| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,2
Renal Unit, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA;
* Pulmonary and Rehabilitation Research Group, Department of Medicine, University Hospital Aintree, Liverpool, L9 7AL, UK;
Department of Agricultural, Food and Nutritional Sciences, University of Alberta, Edmonton, AB T6G 2M7, Canada;
Renal Division, Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA;
Department of Medicine, University of Texas Medical Branch, Galveston Texas, USA; and
Department of Cell Biology, Harvard Medical School, Boston, Massachusetts, USA
2Correspondence: Department of Biology, Harvard Med. School, Boston, MA 02115, USA. E-mail: Alfred_goldberg{at}hms.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: ubiquitin–proteasome pathway • transcription
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Atrophy of specific muscles results from their disuse or denervation. In most types of muscle atrophy overall rates of protein synthesis are suppressed and rates of protein degradation are consistently elevated; this response accounts for the majority of the rapid loss of muscle protein. In a variety of animal models of human diseases [e.g., fasting (2
, 3)
, diabetes (4)
, cancer cachexia (5
6
7)
, acidosis (8)
, sepsis (9)
, disuse atrophy (10)
, denervation (2)
, and glucocorticoid treatment (11)
], most of the accelerated proteolysis in muscle appears to be due to an activation of the Ub–proteasome pathway (12)
. For example, in these diverse conditions the muscles show a two- to fourfold increase in levels of messenger RNA (mRNA) for polyubiquitin and certain proteasome subunits. A similar induction of components of the Ub–proteasome pathway has been found in atrophying human muscle (13
, 14)
. Whereas weight loss in fasting and diabetes involve reduced levels of insulin and elevated glucocorticoid levels, tumor cachexia and sepsis are often associated with increased TNF
and renal failure with metabolic acidosis. Despite the varied physiological or pathophysiological stimuli for muscle atrophy, earlier studies revealed striking similarities in the transcriptional adaptations of genes encoding certain components of the Ub–proteasome pathway. Therefore, we hypothesized that atrophying muscles exhibit a coordinated series of transcriptional adaptations that constitute a common atrophy program (15
, 16)
.
To test whether a common program of transcriptional adaptations indeed occurs in muscle as it undergoes atrophy and to better understand these catabolic states, we used cDNA microarrays to compare mRNA content of normal muscle with atrophying ones from fasted mice and rats with renal failure, cancer, or diabetes. Transcriptional profiling using microarrays is ideal for defining the atrophy-induced changes in mRNA content, with the understanding that these changes may reflect alteration in mRNA stability as well as gene transcription. Many authors have used this or other genomic techniques to study transcriptional changes in muscle in certain conditions (17
18
19
20
21
22)
, although no comparisons of such profiles from different types of muscle atrophy have been performed. The common set of genes induced and suppressed during atrophy has therefore not been investigated.
As an initial step, we studied the transcriptional changes in muscles of fasted mice (20)
. Besides confirming the increases in mRNA levels shown by Northern blot for polyUb and certain 20S proteasome subunits, the study demonstrated differential expression of other genes with diverse functions (20)
. During these studies we identified and cloned a previously unknown muscle-specific Ub ligase that is dramatically induced in muscle wasting not only in fasting, but also in tumor-induced cachexia, diabetes, chronic renal failure, and dexamethasone treatment (16)
. Simultaneously, Bodine et al. showed this same gene is induced in atrophy by denervation or disuse (19)
. We named this gene atrogin-1, as it was the first new gene identified in our attempts to define the atrophy program. Another muscle-specific Ub ligase, MuRF-1, was shown to be markedly induced upon denervation and disuse (19)
. The present study tested whether its level also rises in atrophying muscles due to fasting or catabolic diseases.
In fasting, protein breakdown in muscle provides the organism with a source of amino acids for gluconeogenesis. We have used the microarray approach (20)
to test whether the response to fasting involves similar adaptations to those in muscles atrophying due to cancer cachexia, uncontrolled diabetes, and chronic renal failure. In each of these animal models, atrophy, especially of fast-twitch muscle fibers, has been extensively documented. Overall rates of protein breakdown as measured in isolated muscles in vitro are increased by 40–65% (Table 1
). mRNA for some components of the Ub–proteasome pathway are elevated by two- to threefold (2
3
4
5
6
7
, 23)
and rates of Ub conjugation in cell-free extracts are enhanced (24
, 25)
. In fasting, some changes found in muscle might be specifically related to the inadequate caloric or nutrient intake. Therefore, it was important to compare changes in mRNA in muscles after tumor implantation and renal failure with muscles from pair-fed animals, since anorexia often accompanies the metabolic disturbances under these conditions. This approach allowed us to identify the transcriptional changes resulting directly from the disease process and to avoid the potential complications of nutrient deprivation.
|
Defining a common transcriptional profile in a range of wasting diseases should increase our understanding of the critical adaptations associated with muscle atrophy independent of the cause of the muscle wasting. Beyond helping us understand these important responses, this analysis may identify therapeutic targets for retarding the atrophy process (26)
. Furthermore, this study lays a basis for comparison with the transcriptional changes that occur in specific muscles in denervation or disuse atrophy. We show here that many genes, which we term atrogins, are differentially expressed in multiple types of atrophy and comprise a common atrophy program.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, diabetes mellitus (4
, 27)
, renal failure (23)
, and tumor implantation (5)
. All animal experiments were approved by institutional review boards.
Microarray hybridization and data analysis
RNA extraction from muscles and performance of micoarray hybridizations by Incyte Inc. (St. Louis, MO, USA) were as described previously (20)
. Analysis was performed using Rosetta Resolver (Rosetta Inpharmatics, Seattle, WA, USA), Microsoft Excel, and the relational database Microsoft Access. Raw data files from hybridizations that passed quality control tests applied by Incyte were loaded into Resolver and analyzed using a specific Rosetta error model generated for Incyte microarrays (28)
. Resolver allows the combining of repeated experiments to yield single fold change and significance values for data points common to multiple microarrays.
General design of experiments
For each experiment (defined below), RNA was extracted from a fresh set of muscles from a minimum of two experimental and control animals. Three hybridization experiments were performed with muscles from fasted mice: two using muscles from diabetic rats, two using muscles from rats with renal failure, and two using muscles from tumor-bearing rats. Each experiment involved hybridizing atrophying and control samples to both human (Human UniGEM 1 or 2) and mouse (Mouse GEM 1) microarrays. We previously showed that hybridizing mouse RNA to human cDNA microarrays yields reliable and useful results. Since the cDNA sequences on the human and mouse arrays represent overlapping but different sets of genes, this strategy allowed us to extend the range of genes analyzed (20)
. The extensive sequence similarity between rodent and human genes permits such cross-species hybridization. In one experiment for each condition, the samples previously labeled with Cy5 were labeled with Cy3 (and vice versa) to compensate for any nonlinearity in the emission signal intensity response curve for each fluorophore. One of the diabetes experiments was hybridized to a mouse microarray and failed to give technically satisfactory results at Incyte; it was not included in our analysis.
Defining atrophy-specific genes or atrogins
Results for each array were combined in Resolver to yield a single average fold change (atrophy/control) for each gene in each type of atrophy. Heatmaps shown in the figures were generated using the natural logarithm of this ratio using Heatmap Viewer 1.0 (Chang Bioscience, San Francisco, CA, USA). Increased expression is indicated by an increasing intensity of red and decreased expression by an increasing intensity of green. When results within each atrophy state were combined, Resolver calculated the probability of differential expression for each gene. We define those genes with an average P value of <0.05 in all four types of muscle atrophy as atrogins and other genes as shared (P<0.05 in two or three states only), disease-specific (P<0.005 in one state but P>0.2 in the other three conditions), or unchanged (P>0.05). Cutoffs adopted for the disease-specific genes are arbitrary but were designed to minimize the false-positive rate (0.5%) for the catabolic state in question while controlling the false-negative rate in the three other catabolic states (i.e., to minimize the probability that a gene identified as differentially expressed in only one catabolic state was in fact differentially expressed in other states) (Table 2
).
Analysis of transcription factor binding motifs in the upstream regions of atrogins
Rates of occurrence of the 124 transcription factor binding motifs in the TRANSFAC database were obtained in 3 kb of the 5' region of 31 up-regulated atrogins (Unigene clusters Mm.3238, Hs.61661, Hs.173685, Mm.2159, Mm.930, Hs.87417, Hs.194669, Hs.71819, Hs.25732, Mm.29891, Mm.14638, Hs.7879, Hs.112396, Mm.28548, Mm.28357, Mm.22749, Mm.25311, Mm.41792, Mm.6720, Mm.180499, Mm.28571, Hs.78466, Mm.30097, Mm.21874, Hs.182979, Hs.8765, Hs.28491, Hs.94360, Hs.5308, Hs.183842, Hs.18370) and 17 down-regulated atrogins (Unigene clusters Hs.177584, Hs.172928, Hs.80691, Hs.115285, Hs.750, Hs.287820, Mm.578, Hs.17109, Hs.198951, Hs.2795, Mm.3156, Mm.2060, Mm.30000, Mm.4919, Hs.181013, Mm.147387, Mm.28683). Frequency values (per 1000 bp) were subsequently divided by the frequency of random occurrence of the motifs, calculated by the MatInspector program. For each motif, the occurrence frequency in up-regulated divided by down-regulated atrogins was calculated. Motif frequency was measured in the 5' region of 15 genes not differentially expressed in any of the atrophying muscles (Unigene clusters Hs.118442, Hs.197540, Hs.25450, Hs.26045, Hs.302131, Hs.348412, Hs.37616, Hs.75219, Mm.16373, Mm.1764, Mm.179747, Mm.83615, Mm.2661).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Thus, these muscles were undergoing rapid atrophy. Since microarrays from the 7/8 nephrectomized, streptozotocin-treated, and tumor-bearing animals were compared with ones from pair-fed control animals studied in parallel, changes in mRNA specifically reflect effects of the disease process, and not any associated decrease in food intake. The large number of fasting-specific transcriptional changes argues that our attempts to control for decreases in food intake in the disease models by pair-feeding regimes were largely successful and a unique pattern of transcription in fasting alone was still evident.
The results from hybridizations of muscle RNA to mouse and human microarrays together yielded 16,392 individual gene sequences that could be analyzed in all four catabolic states (Table 2)
. Of these, 133 mRNAs (0.8%) were differentially expressed (i.e., P<0.05) in all four and were designated as atrogins. This number actually comprises 120 unique genes, since some sequences were duplicated on the mouse and human arrays. Expression data for all genes defined as atrogins are included in Supplementary Table 1
. Data for the disease-specific genes, the individual microarray outputs, and Supplementary Tables 1 and 2 can be found at http://agoldberg.med.harvard.edu/muscledatabase.
|
Protein degradation
mRNAs for many genes involved in protein degradation were up-regulated in all four types of atrophying muscle (Fig. 1
). As expected, nearly all these genes encoded components of the Ub–proteasome pathway, including the two polyUb genes, four different subunits of the 20S proteasome and three of the 19S proteasome regulator, some of which have been demonstrated by Northern blot analysis previously (20
, 29)
. An additional four proteasome subunits and one 19S subunit were up-regulated in three of four atrophy states, and others in one or two catabolic states (see Supplementary Table 2 at http://agoldberg.med.harvard.edu/muscledatabase). Surprisingly, five of the thirty-four 26S subunits were not differentially expressed in any of the atrophy conditions, and levels of none was repressed (see Supplementary Table 2). Of note was the 
threefold increase in mRNAs for PA200, a recently described component of nuclear proteasomes that activates peptide hydrolysis and has been proposed to play a role in DNA repair (30)
. Finally, in all types of atrophy, there was an induction of USP14, an isopeptidase that associates reversibly with the 19S complex (31)
, and may be important in recycling polyUb chains back to Ub monomers.
|
A marked induction in mRNA levels for two Ub extension proteins that are carboxyl-terminal fusions between Ub and ribosomal proteins (RPS27A or UBA52) was consistently found in all types of atrophying muscle studied. These fusion proteins probably serve as a source of free Ub, in addition to the polyUb genes UBB and UBC, when proteolysis increases.
Ub ligases
The Ub ligase (E3) atrogin-1/MAFbx was originally cloned because its mRNA was the most highly induced in muscle during fasting and its expression increases between 4- and 15-fold in different catabolic states (16)
. Atrogin is strongly induced in muscles from septic rats (32)
and after denervation or disuse (19)
. Furthermore, mice lacking this gene show reduced rates of disuse atrophy (19)
. Expression of another muscle-specific E3, MuRF-1, increases markedly after denervation or disuse (19)
, but was not present on our microarrays. To test whether it is of general importance in muscle atrophy, the same RNA was analyzed by Northern blot. Indeed, MuRF-1 mRNA was strongly induced in all four catabolic states (Fig. 2
).
|
The dramatic induction of atrogin-1 and MuRF-1 contrasts with the lack of change in expression of many other Ub-conjugating enzymes (see Supplementary Table 2), including E1, many E2s, and E3s, Nedd4, and E6AP. It is noteworthy that no significant change occurred in Ubr1(E3) or E214k (components of the N-end rule pathway), whose mRNA was found by Northern analysis to rise in muscles in diabetes (25)
, fasting (33
, 34)
, and sepsis (35)
. Small increases (1.3 to 3-fold) were observed in mRNAs for the ubiquitination factor E4B, which may act in combination with an E3 to increase the efficiency of Ub conjugation to proteins (36)
. Also increased in all catabolic states was mRNA for another Ub carrier protein (E2), noncanonical Ub-conjugating enzyme 1, whose role in muscle is unclear. Its yeast homologue, Ubc6, is involved in ubiquitinating proteins retro-translocated from the ER (37)
.
An important lysosomal cysteine protease, cathepsin L, was induced two- to threefold in all four catabolic states whereas mRNA for the other lysosomal hydrolases was not (see Supplementary Table 2). Although lysosomes have only a limited role in the bulk of intracellular proteolysis, large increases in cathepsin L mRNA occur in muscle from septic, tumor-bearing, dexamethasone-treated, and fasted rodents (18
, 20)
. Some reports have suggested that cathepsin L may be found extracellularly (38)
; if so, this protease might play a special role in turnover of extracellular components during atrophy. No change in mRNA levels for a range of matrix metalloproteases (MMPs) was found. MMP-2 and MMP-9 are expressed in muscle (39)
and have been proposed to function in remodeling of the extracellular matrix after disuse or denervation (40
, 41)
. However, MMPs are induced late in disuse atrophy (40)
and so may not be a feature of the rapid atrophy studied here.
ATP production and substrate metabolism
Genes encoding some key proteins in mitochondrial energy production, glucose, and ketone body metabolism were differentially expressed in all four catabolic states (Fig. 3
). mRNA for seven different inner mitochondrial membrane proteins as well as mitochondrial creatine kinase, all of which participate in electron transport and/or ATP synthesis, was reduced in all catabolic states examined in this study. mRNAs encoding the 
subunit of the glycogen phosphorylase kinase complex and several enzymes catalyzing later steps in glycolysis were reduced, as well as two components of the pyruvate dehydrogenase complex that regulate whether pyruvate is oxidized by the TCA cycle. Also reduced was expression of malate dehydrogenase, a key component of this cycle, as well as the malate-aspartate shuttle that brings reducing equivalents produced in glycolysis to the mitochondrion. Finally, mRNA for 3-oxo CoA transferase, required for oxidation of ketone bodies, was reduced in all four catabolic states. These changes in gene expression would be expected to suppress muscle’s capacity to utilize glucose and reduce muscle energy production generally. Reduced glucose utilization is consistent with the lack of insulin in fasting and insulin resistance in cancer and renal failure; regulation of these steps had not been reported previously. These changes do not appear to support the idea that cachexia is a purely hypermetabolic response where substrates are used at accelerated rates (42)
. In our analysis, the inducible form of 6-phosphofructo-2-kinase, which is induced in some tumors (43)
, was the only atrogin not similarly regulated in all four catabolic states.
|
Kahn and co-workers recently described the transcriptional profile in muscle from mice made diabetic by prolonged treatment with streptozotocin (21)
and observed coordinate suppression a different set of genes involved in glucose utilization. These workers did not observe changes in many of the genes identified here as atrogins (e.g., Ub, proteasome subunits, metallothionein). However, the muscles studied by Yechoor et al. were from mice (instead of rats) and were treated longer with streptozotocin to create a more chronic, adapted diabetic state. Unfortunately, no analysis of the extent of muscle weight loss or rates of protein degradation was performed in that study; thus, muscle wasting may have ceased at the time of their analysis.
Nitrogen metabolism
Expression of three genes encoding enzymes for purine or polyamine catabolism was consistently increased, including 1) spermidine N1-acetyltansferase, a key enzyme in polyamine catabolism, which was induced fivefold in muscles from uremic animals, as well as 2) IMP dehydrogenase and 3) AMP deaminase 3, which are involved in the purine–nucleotide cycle. This cycle may play a key role in energy production in muscle (44)
and is a source of ammonia derived from amino acid degradation. In muscle, ammonia is used to form glutamine from glutamate in a reaction catalyzed by glutamine synthase. Indeed, glutamine synthase is markedly increased in muscles from all four states. Glutamine production and export by muscle occur in a variety of catabolic conditions, including sepsis and trauma; simultaneously there is increased uptake of glutamine by the liver, lymphocytes, and the kidney (45)
. This inter-organ flux of glutamine appears to help provide substrates for increased gluconeogenesis and urinary ammonia generation in acidosis. No significant changes were noted in mRNAs for enzymes of branched chain amino acid metabolism, which are induced in fasting. However, transcript levels for an amino acid transporter, Slc7a8, did increase. Slc7a8 may be involved in exporting amino acids like glutamine, whose release from muscle increases when there is net proteolysis (46
, 47)
.
Transcription factors
No evidence was obtained for the simplistic view that atrophy involves a general repression of muscle gene expression (Fig. 4
a), though three transcriptional activators associated with rapid growth—JUNB, MAF, and the mouse ortholog of human Snf2-related CBP activator protein (SRCAP)—had reduced mRNA levels in all atrophy states studied and may reflect suppression of growth in wasting muscle. mRNA levels for the transcriptional repressor EZH1, which may stabilize heterochromatin, were increased. These changes would appear to favor a reduction in the transcription of a subset of genes in atrophy. On the other hand, mRNA for MAX, a MYC-related transcriptional activator, increased somewhat under all the conditions studied. Foxo1, a member of the forkhead family of transcriptional activators, was strongly induced. Foxo1 has recently been implicated in the development of insulin resistance in type II diabetes in liver, pancreas, and adipose tissue (48)
. Perhaps the induction of Foxo1 in atrophying muscle may help explain the insulin resistance in these catabolic states. mRNA for the p23 telomerase binding protein Tebp, which binds to DNA and interferes with some hormone-dependent transcription, and Nfe2l2, which responds to oxidative stress, increased in all atrophy states (49)
. ATF4, which regulates amino acid metabolism and resistance to oxidative stress, was increased (50)
. Finally, mRNA for Tgif, a homeobox gene that may repress certain transcriptional activators (51)
, was markedly increased.
|
Extracellular matrix components
In the four catabolic states, mRNA levels were reduced for collagen I, III, V, and XV and the procollagen-C endopeptidase enhancer, which accelerates maturational cleavage of the procollagen I C-propeptide in the extracellular compartment (Fig. 5
). It was recently reported that mRNA for collagen III as well as levels of this protein decrease during disuse atrophy (17)
. mRNA levels for other matrix components, fibrillin and fibronectin, were reduced, as were those for OSF-2 a cell adhesion protein, and galectin 1, LGALS1, which has recently been shown to enhance myoblast fusion and induce myogenic differentiation of fibroblasts (52)
.
|
Genes involved in translational control
In fasting, a surprising increase was found in mRNA for genes encoding certain translation initiation factors and the inhibitor of cap-dependent initiation, 4E binding protein (20)
. These adaptations could help reduce 5' cap-dependent translation that occurs when growth factors and nutrients are decreased while allowing cap-independent translation of other key proteins that appear to be important in stressful conditions (53)
. In all four conditions, mRNAs for translation initiation factors EIF4A2, EIF4G3, and EIF4EBP1, which act through cap-independent mechanisms, increased (Fig. 4b
). Despite the reduction in translation, transcript levels for a 60S ribosomal protein L12 and nucleolin, which are believed to participate in ribosomal assembly, increased in all four catabolic states. Finally, an RNA helicase-related protein containing a DEAD box motif was increased in all four states, with a particularly dramatic rise (up to 100-fold) in diabetes and renal failure. Although the precise role of this protein is unknown, similar RNA helicases are involved in ribosomal assembly, translation initiation, and RNA processing.
Metallothionein and other proteins induced in oxidative stress
Metallothionein was among the most strongly induced genes in all the atrophying muscles on both human and mouse microarrays. Two of the 14 highly homologous human metallothionein genes present on the human microarrays, MT1B and MT1L, were dramatically induced (3- to 20-fold) (Fig. 6
). Mouse Mt1 was induced 1.5- to 2.5-fold (Fig. 6)
; clones of Mt1 and Mt2, present only on the arrays from fasted mice, were induced 5- to 6-fold (see clones 1037652, 334351). Since the mouse Mt1 sequence is highly homologous to that of human MT1L and MT1B (data not shown), it is likely that rodent Mt1 transcripts from the atrophying muscles hybridized to the MT1L and MT1B cDNA fragments on the human arrays.
|
Metallothioneins are induced by heavy metals, glucocorticoids, and oxidative stress and can protect cells against DNA damage from reactive oxygen species (54
, 55)
. The mechanisms by which metallothionein exerts its protective effect are still unclear. A more modest increase in mRNA for the 32 kDa thioredoxin-like protein (56)
, which helps maintain the cytosol in a reduced state, was observed in all four atrophy states. As mentioned above, mRNA for Nfe2l2 and ATF4, transcription factors that regulate genes controlled by antioxidant response elements, increased in all four states (Fig. 4a
). Thus, the atrophy program seems to include elements of a transcriptional response to oxidative stress.
Genes involved in muscle growth and differentiation
Insulin-like growth factor-1 is a major determinant of muscle growth and plays a key role in compensatory hypertrophy (57)
. It stimulates the PI-3-kinase and Akt signal transduction cascade whose activation can combat denervation and disuse atrophy (58)
. Although expression of IGF-1 or downstream signaling molecules did not change in the atrophying muscles, IGF-1 binding protein 5, which binds to the extracellular matrix and modulates the muscle’s response to IGF-1 (59)
, was dramatically down-regulated in all atrophy conditions studied (Fig. 6)
.
The group of atrogins included the proapoptotic gene Bnip3, which was induced threefold in all four types of atrophy (Fig. 6)
. Bnip3 interacts with and can antagonize Bcl-2 (60)
; it is induced by hypoxia and acidosis, when it triggers cardiomyocyte cell death (61)
. However, there is no evidence for cell death in the reversible forms of muscle atrophy studied here.
mRNA levels decreased for two calcium binding proteins: parvalbumin and secreted modular calcium binding protein 2 (Smoc2) (Fig. 6)
. Parvalbumin is found in fast-twitch fibers and binds cytosolic Ca2+, enabling rapid relaxation after contraction. Smoc2 is highly expressed by skeletal and vascular smooth muscle; although its role is unclear, up-regulation of Smoc2 occurs in vascular smooth muscle in response to stretch injury and is associated with smooth muscle proliferation (62)
.
Although large decreases in expression of myofibrillar proteins were anticipated, little or no evidence was obtained for suppression of the transcription of myofibrillar or cytoskeletal genes in atrophying muscle. Only two components of the myofibril—namely, two myosin light chain isoforms—feature in the group of down-regulated atrogins (Fig. 5)
. Only mRNA for microtubule-associated protein Map1lc3 was strongly induced in all states whereas mRNA for LIM domain binding protein 3, which binds actinin and localizes to the Z-band of the myofibril (63)
, was suppressed (Fig. 5)
.
A large proportion of atrogins were unknown genes or genes whose function in muscle is still obscure. Included in this group is the newly discovered gene lipin, which, when mutated, results in lipodystrophy (64)
. mRNA levels for lipin were markedly induced in all types of atrophy (Fig. 6)
. Lipin may regulate lipid biosynthesis in the liver. It is phosphorylated by mTOR in response to insulin (65)
and thus is controlled by the Akt signaling pathway. In addition, the interferon-related developmental regulator-1 (Ifrd1) was induced threefold in all four states (Fig. 6)
. Although Ifrd1 is necessary for normal muscle differentiation (66)
, its role in the development of atrophy is unknown. In contrast, mRNA for P311 was consistently reduced between 3- and 14-fold in these states (Fig. 6)
. P311 is an 8 kDa protein first found in neurons late in brain development and in invasive glioblastoma (67)
. P311 appears to be involved in smooth muscle differentiation in myofibroblasts by driving expression of actin and other muscle-specific genes (68)
.
The largest subgroup of atrogins was that for which no known function has been established. These included several genes with dramatic increases in mRNA levels of 7- to 8-fold (e.g., clones 619799, 651825, 621966) and others with even larger relative decreases in mRNA levels of up to 100-fold (clone 692699).
Transcriptional control of atrogins
To test for coordinate regulation of the group of atrogins defined above, we examined 3 kb of upstream sequences from 31 of the most up-regulated and 17 of the most down-regulated atrogins for common transcription factor binding motifs using the TRANSFAC database (http://transfac.gbf.de/TRANSFAC/). For each of the 124 motifs examined in that database, the frequency of occurrence was calculated and compared between the up- and down-regulated atrogin groups. These results were compared with a group of 16 randomly selected genes that were not differentially expressed in any of the four atrophy states. No motifs were found solely in either up- or down-regulated genes (Fig. 7
). The frequency of occurrence of all motifs was similar in genes whose expression was up-regulated, down-regulated, or unchanged. Glucocorticoids alone can induce muscle atrophy (69)
and appear to be essential for atrophy during fasting, renal failure, and diabetes. We therefore recorded the frequency of glucocorticoid response elements and the frequency of binding sites for other transcription factors, such as Sp1 (70)
and C/EBP (71)
, which are regulated by glucocorticoids and have been proposed to regulate proteasome and Ub expression in atrophying muscle. In the upstream regions of the atrogin genes, the frequency of occurrence of GREs, Sp1, and C/EBP binding sites did not differ among the up-regulated, down-regulated, and control groups. Finally, when atrogins were grouped by function (e.g., degradation-related), no difference in frequency of appearance of transcription factor binding motifs was apparent when compared with other atrogins or genes that were not differentially expressed. It remains likely there are as yet undefined transcriptional modulators that activate the changes in gene expression in atrophying muscle.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. A number of new and unexpected features of the atrophy process and its regulation are suggested by these atrogins.
Many adaptations enhance capacity for protein degradation
The present findings provide further evidence that the accelerated proteolysis underlying muscle atrophy is due largely to activation of the Ub–proteasome pathway. In fact, increases in mRNAs for polyUb and several proteasome subunits in muscles upon denervation and fasting (2)
provide the first clue that different types of atrophy might involve a common set of transcriptional adaptations. This study confirms that mRNAs for polyUb and multiple 26S proteasome subunits rise in atrophying muscles, but also demonstrates that mRNAs encoding two Ub ribosomal protein fusion genes, RPS27A and UBA52, generally increase. These Ub fusion proteins had been thought to function constitutively as a source of Ub monomers during growth (72)
, but since they are induced coordinately with the polyUb genes UBB and UBC, they appear to serve as an additional source of Ub when overall proteolysis rises. Although transcription of several subunits of the 19S and 20S proteasome increase coordinately, some did not change in any catabolic state. Thus, expression of certain proteasome subunits may be subject to tighter transcriptional control than others and may be rate-limiting in assembly of the mature complex. These findings suggest that in muscle, in contrast to findings in yeast (73)
, different transcription factors or coregulators appear to affect the expression of subgroups of proteasome subunits.
Our findings rule out the simplest model for atrophy in which the general acceleration of proteolysis and Ub conjugation results from increased expression of all or many Ub-conjugating enzymes in muscle. In fact, mRNAs for the vast majority of Ub-conjugating enzymes did not change (supplementary data, Table 2) whereas mRNAs for two Ub ligases, atrogin-1 and MuRF-1, were dramatically increased. Among the enzymes that did not rise were E214k (E2A/B) and E3 (Ubr1), which comprise the N-end rule pathway. mRNAs for these factors had been found to increase up to twofold in muscle from fasted, diabetic, and tumor-bearing animals (7
, 25
, 33)
; the N-end rule pathway has been found to account for most of the increased Ub conjugation in soluble extracts from septic and tumor-bearing animals (24)
. However, upon fasting, mice lacking E214k undergo muscle atrophy like control animals (74)
. mRNA for other E2s has been reported to increase in various models of atrophy [e.g., UbcH2 after TNF administration in muscle cultures (75)
and E2G after glucocorticoid administration to rats (76)
]. Our results suggest these changes are not general features of atrophying muscles, although it is possible that if a larger number of muscles were analyzed, some of these borderline changes (e.g., for E2G, see Supplementary Table 2) would have reached statistical significance.
Suppression of cell growth and extracellular matrix in atrophying muscle
Stimulation of the Akt pathway in muscle causes hypertrophy (57)
and suppression of this pathway may trigger atrophy (26
, 77)
. IGF-1-induced hypertrophy occurs via Akt; in all the atrophying muscles studied, mRNA for IGFBP-5, which enhances the effects of IGF-1 (59)
, fell dramatically, suggesting that activity of the Akt pathway is reduced in these muscles. It is noteworthy that expression of the forkhead transcription factor Foxo-1 increased in all four catabolic states; this family of transcription factors is sequestered and inactivated in the cytoplasm by Akt phosphorylation (78)
. Reduced Akt phosphorylation would leave Foxo-1 in its underphosphorylated, active form, which can induce programmed cell death and insulin resistance in several tissues in type II diabetes (48)
. Insulin resistance is also a prominent feature of muscles in uremia, cancer cachexia, and fasting and should lead to enhanced proteolysis and reduced translation of new proteins (79)
. Thus, the up-regulation of Foxo-1 may contribute in multiple ways to the atrophy process.
In addition to Foxo1, another proapoptotic gene, Bnip3, was induced threefold in the four catabolic states. Bnip3 interacts with Bcl-2 and can antagonize its prosurvival function (60)
. In the ischemic heart, Bnip3 is induced by hypoxia and acidosis and triggers myocyte death (61)
. The finding that two proapoptotic genes are also atrogins suggests that both function in each process as part of a growth suppression program.
The extracellular matrix in muscle is generally assumed to be stable. Nevertheless, a rapid decrease in mRNA occurred for many components of the extracellular matrix in these catabolic states. Previous studies have shown a loss of collagen proteins in disuse atrophy (17)
, and the marked reduction in mRNAs for several extracellular proteins suggests that reduced synthesis of the extracellular matrix is linked to loss of intracellular protein and presumably contractile load.
Control of transcription and translation during atrophy
In addition to the acceleration of overall proteolysis, these systemic forms of muscle atrophy involve a suppression of overall protein synthesis. However, even when protein synthesis is reduced, the atrophying muscles must maintain or even increase the expression of certain key proteins. Indeed, over half of the atrogins are increased in the atrophying muscles at the mRNA level. By contrast, changes in mRNA for certain transcriptional regulators suggests both activation and repression of gene transcription. Whereas expression of activators such as Foxo1 and MAX are increased, the expression of other proto-oncogenes (growth promotors) such as JUNB (80)
fall. Several up-regulated atrogins encode translation initiation factors that could indicate a reduction in translation in muscle. The strong induction of EIF4EBP1, an inhibitor of translation of capped mRNA, should reduce overall rates of translation. Simultaneously, mRNA levels for EIF4A2 and EIF4G3 increase, which suggests a mechanism for enhancing translation of the subgroup of mRNAs with internal ribosome entry sites, which tend to be important in stressed cells (53)
. Together, these transcriptional changes indicate ways by which the levels of key proteins may be maintained when overall transcription and translation decrease.
An important regulator of gene expression in muscle that may account for many of the adaptations shown here are glucocorticoids. Excessive secretion of glucocorticoids as occurs in Cushing’s syndrome or administering pharmacological doses can cause muscle wasting (81)
. Adrenal steroid production is required for muscle atrophy in fasting (3)
, diabetes (27)
, acidosis (82)
, and sepsis (83)
. Among those atrogins induced most dramatically were glutamine synthase and metallothionein, which contain glucocorticoid response elements in their promoter regions (54
, 84)
. However, no GREs were found in the promoters for the majority of up-regulated atrogins, including some (e.g., atrogin-1) that are inducible by this hormone. Presumably glucocorticoids act indirectly, perhaps by inducing the expression of a small number of key proteins (e.g., C/EBP), which in turn activate genes induced during atrophy (71)
.
A reduced circulating level of insulin is a characteristic feature of fasting and type I diabetes, and this lack of insulin can accelerate muscle proteolysis (27)
. As discussed above, insulin resistance is an important feature of the systemic diseases studied here and probably contributes to muscle wasting. Insulin resistance can be induced by glucocorticoids as well as TNF, whose production rises in many types of cancer cachexia and sepsis. Induction of the transcription factor Foxo1 may contribute to the insulin resistance characteristic of these atrophying muscles (48)
.
The present findings raise the possibility that reactive oxygen species play an important general role in atrophy, presumably in the initiation or regulation of this response. mRNAs for a thioredoxin-like protein, as well as ATF4 and Nfe2l2, transcription factors that promote expression of oxygen-stress response genes (49
, 50)
, were increased in all these atrophying muscles. Markedly increased were transcripts encoding metallothionein-1, which can play a role in combatting oxidative stress (55)
. The production of reactive oxygen species has been proposed as a mechanism by which TNF might damage muscle and regulate gene expression via activation of redox-sensitive transcription factors such as NF-
B (85)
. Reactive oxygen species produced after burn injury have been proposed to contribute to the loss of muscle at distant sites (86)
. Clearly, the role of oxygen radicals in atrophy merits in-depth study.
We have studied atrophying muscles in pathophysiological states where systemic muscle wasting is triggered by circulating factors. It remains to be seen whether the group of atrogins identified here change in denervation and disuse atrophy, where contractile activity is reduced in specific muscles, and where slow-twitch fibers show greatest weight loss. In the systemic diseases studied here, fast-twitch fibers are preferentially lost. While mRNAs for polyUb, some proteasome subunits, MuRF-1, and atrogin-1 increase upon denervation/disuse, certain mRNAs found to decrease (IGFBP-5 and parvalbumin) increase upon denervation or unloading (unpublished results and refs 87
, 88
). It is likely that further important differences will emerge between the wasting induced by systemic diseases and decreased contractile activity.
The demonstration of a transcriptional program in muscle common to a variety of wasting diseases has already suggested novel therapeutic targets to combat muscle wasting (e.g., atrogin-1 and MuRF-1) but has uncovered several unexpected adaptations. A major surprise was that muscle atrophy was not associated with marked reduction in expression of the contractile apparatus. The loss of these components must be due largely to accelerated degradation or reduced translation. A full understanding of the atrophy process will require in-depth analysis of the physiological importance of these responses, especially the decreased expression of glycolytic enzymes or of extracellular matrix and the induction of transcription factors and proteins related to oxidative stress and NH3 metabolism. Of obvious importance will be identification of the structure and function of the many other ORFs differentially expressed in these muscles.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication July 1, 2003. Accepted for publication August 8, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
