| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Cell Biology, Harvard Medical School, Boston, Massachusetts, USA;
School of Nursing and Health Studies, Department of Human Science, Georgetown University, Washington, D.C., USA;
Pulmonary and Rehabilitation Research Group, Department of Medicine, University Hospital Aintree, Liverpool, UK;
Department of Physiological Sciences and Brain Research Institute, University of California, Los Angeles, California, USA; and
|| Renal Unit, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA
2Correspondence: Renal Unit, Beth Israel Deaconess Medical Center, 330 Brookline Ave., Boston, MA 02215, USA. E-mail: slecker{at}bidmc.harvard.edu or alfred_goldberg{at}hms.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and PGC-1ß coactivators (15-fold). When atrophy slowed (day 14), the expression of 92% of these atrogenes returned toward basal levels. At 28 days, the atrophy-inducing transcription factor, FoxO1, was still induced and may be important in maintaining the "atrophied" state. Thus, 1) the atrophy associated with systemic catabolic states and following disuse involves similar transcriptional adaptations; and 2) disuse atrophy proceeds through multiple phases corresponding to rapidly atrophying and atrophied muscles that involve distinct transcriptional patterns.—Sacheck, J. M., Hyatt, J-P. K., Raffaello, A., Jagoe, R. T., Roy, R. R., Edgerton, V. R., Lecker, S. H., Goldberg, A. L. Rapid disuse and denervation atrophy involve transcriptional changes similar to those of muscle wasting during systemic diseases.
Key Words: atrogene • PGC-1 • ubiquitin • proteasome • muscle wasting • inactivity • muscle disease
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. It has been proposed that these various types of atrophy occur through common cellular mechanisms involving similar biochemical and transcriptional adaptations (3
4
5)
. However, there have been no systematic investigations of the extent to which transcriptional changes are similar in the muscle atrophy that results from inactivity with or without denervation and the atrophy occurring in various catabolic states.
Early studies established that protein degradation increases during denervation atrophy, especially the breakdown of myofibrillar proteins, the primary cause of the loss of fiber mass (6
, 7)
. Similar changes in proteolysis have been demonstrated upon unloading (8
9
10
11)
, or inactivation (24)
of innervated muscles. In rodent muscle, most proteins, including the long-lived myofibrillar components, are degraded by the ATP-dependent process requiring ubiquitin and the proteasome (12)
. In various types of atrophy, this process appears to be generally activated since the accelerated proteolysis is sensitive to proteasome inhibitors and requires ATP (13)
. In addition, the atrophying tissues show an elevated content of mRNA for ubiquitin and several proteasome subunits (14)
, an accumulation of ubiquitin-conjugated proteins (15)
, and enhanced rates of ubiquitin conjugation (12
, 16)
.
The strongest evidence for a specific transcriptional program for muscle atrophy has come from our use of cDNA microarrays to identify a common set of changes in the mRNA content of muscles from fasted mice and from rats with cancer cachexia, streptozotocin-induced diabetes mellitus, and uremia (4
, 5)
. These transcriptional changes were found under conditions where the muscles were undergoing rapid atrophy and had high rates of protein degradation (4
, 5)
. Approximately 120 genes were shown to be coordinately induced or suppressed in the muscles in these different catabolic states and were termed "atrogenes" (4
, 5)
. Among the most strongly induced atrogenes were components of the ubiquitin-proteasome pathway including genes for polyubiquitin and ubiquitin-fusion proteins, multiple subunits of the 20S proteasome, and especially the ubiquitin ligases (E3s), atrogin-1/MAFbx and MuRF1 (3
, 17)
. Atrogin-1 and MuRF1 mRNA are induced early during the atrophy process, and upon fasting, the rise in atrogin-1 expression precedes the loss of muscle weight (3)
. In cultured cells, the level of atrogin-1 mRNA is tightly correlated with increases in protein breakdown (18)
. Atrogin-1 and MuRF1 are both also induced in muscle in other catabolic states such as cardiac failure (19)
and sepsis (20)
, and in knockout mice lacking either of these E3s, the rate of denervation atrophy is reduced (17)
. During systemic atrophy, there are characteristic changes in mRNAs for a number of genes involved in energy production and formation of extracellular matrix (ECM), but surprisingly little or no change in mRNAs for contractile proteins.
A primary goal of the present study was to determine whether this same set of transcriptional adaptations responds similarly during muscle atrophy when contractile activity is reduced, such as after nerve injury, section of the motor nerve (denervation), spinal cord injury, immobilization, or unloading. Denervation not only prevents contractile function but also leads to marked changes in muscle membrane properties, resulting in such effects as fibrillation or supersensitivity to acetylcholine (Ach) (21)
. These results appear to be due to transcriptional changes that have often been attributed to the loss of neurotrophic factors, but the possible effects of inactivity on membrane properties have not been investigated systematically. In fact, there have been few studies of pure disuse atrophy, although several groups have reported transcriptional changes after hind limb unloading (a form of disuse in which local reflexes still function) (11
, 22)
. Roy and colleagues (23)
have described a spinal cord isolation model of near-complete disuse, where supraspinal, infraspinal, and peripheral inputs are surgically eliminated while maintaining neuromuscular connectivity. In effect, all activity-dependent influences on the motor neurons and associated muscles are eliminated whereas the activity-independent influences are preserved. This procedure leaves the motor neuron-muscle connectivity intact, but results in electrical silence in the neuromuscular unit and a rapid loss of muscle weight within several days. This rapid atrophy is followed by a pronounced slowing of weight loss (in rats by 2 wk) after operation (23
24
25)
. Atrogin-1 mRNA expression increases during the first 4 days after spinal cord isolation (26)
, as it does after denervation. The resulting weight loss after spinal cord isolation, however, is less extensive than that seen after nerve section, especially with longer times after denervation, for reasons that are unclear (27)
. These different time courses have enabled us to investigate whether the transcriptional adaptations in the disused muscle correlate with the rate of atrophy or are similar in all inactive muscles.
The present study of denervated muscle and normally innervated, but inactive, muscle were undertaken to determine whether these processes involve the same signaling pathways implicated in systemic wasting of muscle, most of which require glucocorticoids (28)
and involve reduced levels of insulin or insulin-like growth factor 1 (IGF-1), resistance to these hormones, or increased TNF- (29
, 30)
. By contrast, disuse atrophy involves a simple loss of contractile activity in a specific muscle without any clear requirement for these circulating systemic factors. While some studies have demonstrated increased numbers of glucocorticoid receptors (31)
and even circulating glucocorticoid levels (32)
in muscle disuse, inhibition of glucocorticoid action or adrenalectomy does not prevent development of disuse atrophy (32)
. A potential unifying factor between disuse and these systemic catabolic states is IGF-1, which besides being a circulating growth factor, is produced locally in muscle with contractile activity (33)
and alters the muscle’s sensitivity to insulin (34)
. IGF-1 and insulin act through the PI3K/Akt pathway to inactivate forkhead (FoxO) transcription factors (35
, 36)
and reduce the expression of several atrogenes in cultured myotubes and adult muscle (18
, 19
, 35
, 37)
. Although activation of FoxO transcription factors is critical in the induction of atrogin-1 and atrophy (35)
, activation of NF-
B signaling pathways appears to play a role in disuse atrophy (38)
and can cause MuRF1-dependent atrophy without induction of atrogin-1 (39)
.
This study addresses the following fundamental questions: 1) Do the genes identified as atrogenes during systemic muscle wasting in fact comprise a general "atrophy program" that is also activated by eliminating contractile activity and/or neurotrophic factors? 2) How does the expression of the critical ubiquitin ligases, atrogin-1 and MuRF1, and other atrogenes change as the extent of atrophy proceeds? 3) Do these changes in mRNA content correlate with the rate of weight loss? 4) Is there is a distinct set of transcriptional changes associated with inactive muscles? These present findings should help identify the gene products, and eventually the regulatory systems and transcription factors that are important in onset and progression of the atrophy process and in the maintenance of the "atrophied" state. These adaptations may serve as useful biomarkers for monitoring the progression of this debilitating loss of muscle mass and eventually in its treatment.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
200–235 g) were examined at 1, 3, 14, or 28 days after nerve section or spinal cord isolation and in age-matched control rats. Surgical procedures and animal care were described previously (25)
. Briefly, hind-limb muscles were denervated by removing a 5–10 mm section of the sciatic nerve in the thigh region of hind limbs and denervated muscles compared with those of control rats. For the rat denervation and spinal isolation experiments, control rats were not sham-operated because extensive pilot studies in this model have shown the muscle weights of sham-operated rats to be similar to those of control rats. Denervation was performed bilaterally to better mimic the bilateral effects of spinal isolation. An identical surgical procedure was used for a denervation experiment performed on 3-month-old CD1 mice, except that a smaller section of the sciatic nerve was excised (2–3 mm) and the contralateral nondenervated hind limb served as an internal control in subsequent analyses, which were carried out 1, 3, 7, or 14 days after the operation. This procedure does not affect the animal’s ability to ambulate, and food intake is internally controlled since the nondenervated limb from the same animal is used as the control muscle.
The spinal isolation procedures are a modification of the original protocols of Tower (40)
and have been described in detail before (23
, 41)
. Briefly, a longitudinal midline skin incision was made over the spinal column from the T6 to the S1 vertebral levels and a partial laminectomy was performed between vertebral levels T7 and S1 with section of the dorsal roots bilaterally at each level. The spinal cord was then completely transected at midthoracic and high sacral spinal cord levels. The rats were allowed to fully recover from anesthesia in an incubator and were given lactated Ringers solution for volume expansion. PolyFlex (G.C. Hanford Manufacturing Co., Syracuse, NY, USA), a broad-spectrum antibiotic, was administered during the first 3 days of recovery. Postsurgical care involved ensuring bladder emptying on a twice-daily basis, the hind limbs were manipulated passively through a full range of movement to maintain joint flexibility, and reflexes in the hind limbs were continually assessed (i.e., withdrawal reflex and toe spread response). Throughout the study there was no response to reflex testing or toe pinching, and the hind limbs remained completely flaccid. Rats were supplied rat chow and water ad libitum. The experimental rats received proper nutrition and hydration since the body weights of the rats begin to increase in parallel to that of age-matched control rats after the immediate postoperative period. While there is a short period of "spinal cord shock" after the procedure, it is unlikely that cord shock (or the general effects of surgical trauma) causes or contributes significantly to the muscle atrophy, which closely resembles that seen upon denervation following section of the sciatic nerve, which causes disuse but no demonstration systemic trauma. In classical spinal shock, one cannot easily activate the neural circuits. In this model of spinal isolation, axon potentials can be readily induced via the motor axons, and nearly normal specific tension can be generated in muscles (and motor units) after weeks or months of spinal isolation.
At each experimental time point, the medial gastrocnemius muscle was dissected from both hind limbs of each animal, quick frozen in melting isopentane cooled in liquid nitrogen, and stored at –70°C until RNA extraction. All animal experiments were approved by institutional review boards. Total RNA was prepared using TRIZOL® according to the manufacturer’s instructions. Equal amounts of total RNA from each animal in each experimental group were combined to form a pool of RNA that was then used to perform Northern blots, real-time polymerase chain reaction (PCR) studies, or microarrays.
RNA content
Total RNA was extracted from the muscles with phenol/chloroform as described previously (3)
. The purity of the extracted RNA was confirmed by determining the ratio of optical density (OD) at 260 and 280 nm (ratios were between 1.8 and 2.0). The concentration of total RNA was estimated by multiplying OD at 260 nm by 40. Total RNA content of the medial gastrocnemius was calculated by taking the RNA concentration of each sample divided by the weight of the muscle and multiplying it by the relative amount of atrophy of the muscle compared with the control animals.
Northern blot analysis
For Northern blot analysis, RNA was electrophoresed on 1% formaldehyde-agarose gels, transferred to a Zeta-probe membrane (Bio-Rad, Hercules, CA, USA), and UV cross-linked. Atrogin-1 and MuRF1 probes were prepared as described previously (3)
. To generate FoxO1 and lipin cDNAs, primers for the full-length mouse FoxO1 (NM_019739) and lipin (NM_015763) genes were purchased from Sigma Genosys (The Woodlands, TX, USA). A plasmid-based mouse skeletal muscle cDNA library was used as the PCR template as described previously (3)
. PCR products were labeled by random priming (Ambion, Austin, TX, USA, and Stratagene, La Jolla, CA, USA). Hybridization was performed by the method of Church and Gilbert at 65°C overnight (42)
. Hybridized membranes were analyzed by using a Fuji PhosphorImager with QuantityOne® software (Bio-Rad). The same blot was stripped and rehybridized with each probe, and finally with a mouse GAPDH probe (Ambion) to ensure equivalent gel loading.
Real-time PCR
To measure more precisely the time course of atrogin-1 and MuRF1 gene expression after denervation and spinal cord isolation, we used real-time PCR. Total RNA was subjected to DNase digestion using an RNeasy kit (QIAGEN Inc., Chatsworth, CA, USA), and the amount and purity of the RNA were reassessed by measuring OD. Total RNA (0.5 µg) was reverse-transcribed using a cloned murine leukemia virus reverse transcriptase and random hexamers in a 20 µl reaction, according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA).
Primers for atrogin-1 and MuRF1 were designed using the Primer ExpressTM 1.5 Program (Applied Biosystems) and produced in an automated synthesizer according to the manufacturer’s protocol. Sequences of the forward primers and reverse primers are as follows: rat atrogin-1 forward: 5'-aga aaa gcg gca cct tcg t-3'; reverse: 5'-ctt ggc tgc aac atc gta gtt c-3', rat MuRF1 forward: 5'-gag aac ctg gag aag cag ctc at-3'; reverse: 5'-ccg cgg ttg gtc cag tag-3', primers for PGC-1 and PGC-1ß were chosen using PRIMER 3 software: mouse PGC1 forward: 5'-cgc tgc tct tga gaa tgg at-3'; reverse: 5'-cgc aag ctt ctc tga gct tc-3', mouse PGC-1ß forward: 5'-aga agc gct ttg agg tgt tc-3'; reverse: 5'-cca tag ctc agg tgg aag ga-3'. To normalize target cDNA values, GAPDH cDNA levels were quantified using commercially available rodent GAPDH primers (Applied Biosystems).
SYBR Green® RT-PCR was performed on the LightCycler instrument (Roche Diagnostics, Nutley, NJ, USA) using the following cycle parameters: denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 10 s and 55°C for 5 s for annealing and 72°C for 12 s for extension. For each gene, first-strand cDNAs were amplified utilizing the FastStart DNA Master SYBR Green kit and protocol (Roche). Briefly each 20 µl reaction contained 2 µl FastStart Reaction Mix with enzyme, 3 mM MgCl2, target cDNA-specific primers (5 µmol final concentration), and 4 ng of first-strand cDNA sample. Each sample was run in duplicate, and all runs included PCR grade water as a negative control. Quantification of the mRNA for atrogin-1, MuRF1, or GAPDH was performed relative to a standard curve of the cDNA of that gene, which also served as the positive control.
For PGC-1 and -1ß analyses, total RNA from each mouse muscle was pooled at each time point, and was reverse transcribed with SuperScript II and oligo-dT. Diluted cDNA was amplified in 10 µl PCR reactions in a GeneAmp 9600 thermocycler, coupled with a GeneAmp 5700 Sequence Detection System (Applied Biosystems). For each time point, we amplified multiple serial dilutions of the cDNA input. For mouse PGC-1 real-time PCR analyses, SYBR green chemistry was used as described previously (43)
. Differences in gene expression were evaluated by a relative quantification method, as described by Pfaffl et al. (44)
. Values were normalized to the expression of ß2 microglobulin, an internal reference whose abundance did not change under our experimental conditions. Normalized ratios were converted to a logarithmic scale, and SD were calculated as described by Marino et al. (45)
.
Gene microarray analyses
We performed human cDNA microarrays on a pool of equivalent quantities of total RNA prepared from 5–7 individual rat medial gastrocnemius muscles and age-matched controls at both 3 and 14 days after denervation and spinal cord isolation. This approach implies that we are unable to describe the variation in gene expression within animals in each group. Using this method, expression levels for each gene for individual animals within the group are averaged. The pooled sample thus provides a robust method of obtaining the group mean gene expression and simplifies comparison with other experimental groups. To compare these data with those obtained in various types of atrophy, the same human cDNA chips were used as in earlier studies from this lab (4
, 5)
. The mouse cDNA chips studied previously were not used in the present analysis due to changes in the genes represented and their annotations in the currently available mouse cDNA arrays. However, because forkhead box O1 was not on the human chips, we have incorporated the results for FoxO1 due to its high relevance to this work. Our results therefore refer to the 9172 analyzable sequences on the human cDNA microarrays plus the FoxO1 sequence from the mouse cDNA microarrays.
Microarray hybridization and analysis
RNA extraction from muscles, isolation of poly(A)+-containing RNA, and microarray hybridizations (Incyte cDNA chips, St. Louis, MO, USA) were performed as described previously (4)
. Reverse transcription with amino-allyl dUTP and coupling to NHS-cyanine dye for the cDNA microarrays was performed by a standard protocol developed by the Harvard Center for Genomics Research (Cambridge, MA, USA). Experiments were run as follows: at least two hybridizations per time point of denervation and control or spinal cord isolation and control samples to human cDNA chips (UniGene 1.44, 1.46). Replicates were run on reversed flourophore platforms.
Microarray data analysis was performed using GEMTools software (Version 2.4.1, Incyte, Inc.), MicrosoftExcel, Microsoft Access, and Rosetta Resolver (Rosetta Inpharmatics, Seattle, WA, USA). After quality control tests from Incyte were applied to the data, data were imported into the Resolver software for further analysis (5)
using a specific Rosetta error model generated for Incyte microarrays (46)
. Results for duplicate or triplicate hybridizations were combined in Resolver to obtain a single average fold change and significance value since Resolver allows the combining of repeated experiments to yield single fold change and significance values for data points common to multiple microarrays. Initially, denervation and spinal cord isolation results from 3 days postintervention were combined to identify those mRNAs showing significant changes common to disused muscles (P<0.05). We compared these data with those from systemic atrophy states induced by fasting or disease (fasting: 2-day food-deprived mice; tumor implantation: rats bearing Yoshida hepatoma for 6 days; uremia: 7/8 nephrectomized rats (24 days); and diabetes: 3 days streptozotocin-treated rats) (5)
. All muscles were examined at a time when they were undergoing rapid atrophy (5)
. Please visit http://agoldberg.med.harvard.edu/muscledatabase to view the individual microarray outputs and the supplemental tables.
Other statistical analyses
Differences in muscle weight over time and between interventions were compared by ANOVA and Tukey’s post hoc analysis. Statistical significance was established at P < 0.05. This analysis was not performed for RNA content or mRNA expression levels, as the samples were internally controlled through pooling of individual samples from animals of the same intervention and time point.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In both cases, two distinct phases of muscle weight loss could be distinguished. A rapid weight loss occurred over the first 14 days, followed by a slower weight loss over the next 2 wk (Fig. 1
A). When these changes in muscle weight were corrected for the increases in body weight that occurred in the control rats, denervation and spinal cord isolation led to a 12% loss of muscle weight at 3 days (Fig. 1B
). Subsequently, weight loss continued in both models of atrophy, albeit more slowly in the case of pure disuse. By 14 days, the weight loss was 25% greater after nerve section than after spinal cord isolation. After 14 days of spinal cord isolation, there was almost no further weight loss whereas denervated muscle continued to atrophy through 28 days, when its weight was only 25% of control muscle. To better visualize the rates of weight loss in these muscles at different times, the approximated mean daily weight loss (average %/day) between surgery and day 3, days 3 and 14, and days 14 and 28 were calculated from Fig. 1B, C
. This analysis clearly demonstrates that after denervation and spinal cord isolation, there is a phase of rapid weight loss followed by a second, slower phase, then weight loss virtually ceases.
|
Although muscle weight had decreased significantly by 3 days of disuse, RNA concentration and content were unaffected (Fig. 2
), unlike in certain systemic types of atrophy (47
48
49
50)
. In our experiments RNA concentration (mg RNA/g muscle) remained unchanged 3 days after spinal cord isolation and, surprisingly, increased by 78% at 14 days and 96% at 28 days (Fig. 2A
). After denervation, RNA concentration rose even higher and was 4- to 6-fold control levels after 14 days (Fig. 2A
). This increase in RNA concentration was due in part to the differential loss of muscle protein. When the total amount of RNA in the medial gastrocnemius was calculated after spinal cord isolation (RNA concentration multiplied by the relative amount of atrophy of the muscle compared with the control animals), RNA content was slightly decreased at 3 days, then rose by 25% by 14 days (Fig. 2B
). In the denervated muscle, there was a net increase in total RNA content at 14 days, and this value remained elevated thereafter. Thus, the rapid loss of muscle mass after denervation and spinal cord isolation does not result in decreases in gastrocnemius RNA content, as has been observed in other catabolic states. This increased RNA content may reflect increased ribosomal content and higher rates of protein synthesis in these muscles, but we did not find a consistent increase in translational apparatus in our array analyses (see below).
|
Atrogin-1 and MuRF1
We next investigated how weight loss during disuse atrophy correlates with gene expression. Atrogin-1 and MuRF1, which have been shown using knockout animals to be essential for rapid atrophy (17)
, are among the mRNAs induced most dramatically in the atrophying muscles. Using Northern blot and RT-PCR techniques, we found that atrogin-1 and MuRF1 mRNAs rose by 1 day of disuse, reached a peak at 3 days, then fell toward basal levels at 14 and 28 days (Fig. 3
A). The peak expression at 3 days for these Ub-ligases was nearly 50-fold for atrogin-1 and 20-fold for MuRF1 (Fig. 3A, B
). Others have reported similar, although less dramatic, changes after 3 days of denervation (17)
. In the present study, atrogin-1 mRNA decreased after 14 days of denervation but was still significantly elevated above basal levels (4.2-fold). MuRF1 mRNA was expressed at low levels after spinal cord isolation at 14 days (2.6-fold) and at 28 days (2.8-fold), but was above basal levels 14 days (9.5-fold) after denervation. These changes do not appear to be fiber type-specific since similar changes in these E3s were observed in the tibialis anterior, a predominantly fast flexor, and soleus, a predominantly slow extensor (data not shown). Maximal increases in the mRNA of these critical E3s coincided with the greatest rates of muscle weight loss. Bimodal changes in the expression of these E3s provide further strong support for the existence of distinct early rapid and later slow phases of the atrophy process.
|
PGC-1 and PGC-1ß
We investigated whether these dramatic changes in muscle size and atrogin-1 and MuRF1 mRNA levels correlate with expression of the PGC-1 family of the transcriptional coactivators (PGC-1 and PGC-1ß), which are important for mitochondrial production and oxidative metabolism. In skeletal muscle, PGC-1 functions as a metabolic sensor that reflects the level of contractile activity (51)
and is strongly induced with exercise (52
53
54)
. Initially, PGC-1 levels were high in the medial gastrocnemius (although not as high as in the dark soleus), but by 3 days after denervation or spinal cord isolation in the rat, the amount of PGC-1 mRNA measured by Northern blot analysis and real-time PCR was much lower than in control muscle—so low, in fact, it could not be accurately quantitated (data not shown). To further study the fall in expression of this coactivator, we analyzed the mRNAs for PGC-1 and PGC-1ß by real-time PCR in mouse muscles at shorter times after denervation (Fig. 4
). PGC-1 and PGC-1ß mRNAs fell by 80% within 1 day and were suppressed to less than 10% of control levels at 3 days postdenervation. This very large (>15-fold) reduction in PGC-1 mRNA in mouse muscles thus confirms the dramatic disease in its content in rat tissues after denervation or spinal isolation. PGC-1 mRNA then rose slightly but remained at 20% of wild-type levels through 14 days, while mRNA for PGC-1ß remained very low for 7 days before rising slightly at 14 days. This is the first report that PGC-1ß mRNA levels, like PGC-1 content (15)
, reflect muscle activity. Thus, the dramatic fall in PGC-1 and PGC-1ß early after nerve section or spinal isolation may help signal the other transcriptional changes in these types of atrophy.
|
Transcriptional profiling of atrogenes using cDNA microarrays
Of the 9000 genes present on the cDNA arrays used in this study, 8–10% were differentially expressed after 3 days of denervation or spinal cord isolation (720/8897=8% and 840/8839=9.5% for denervation and spinal isolation, respectively; Table 1
), when rates of protein loss, and expression of atrogin-1 and MuRF1 were greatest. Nearly 70% of these differentially expressed mRNAs were up-regulated. 517 (6%) were differentially expressed in a similar manner in denervation and spinal cord isolation (i.e., "shared muscle disuse genes," Table 1
). In earlier studies we had identified a set of 63 atrogenes (5)
present on similar cDNA microarrays that were consistently induced or suppressed in atrophying muscles from food-deprived (4)
, diabetic (55
, 56)
, uremic (57)
, and tumor-bearing cachectic rodents (47)
. Remarkably, 49 of 63 of these atrogenes (78%) were differentially expressed at 3 days of muscle disuse and have been termed common atrogenes.
|
Atrogenes common to both disuse and systemic atrophy
Many of the atrogenes encode proteins involved in protein breakdown, mainly by the Ub-proteasome system. Among the induced genes comprise several but not all components of the 26S proteasome, including subunits of the 19S complex (Rpn12 and Rpn6) and of the 20S core particle (PSMA1, PSMB3, and PSMB4), proteins that encode polyubiquitin (UBB, UBC), ubiquitin-fusion proteins (RPS27A and UBA52), and the ubiquitin-protein ligase, atrogin-1 (Fig. 5
A). Induction of these genes probably contributes to the increased protein degradation seen in these atrophying muscles. The lysosomal protease, cathepsin L, but not other lysosomal hydrolases, was highly induced, possibly denoting a role for this organelle in the atrophy process.
|
Stimulation of the Akt pathway in muscle causes hypertrophy, and suppression of this pathway appears to be critical in many forms of atrophy. Akt phosphorylates and suppresses activation of the FoxO transcription factors (35
, 36)
, which become active in atrophying muscle. FoxO1 mRNA is up-regulated after denervation or spinal cord isolation as it is in the systemic wasting states (Fig. 5C
), suggesting a fundamental role of FoxO1 in all types of atrophy.
The majority of down-regulated atrogenes were for proteins involved in glucose metabolism (lactate dehydrogenase, phosphoglycerate mutase 1 and 2) and mitochondrial ATP synthesis (ATP synthase, malate dehydrogenase, NADH dehydrogenases) (Fig. 5B
). There was also a decrease in mRNAs for ECM proteins, such as collagen I and V and fibronectin (Fig. 5D
), which may be related to the loss of protein that occurs with atrophy. Denervation did not result in the same changes in RNA expression of some of the extracellular matrix genes as did spinal cord isolation. For example, upon denervation, increased expression of mRNAs for galectin and fibronectin was noted (Fig. 5D
), but the expression of these genes is strongly suppressed in spinal cord isolation and other forms of systemic atrophy. The galectins are a family of ß-galactoside binding proteins implicated in modulating cell-cell and cell-matrix interactions. LGALS1 may act as an autocrine negative growth factor that regulates cell proliferation (58)
. Galectin is also up-regulated in aging in mice and has been proposed to protect against cellular damage and destruction (59)
.
During the rapid phase of muscle atrophy, there were decreases in expression of proteins associated with cell growth and differentiation, such as the transcription factor JUNB and P311 (Fig. 5C, E
). JUNB is a growth-promoting transcription factor that is down-regulated in all systemic atrophy states (5)
. The most suppressed mRNA, P311, is also involved in smooth muscle differentiation (60)
and has been reported by others to be down-regulated in skeletal muscle after denervation (61)
(Fig. 5E
). Several up-regulated atrogenes such as EIF4EBP1 inhibit translation initiation factors and could indicate a reduction in translation in muscle (Fig. 5C
). Additional genes involved in cap-independent translation (EIF4EBP1, EIF4A2, EIF4G3; Fig. 5C
) are generally increased in all atrophy conditions. Cap-independent translation enhances translation of a subgroup of mRNAs with internal ribosome entry sites, which tend to be important in stressed cells. 5'TG3'-interacting factor (TGIF) is induced in all types of atrophy, and is another transcriptional target of TGF-ß that may act as a corepressor that associates directly with Smad proteins and inhibits Smad-mediated transcriptional activation (62)
.
Several of the atrogenes found to change in muscle in catabolic states had not previously been implicated in muscle wasting, and some are of unclear physiological significance (5)
. One of these, metallothionein, binds several heavy metals (zinc, copper, and cadmium) but may also function as an antioxidant (63)
, and is induced by oxidative stress and heavy metals. Metallothionein-1L and B were the most dramatically induced (up to 68-fold after 3 days of denervation and up to 30-fold after 3 days of SI; Fig. 5E
). Induction of metallothionein has been noted in various disease states, although its exact physiological function is uncertain (64)
. Metallothionein is highly induced in cultured myotubes after treatment with dexamethasone, and its expression is suppressed by IGF-1 (18
, 37)
. In fact, in cultured myotubes its induction far exceeds that of atrogin-1, and rises sooner (J. M. Sacheck, unpublished observations). The transcription factor ATF4, which promotes the expression of oxidative stress responsive genes, is induced in all atrophy states (65)
(Fig. 5C
). Together, these changes suggest a role for activation of an oxidative stress response in atrophying muscles, as has been suggested by others (66
, 67)
.
Atrogene program is reversed when the rate of muscle loss decreases
To investigate whether changes in the expression of these 49 atrogenes is related to the rate of muscle weight loss, we performed additional cDNA microarrays at 14 days, when the rate of atrophy had slowed markedly (Fig. 1C
) and atrogin-1 and MuRF1 mRNAs had returned to near basal levels (Fig. 3)
. This approach was used to differentiate genes important to the development of atrophy (3 days) and maintenance of the atrophied state (14 D). Even though total RNA content was increasing at 14 days, 92% of the atrogenes were no longer differentially expressed to a significant degree 14 days after spinal cord isolation (Fig. 6
A, B). The majority of these mRNAs (57%) returned toward basal levels by 14 days of denervation (Fig. 6A
), including all of the genes encoding proteins involved in protein breakdown and translation initiation. A subgroup of atrogenes remained differentially expressed after 14 days of denervation, although they returned toward basal levels after 14 days of spinal cord isolation (Fig. 6B
). Many of these genes are involved in glycolysis and ATP synthesis (e.g., ATP synthase, phosphoglycerate mutase 2, lactate dehydrogenase). This difference in expression patterns between denervation and spinal cord isolation at 14 days (57% vs. 92% of mRNAs returning to basal levels) correlates with the rate of muscle weight loss, which is still significant 14 days after nerve section (Fig. 1C
), but not spinal cord isolation. These findings indicate that during disuse atrophy, the pattern of mRNA changes correlate with muscle wasting and are generally not maintained in the atrophied tissue. This pattern of atrogene expression is further evidence for a function of these genes in the loss of tissue mass.
|
Although the majority of atrogenes change in one direction during the initial 3 days of muscle disuse, then reverse their direction of expression after 14 days of disuse, five atrogenes remained differentially expressed after 14 days of denervation or spinal cord isolation (Fig. 7
A). These genes included metallothionein-1L and -IB, the most highly up-regulated of the atrogenes, and FoxO1 (35
, 36)
. As shown in Fig. 7B
, FoxO1 expression (assayed by Northern blot) persists at high levels even 28 days after surgery. Thus, in addition to playing a role in initiating muscle atrophy, in part through catalyzing the transcription of atrogin-1, FoxO1 may also play a role in the maintenance of muscle in an atrophied state.
|
Genes unique to systemic or disuse atrophy
Only 6 of the 63 originally identified atrogenes on the human cDNA arrays are unique to systemic atrophy states (Fig. 8
). One of these genes, lipin, is involved in adipogenesis (68)
, and mutations in it result in lipodystrophy, high blood triglycerides levels, and insulin resistance. The induction of lipin during systemic wasting therefore may be related to the decreased insulin sensitivity and changes in energy metabolism in these catabolic states. Further studies are required to understand what specific role activation of these genes has in the muscle wasting associated with these cachetic states.
|
Of the 279 genes differentially expressed after 3 days of denervation and spinal cord isolation but not in the systemic atrophy conditions, 75% were up-regulated. Figure 9
lists the 30 genes that changed most significantly (P<0.001 with a fold change>2.0; see supplemental table with all 279 genes). Most notably, several of these genes are components of the neuromuscular junction or are membrane-associated proteins (-cholinergic receptor, neural cell adhesion molecule (CAM) 1, dysferlin). Many others, at least in the denervation model, may be related to denervation supersensitivity, in which denervated muscle fibers become supersensitive to Ach, or fibrillation due to insertion of newly synthesized Ach receptors (21)
. Sarcolipins, inhibitors of the sarcoplasmic reticulum Ca2+-ATPase (69)
, are induced along with dibasic and neutral amino acid transporters (SLC3A1), which are involved in glutamine synthesis and export. Genes encoding proteins for mitochondrial ATP synthesis, such as L-3-hydroxyacyl-coenzyme A dehydrogenase (COX7B), are suppressed. The most suppressed gene in this group was kinesin family member 21B, a microtubule-dependent motor protein involved in the transport of essential cellular components along axonal and dendritic microtubules in neurons (70)
.
|
Some of the most up-regulated genes, such as the -cholinergic receptor and the transcription factor myogenin, are modulated by electrical activity in the membrane (71)
, and myogenin has been identified as being highly induced during atrophy, resulting from denervation and spinal cord isolation (25)
. Meanwhile, denervation disrupts the normal neuromuscular transmission and is associated with a variety of neuromuscular adaptations (such as alterations in expression and distribution of Ach receptor subunits and molecules associated with neuromuscular junction that seem to be unrelated to the decrease in fiber size). Thus, not surprisingly, genes identified as unique to disuse are rich in functions at the neuromuscular junction and in excitation-contraction coupling. A more thorough analysis of these disuse-induced genes will be examined in subsequent papers.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 5)
, and the key role of FoxO transcription factors in muscle wasting (35
, 36)
, strongly support the idea that the loss of muscle proteins involves a specific transcriptional program. This study extends the identification of a set of atrogenes to include ones induced and suppressed during atrophy induced by disuse of specific muscles. Of the atrogenes identified in systemic catabolic states, the great majority (78%) changed in a similar manner in the two experimental approaches to prevent voluntary muscle activity, section of the motor neuron, and spinal cord isolation. This finding is remarkable considering that the various forms of muscle wasting are activated in quite different ways. For instance, in fasting and disease, circulating factors such as TNF-, low insulin, and glucocorticoids trigger muscle wasting (28
29
30)
whereas lack of motor neuron function, presumably by preventing contractile activity, signals the atrophy associated with disuse. Identification of these common atrogenes lends strong support for a final common biochemical program leading to loss of fiber protein regardless of the physiological or pathological etiology and signaling mechanism. Furthermore, the existence of this very general atrophy-specific transcriptional program should be a powerful tool for detecting the earliest regulatory responses signaling these changes in the atrophying muscles.
Among the identified atrogenes are multiple components of the ubiquitin-proteasome pathway, including the two muscle-specific ubiquitin-protein ligases, atrogin-1 and MuRF1. Induction of these factors is not surprising since stimulation of protein degradation, mainly by the ubiquitin-proteasome pathway, accounts for the majority of protein loss in denervation (6
, 7)
. Others studying atrophying muscles after hind limb unloading and denervation have found an induction of atrogin-1, MuRF1, and cathepsin L mRNAs using microarray analyses (11
, 17)
. However, the early and dramatic increases as reported here (a nearly 40-fold induction of atrogin-1 mRNA and 20-fold increase in MuRF1 mRNA) have not been observed previously. In addition, in all these forms of muscle atrophy as well as hind limb unloading (11)
, cathepsin L was induced, which supports earlier findings that lysosomal proteolysis is activated upon denervation atrophy (7)
. However, the differential expression of a single lysosomal protease and not others remains surprising.
It is noteworthy that the largest changes in these atrogenes were seen early (e.g., 3 days), when rates of muscle weight loss were the highest, then fell as the rate of atrophy slowed (see below). This correlation of the levels of atrogene expression with the rate of atrophy was not noted before, perhaps because prior studies focused on acute conditions that are lethal if prolonged (e.g., fasting, cancer cachexia). The major general features of the atrophy-related transcriptional program identified in the acute systemic disease models (4
, 5)
are apparent in these disuse models as well. Expression of a number of genes for energy production and for overall protein synthesis (i.e., translational initiation) was suppressed, as well as components of the ECM. Finally, in all these atrophying muscles there are suggestions of oxidative stress, since all showed increased expression of ATF4, and thioredoxin-like protein as well as metallothionein, which has been proposed to help protect against oxidative stress.
Mechanisms regulating atrogene expression: suppression of PGC-1 and activation of forkhead factors
The dramatic fiber atrophy upon disuse is associated with changes in contractile properties and myosin heavy chain expression, as well as a shift toward a more glycolytic pattern of metabolism (23
, 27
, 72)
. These adaptations correlate with a preferential atrophy of oxidative, slow-twitch fibers, but could also be due to a transition from slow to fast fiber phenotypes (23
, 72)
. On the other hand, during the systemic wasting states, fast-twitch glycolytic fibers atrophy preferentially. The sizes of these fast-twitch fibers are more sensitive to glucocorticoids (73
, 74)
, and an elevation in these adrenal hormones is required for most systemic types of muscle atrophy (28)
. Thus, it is surprising that the transcriptional changes in disuse atrophy and these catabolic states are both quite similar, especially the general reduction in mRNAs for enzymes associated with glycolysis and mitochondrial ATP production. It is unclear whether expression of these atrogenes is suppressed similarly in both oxidative type I and glycolytic type II fibers, but our data suggest an early shift in disuse atrophy to a more glycolytic phenotype, which then sensitizes the muscles to these transcriptional changes. The dramatic suppression of PGC-1 and ß expression in denervated muscles may be an important mechanism that helps initiate the atrophy process, since the large decreases in expression of these coactivators occurred at the earliest times tested after denervation, and similar effects occur rapidly in various types of systemic atrophy examined (75). Muscles overexpressing PGC-1 show a marked increase in mitochondrial content and develop a more oxidative phenotype (increased type I and IIa fibers) (76)
, and loss of this factor presumably promotes the development of the opposite, glycolytic, phenotype.
The forkhead family of transcription factors have recently been identified as key activators of the atrophy process downstream of AKT in the IGF-1 signaling pathway (35
, 36)
. The increased expression of the FoxO1 gene (Fig. 7)
should complement the dephosphorylation/activation of the FoxO gene products and amplify FoxO-dependent transcription. In the present study, we find that FoxO1 is also induced in denervation and spinal cord isolation, where circulating insulin and IGF-1 levels should not fall, although disuse has been associated with insulin resistance (8)
and exercise in muscle stimulates paracrine production of IGF-1. In addition, shifts to a faster fiber type are associated with increased FoxO1 expression (77)
and are consistent with patterns of gene expression. The induction of FoxO1 gene expression as well as the fall in PGC-1 during all types of muscle wasting is consistent with the findings that the atrophying fibers are generally transitioning toward faster types (e.g., from type I and IIa to IIx or IIb fibers).
Disuse atrophy proceeds through two phases
We chose to study two well-defined disuse models to illustrate the effects of the loss of contractile activity independent of systemic factors, one through the loss of the motor neuron (denervation) and the other solely due to the loss of neural activity (spinal isolation). Denervation not only prevents normal use but also leads to marked changes in the muscle membrane, in part due to loss of neurotrophic factors released by the motor neuron (21)
. Spinal cord isolation maintains motor neuron-muscle connectivity and thus appears to be a useful model of pure inactivity or disuse (23)
. It is noteworthy that the atrophy was more pronounced and prolonged after denervation than after spinal cord isolation. Nevertheless, the patterns of atrogene expression were quite similar in these two models of inactivity, especially during the initial rapid phase of muscle atrophy. This initial rapid weight loss lasting several days was followed by a slower, prolonged atrophy period that continued for more than a month in the denervated muscle, but was shorter after spinal isolation. Accordingly, nearly all the atrogene mRNAs returned to basal levels by 14 days, while some atrogenes maintained their differential expression for a longer period in the denervated muscle. This study uncovered a large number of transcriptional changes in muscle resulting from disuse that were not seen in systemic muscle wasting, especially changes in membrane components and energy metabolism. Presumably, these changes represent important adaptations to disuse but clearly are not essential to rapid fiber atrophy. Aside from the dramatic induction of cholinergic receptors, which has often been reported in denervated muscle (71)
, for most of this group (Fig. 9)
the relationship of these changes in expression to a lack of electrical or contractile activity in the fibers is unclear and an important question for future research. Because the present study has focused specifically on comparisons of atrogene expression during catabolic states and disuse, we plan to explore further the differences in gene expression between these two types of disuse atrophy in a subsequent paper.
The biphasic nature of the expression of the atrogenes illustrates why many of these changes in mRNA and other biochemical correlates of the atrophy process may have been missed in prior studies where the muscles were examined later in the course of atrophy. Presumably, the return of these mRNAs to baseline is followed by the gradual decrease in the levels of these gene products, which accounts for the slowing of atrophy process. It is notable that FoxO1 is one of the few atrogenes that remains significantly increased throughout both rapid and slow phases of atrophy after denervation and spinal cord isolation. Therefore, besides playing a role in initiating muscle atrophy (36)
, FoxO1 may play a role in maintaining the atrophied state; if so, however, it is not acting through expression of the general atrogene program. Whether a similar two-phase atrophy response occurs in prolonged systemic forms of muscle wasting remains to be investigated, but in the acute experimental models studied so far (fasting, cancer cachexia, untreated diabetes), rapid wasting was associated with death of the rodents, and prolonged studies were not possible.
In most types of muscle atrophy, overall rates of protein synthesis are suppressed, rates of protein degradation are elevated, and total RNA content decreases (47
48
49
50)
. Surprisingly, even though others have reported a decrease in RNA content after denervation or disuse in rat tissues (14
, 24)
, we found that with disuse, total RNA content of the muscles does not change at 3 days; RNA concentration and total muscle RNA content steadily increase later, especially after denervation. This increase in RNA content may indicate that protein synthesis is not suppressed in these muscles and that muscle atrophy results from even more dramatic increases in protein degradation. Indeed, Furuno et al. (7)
found that protein synthesis in isolated rat muscles increased by 25% 3 days after denervation, but the rise in protein degradation was much greater at these times (83%). Another possibility that may contribute to the increase in muscle RNA content in the weeks after denervation or pure disuse could include the infiltration of macrophages (78)
or increased satellite cell proliferation and differentiation after denervation (25)
. Consistent with this hypothesis, we found a larger increase in RNA content after denervation than after spinal cord isolation.
In conclusion, the majority of genes identified as atrogenes during muscle atrophy induced systemically by circulating factors in untreated diabetes, tumor implantation, renal failure, or starvation changed similarly during spinal cord isolation or denervation. The time of expression of these genes correlates closely with the time course of muscle weight loss; these genes are highly induced when weight loss is rapid, and their expression returns to control levels as weight loss slows. Thus, we have begun to define which genes play an important role in the initiation of atrophy as well as in the maintenance of the atrophied state. By extension, these studies should allow us to identify those transcriptional changes induced by disuse that are specific adaptations to the loss of contractile activity and not related to the loss of muscle mass.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication May 25, 2006. Accepted for publication August 14, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
