shows representative image sets from muscle fibres transfected with GFP under the control of the TnIf2100 promoter and subjected to one of the four conditions described above. Figure 4B represents the corresponding quantitative analysis. In normal innervated muscles, fluorescence increased on average by 20% between the two time points (Fig. 4Aa and b, and B). This observation is in agreement with our previous results (Utvik et al. 1999), and is probably due to continued accumulation of GFP over time. The values ranged from 60% decrease to 110% increase, indicating some variability between different muscle fibres. Denervation resulted in a decrease in GFP fluorescence that was significant when compared to innervated control muscles, suggesting that nerve activity contributes to the basal activity of TnIf2100 in soleus independent of a distinct high-frequency activity pattern (Fig. 4Ac and d). In fibres stimulated with slow-patterned electrical activity (Fig. 4Ae and f, and B) we found a 30% reduction in fluorescence. This reduction was significant when compared against normal and fast-stimulated, but not denervated, muscles. The observation that slow-patterned stimulation and natural slow motoneurone activity triggered different responses in reporter gene expression was unexpected, since the slow paradigm was designed to mimic the frequency of slow motoneurone discharges. Because our test animals were not exercised or physically challenged in any way, it is conceivable that endogenous motoneurone activity was low compared to the applied electrical stimulation paradigm. Under such circumstances artificial stimulation could have had a stronger effect than normal innervation. Conversely, in muscles that were subject to a high-frequency depolarization paradigm (Fig. 4Ag and h, and B), GFP fluorescence was augmented significantly over denervated muscles with a mean increase of ∼50%.
In addition to fibres that displayed measurable changes in fluorescence before and after stimulations, we also frequently observed the appearance of fluorescence in fibres as a result of high-frequency stimulation or their disappearance after low-frequency stimulation. For obvious reasons these fibres could not be included in the quantitative analysis as shown in Fig. 4B and are therefore presented separately in Fig. 4C. The data are expressed as the relative change (%) in the number of fluorescent fibres regardless of intensity. Muscles that were stimulated with a high-frequency pattern showed an average 73% increase in the number of fluorescent fibres while the low-frequency pattern resulted in a 76% decrease. This result adds further weight to the quantitative analysis. Based on the comparison between high and low-frequency stimulation paradigms, we conclude that DNA regulatory sequences present in TnIf2100 selectively respond to distinct stimulation patterns in a manner that is consistent with the expression of the endogenous TnIf gene.
A corresponding set of experiments was carried out to assess the role of patterned activity in regulating the transcriptional activity of the TnIs promoter. As shown in Fig. 5Aa and b, and B, GFP fluorescence in normal soleus muscle fibres increased during the experiment on average by 40% probably due to ongoing accumulation of GFP (see above). Unlike TnIf2100, the response to denervation in fibres transfected with the TnIs2700 promoter segregated between two distinct populations. One population, consisting of 27 fibres (16 muscles), showed an average decrease of 20%, consistent with the down-regulation of the TnIs gene in denervated muscles (see Fig. 1B), while fluorescence in the second smaller population (represented by 9 fibres/4 muscles) increased by 79%; both values were significantly different from normal muscle (see Table 1). A similar effect was observed in the cohort of muscles that were stimulated with a slow pattern (Fig. 5Ae and f). A group of 17 fibres (6 muscles) showed a modest average increase of 15% that was not significantly different from normal muscle fibres (Fig. 5B). Fluorescence in the second group, however, increased by 148% (8 fibres/3 muscles), suggesting that in these muscles electrical stimulation was more effective than endogenous slow nerve activity (see above). At this time, the reason for the observed bimodal distribution of data points in these two conditions is unclear. Notably though, fibres from any given muscle consistently segregated into the same group, indicating that the underlying processes affected entire muscles rather than individual fibres. For this reason, and because soleus muscles from rats consist primarily of type I fibres (∼85% for the Wistar strain used in this study), it is statistically unlikely that this effect is due to preferential transfection of distinct fast or slow fibre types. The number of fluorescent fibres in the denervated group decreased by 26%, and increased modestly by 17% in the group that was stimulated with a slow pattern (Fig. 5C).
High-frequency stimulation (Fig. 5Ag and h, and B) resulted in an average decrease in fibre fluorescence of 20%. This effect was significantly different from the response observed in normal innervated but not denervated soleus (both considering the entire cohort as well as the major subgroup). Likewise, the decrease in the number of fluorescent fibres (33%) was similar for both treatments (Fig. 5C). This finding suggests that fast-patterned electrical activity does not act on the TnIs promoter as an instructive repressive signal. Rather, the absence of slow-patterned activity seems largely responsible for the down-regulation of reporter expression.
Taken together, the effect of patterned electrical activity on transcription from the TnIs2700 and TnIf2100 promoters is consistent with the pattern-dependent expression of the corresponding endogenous genes. Therefore our findings strongly suggest the presence of pattern-responsive DNA regulatory elements in both TnI promoters.
In this report we describe a novel combination of in vivo transfection and live muscle fibre imaging to assess the role of distinct patterns of action potentials in the regulation of TnI promoter sequences. To our knowledge, this is the first time that responsiveness of fibre type-specific muscle gene promoters to defined patterns of action potentials has been directly addressed. The results suggest that promoter elements of TnI fast and slow isoform genes react to distinct attributes of fast and slow patterns of action potentials, respectively. The elements confer positive regulation when isoform and pattern were matched. Because removal of nerve-evoked activity by denervation led to reduced transcription from both promoters, we postulate that TnI expression is positively regulated by distinct fast and slow pathways, and not through suppressive effects of mismatching activity patterns.
Different troponin I promoters respond distinctively to fast and slow activity patterns
Previously, promoters for slow and fast muscle genes such as myosin light chains 2v and 1s, β-myosin heavy chain, and aldolase were tested in transfected regenerating and adult muscle fibres and in transgenic mice with regard to the role of innervation per se (Jerkovic et al. 1997b; Lupa-Kimball & Esser, 1998; Spitz et al. 1998; Murgia et al. 2000; Bertrand et al. 2003; Huey et al. 2003). Although the TnIs enhancer SURE has previously been implicated in calcineurin-dependent slow-specific transcriptional regulation (Chin et al. 1998; Wu et al. 2000), this is the first study to directly demonstrate the ability of TnI promoters to differentially respond to distinct activity patterns. Based on studies measuring RNA and protein levels in neonatal and regenerating adult rat muscles, it has been proposed that fast-specific contractile protein expression represents a default state that prevails in the absence of innervation (Butler-Browne et al. 1982; Esser et al. 1993; Jerkovic et al. 1997a). Our data suggest the existence of a distinct ‘fast’ pathway activating fast genes by mediating signals from a high-frequency pattern of action potentials (Calvo et al. 1996). Nerve removal, on the other hand, merely reactivates an intrinsic embryonic muscle gene expression programme including the up-regulation of myogenic regulatory factors (Eftimie et al. 1991; Buonanno et al. 1992) and embryonic myosin heavy chains (Schiaffino et al. 1988), and lower levels of transcription for both of the troponin I isoforms (this paper).
Activity represents an instructive signal to the muscle fibre to adapt its gene expression programme to changing functional demands (Gundersen et al. 1988; Pette & Vrbova, 1992; Schiaffino & Reggiani, 1996; Gundersen, 1998). The effects are most often attributed to the action potentials per se, but our understanding of intracellular pathways connecting action potentials and gene expression is still rudimentary. The contribution of other phenomena related to action potentials, such as stretch, cannot be exluded. The number of action potentials delivered is among the most important signals regulating MHC and shortening velocity (Gundersen et al. 1988; Gundersen & Eken, 1992), and it has been proposed that low-amplitude elevations of [Ca2+]i trigger slow signalling pathways (Chin et al. 1998). Twitch duration, on the other hand, has to be matched to the prevailing firing frequency of the motoneurone in order to ensure sensitive force regulation, and this parameter has been shown to be dependent also on the instantaneous frequency of the action potentials (Gundersen et al. 1988; Gundersen & Eken, 1992). Troponin I is the regulatory component of the troponin complex, and probably influences the rate of force generation and relaxation during a twitch (Squire & Morris, 1998). We showed that delivery of equal amounts of stimuli at different frequencies selectively regulates TnIs and TnIf expression (Calvo et al. 1996). Studies of TnI promoter elements are therefore well suited to identify signalling pathways that are dependent on the frequency of action potentials.
The DNA responsive elements of all putative activity-dependent pathways are yet to be identified. In the sequences used in this study, the only bona fide regulatory regions identified in cultured cells and transgenic mice are the TnIs SURE and TnIf FIRE as well as the core promoters immediately upstream of the respective transcription initiation sites. Myogenic regulatory factors, MEF-2 and NFAT proteins have been implicated in fibre type-specific gene regulation (Hughes et al. 1993; Chin et al. 1998; Hughes et al. 1999; Wu et al. 2000; Schiaffino & Serrano, 2002; Ekmark et al. 2003). However, two pieces of evidence argue against a role of these factors in conferring responsiveness of TnI expression to patterned electrical activity. First, both SURE and FIRE contain binding sites for all three families of transcription factors. Second, experiments using 5′-deleted and chimeric FIRE–SURE enhancer constructs in transgenic mice have unequivocally demonstrated that the downstream half of the SURE, which includes cis-elements for all these factors, confers pan muscle transcriptional activity and lacks slow muscle specificity. The upstream sequences in both SURE and FIRE are necessary to restrict transcription to distinct fibre types (Calvo et al. 1999; Calvo et al. 2001). The factors that bind these sequences in mature muscles are not known. Alternatively, as yet unidentified DNA elements that respond to patterned activity could be located within promoter sequences analysed in this study. These sequences could work in conjunction with, or independently of, SURE and FIRE enhancers. Future experiments will address these scenarios by selectively deleting or mutating SURE and FIRE enhancers in the context of their cognate promoter sequences.
The relative importance of nerve-derived and developmental factors for TnI expression
Our expression analysis of slow and fast TnI genes in normal and denervated adult rat soleus and EDL muscles revealed the contribution of nerve-dependent and -independent regulatory components. Approximately 80% of TnIs gene transcription in the adult is nerve dependent as estimated by the extent of TnIs mRNA down-regulation 4 days following denervation; for TnIf, innervation accounts for 60% of steady-state mRNA levels in the adult. Strikingly, the residual TnI expression levels seen after prolonged denervation are similar to those found around birth. Although overall TnI transcript levels are reduced in denervated muscles, both TnIs and TnIf genes continue to be expressed preferentially in distinct fibre types. This finding is in agreement with the notion that intrinsic and extrinsic factors are necessary to establish fibre type-specific muscle gene expression (DiMario & Stockdale, 1997). In this regard it is interesting to note that a sequence element in the SURE necessary for fibre type-specific reporter expression in transgenic mice binds the nuclear factor GTF3 (Calvo et al. 2001). Based on its early expression pattern, it could be involved in the establishment of slow muscle-restricted TnIs expression during fetal development.
While the initial prenatal phase of fibre type specification of TnI expression probably occurs in the absence of distinct nerve-evoked activity pattern, such signals play a critical role in the control of TnI expression in adult muscle. This strongly suggests that the nerve-independent component represents an early developmental contribution to fibre type specification of TnI gene expression. Although not directly tested in this study, it is tempting to speculate that the postnatal increase in TnI expression is mediated by nerve activity. In agreement with this idea, motor units in the mouse soleus begin to fire with distinct pattern only after the first postnatal week (Personius & Balice-Gordon, 2001). The corresponding transition includes a qualitative change of motor unit activity from sustained low-frequency (0.5 Hz) activity to intermittent periods of higher-frequency activity (9–12 Hz) at P14–15. After P15, firing patterns resemble those of adult soleus muscle fibres (Hennig & Lømo, 1985).
How should activity-dependent gene expression be studied?
Because pattern-dependent muscle gene regulation commences only late during muscle maturation (see above), recapitulation of these processes in cultured primary cells or cell lines that lack many of the environmental cues that normal muscles are exposed to has been a technical challenge. Therefore, the most convincing experiments thus far have been conducted in situ using either transgenic or in vivo transfection approaches. A major advantage of the approach used here, i.e. combining in vivo transfection with in vivo imaging, is that it allows single fibre resolution of reporter expression that can be monitored before and after electrical stimulation, in contrast to other widely used reporter genes such as luciferase, chloramphenicol acetyltransferase and β-galactosidase whose activities are normally quantified as single data points in lysates of whole muscles. In vivo imaging was pioneered by Lichtman and colleagues to analyse synaptic plasticity (Lichtman et al. 1987; Balice-Gordon & Lichtman, 1990). GFP imaging in electrically stimulated dissociated muscle fibres in vitro has been used previously to monitor nuclear trafficking of NFAT (Liu et al. 2001), but to our knowledge this is the first time such an approach has been used in vivo to follow changes at the level of transcription in single cells over time. This approach requires an array of precautions including the in situ application of GFP calibration beads and a minimization of scattered light from deep transfected fibres contributing to the signal recorded from peripheral fibres.
It is noteworthy that the magnitude of transcriptional changes observed for the endogenous genes in response to denervation or electrical stimulation was significantly bigger than those obtained for the corresponding GFP reporter constructs. However, there are technical reasons that are likely to contribute to these quantitative differences. For example, because of the relative stability of GFP in cells, the effect of treatments that lead to a down-regulation of reporter expression such as denervation were probably underestimated (see below). The discrepancy between the rate of transcription and steady-state protein levels can be dramatic. For example, the parvalbumin mRNA is quickly down-regulated in fast-twitch muscles following denervation or during low-frequency stimulation, while the corresponding protein persists in muscle fibres for several weeks (Huber & Pette, 1996). The future use of destabilized versions of GFP could be one way to reduce the lag time between transcriptional changes and their manifestations at the protein level, provided that the inevitable reduction in steady-state GFP expression does not preclude quantitative fluorescence emission analysis.
The fibre type preference of both reporter constructs used in this study is somewhat smaller than that of the corresponding endogenous genes. While TnIs mRNA is ∼50-fold more abundant in soleus than in the EDL, and TnIf mRNA is ∼12.5-fold more abundant in EDL than in soleus (see Fig. 1), the ratios for the TnIs2700 and TnIf2100 constructs are 11 and 7, respectively (based on luciferase reporter variants of the plasmids used in this study; D. Vellhorst, unpublished observations). This reduction in ‘tightness’ of transcriptional control has also been observed with other muscle gene promoters such as the pM-310 of the aldolase gene (Bertrand et al. 2003). It appears to be largely due to dysregulation of episomal plasmid DNA since the identical DNA construct confers tight regulation when embedded in chromatin in transgenic mice. Because all these factors are likely to influence the steady-state levels of GFP, we strongly believe that it is the quality and not the magnitude of the observed changes in GFP fluorescence in response to stimulation or denervation that is a useful measure to correlate the properties of TnI promoters to the expression of the corresponding endogenous genes.
In conclusion, slow and fast stimulation paradigms elicit opposite transcriptional responses on both promoters. The high-frequency pattern activated TnIf2100 and the low-frequency pattern activated the TnIs promoter. Future experiments will focus on delineating the DNA elements within these promoter regions that respond to distinct activity pattern, and to identify the nuclear and signalling factors that relay the signal from the nerve.
This work was supported by NICHD, NATO grant CRG.CRG 973079, and by the European Commission (contract no. QLK6-CT-2000-00530). D.V. was supported by a fellowship from the German Research Council (DFG).