| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Cell Biology, Harvard Medical School, Boston, Massachusetts, USA; and
* Renal Unit, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA
2Correspondence: Department of Cell Biology, Harvard Medical School, 240 Longwood Ave., Boston MA, 02115, USA. E-mail: alfred_goldberg{at}hms.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: cDNA microarray • ubiquitin • proteasome • cachexia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. These diverse physiological and pathological conditions appear to trigger muscle atrophy through distinct extracellular stimuli; for example, low insulin levels and glucocorticoids activate muscle protein loss in fasting (3)
, tumor necrosis factor 
and other cytokines do so in cancer cachexia (4)
, and the lack of contractile activity is critical upon denervation and in various neuromuscular diseases. Despite these very different physiological signals, the resulting biochemical changes in atrophying muscle share many common features (1
, 2)
.
In fasting and most other conditions when muscles atrophy, overall protein synthesis in muscle is reduced, but the rapid loss of muscle protein results mainly from increased degradation of cell proteins (5)
, especially of contractile proteins, which comprise most of the muscle mass. In all experimental models of muscle wasting studied so far, this increased protein degradation results primarily from an activation of the proteolytic system involving ATP, ubiquitin (Ub), and the 26S proteasome (1)
. In these various types of atrophy, muscles show an increased content of ubiquitin-conjugated proteins, as well as increased rates of ubiquitination of cell proteins, and the accelerated proteolysis is sensitive to proteasome inhibitors (6)
. There is increased expression of genes for key components of the ubiquitin-proteasome pathway. For example, in fasting (7)
and all other experimental models of muscle wasting studied (6)
, the accelerated proteolysis is accompanied by a two- to fourfold increase in mRNA levels for polyubiquitin and certain subunits of the 20S proteasome; these changes in mRNA levels appear to result from enhanced rates of transcription (8
9
10)
.
Therefore we have proposed that in these different catabolic states, the atrophying muscles show a common program of changes in gene transcription that results in cessation of normal growth and net protein loss, including changes leading to activation of the ubiquitin-proteasome pathway (1
, 2
, 6)
. To identify the key transcriptional adaptations causing muscle atrophy, we have begun a series of microarray analyses to compare changes in mRNA content that occur in skeletal muscles in various catabolic states. In the first phase of this project, we focused on the transcriptional changes in muscle during fasting, because the alterations in muscle protein turnover and other metabolic adaptations to food deprivation are especially well defined, as are the endocrine factors signaling these responses (11)
.
After food deprivation, there is a net breakdown of muscle protein to provide the fasting organism with amino acids, which are either oxidized in muscle or released for hepatic gluconeogenesis (12)
. Besides the marked enhancement of muscle proteolysis, especially of contractile proteins (13)
, there is a general reduction in protein synthesis (14)
resulting from a reduction in the rate of translation (15)
that is due to impaired initiation (16)
and a decrease in total cellular ribosome content (17)
. As a result of the decreased levels of insulin in fasting, glucose uptake and oxidation are reduced rapidly in muscle, which instead oxidizes free fatty acids and ketone bodies (18)
. With a more prolonged fast, there is a global reduction in oxidative metabolism in mammals (11
, 19)
. Although most studies of muscle’s adaptations to fasting have focused on changes in metabolic flux and enzyme activities, many (perhaps most) of these responses are likely to be due to changes in gene transcription.
Although several investigators have reported altered levels of certain mRNAs in muscle upon fasting (20
21
22)
, the full extent and importance of alterations in gene transcription in muscles in fasting are unknown, and no systematic analysis of the changes in muscle mRNA in fasted organisms has been reported. To investigate these questions, we used cDNA microarrays to acquire a more comprehensive picture of the fasting-induced transcriptional adaptations in muscle. These experiments were undertaken to learn whether 1) genes encoding all or only specific components of the ubiquitin-proteasome pathway are induced, 2) expression of genes for the various myofibrillar proteins and glucose metabolism are reduced coordinately, 3) there are particularly extensive changes in mRNAs encoding key regulatory proteins, 4) new genes are identified that may have a particularly important role in the disassembly or degradation of the myofibrillar apparatus, and 5) fasting affects the expression of other cellular or extracellular components in unexpected ways. The main goal of this analysis was to define the transcriptional changes that characterize the response of muscle to the fasted state, and thereby to lay a basis for defining which of these adaptations are specific to the atrophy process.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Extraction of RNA
Frozen muscles were placed in TRizol (Life Technologies, Grand Island, NY) and homogenized before extraction of total RNA. After extraction, an aliquot was used to confirm the absence of significant degradation by electrophoresis in a 0.8% agarose gel and visualization by staining with ethidium bromide. Poly(A)+-containing RNA was then isolated using oligo-dT binding resin (Oligotex-midiprep, Qiagen, Valencia, CA) and the yield of poly(A)+ RNA was measured using Riboquant (Molecular Probes, Eugene, OR).
Microarray hybridization
Labeling, hybridization, and scanning of microarrays were performed by Incyte, Inc. (St. Louis, MO). For each experimental sample, poly(A)+ RNA was reverse transcribed with 5' Cy3- or Cy5-labeled random 9-mers (Operon Technologies, Inc., Alameda, CA) to yield first strand cDNA copies of fed (reference) and fasted samples incorporating different fluorophores. The paired, labeled cDNA samples then underwent simultaneous competitive hybridization to the same microarray at 60°C for 6.5 h. Each spot on the microarray (maximum 10,000) was a PCR-amplified, double-stranded cDNA clone of 500-1000 base pair sequence from the 5'end of genes or ESTs selected from public (Institute of Genomic Research, GenBank, or Unigene) or proprietary databases. After hybridization, the relative amounts of labeled RNA from fed and fasted samples hybridized to each spot were determined by using the strength of the emission signal from Cy3 and Cy5 after laser excitation with the appropriate wavelength (Axon GenePixTM scanner, Foster City, CA). For each hybridization, emission signal data were normalized by multiplying the Cy5 signal value by the ratio of the means of the Cy3 and Cy5 signal intensity for all spots on the array.
Data analysis
Analysis was performed using GEMtools software (v2.4.1, Incyte Genomics), a spreadsheet program (Excel 2000, Microsoft Corp.), and a relational database (Access 2000, Microsoft Corp.). To determine whether results from an individual spot were satisfactory technically, the following criteria were adopted. 1) PCR amplification of the sequence spotted on the array was deemed acceptable only if the amplification was confirmed and a single size product was obtained. 2) Accurate printing of each spot was required as shown by emission signal from >40% of the spot area. 3) The signal-to-background ratio had to be >2.5 and the emission signal from the fluorophore label of either the fed or fasted sample had to be >699.
For each gene represented on the microarray, the ratio of normalized emission signals (corresponding to the ratio of individual mRNAs in fed to that in fasted samples) was expressed as the greater divided by the lesser signal value. The ratio was designated positive if fasting caused an increase in expression (i.e., the greater signal value was from the fasted sample) and negative if fasting led to a reduction of expression. The basis for deciding whether genes were differentially expressed in this study were the extensive reproducibility studies provided by the manufacturer (23)
.
We adopted a cutoff of 1.8-fold change based on their finding (23)
that the limit for detectable differential expression was 1.75 (99.5% limits for detection) across a wide range of signal intensities. Subsequent studies using self-self hybridizations with human RNA from a variety of different tissues, including heart muscle, showed that the 99% limits of detection were 1.4-fold. We elected to continue to use the more stringent 1.8-fold criteria because some of our experiments used early builds of the mouse and human microarrays not included in the later technical reports.
Supplementary data
Additional information and data files pertaining to this study can be found at our website: http://agoldberg.med.harvard.edu/muscledatabase/fasting. The file "GeneList" appears there and contains a compilation of all differentially expressed genes from the microarrays used in this study. The site also contains Supplementary Fig. A and primary microarray data formatted as Microsoft Excel spreadsheets.
General design of experiments
Three independent food deprivation experiments were performed. In each experiment, at least 10 fed control mice were used and two groups of 
10 mice were fasted for 1 or 2 days. To identify genes that were differentially expressed in muscle during fasting, comparisons were made between muscle RNA samples from the fed and 1-day fasted, and between the fed and 2-day fasted animals from each experiment. These samples were compared with human (Human UniGEM 1 or 2) and mouse (Mouse GEM 1) microarrays (i.e., 4 microarrays per experiment). In one experiment, the labeling of fed and fasted samples was reversed (i.e., fed samples were labeled with Cy5 and fasted with Cy3) to compensate for any nonlinearity in the emission signal intensity response curve for each fluorophore.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
); by 2 days the total RNA content of the muscles fell by almost 50%. Total mRNA (i.e., poly(A)+-containing molecules) appeared to decrease markedly, but these changes did not reach statistical significance. In any experiment, only about half of the 9000 individual spots printed on each array yielded data that were technically acceptable by our criteria (see Materials and Methods and legend to Table 2
). In these analyses, only those genes whose mRNA levels were increased or reduced by 1.8-fold or more in at least two of three experiments were designated as differentially expressed genes. Our preliminary studies revealed that 70% of the genes differentially expressed in one experiment [those that showed an absolute fold change 
1.8 (23)
] were differentially expressed in the subsequent experiments. Similar variability in cDNA microarray hybridizations has been noted by others (24)
.
|
|
The fraction of genes found to be differentially expressed during fasting using the human microarray was lower than when using the mouse microarray (e.g., 2.0% vs. 5.9% after 48 h fasting; Table 2
), probably because the experimental samples from the mouse muscles had greater affinity for the cDNAs spotted on the mouse microarray. The subset of genes represented on the mouse array differed from those on the human array, which contained a higher proportion of unique genes and fewer ESTs. Nevertheless, results obtained from human and mouse microarrays consistently gave similar results in the many instances when the same gene was represented on both chips.
It was striking that the expression of >94% of genes on the microarrays did not change on either day. Thus, though mRNA levels in muscle appeared to fall by 2 days of fasting (25)
(Table 1)
, relative amounts of the vast majority of mRNAs remained unchanged. The finding that only a small fraction of mRNAs was differentially expressed suggests that these changes represent a specific response of the muscle to the fasted state. To test whether the differential expression of such a small proportion of the cell’s mRNA represented a reproducible pattern of adaptations, we compared transcriptional changes induced by fasting for 1 or 2 days. If a gene was differentially expressed on day 1, the probability of its being differentially expressed on day 2 (positive predictive value) was 90% on the human and 79% on the mouse arrays. As the duration of the fast increased, the changes in transcription became more pronounced, i.e., the number of differentially expressed genes and the magnitude of changes in their mRNA levels were greater at 48 than at 24 h. Thus, the pattern of alterations in gene expression induced rapidly by food deprivation was generally sustained and became more marked with a more prolonged fast.
Genes involved in proteolysis
Components of the ubiquitin-proteasome pathway
Earlier studies of rat muscle using Northern analysis indicated that fasting causes increased mRNA levels for polyubiquitin by two- to fourfold (7)
, and our microarray analysis demonstrated three- to fourfold increases in levels of mRNA for two polyubiquitin genes (UBB and UBC) (Fig. 1
a). An unexpected finding was that mRNA levels for the two genes in which the carboxyl end of ubiquitin is fused to ribosomal proteins (RPS27A, UBA52) were increased (Fig. 1a
). The expression of these fusion proteins was not previously known to be regulated and was believed to only serve as a source of ribosomal subunits and ubiquitin when cell growth is rapid.
|
In addition to increased levels of polyubiqutin mRNA, Medina et al. demonstrated that fasting enhanced expression of five subunits of the 20S proteasome (four , one ß) (7)
. Our analysis yielded reproducible results on mRNAs for 13 of the 20S proteasome subunits, 7 of which were significantly increased by fasting (Fig. 1b
). mRNAs for 17 subunits of the 19S regulatory complex of the proteasome were assayed; 10 of these were increased after food deprivation (Fig. 1c
). At the same time, mRNA levels for many subunits of the 20S (two and four ß) and of the 19S particle were not altered significantly. Thus, surprisingly, transcription of some of the 30 genes encoding subunits of the 26S proteasome is not coordinately regulated. The identities of the proteasome components most highly induced do not reflect any obvious functional or structural subassemblies. Perhaps the expression of certain subunits is highly regulated because their levels are rate-limiting during proteasome assembly.
For most proteins, degradation by the proteasome requires their ubiquitination by a specific ubiquitin-protein ligase (E3). E3s play a critical role in determining which proteins are degraded and the rate of proteolysis in the cell. One EST on the human microarray was the most strongly induced in all fasting experiments (clone 1723142, Fig. 1d
). We have cloned this gene, which we have named atrogin-1, and established that it is a muscle-specific F-box protein that functions as part of an SCF E3 complex (26)
. mRNA for cullin-1, another component of this SCF complex, was significantly increased by 48 h fasting (Cul1, Fig. 1d
). In related studies, we found that atrogin-1 mRNA levels are strongly induced in rat muscles atrophying due to cancer cachexia, diabetic acidosis, uremia (26)
, and dexamethasone treatment (R. T. Jagoe, S. H. Lecker, and A. L. Goldberg, unpublished data); recent studies by others have indicated that atrogin-1 is essential in the rapid atrophy induced by denervation (27)
. Thus, atrogin-1 and (probably) cullin 1 (as part of atrogin-1 or other SCF E3s) are likely to have a key role in the acceleration of proteolysis in fasting. In contrast, no other E3 on the chips was found to be induced. These included E3 (Ubr1), which has been shown to increase to a small degree by Northern analysis in rat muscle wasting due to fasting (28)
, diabetes (29)
, and sepsis (30)
.
Expression of many other genes involved in the ubiquitin-proteasome pathway was not significantly altered, including subunits of the 11S proteasome regulator PA28, several ubiquitin carrier proteins (E2s), and de-ubiquitination enzymes. One mRNA transcript (1.2-kb) of E214k, the E2 which functions with E3 to ubiquitinate muscle proteins in the N-end rule pathway (31)
, was previously found to increase by two- to threefold in muscle from rats fasted for 48 h using Northern analysis (32)
. However, neither of the two isoforms of E214k were differentially expressed 1.8-fold in the present study.
Genes for other proteases
Although the proteasome is responsible for most of the increased proteolysis in muscle in fasting and other types of muscle atrophy (5)
, degradation of certain classes of muscle proteins may involve other proteases such as cathepsins and calpains. Low insulin levels (as seen in fasting) can lead to an increase in intralysosomal proteolysis (i.e., autophagic vacuole formation) (33)
. Although mRNA for 11 lysosomal enzymes including cathepsins B and D did not change, mRNA for one lysosomal protease, cathepsin L, increased twofold within 24 h of food removal (see Supplementary Data, GeneList, entries 50 and 51). Other authors have recently reported that levels of cathepsin L mRNA and protein increase at an early stage in muscle wasting due to sepsis (34)
and that cathepsin L mRNA levels increased in muscle atrophy due to dexamethasone treatment and tumor cachexia (34)
.
Calpain-3 is a muscle-specific calcium-dependent protease, and mutation of this enzyme causes type 2A limb girdle muscular dystrophy (35)
. We found that calpain 3 mRNA (but not 6 other calpains) fell by 2 days’ fasting (see Supplementary Data, GeneList, entry 49). Reduced calpain-3 expression has been found in muscles in cancer cachexia (36)
and after denervation (37)
; thus its fall, like the rise in cathepsin L, seems to be a general feature of atrophying muscles.
Amino acid metabolism: glutamine synthase
The end products of protein breakdown are free amino acids, most of which are released directly from muscle, although leucine, isoleucine, and valine can be oxidized by muscle for energy, especially in fasting (12)
. Amino groups released during oxidation of the branched chain amino acids are transferred to -ketoglutarate to form glutamate, which can be further aminated by glutamine synthetase to form glutamine. Muscle releases large amounts of glutamine, which is metabolized for fuel by the intestine and kidney, where it contributes to acid base regulation. During starvation, oxidation of branched chain amino acids in muscle increases to help spare glucose (38)
and glutamine production by glutamine synthetase rises (39)
. Accordingly, mRNA levels for glutamine synthetase increased threefold by 24 h and remained high at 48 h (Fig. 2
d).
|
Glucose and glycogen metabolizing enzymes
Glycolysis
One of the most important metabolic adaptations to fasting is the sparing of glucose by skeletal muscle to ensure continued glucose supply to the brain (40)
. Due to the low insulin levels, glucose uptake via the GLUT4 transporter is reduced (41)
and there is decreased oxidation of pyruvate produced by glycolysis (18)
. Instead, the three-carbon products of glycolysis (pyruvate, lactate, and alanine) are released by muscle for use in hepatic gluconeogenesis (the Cori or alanine cycles) (12)
. One possible way by which glucose utilization in muscle might be limited during a prolonged fast is by decreased transcription of genes encoding glycolytic enzymes. Although no consistent changes were seen in mRNAs for most glycolytic enzymes, including hexokinase or the highly regulated phosphofructokinase 1, mRNAs for four enzymes that catalyze late steps in glycolysis were reduced. These included enolase and phosphoglycerate mutase, whose mRNA levels fell by 24 h of fasting (Fig. 2a
). mRNAs encoding three enzymes that act on intermediates or products of glycolysis were reduced by 48 h fasting: glycerol phosphate dehydrogenase, 2,3 bisphosphoglycerate mutase, and lactate dehydrogenase (see Supplementary Data, GeneList, entries 106, 119, 120, 125).
Enzymes for pyruvate oxidation
A key adaptation to low insulin states that favors the preservation of gluconeogenic precursors is inhibition of pyruvate dehydrogenase, which catalyzes the conversion of pyruvate to acetyl CoA in the mitochondrion (18)
. Decreased mRNA levels for two components of the pyruvate dehydrogenase complex were noted (see Supplementary Data, GeneList, entry 130). The activity of pyruvate dehydrogenase is inhibited by phosphorylation by pyruvate dehydrogenase kinase; previous studies have shown increases in mRNA and protein levels of pyruvate dehydrogenase kinase isoenzyme 4 (PDK4) in skeletal muscle during starvation (42
, 43)
and diabetes (44)
. In agreement with these results, we found a dramatic fivefold increase in expression of PDK4 by 1 day of fasting (Fig. 2b
). In one experiment, mRNAs for PDK3 were raised by fasting (Fig. 2b
). This isoform is known to have the highest specific activity, but previous studies had suggested its expression was largely restricted to the testis (45)
.
Glycogen metabolism
An additional source of glucose in muscle is stored glycogen, whose breakdown is stimulated by epinephrine and contractile activity (i.e., increased free Ca2+), which activate phosphorylase kinase (40)
. Phosphorylase kinase is a large complex made up of regulatory subunits (,ß, 
), the subunit being the Ca2+ binding protein calmodulin, and a catalytic (
) subunit that phosphorylates and activates glycogen phosphorylase b and inactivates glycogen synthase a (46)
. The continued presence of muscle glycogen even after prolonged starvation implies there are adaptive mechanisms preventing complete depletion of the muscle glycogen pool despite decreased glucose uptake (47)
. We found a reduction in mRNAs encoding three of the subunits of phosphorylase kinase (, , ) and phosphoglucomutase, which interconverts glucose-1-phosphate and glucose-6-phosphate in the last step of glycogen breakdown (Fig. 2c
). mRNA levels increased for UDP-glucose pyrophosphorylase, which catalyzes an early step in glycogen synthesis from glucose (Fig. 2c
). These complementary changes in mRNAs for enzymes involved in breakdown and synthesis of glycogen suggest adaptations that may limit the loss of muscle glycogen in fasting.
A small proportion (5%) of muscle glycogen is in the lysosomes (48)
, and lysosomal glycogen stores fall during starvation (47)
. In contrast to the changes in mRNAs that appear to favor preservation of cytosolic glycogen, expression of a lysosomal glycogenolytic enzyme, acid glucosidase , was increased 2.2-fold (see Supplementary Data, GeneList, entry 118).
Enzymes involved in fatty acid oxidation
Transport of fatty acids into muscle cells and mitochondria
As glucose oxidation falls in muscle during starvation, there is increased oxidation of free fatty acids (22)
. We were surprised to find that mRNA levels for most enzymes involved in fat metabolism were not differentially expressed, including lipoprotein lipase and long, medium, and short chain acyl-CoA dehydrogenases, which had been reported to rise in muscle of fasting rats (22
, 49)
. However, mRNAs encoding CD36 (see Supplementary Data, GeneList, entry 108), a cell membrane protein responsible for the transport of the bulk of long chain fatty acids into the cell (50)
, increased by twofold. Similar increases were seen in expression of malonyl-CoA decarboxylase (see Supplementary Data, GeneList, entry 127), which promotes mitochondrial fatty acid uptake via carnitine palmitoyl transferase (51)
. Thus, our findings suggest that the most prominent transcriptional adaptations affect genes involved in fatty acids transport into the cell and mitochondria. Our results may differ from earlier studies because the response to fasting depends on body composition (52)
, and adaptations to fasting in mice may differ from those of more obese and older rats or larger animals.
Peroxisomal metabolism of fats
The activity and number of peroxisomes increase in muscle during fasting (53)
, but it is unclear to what extent fatty acid oxidation in peroxisomes contributes to overall energy production in normal muscle. These organelles are thought to be of particular importance for metabolism of very long chain fatty acids (C22 or longer), which cannot be transported into the mitochondria via the carnitine shuttle (54)
. We found that mRNA levels for peroxisomal biogenesis factor 16, believed to be important in the synthesis of new peroxisomes, was increased by 48 h fasting. Delta 3, delta 2-enoyl-CoA isomerase, a key enzyme in peroxisomal fatty acid oxidation, was increased by 24 h (see Supplementary Data, GeneList, entries 134 and 135). These transcriptional changes agree with earlier reports of increased peroxisomal activity in muscle on fasting (53)
.
Proteins involved in mitochondrial energy production and respiration
Inner mitochondrial membrane proteins
Another adaptation to prolonged starvation is a reduction in total oxidative metabolism and basal metabolic rate (19)
. In our study, very few mRNAs encoding mitochondrial proteins involved in oxidative phosphorylation and ATP synthesis decreased on fasting. Two of those for which mRNA levels were reduced were the mitochondrial membrane protein involved in ubiquinone biosynthesis, dimethyl-Q 7, and NADH dehydrogenase (ubiquinone) (see Supplementary Data, GeneList, entries 111 and 128). Uncoupling proteins (UCP) are present in the inner mitochondrial membrane and use the proton gradient to produce heat (55)
. In accord with earlier studies, expression of UCP-3 (see Supplementary Data, GeneList, entry 143) was increased by food deprivation; in contrast, no change in UCP-2 expression was found (56)
.
Creatine kinase and carbonic anhydrase III
Fasting led to unexpected two- to threefold reductions in mRNAs for two proteins that play major roles in energy metabolism: creatine kinase, a key protein involved in buffering ATP content, and cytosolic carbonic anhydrase III, which facilitates transport of CO2 produced by oxidative phosphorylation (57)
(Fig. 2d
). These proteins are expressed at very high levels in muscle, but adaptive changes in their levels in response to fasting or other types of atrophy have not been reported before.
Myofibrillar proteins
Because muscle wasting and weakness results primarily from loss of myofibrillar proteins, any change in mRNA levels for these proteins is of particular interest. Surprisingly, expression of most myofibrillar mRNAs, including several encoding thin filament components or titin-associated proteins, did not change. However, mRNAs for four components of the thick filament were significantly reduced (Fig. 3
). Of these, the myosin binding protein H (Mybh) mRNA decreased 3.3-fold within 24 h of food deprivation and 6-fold by 48 h. The mRNA for this same gene on the human microarray (MYBH) did not decrease below the 1.8-fold threshold. Therefore, a Northern blot was performed using the mouse cDNA probe, which confirmed that myosin binding protein H mRNA falls sharply in fasting (data not shown). Myosin binding protein H is a component of the thick filament, but its function is unclear. The rapid, significant reduction in its mRNA raises the possibility that it plays a special role in the assembly of the myofibrillar apparatus, and a reduction in its synthesis may speed myofibrillar disassembly and accelerate degradation of contractile proteins. Other genes that were significantly reduced (though only at 48 h) included certain myosin light chains, embryonic myosin heavy chain genes, and an EST similar to myosin binding protein C. The magnitude of the changes in their mRNA levels was similar to the 50–80% reduction in mRNAs for myosin heavy chain and myosin light chain reported earlier (20)
. Thus, contrary to expectations, a coordinate reduction in transcription of myofibrillar proteins is not a prominent feature of this type of muscle atrophy, even though one component, myosin binding protein H, appears to be regulated particularly tightly.
|
Extracellular matrix proteins
Muscle atrophy involves primarily a loss of intracellular proteins, and possible changes in the production of extracellular matrix components during atrophy have received little attention. We were surprised to find that mRNAs for a number of extracellular matrix proteins, including several collagens and fibronectin, were suppressed by fasting (Fig. 4
). In a recent report, muscle disuse led to reduced mRNA for procollagen, type III, alpha 1 (Col3a1), and a decrease in collagen III protein levels (58)
. These adaptations suggest a general reduction in synthesis of matrix proteins as the muscles atrophy; similar effects have been found in other types of muscle wasting (unpublished observation).
|
Muscle cell growth and survival
Insulin-like growth factor binding protein 5
Insulin-like growth factor binding protein 5 (IGFBP-5) is one of a family of extracellular proteins that bind and modulate the activity of insulin-like growth factors 1 and 2 (IGF-1, IGF-2), which are key stimuli for muscle growth (59)
. IGFBP-5 is expressed early in myogenesis (60)
and is secreted from muscle cells, where it binds to the extracellular matrix and stabilizes IGF and enhances growth (61)
. We found that mRNA levels for IGFBP-5 were strongly reduced (3- to 4-fold) by 24 h fasting (see Fig. 6
), which is likely to favor the atrophy process.
|
Regulators of transcription
Some of the differentially expressed genes were themselves regulators of transcription, which might cause gene-specific transcriptional adaptations or the general fall in cellular RNA. The changes in mRNAs for genes likely to influence overall transcription rates did not follow a consistent trend; some general repressors and activators were found to rise and some to fall (Fig. 5
a). Changes in expression of three histones including linker (H1) and core (H2, H3) histones were induced by fasting (Fig. 5a
). Linker (H1) histones help maintain higher order chromatin structure and inhibit movement of nucleosomes (62)
; mRNA levels for one of these (H1f2) was increased by 24 h fasting. One of the core histones whose expression fell during fasting (H2afz) is essential for chromatin integrity and viability in yeast, where it is involved in recruitment of RNA polymerase II and TATA binding protein (63)
.
|
Several transcription factors that bind to specific promoter sequences were differentially expressed after fasting. Of particular interest was the fall in mRNA for myocyte enhancer factor 2C (MEF2C) (Fig. 5a
). The MEF family of transcription factors functions in terminal differentiation of myocytes and skeletal muscle growth (64)
and in combination with other factors (65)
, MEFs control the expression of several muscle-specific proteins, including contractile proteins (66)
. Thus, the fall in MEF2C mRNA may account for the reduction in mRNAs for certain contractile proteins noted above. Also reduced were mRNAs of growth or differentiation-inducing transcription factors such as JUNB and MAF (Fig. 5a
). However, mRNA levels for some transcription factors increased on fasting, including forkhead box O1 (Foxo1) and heterochromatin protein 1 binding protein 3 (Hp1bp3). Although intriguing, the physiological significance of these transcriptional adaptations is presently unclear, since the targets of these factors and their interactions with other coactivators or suppressors are unknown.
Translation initiation
In fasting, protein synthesis falls rapidly (14
, 67)
, mainly through reduced initiation (16)
. This is believed to be due to reduced activity of the Akt/mTOR pathway, which in the presence of growth factors and nutrients stimulates translation by enhancing 5' cap-dependent initiation (68)
. Formation of the initiation complex at the 5' cap requires initiation factor eIF4E; one mechanism by which initiation is enhanced by the Akt/mTOR pathway is by phosphorylation of the inhibitory eIF4E binding protein, which reduces its affinity for eIF4E. Earlier evidence for transcriptional regulation of the translation initiation factors is lacking. Fasting for 24 h caused a twofold increase in mRNAs for the inhibitory eIF4E binding protein (EIF4EBP1, Fig. 5b
), which presumably serves to reinforce the inhibition of cap-dependent translation resulting from inactivation of Akt/mTOR. Some mRNAs, particularly those translated under stress conditions, have internal ribosome entry sites enabling cap-independent initiation (69)
. Fasting for 48 h induced increases in mRNA for the initiation factors eIF2B, eIF3, and eIF4A (Fig. 5b
). An increase in the expression of these initiation factors, but not eIF4E, may be important during fasting to preserve the cap-independent translation of a particular subset of mRNAs essential for cell survival.
Proteins associated with apoptosis and/or proliferation
Although it is widely assumed that apoptosis does not occur in skeletal muscle, in muscles atrophying due to denervation and heart failure (70
71
72)
, and in limb girdle muscular dystrophy (73)
, apoptotic nuclei are frequently observed, and mRNAs for proapoptotic proteins Bcl-2 and caspase-3 are increased. Curiously, in fasting there were clear increases in mRNAs for proteins that have proapoptotic (BaK1 and Bnip3) and antiapoptotic (PEG3 and BCl2l) functions.
Similarly, extensive changes occurred in the expression of genes that influence cell growth and cell cycle progression. For example, mRNA levels of Cdkn1/P21, which inhibits proliferation, increased fourfold by 24 h fasting; over the same time period, mRNA encoding S-adenosylmethionine decarboxylase 1 (AMD1), a key enzyme in polyamine synthesis and nucleic acid synthesis, was reduced two- to fourfold. On the other hand, two proteins believed to have antiproliferative functions had reduced mRNA levels: Tob1, the transducer of ErbB2.1, and Dusp3 EST, a cDNA highly similar to dual-specificity phosphatase 3, which negatively regulates members of the mitogen-activated protein kinase superfamily. More surprising, mRNA for Akt1, whose phosphorylation mediates insulin and IGF-1 actions and muscle hypertrophy (74)
and which is dephosphorylated in serum-starved cells, increased by 48 h fasting (Fig. 6
). One possible explanation for these seemingly contradictory changes in expression of genes related to apoptosis and growth suppression is that different cells or nuclei within fasting muscle respond differently (e.g., nuclei in satellite cells and those in muscle cells). Alternatively, some of these changes may appear surprising because our understanding of the protein’s function is incomplete.
Cytoskeleton-related proteins
By definition, muscle atrophy involves a reduction in the size of muscle cells, primarily through a loss of contractile proteins. However, a possible role for noncontractile cytoskeletal proteins in the reduction in size of muscle cells has not been studied. Microtubules could play a role in the accelerated proteolysis characteristic of atrophying muscles. Inhibitors of microtubule assembly reduce protein degradation in heart muscle (75
, 76)
and, under certain conditions, microtubules may help transport protein substrates to proteasomes. Although no change was seen in the expression of several major cytoskeletal proteins (e.g., tubulin, cofilin), mRNAs for many cytoskeletal-associated proteins were differentially expressed (Fig. 7
). Filamin and ankyrin, which anchor membrane proteins to the actin cytoskeleton, were up-regulated, as were some microtubule-associated proteins, including Map1lc3. On the other hand, there was reduced expression of some mRNAs for other structural proteins such as paladin, which colocalizes with -actinin in stress fibers.
|
Metallothionein
Among the unexpected findings were a large (up to 6-fold) increase in metallothionein I mRNA levels on fasting (see Supplementary Data, GeneList, entries 173 and 174). The expression of metallothionein is induced by a variety of stressful conditions and appears to reduce cellular damage. For example, recently overexpression of metallothionein was shown to reduce the development of diabetic cardiomyopathy in mice (77)
. Metallothioneins bind various metals (e.g., cadmium, mercury, copper, and zinc), which can induce metallothionein gene expression via specific metal response elements in the promoter. Perhaps increased metallothionein expression in atrophying muscle may result from the need to detoxify metals released in the breakdown of muscle proteins such as myoglobin. However, other roles for metallothioneins have been proposed, such as modulating the activity of enzymes by donating or binding zinc (78)
and maintaining levels of reduced glutathione (77
, 79)
. The marked induction of metallothionein in fasting and other types of atrophy (unpublished results) may be triggered by glucocorticoids, which are potent inducers of this protein (80)
and essential for the weight loss and acceleration of muscle proteolysis in fasting (3)
.
Proteins involved in excitation-contraction coupling
Nerve impulses to muscle lead to membrane depolarization that is sensed by the voltage-dependent dihydropyridine receptor (DHPR) in the transverse tubules. The DHPR then triggers release of Ca2+ from the endoplasmic reticulum (ER) though the ryanodine receptor, resulting in contraction of the myofibrils (81)
. One subunit of the DHPR (the ß1 subunit, Cacnb1) is required for assembly and targeting of the DHPR complex to the triad junction and for excitation-contraction coupling (82)
. The role of the subunits of the DHPR complex is less well understood, but studies with one isoform (1) have shown that it limits the L-type Ca2+ current, thus reducing the influx of extracellular Ca2+ (83
, 84)
. We found that mRNAs for several of these components of the excitation-contraction apparatus were differentially expressed. mRNAs for the ß1 subunit of the DHPR (Cacnb1), anchoring proteins at the triad junction (Trdn, Mg29), and calcium binding proteins in the ER (Casq2) and cytosol (PVALB) were reduced, whereas expression of the 6 isoform (Cacng6) was increased (Fig. A, see Supplementary Data). These coordinate transcriptional adaptations would appear to favor inhibition of Ca2+ influx and ER Ca2+ release by the DHPR. In combination with the reduction in mRNAs encoding parvalbumin (PVALB) in fast fibers, these early changes might lead to slowing of the twitch characteristics of the muscle and may reflect the selective atrophy of fast fibers during fasting.
Highly induced or suppressed genes with unknown function
Of 315 unique genes that were differentially expressed by fasting, 146 were incompletely characterized or had no known function in muscle (see GeneList, Supplementary Data). Included in this group were some of the most highly induced genes, such as atrogin-1 (see above and Fig. 1d
), and myeloid differentiation primary response gene 116, which increased four- to fivefold after 24 h fasting (see Supplementary Data, GeneList, entry 175). Expression of other partially characterized genes was strongly suppressed by fasting, including a putative growth-related protein, P311 (85)
, whose mRNA levels fell by eightfold (see Supplementary Data, GeneList, entries 254–257).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 34
, 58
, 86
, 87)
. Though some of the changes described here were seen in previous genomic analyses of other types of atrophy (34
, 58
, 87)
, our more systematic analysis has defined some new transcriptional adaptations and uncovered intriguing trends. In this study, we have attempted to increase the power of our analysis by using microarrays generated from human and mouse cDNA. Nevertheless, many genes of interest were not present on either array, and further studies using whole genome arrays will undoubtedly extend or test the trends uncovered here.
Despite a complete lack of nutrients and the rapid weight loss, expression of only a small subset of muscle mRNAs (2–4% of the analyzable genes) was altered (Table 2)
. Roughly half of these mRNAs increased, even though the total RNA content in the muscles fell after food deprivation, and the total amount of mRNA appeared to fall sharply by the second day (Table 1)
. Thus, the adaptation of muscle to reduced food intake does not involve simply a general suppression of transcription or a selective reduction in the expression of a particular subset of growth-related genes. Instead, food deprivation appears to elicit a specific program of changes in gene expression, positive and negative, which became more pronounced with time (i.e., changes found at 24 h were generally smaller and less numerous than those at 48 h). Our earlier studies and those of others have emphasized that in experimental models, atrophy proceeds mainly through enhanced degradation of muscle proteins (see below) and that protein synthesis, though decreased in most types of atrophy, contributes to a lesser extent to the rapid loss of tissue protein (67)
. These conclusions were based on an analysis of overall protein balance in the atrophying muscles. However, even if synthesis of most proteins falls in fasting, the expression of certain regulatory proteins may increase (e.g., to suppress protein synthesis to induce disassembly of myofibrils or enhance the cell’s proteolytic capacity). The complex pattern of the transcriptional changes found here in fasting clearly indicates a coordinated modulation of diverse cellular processes; as discussed below, we have attempted to identify which of these changes can contribute to the rapid reduction in muscle mass, to the conservation of metabolizable fuels, especially glucose, or to other adaptations that enhance survival in the absence of exogenous nutrients. By comparisons of these patterns with the transcriptional changes in other conditions where there is rapid muscle wasting but normal food intake, we have been able to identify a subset of changes specifically associated with muscle atrophy (unpublished results).
Adaptations likely to favor net loss of muscle protein
Because activation of the ubiquitin-proteasome pathway plays a key role in the atrophy process, we were especially interested in obtaining a more complete analysis of components of this pathway in muscles from fasted mice (1
, 6)
. Several important results have emerged from this analysis. In response to fasting, the muscles showed rapid early increases in mRNA not only for polyubiquitin, but for the cell’s two ubiquitin extension proteins (Fig. 1a
), which had not been known to be regulated. It had been assumed that the two ubiquitin extensions were expressed constitutively and served primarily as chaperones in ribosome assembly as well as a prime source of ubiquitin during growth (88)
. Enhanced expression of these Ub extension genes along with polyubiquitin in fasting when there is a decrease, not a rise, in ribosome production suggests that all the cell’s genes for ubiquitin are expressed concomitantly so that ubiquitin levels do not become limiting when proteolysis is accelerated.
The muscles in fasting showed marked increases in mRNA for 7 of 13 subunits of the 20S proteasome and 10 of 17 subunits of the 19S regulator (Fig. 1b, c
). However, the levels of mRNA for several proteasome subunits clearly did not change in these atrophying muscles. Because proteasomes have a fixed subunit composition and the production of new particles must involve the assembly of many subunits in specific stoichiometric ratios, the increased expression of mRNAs for only some 19S and 20S subunits is surprising. The identities of the subunits that are most highly induced do not reflect any obvious functional or structural subassemblies, but perhaps the expression of certain subunits is highly regulated because their levels are rate-limiting for proteasome assembly. In any case, these findings strongly suggest that post-transcriptional mechanisms are important in regulating production of new proteasomes. mRNA levels for the 11S regulator/PA28 subunits were not differentially expressed on fasting. This activator is induced by -interferon and increases in muscle in some catabolic states (89)
, perhaps only in those states where -interferon levels are high (e.g., sepsis or cancer).
At the sensitivity limit we chose, levels of mRNA for many ubiquitin-conjugating enzymes were not found to change, with the notable exception of the E3, atrogin-1, which increased 10-fold after food deprivation (Fig. 1d
). Atrogin-1 thus appears to be uniquely regulated or at least to be one of only a small set of ubiquitin-protein ligases that play a key role in the accelerated proteolysis in atrophying muscles (26)
. Another E3, MuRF1, has recently been found to be up-regulated markedly and to be important in disuse atrophy (27)
. Although MuRF1 was not contained on our microarray, it appears to be induced severalfold in the muscles from food-deprived mice (S. H. Lecker, M. Gomes, and A. L. Goldberg, unpublished results). We had previously noted an important role of the N-end rule and the Ub ligase E3 in Ub conjugation to soluble proteins; in the present study, however, any possible change in E3 (Ubr1) did not reach statistical significance.
One surprising finding was that the lysosomal cysteine protease cathepsin L, but not other lysosomal enzymes (e.g., cathepsin B and D), was induced in our study. Earlier studies have minimized the importance of lysosomal proteases in the atrophy process, because inhibitors of lysosomal proteolysis reduce protein degradation rates only slightly in normal or atrophying muscle (90)
, in contrast to the marked inhibition by proteasome inhibitors (5)
. Recent studies have demonstrated a strong induction of cathepsin L in muscles from septic and tumor-bearing animals (34)
. Thus, this enzyme may play an important, though as yet undefined, role in the atrophy process, though probably not as part of lysosomes. In the lung, cathepsin L is released by alveolar macrophages and is thought to contribute to smoking-related lung damage by cleaving extracellular matrix proteins (91)
. Therefore, it is possible that this enzyme is involved in the breakdown of critical extracellular proteins in atrophying muscles. In fasting and other types of muscle atrophy (unpublished results), there was an unexpected, significant reduction in the expression of many collagens and extracellular components. These findings suggest a general remodeling of the extracellular matrix, although the significance of this new feature of the atrophy process remains unclear.
The accelerated degradation of myofibrillar proteins, which usually comprise 50–70% of muscle protein, accounts for most of the increase in proteolysis on fasting (13)
. Yet it is still unclear how parts of the contractile apparatus are dismantled and individual components hydrolyzed. It is likely that reduced production (or selective breakdown) of certain key myofibrillar components in an atrophying muscle may promote net disassembly of myofibrillar proteins. Our findings suggest a possible candidate for such a key component, myosin binding protein H, a small component of the thick filament whose mRNA levels were reduced threefold by 24 h. In related studies, a similar decrease in expression of this gene was found in muscle atrophy due to uremia and tumor cachexia (R. T. Jagoe and S. H. Lecker, unpublished observation). By contrast, there was no change in expression of components of the thick filaments and only small changes in subunits of the thick filament.
Calpains have often been proposed to play a key role in atrophy. For example, clipping of myofibrillar proteins by calpains has been proposed to release fragments for subsequent degradation in atrophy due to sepsis (92)
. However, in fasting and a variety of other wasting conditions, inhibition of the calpains in isolated muscles had no effect on overall proteolysis or myofibrillar protein degradation (3)
. In fasting as well as other catabolic states (36
, 37)
, expression of calpain 3 (but not other calpains) is actually reduced. Therefore, this enzyme may in some way play a role as a negative regulator of muscle protein loss.
Another intriguing adaptation that may be critical in the net loss of muscle mass was the suppression of expression of IGF binding protein 5, a key factor released by muscle cells that binds circulating IGF-1. This hormone enhances overall protein synthesis and reduces proteolysis in skeletal muscle; it is required not only for normal postnatal growth of muscle, but functions in an autocrine growth stimulant in work-induced hypertrophy (93)
. Because IGF binding proteins are likely to determine the peripheral sensitivity to IGF-1 (61)
, it seems likely that the fall in IGF binding protein 5 may reduce the anabolic effect of circulating IGF-1 as well as the IGF-1 released by muscle cells that function as an autocrine growth factor.
In fasting and some other cachectic states, there is a marked fall in protein synthesis that contributes to the loss of muscle protein and helps spare amino acids, which can be used for gluconeogenesis. Decreased translational efficiency seems to result from a loss of the stimulation of initiation by insulin, which acts largely by causing phosphorylation of eIF2. The present studies uncovered an unexpected increase in mRNAs for several components of the initiation complex, especially the inhibitory factor eIF3 binding protein. As overall translation falls, these changes in gene expression could favor continued translation of certain mRNAs through cap-independent initiation at internal ribosome binding sites, a process that seems to be particularly important in various stressful conditions (69)
.
Muscle adaptations that may help preserve glucose
These studies uncovered many other interesting changes in transcription in muscle that warrant in-depth study, since they are likely to be important in the conservation of body energy reserves. The dramatic increase in mRNAs for pyruvate dehydrogenase kinase 4, as well as coordinate decreases in mRNAs for four late enzymes in glycolysis (none of which had been previously known to be key regulatory points), are likely to be important in glucose sparing by muscle. These adaptations, especially the large induction of this kinase, a key inhibitor of pyruvate dehydrogenase, probably contribute to the decreased oxidation of pyruvate in mitochondria and favor the release from muscle of pyruvate, lactate, and alanine, the primary substrates for hepatic gluconeogenesis in fasting.
The marked induction of glutamine synthase appears to be important in energy conservation, since glutamine production enables muscle to burn certain energy-rich amino acids (e.g., leucine) as an alternative to glucose. Similarly, the increase in expression of genes involved in muscle uptake of fatty acids and in the peroxisomal oxidation of long chain fatty acids probably contributes to glucose sparing and the preferential utilization of lipids as alternative fuels. Despite the decreased glucose uptake and decreased metabolism of muscle glycogen in fasting, we found coordinate changes in the expression of genes for glycogen metabolism (e.g., decreased expression of several enzymes for glycogen breakdown), which are of particular interest as they may account for the maintenance of muscle glycogen in fasted organisms.
Limits of this analysis
In analyzing the results of the microarrays, we used ±1.8 as the threshold for defining a change in expression as statistically significant, which defines the sensitivity of the analysis. This stringent threshold is based on the manufacturer’s detailed analysis of the sensitivity and reproducibility of these cDNA microarrays (23)
, though more recent evidence suggests that thresholds as low as ± 1.4 can be used. We elected to continue to use the more stringent 1.8-fold criteria because some of our experiments were done using early builds of the mouse and human microarrays. As a result, we may have significantly underestimated the number of differentially expressed genes. Significant changes in gene expression of the magnitude uncovered here (3- to 10-fold) have seldom been reported in fully differentiated muscle. Therefore, the changes in expression of these highly regulated mRNAs are likely to represent physiologically important adaptive responses. However, smaller changes in the levels of some highly expressed mRNAs may have broad effects on the composition of muscle. For example, myosin heavy and light chains and skeletal -actin are major muscle components; earlier Northern analysis of muscles in rats showed a reduction of 55–60% in their levels by three days fasting (20)
. In our study, these mRNAs appeared to fall in 2 days, but by <1.8-fold.
The changes in mRNA levels uncovered here, of course, may not reflect changes in the cell’s content of the corresponding proteins. The synthesis of some myofibrillar proteins appears to be predominantly regulated at the level of translation (94)
, and changes in the concentrations of such mRNAs may have little effect on the production of these major cell constituents. Therefore, in-depth studies of changes in the content of these various intracellular and extracellular proteins during fasting, especially changes in the products of the highly regulated genes, are particularly important and a challenge for future studies. It should be noted that in atrophying muscles, where overall rates of proteolysis are accelerated, decreases in synthesis and enhanced proteolysis are likely to have synergistic effects on levels of many proteins. Similarly increased expression of a protein in an atrophying tissue (e.g., ubiquitin or proteasome subunits) may just be a mechanism to maintain the high levels of these proteins in the face of rapid proteolysis.
The present studies of muscle illustrate several ways that the study of global patterns of transcription by gene microarrays can lead to novel biochemical and physiological insights. 1) Grouping coordinately regulated genes with related functions can suggest aspects of a physiological response previously unexpected. In fasting, for example, we have found surprising reductions in the expression of genes for extracellular matrix components and complex changes in mRNAs for proteins involved in excitation-contraction coupling, as well as certain cytoskeleton-related proteins, none of which had been linked to atrophy or fasting. 2) Discovering tight transcriptional regulation of an unknown or uncharacterized gene can lead to identification of new proteins with major roles. For example, atrogin-1 first came to our attention as an EST, which was the most highly regulated gene on the microarray in fasting and other catabolic states (26)
. The present study has uncovered many other ESTs and unknown full-length cDNAs that are strongly induced or suppressed in fasting. Their tight regulation suggests they are important enough to merit in-depth study. 3) Discovery of unsuspected expression patterns of known gene products can suggest novel functions—for example, myosin binding protein H or metallothionein, whose dramatic induction in fasting and in all other forms of atrophy examined (unpublished results) is hard to reconcile with a simple role in metal binding. 4) Transcriptional profiling can provide insight into key transcriptional networks. For instance, the intriguing fall in mRNA for MEF2C (Fig. 5a
) that controls the expression of several muscle-specific genes may be responsible for the reduction in mRNAs for myosin light chain and myoglobin. Similarly, the discovery that translation initiation factors are regulated at the level of gene expression is new and unexpected. 5) This type of global analysis can serve to clarify the responses to different circulating factors: the enhanced proteolysis in fasting results at least in part from glucocorticoid-mediated changes in gene expression and the fall in insulin levels (3)
. In fact, several genes shown to be up-regulated here have long been known to be induced by adrenal steroids in cultured cells (e.g., glutamine synthase, metallothionein). Glucocorticoids are essential for activation of proteolysis in several other catabolic states (95
96
97)
. It will be interesting to learn whether many of the changes in mRNA observed here are seen in adrenalectomized animals upon fasting or may be increased by administration of these hormones to fed animals. Presumably, many of the other changes are due to decreases in the insulin and IGF-1 levels characteristic of fasting (98)
. The importance of these factors (and others) in signaling these various atrophy-inducing and energy-conserving adaptations can now be tested. 6) Finally, this analysis of fasting has provided a basis for detailed comparisons with transcriptional changes occurring in muscle atrophy induced by other systemic diseases (i.e., cancer cachexia, renal failure, diabetes). We are conducting such a systematic analysis that will allow us to distinguish a distinct subset of genes whose expression is altered in all these catabolic states, which we have termed atrophy-specific genes or atrogins.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication April 24, 2002. Accepted for publication June 21, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-1.8), on two occasions only. Red or darker green indicate that differential expression was observed in all experiments or in only two but with reversed fluors (see Materials and Methods). Black shading indicates no data. Where possible, standard (Human Genome Nomenclature Committee or mouse equivalent) abbreviations have been adopted. A fuller description of the gene name is given in Supplementary Data, GeneList.
