|Home | About | Journals | Submit | Contact Us | Français|
Muscle wasting (cachexia) is a consequence of chronic diseases, such as cancer, and is associated with degradation of muscle proteins such as MyoD. The cytokines tumor necrosis factor alpha and gamma interferon induce muscle degeneration by activating the transcription factor NF-κB and its target genes. Here, we show that a downstream target of NF-κB is the nitric oxide (NO) synthase gene (iNos) and suggest that NO production stimulates MyoD mRNA loss. In fact, although cytokine treatment of iNos−/− mice activated NF-κB, it did not trigger MyoD mRNA degeneration, demonstrating that NF-κB-mediated muscle wasting requires an active iNOS-NO pathway. The induced expression of iNOS by cytokines relies on both transcriptional activation via NF-κB and increased mRNA stability via the RNA-binding protein HuR. Moreover, we show that HuR regulates iNOS expression in an AMP-activated protein kinase (AMPK)-dependent manner. Furthermore, AMPK activation results in HuR nuclear sequestration, inhibition of iNOS synthesis, and reduction in cytokine-induced MyoD loss. These results define iNOS and HuR as critical players in cytokine-induced cachexia, establishing them as potential therapeutic targets.
Cachexia (muscle wasting) is a condition that leads to the alteration of several physiological and behavioral attributes, ranging from fatigue and fever to excessive weight loss (32). Unlike anorexia, where weight loss is due to a decrease in fat content, the detrimental effects of cachexia occurs as a consequence of excessive wasting of skeletal muscle tissue (55). This massive effect appears to be mediated by a strong activation of the ubiquitin- and proteasome-dependent pathways and an increase in the rate of protein decay, ultimately resulting in death (54). Under these diverse conditions, the expression of many markers of differentiated skeletal muscle cells, such as the myogenic bHLH transcription factor family (MyoD, Myf5, myogenin, and MRF2), is largely affected, resulting in muscle atrophy (1, 23). However, under normal conditions, these transcription factors implement a strict control on the skeletal muscle differentiation program and participate in the maintenance of differentiated fibers (44).
It is well established that muscle atrophy requires the activation of transcription factors such as NF-κB (10) and Foxo-3 (50), leading either to the rapid decrease of MyoD mRNA (23) or to the overexpression of the ubiquitin ligase, atrogin-1 (50). It was demonstrated that NF-κB-dependent muscle wasting is activated by late-stage cancer or by chronic inflammation; however, Foxo-3/atrogin-1 pathway induces atrophy upon food starvation (40). Furthermore, NF-κB stimulates the proteasome machinery by activating the muscle specific E3 ligase, MuRF1, leading to protein decay and muscle collapse. The depletion of the Murf1 gene, however, does not fully protect mice against muscle wasting, suggesting that NF-κB also triggers atrophy by activating other critical, as yet unidentified pathways.
One of the main activators of the NF-κB pathway is the cytokine tumor necrosis factor alpha (TNF-α) (8, 23, 28), which was shown to play a role in muscle degeneration. However, several studies have suggested that TNF-α needs other factors to promote massive protein loss. Gamma interferon (IFN-γ) has been demonstrated to potentiate TNF-α effects in many cell types, including muscle cells (23, 34, 45). Therefore, it is likely that transcription factors activated by these two cytokines induce common genes encoding the downstream effectors. In spite of the recent progress in delineating the role of TNF-α and IFN-γ in cachexia, the downstream effectors and their implication in the process are still elusive.
NF-κB regulates the expression of a wide variety of genes, including those encoding cytokines, chemokines, adhesion molecules (e.g., ICAM, VCAM, and E-selectin), and inducible effector enzymes (e.g., inducible nitric oxide synthase enzyme [iNOS] and COX-2) (20). It was previously shown that TNF-α induces the expression of iNOS, leading to the production and release of nitric oxide (NO) and, thus, oxidative stress in the skeletal muscles of cachectic animals (8, 59). The implication of iNOS-NO in TNF-α-induced muscle wasting was demonstrated by the use of a specific NO inhibitor, which was shown to prevent the onset of cachexia in nude mice injected with CHO cells expressing TNF-α (8). NO has several important biological roles, such as mediating the ability of a host to defend itself from microbial pathogens (42, 43, 59). The release of NO by different cell types depends on the transcription of the iNos gene. The murine and human iNos promoters contain several binding sites for transcription factors such as NF-κB, Jun/Fos heterodimers, and some members of the C/EBP-, CREB-, AP-1-, and IFN-γ-activated STAT family (3). Regulation of iNOS via the NF-κB pathway is an important mechanism in inflammatory processes and constitutes a potential target to combat inflammation-related disease.
Although transcriptional regulatory mechanisms are important for the expression of cytokine-inducible genes such as iNos and Cox2, transcription alone does not explain the dramatic increase in the production of these proteins since the expression levels of the messages encoding them remain modest. It was observed that cytokines induce the transcription rate of iNos two- to fivefold, whereas the levels of iNOS mRNA increase 20- to 100-fold (12, 37), consistent with posttranscriptional regulatory events. A classical destabilizing sequence, the AU-rich element (ARE), which regulates the expression of the human iNOS mRNA, was found in its 3′ untranslated region (3′UTR). This ARE, with a typical repeat of the pentamer AUUUA, was identified as the binding sequence for the stabilizing RNA-binding protein HuR (48). The disruption of HuR's expression using antisense oligonucleotides led to the rapid decrease of iNOS mRNA in cytokine-treated human adenocarcinoma DLD-1 cells (48). In general, ARE-containing mRNAs are labile, ensuring they are maintained within critical expression levels to allow proper cell growth, differentiation, and response to external stimuli (5, 6, 27). HuR has been shown to affect the expression of these short-lived mRNAs, regulating their cellular turnover and nucleocytoplasmic movements (14, 19, 29, 30).
To define the role of cis-acting elements, such as AREs, in association with their RNA-binding proteins such as HuR in the expression of inducible cachectic effectors, we used a cytokine-treated muscle cell line model. Using genome-wide cDNA microarray screening, we observed that iNOS mRNA is one of the main messages that are induced upon cytokine treatment of myotubes. Our results suggest that iNOS is a muscle-wasting inducer both in vitro and in vivo that triggers the loss of MyoD mRNA. Our observations reveal new insight into the molecular mechanisms by which muscle atrophy occurs.
The following reagents were used: actinomycin D and aminoguanidine were obtained from Sigma, IFN-γ and TNF-α from R&D Systems, FeTPPS from Calbiochem, and AICAR (5-aminoimidazole-4-carboxamide 1-β-ribofuranoside) from Toronto Research Chemicals.
C2C12 cells (ATCC) were grown and induced for differentiation as previously described (56). HcNeoSR (stably expressing a phosphorylation-defective mutant of human IκΒα) as well as the vector control transfected HcNeo cell lines, generously provided by D. Guttrigde, were generated and maintained as previously described (23). On the third day of differentiation, cells were treated with or without 100 U/ml IFN-γ and 20 ng/ml TNF-α for the indicated periods of time. When indicated, aminoguanidine (100 μM), FeTPPS (50 μM), or AICAR (2 to 4 mM) was added to cells simultaneously in the presence or absence of IFN-γ and TNF-α.
Microarray experiments were performed at the UAlbany Center for Functional Genomics Microarray core facility (Rensselaer, NY) as previously described (52) using Affymetrix Mouse MU74Av2 oligonucleotide arrays and following the standard Affymetrix protocol (see the manufacturer's instructions). Briefly, total RNA (5 μg) was first converted to single-stranded cDNA by use of Superscript II reverse transcriptase (Invitrogen) and a GeneChip T7 promoter primer kit. The single-stranded cDNA was then converted to double-stranded cDNA by use of DNA polymerase I, DNA ligase, and RNaseH from Escherichia coli. The double-stranded cDNA was purified using a cleanup kit from Affymetrix and converted to cRNA by in vitro transcription using biotinylated ribonucleotides and T7 polymerase (ENZO Bioarray High-yield RNA transcript labeling kit). The labeled cRNA was subsequently purified and fragmented by metal-induced hydrolysis to yield 35 to 200 base fragments that can be hybridized to the Affymetrix Mouse Genome MU74Av2 oligonucleotide array. After hybridization, the chip was washed and stained with streptavidin-phycoerythrin before being scanned. An antibody amplification staining protocol that used biotinylated goat immunoglobulin G (IgG) followed by a second streptavidin-phycoerythrin staining increased the sensitivity of the assay. The chip was then scanned, and images were analyzed qualitatively using Affymetrix GCOS software. Further quantitative data analysis to identify changes in gene expression patterns was done using GeneSpring (Silicon Genetics). The normalizations used include “per chip” and “normalize to control samples” procedures. Samples were classified into three groups for analysis based on time of exposure to IFN-γ and TNF-α. Data was filtered to obtain lists of genes with good signal-to-noise ratios. These genes were further filtered to include only genes that were present or marginally present in two of three samples. Analysis of variance with a 95% confidence interval was done to compare the different time points. No multiple testing corrections were performed. The statistically significant genes were then filtered to obtain lists based on expression levels (i.e., twofold up or down, etc.).
Total and nuclear, as well as cytoplasmic, cell extracts were prepared as previously described (13, 56). Western blots performed as previously described (7, 56) were probed with the following antibodies: anti-HuR (17), anti-MyoD, antitubulin, anti-IκΒα (Santa Cruz), anti-iNOS (BD Biosciences), anti-Cox-2 (Caymen Chemicals), and anti-MF-20 (Developmental Studies Hybridoma Bank).
Immunofluorescence was performed as previously described with an anti-HuR antibody diluted 1:1,500 in 1% goat serum-phosphate-buffered saline and a fluorescein isothiocyanate-labeled secondary antibody diluted at 1:500 (17). Nucleus detection was obtained by staining of cells with DAPI (4′,6′diamidino-2-phenylindole) for 15 min after incubation of cells with the secondary antibody. HuR localization was then analyzed using a Zeiss Axiovision 3.1 microscope (40× oil objective) and an Axiocam HR (Zeiss) digital camera.
Quantification of NO in media secreted by both untreated cells and cells subjected to IFN-γ treatment (IT) and TNF-α treatment was performed using GREISS reagent as previously described (41).
Northern blot analyses were performed as previously described (60) using 10 μg of total RNA extracted from cells with TRIzol reagent (Invitrogen). Upon transfer of the RNA to a Hybond-N membrane, blots were hybridized with [32P]dCTP-labeled probes. After hybridization, blots were washed and exposed to Biomax films.
Immunoprecipitation of HuR and the subsequent extraction of RNA pulled down with the anti-HuR antibody was performed as previously described (53, 56). An anti-IgG antibody was included in the experiments as a negative control. Briefly, cell extracts from untreated as well as IFN-γ- and TNF-α-treated myotubes were incubated with either anti-HuR or anti-IgG antibody preincubated with protein A-Sepharose beads. The RNA transcripts associated with HuR were analyzed by reverse transcriptase PCR (RT-PCR) as previously described by use of iNOS (26)- and MyoD (47)-specific conditions and primers. The sequences of the Cox-2 primers used in the study are as follows: forward, 5′-GGC GGA TCC GCT GTA AAA GTC TAC TGA CCA-3′; reverse, 5′-GGC AGA TCT AAC TTG GAC CCC TTT GTT TG-3′. PCR products were then analyzed on a 2% Tris-acetate-EDTA agarose gel.
The iNOS cRNA was generated from a synthetic oligonucleotide (spanning the last 39 bases of the iNOS 3′UTR; ASSC BC062378) fused to a T7 promoter (5′-GAATTGTAATACGACTCACTATAGGGCGA-3′) by an in vitro transcription reaction as previously described (18). Gel-shift/supershift assays were then performed using the [32P]UTP-labeled iNOS cRNA probe incubated with C2C12 total cell extract and an anti-HuR antibody (for detection of complexes containing HuR) as previously described (13). An antivimentin antibody was included in the supershift assays as a negative control.
C2C12 myoblasts were transfected with either the control (HuSi-C) or the HuR (HuSi-1) small interfering RNA (siRNA) when cells were 20 to 30% confluent as previously described (56). The procedure was repeated 24 h later. Transfected cells were then treated 8 h after the second transfection with or without IFN-γ and TNF-α for an additional 12 to 24 h. RNA or proteins were subsequently extracted from the cells in order to assess the effect of HuR on iNOS mRNA and protein levels. In order to rescue endogenous HuR protein levels in the knockdown cells, 50 nM of a recombinant AP (antennapedia cell-permeable peptide)-HuR-glutathione S-transferase (GST) protein was added to the mock- or siRNA-transfected cells 6 h prior to stimulation of cells with IFN-γ and TNF-α. An AP-GST recombinant protein was used as a negative control in the rescue experiments.
The stability of the iNOS mRNA was assessed by the addition of 5 μg/ml of actinomycin D (a transcriptional inhibitor) to the treated and untreated cells for the indicated periods of time. iNOS mRNA levels were then determined by Northern blot analysis as described above.
This study was approved by the Animal Care and Handling Committee of Laval University, where animals were cared and handled in accordance with the Canadian Guide for the Care and Use of Laboratory Animals. Breeding pairs of iNOS wild-type mice (C57BL/6J) and iNOS knockout mice (C57BL/6-NOS2tm1Lau) were purchased from Jackson Laboratories (Bar Harbor, Maine). The mice were bred under pathogen-free conditions at the animal facility of Laval University Hospital research center. Three-week-old mice were weaned and housed in pairs in plastic cages in a room kept at 23 ± 1°C with a 12-h light/12-h dark cycle and had free access to water and food (sterilized Global 18% Rodent; Harlan Teklad, Madison, Wisconsin). The proportions of calories derived from fat, carbohydrates, and protein were 14%, 64%, and 21%. Experiments were carried out with 10- to 12-week-old mice as described previously (23) with the following modifications. At time 0, the gastrocnemius muscle was injected intramuscularly with 30 μl of a solution containing 5,000 U of IFN-γ and 2 μg of TNF-α or with 30 μl of saline. Injections were repeated 8 and 16 h after the first injection for a total of three injections per muscle. At 24 h after the first injection, mice were sacrificed by CO2 inhalation and gastrocnemius muscles were dissected and quick-frozen in liquid nitrogen for future biochemical analysis.
To define genes affected in myotubes during cytokine-dependent muscle fiber loss, we hybridized total mRNAs collected from TNF-α- and IFN-γ-treated (IT) myotubes to Affymetrix Mouse MU74Av2 oligonucleotide arrays. As previously reported, we observed that IT induces massive loss of differentiated myotubes 24- to 48-h posttreatment (Fig. (Fig.1A)1A) (23). Furthermore, the levels of MyoD as well as the myosin heavy chain (MF20) were significantly reduced under these conditions (Fig. (Fig.1B,1B, lane 4). Total mRNAs from these cells were collected and analyzed using the standard Affymetrix protocol. We found that levels of only a limited number of genes were affected (100 to 162 out of 12,487 genes screened) (Fig. (Fig.1C)1C) (see Fig. S1 in the supplemental material). Moreover, we observed that many of the classic gene targets for IFN-γ and TNF-α (interferon-activated gene 204, 202A, IRF-1, STAT1, alpha-induced protein 3, tumor necrosis factor receptor super family, member 11b osteoprotegerin) were significantly induced fourfold or more (see Fig. S1B and C in the supplemental material), indicating that our system reflects an IT-dependent effect. Additionally, we found that MyoD mRNA is down regulated twofold at 24 h post-IT (Fig. (Fig.1D)1D) (see Fig. S1A in the supplemental material). These observations are consistent with the reported cytokine-induced muscle fiber degeneration, leading to activation of the proteasome pathway, as well as the decay of myogenic regulatory transcription factor (MRF) messages (1, 23, 33, 38).
The fact that expression of iNOS mRNA, a cytokine-dependent muscle-wasting inducer (2, 8), was enhanced sixfold early upon IT (12 h) and remained high for more than 24 h (Fig. (Fig.1D)1D) suggested a potential link between iNOS-dependent NO release and the loss of MRF messages such as MyoD. Indeed, Northern blot analysis using total RNA extracted from myotubes subjected to IT showed a >25-fold increase in iNOS mRNA levels associated with NO release, while MyoD mRNA levels decreases over time (Fig. (Fig.2A,2A, lanes 8, 12, and 16, and 2B, lane 2). Similar results were obtained with undifferentiated myoblasts (data not shown). In our analysis we note that although MyoD mRNA is the first message to be down regulated (<4-fold at 24 h) upon IT, the Myogenin mRNA is also decreased, albeit much later (Fig. (Fig.2A,2A, lane 16). Moreover, and as previously reported (34), we observed a rapid and massive loss of MyoD mRNA in muscle cells treated with TNF-α and IFN-γ simultaneously (Fig. (Fig.2A).2A). These results are consistent with the fact that both cytokines together are required to trigger rapid wasting of skeletal muscle.
The role of endogenously produced NO in the autocrine-mediated loss of myotubes was assessed by the use of aminoguanidine (2-AMG), an inhibitor known to specifically block the enzymatic activity of iNOS protein without affecting its expression levels (2, 8). Addition of 2-AMG to myotubes subjected to IT was found to reduce NO release (Fig. (Fig.2B)2B) and prevent the loss of myotubes (Fig. (Fig.2C)2C) and MyoD mRNA (Fig. (Fig.2D).2D). Although these results clearly implicate iNOS-NO in MyoD mRNA and muscle fiber loss, they do not define whether this effect depends directly on NO itself. In fact, it is well established that cytokine treatment of macrophages induces the release of both NO and superoxide (O2−) leading to the formation of peroxynitrite (ONOO−). It was also demonstrated that the activity of peroxynitrite is linked to inflammatory-induced pathologies (39), mediating the downstream negative effects of NO. Consistent with these observations, when we treated myotubes directly with only an NO donor, we did not induce loss of muscle fibers (data not shown). However, when we treated myotubes with both IFN-γ and peroxynitrite scavenger (FeTPPS), we prevented the loss of muscle fiber (Fig. (Fig.2C;2C; compare panels 4 to 6) as well as MyoD mRNA (Fig. (Fig.2D,2D, lanes 2 and 6), without affecting significantly the rate of NO release (data not shown). These results demonstrate a direct link between NO release, peroxynitrite formation, and cytokine-mediated MyoD mRNA loss in muscle cells. Our subsequent experiments demonstrated that this is also true in an animal model. Indeed, disruption of the iNos gene appears to protect MyoD mRNA against cytokine-mediated loss (Fig. (Fig.2E)2E) despite a significant activation of the NF-κB pathway demonstrated by the induction of MurF1 mRNA (Fig. (Fig.2E).2E). These data provide clear genetic evidence that the loss of MyoD mRNA during the early stages of cytokine-induced muscle wasting in mice is a direct consequence of the activation of the iNOS-NO pathway.
The stimulation of NO release in muscle cells, as well as in other cell systems, depends on the activation of the NF-κB pathway (3). To define the implication of NF-κB in iNOS-NO-dependent myofiber decay, we used a muscle cell line in which the IT-dependent loss of MyoD mRNA was prevented by stably expressing a NF-κB dominant-negative mutant (myoblasts constitutively expressing the active IκΒα superrepressor) (23). We observed that the inhibition of NF-κB pathway in myotubes subjected to IT prevents the expression of iNOS mRNA and protein, as well as NO release (Fig. 3A to C). Furthermore, we demonstrated that under these conditions, these cells do not undergo muscle fiber loss or show diminished MyoD mRNA levels (Fig. (Fig.3E,3E, panel 2, and 3F). These results clearly demonstrate that iNOS mRNA expression during IT-induced muscle wasting depends on activation of the NF-κB pathway.
NF-κB is not the only pathway known to trigger muscle wasting. It was shown that starvation induces muscle atrophy both in mice (21) and in cultured myotubes (50) without activating NF-κB. Since the expression of iNOS mRNA could be stimulated by alternative pathways (3), we tested the effects of nutrient starvation on cultured myotubes defective in NF-κB activity. After incubation for 6 h in serum- and glucose-free media at 37°C, these myotubes presented the same dramatic fiber loss (Fig. (Fig.3E,3E, panel 4) seen in the control cells (Fig. (Fig.3E,3E, panel 2). However, Northern blot analysis did not reveal any expression of iNOS mRNA (Fig. (Fig.3D).3D). These results strongly suggest that starvation induces muscle atrophy in an NF-κB- and iNOS-NO-independent manner.
Since the expression of iNOS mRNA has been shown to be regulated posttranscriptionally via HuR protein in other cell systems (48), we decided to determine whether this could also be the case for myofibers subjected to IT. Total protein extracts from untreated myotubes as well as myotubes subjected to IT were collected and used for immunoprecipitation experiments utilizing a monoclonal anti-HuR (3A2) or -IgG antibodies. RT-PCR analysis of the precipitated RNA, by use of iNOS- and COX-2- as well as MyoD-specific primers located on two separate exons, indicates that HuR associates with both iNOS and COX-2 mRNA only in treated cells (Fig. (Fig.4A,4A, lane 2). The interaction of HuR with the MyoD mRNA, however, was detected only in untreated myotubes (Fig. (Fig.4A,4A, lane 1). Our observations suggest that HuR's interaction with these messages is likely to depend on their availability, since in both treated and untreated myotubes the amount of immunoprecipitated HuR seems to be the same (Fig. (Fig.4B4B).
When we analyzed the primary sequence of mouse iNOS mRNA we detected the existence of an ARE (miNOS-ARE) in its 3′UTR (Fig. (Fig.4C)4C) homologous to the ARE previously described for its human counterpart (48). To define the implication of this element in HuR-iNOS mRNA association, we performed an RNA gel shift analysis using a radiolabeled probe of the miNOS-ARE incubated with extracts prepared from C2C12 myotubes. We observed the formation of two RNA-protein complexes (complexes A and B) (Fig. (Fig.4D,4D, lane 2). Addition of a HuR antibody to the assay causes a super shift of one of the two complexes (complex B to HuRc) (lane 3). Therefore, the observed HuR-iNOS interaction is likely to be mediated by the miNOS-ARE, suggesting that HuR could play a key role in regulating posttranscriptionally iNOS mRNA expression during cachexia.
To define whether HuR protein is essential for the expression of iNOS mRNA, we used an RNA interference (RNAi) approach to deplete its expression. Transfecting C2C12 cells with siRNA duplexes against HuR (HuSi1) but not the control siRNA (CTL) (56) resulted in a significant decrease in HuR (>80%) and iNOS (>70%) protein levels and NO release (>45%) (Fig. (Fig.5A,5A, lanes 2 and 5) (see Fig. S3A in the supplemental material). Interestingly, the knockdown of HuR did not cause a reduction in COX-2 protein (Fig. (Fig.5A,5A, lanes 4 to 6), demonstrating that in muscle cells stimulated by IT, HuR specifically regulates the expression of its mRNA target iNOS but not COX-2. RT-PCR analysis of total RNA prepared from RNAi-transfected cells showed that decreasing HuR levels lead to a significant (>73%) reduction in iNOS mRNA expression (Fig. (Fig.5B;5B; compare lanes 4 and 6 to lane 5). A Northern blot analysis of the same samples showed even more dramatic down regulation (>88%) (Fig. (Fig.5B,5B, lanes 7 and 8). These results demonstrate that HuR is required for iNOS expression in muscle cells subjected to IT.
Rescue experiments employing either an AP-conjugated HuR or GST fusion protein (56) (AP is the antennapedia cell-permeable peptide that allows cellular uptake of the chimera with > 90% efficiency) (19) were subsequently used to ensure that the reduction in iNOS expression is a direct consequence of HuR depletion. Addition of the AP-HuR-GST, but not the AP-GST-negative control, to the HuR knockdown cells efficiently restores HuR levels and rescues iNOS protein expression in the muscle cells subjected to IT (Fig. (Fig.5C,5C, lanes 3 and 4). Furthermore, treating control cells with AP-HuR-GST increases the expression of iNOS protein by at least 2.5-fold (Fig. (Fig.5C,5C, lanes 1 and 2). Together, these results implicate HuR in regulating the expression of the iNOS message in muscle cells subjected to IT.
Recent observations suggested that the synthesis of cytokine-induced iNOS protein in muscle cells is blocked by stimulating the AMP-activated protein kinase (AMPK) (46). The AMPK pathway is known as the monitoring system for energy control (11), and its role as a sensor for ATP levels is well defined (11). AMPK activation is triggered by a decrease in ATP and an increase in AMP levels (AMP/ATP > 1) (24). It was shown that activating AMPK in human colorectal carcinoma RKO cells rapidly reduces the cytoplasmic levels of HuR and the stability of some of its mRNA targets (57, 58). Thus, it is reasonable to assume that, in muscle cells, the cellular distribution of HuR could be affected by energy variation. To test this possibility, we activated AMPK using an adenosine analogue, AICAR (5-aminoimidazole-4-carboxamide 1-β-ribofuranoside) (data not shown) (24) in C2C12 cells subjected to IT and performed a Northern blot analysis using an iNOS cDNA probe. We observed that the steady-state level of the iNOS message is decreased with 4 mM AICAR (Fig. (Fig.5E).5E). Furthermore, addition of AICAR to myotubes subjected to IT prevented the loss of MyoD mRNA (data not shown) seen in Fig. Fig.2A.2A. Both immunofluorescence and cellular fractionation experiments showed, furthermore, that treatment of cells with 2 to 4 mM AICAR along with IT induced a significant decrease in the cytoplasmic levels of HuR protein (Fig. 5D and F). Our data strongly suggest that the cytokine-dependent synthesis of iNOS in muscle cells depends on active AMPK and requires HuR's localization in the cytoplasm.
Our results described above clearly show that the expression of IT-dependent iNOS mRNA and protein, as well as NO release, increases exponentially with time in muscle cells (Fig. (Fig.2A2A and 3A to C). Since the loss of MyoD levels is maximal only when cells are treated with both cytokines simultaneously, we decided to compare their effects (individually and together) on NO release. Differentiated myotubes were treated for various lengths of time with IFN-γ and/or TNF-α, and the amount of NO release was measured using GREISS reagent (41). The level of NO in the medium increases with time upon treatment with IFN-γ or TNF-α individually. However, the addition of both cytokines simultaneously synergistically enhanced NO secretion (18- to 36-fold) (Fig. (Fig.6A).6A). Under these conditions, treatment of the myotubes with the two cytokines together induces an additive (rather than synergistic) up regulation of the iNOS mRNA (Fig. (Fig.2A).2A). These results suggest that the expression of iNOS mRNA and protein is induced by these cytokines, reflecting transcriptional and/or posttranscriptional regulations.
To address the role of posttranscriptional mechanisms, we tested the stability of iNOS mRNA by performing an actinomycin D pulse-chase experiment. Differentiated myofibers were treated for 12 h with TNF-α and/or IFN-γ and then incubated in medium containing 5 μg/ml of actinomycin D for various periods of time. Total RNA was collected and analyzed by Northern blotting using radiolabeled probes against iNOS, MyoD, and GAPDH messages. Our results indicate that the half-life of iNOS mRNA is significantly (approximately twofold) increased when myotubes are treated with both IFN-γ and TNF-α in comparison to cells treated with each of them individually (Fig. 6B and C). The half-life of iNOS mRNA in cells treated with a single cytokine is ~6.5 h; however, it increases twofold (~9 to 10 h) when both IFN-γ and TNF-α are used (Fig. (Fig.6C).6C). As previously reported (23), MyoD mRNA is rapidly degraded (less then 3 h) under the same conditions in cells treated with or without IFN-γ and TNF-α (Fig. (Fig.6B).6B). The iNOS mRNA stability was furthermore enhanced in muscle cells treated with AP-HuR-GST (data not shown). Together these results suggest that iNOS expression is regulated at the level of mRNA stability during cytokine-induced muscle fiber loss. The fact that HuR protein is required for iNOS mRNA expression (Fig. (Fig.5)5) suggests the involvement of an HuR-mediated iNOS mRNA stabilization pathway in muscle fibers undergoing wasting.
In this paper, we demonstrate that the iNOS-NO pathway plays a direct role in MyoD mRNA decay during cytokine-NF-κB-induced muscle fiber loss. Previous studies had shown that treatment of myotubes with IFN-γ and TNF-α results in the loss of MyoD mRNA and muscle fibers in an NF-κB-dependent manner (22). Although the implication of these two cytokines in iNOS-NO-mediated muscle wasting was previously suggested (8, 9) the effect of NO release on NF-κB-dependent MyoD loss was unknown. Here we delineate the molecular mechanisms by which iNOS mRNA is expressed under cachectic conditions and demonstrate its direct role in the loss of the MyoD message. Our results clearly implicate the RNA-binding protein HuR, which associates with, and posttranscriptionally regulates, the iNOS message in an ARE-dependent manner (Fig. (Fig.5),5), in this process. Therefore, we suggest a model by which IFN-γ- and TNF-α-dependent muscle wasting involves the following events: activation of the NF-κB pathway, which in turn drives the transcription of the iNos gene. The de novo transcribed message binds to HuR for stability and probably for rapid export to sites for translation in the cytoplasm (HuR was shown to serve as an adaptor for mRNA export) (19). These events consequently induce iNOS protein synthesis, which in turn (through its enzymatic activity) leads to the production and release of NO gas. NO will likely react with superoxide (O2−) to form peroxynitrite (ONOO−) (39), which will ultimately mediate both MyoD mRNA decay and muscle atrophy (Fig. (Fig.77).
The formation and maintenance of muscle fibers depends on a variety of factors, such as the myogenic regulatory transcription factors (MRF), MyoD, and myogenin (49). Since IFN-γ and TNF-α have different downstream effectors (TNF-α activates NF-κB , while IFN-γ stimulates STATs/IRFs ) but do not have any direct role in modulating the transcription of the MyoD gene, it was reasonable to assume that these two cytokines participate in the activation of common genes, which in turn trigger muscle wasting. Our cDNA microarray experiments showed that both cytokines stimulate the expression of genes known to be induced by the two cytokines, as well as cachectic-dependent messages (such as those belonging to the ubiquitin-proteasome pathway) (see Fig. S1A and D in the supplemental material). While searching for common target genes, we identified the NO synthase (iNOS) mRNA, which is up regulated sixfold 12 h following IT (Fig. (Fig.1D1D and and2A)2A) (see Fig. S1 in the supplemental material). These observations are consistent with the fact that the iNos promoter contains binding sites for both NF-κB and STAT1 (31). Interestingly, when we treated myotubes with these two cytokines, we observed that the loss of MyoD mRNA correlated with a significant increase in iNOS mRNA levels as well as NO release (Fig. 2A and B). Since IFN-γ and TNF-α individually induce only modest levels of both iNOS mRNA and NO and have a small effect on MyoD mRNA expression (24 h post-IT) (Fig. (Fig.2A2A and and6A),6A), we concluded that the iNos gene could be a common target activated synergistically by these cytokines to trigger muscle fiber loss. These observations support previous findings demonstrating that IFN-γ is needed to potentate the NF-κB-dependent MyoD loss in response to TNF-α in muscle fibers (23, 34). The fact that MyoD mRNA decay was prevented by specific inhibitors of the iNOS-NO pathway in muscle fibers (Fig. 2B to D), or by disrupting the iNos gene in mice (Fig. (Fig.2E),2E), clearly demonstrates that NO release is absolutely required to trigger IT-dependent muscle wasting both in vitro and in vivo. This conclusion was further supported by our experiments using muscle fibers expressing the NF-κB dominant negative mutant IκBαSR, which do not produce iNOS mRNA and protein and are protected against muscle wasting when subjected to IT (Fig. 3A to C and and3E3E).
Although recent reports confirmed a role for the NF-κB pathway in cytokine-induced cachexia both in myotubes and in mice (10, 23), it was also demonstrated that starvation triggers this debilitating condition in an NF-κB-independent manner (50). Therefore, and as expected, incubating myotubes stably expressing IκBαSR in serum-free and nutrient-free media led to rapid atrophy of fibers without inducing iNOS mRNA expression (Fig. 3D and E). Thus, we concluded that starvation and cytokine-dependent NF-κB activation induce muscle wasting using two different pathways. This assumption is further supported by the fact that IT did not activate the specific starvation-dependent ubiquitin ligase, atrogin-1/MAFbx (10) (see Fig. S1 in the supplemental material). Furthermore, active NF-κB triggers muscle wasting by specifically inducing the muscle E3 ubiquitin-ligase MuRF1 (10). However, the transgenic mice expressing the active NF-κB isoform were protected against muscle wasting by only 50% when crossed with MuRF1−/− knockout mice (10). This partial rescue clearly indicates that NF-κB also stimulates another pathway, which we identify here as NF-κB-iNOS-dependent NO release (Fig. (Fig.7).7). The link between NO and muscle fiber decay was previously suggested by several studies of muscle cell lines as well as of mice (8, 9). However, the molecular mechanisms leading to iNOS expression, as well as the downstream targets of NO in muscle cells undergoing cachexia, were unknown. Our data identify the MyoD message as an NO target and clearly define that the stability; thus, expression of iNOS mRNA is regulated posttranscriptionally. We showed that the iNOS message interacts with HuR protein through its destabilizing element miNOS-ARE (Fig. (Fig.4).4). The importance of HuR protein in muscle cells was underscored in previous studies by its vital role as a stabilizer for MyoD mRNA during myogenesis (15, 56). Since its expression level is not affected by IT (Fig. (Fig.2A),2A), it is reasonable to assume that under these conditions, HuR stabilizes newly synthesized ARE-containing messages, such as iNOS.
The presence of an ARE in the 3′UTR of the iNOS message (miNOS-ARE) prompted us to verify whether HuR in myotubes could bind and, thus, regulate the expression of iNOS mRNA. Immunoprecipitation experiments using both untreated and treated muscle cells coupled to RT-PCR showed that HuR associates with iNOS mRNA in both myoblasts (data not shown) and myotubes (Fig. (Fig.4A),4A), indicating that the HuR-iNOS complex is not dependent on the myogenic process. Interestingly, under the same conditions, HuR loses its ability to bind MyoD message (Fig. (Fig.4A,4A, lane 2), which is one of its known primary mRNA targets during the myoblast-to-myotube transition (56). The straightforward explanation for the absence of this binding is the down regulation of the MyoD mRNA under these conditions. However, it is also possible that, as in the case of lipopolysaccharide-treated human macrophages, these cytokines induce posttranslational modification (such as methylation) of the HuR protein (36), thus affecting its cellular movement and/or interaction with the MyoD transcript. Exploring this possibility in muscle cells will further uncover the molecular mechanism by which HuR triggers muscle fiber loss. Likewise, the miNOS-ARE we identified in this study (Fig. (Fig.4C)4C) presents the same structural features as well as binding affinity to HuR (Fig. (Fig.4D)4D) as its human counterpart (48). This observation indicates the generality of HuR's implication in iNOS expression, making the identification of the specific HuR-iNOS-NO pathway inducers in muscle cells undergoing wasting a crucial step in better understanding this deadly disease.
The importance of HuR in the regulation of iNOS mRNA was evident from our experiments assessing the effect of its depletion in C2C12 cells subjected to IT (Fig. (Fig.5).5). Unfortunately, and for technical reasons, we were not successful in knocking down HuR's expression in differentiated myotubes (data not shown). Therefore, since HuR associates with iNOS messages equally well in both differentiated and undifferentiated C2C12, we used myoblasts for our subsequent experiments. The fact that recombinant AP-HuR-GST rescues the expression of the iNOS protein (Fig. (Fig.5C)5C) in HuR-depleted C2C12 cells subjected to IT (Fig. 5A to C) is a clear indication that the HuR protein is absolutely required for the cytokine-induced iNOS-NO pathway. Likewise, introducing AP-HuR-GST protein into muscle fibers subjected to IT (mimicking overexpression) induces stabilization of the iNOS mRNA (data not shown), thus increasing iNOS protein production (Fig. (Fig.5C)5C) and NO release (data not shown). These results strongly indicate that, as in the case of myoblasts, HuR plays a prominent role in iNOS stabilization, leading to high iNOS protein levels in myofibers subjected to IT. The fact that TNF-α and IFN-γ individually induce only a modest level of NO release (Fig. (Fig.6A),6A), and that both of them are required for the stabilization of iNOS mRNA (Fig. 6B and C) and NO-dependent muscle fiber loss (Fig. (Fig.1A),1A), is a clear indication that posttranscriptional regulatory mechanisms are prominently involved in cytokine-induced cachexia.
Recent observations suggested that the expression of the iNOS protein depends on the maintenance of high ATP levels in muscle cells (46). Indeed, the activation of AMPK in muscle cells, which mimics low ATP levels, prevents the synthesis of iNOS protein without affecting the abundance of its message (46). Interestingly, activation of this enzyme has also been shown to alter the cytoplasmic localization of HuR in the human colorectal carcinoma cell line (RKO), leading to a decrease in the steady-state levels and stability of many ARE-containing messages such as p21waf1, cyclin A, and cyclin B1 (57). Here we demonstrate that activation of AMPK in muscle cells subjected to IT sequesters HuR in the nucleus (Fig. 5D and F) and affects the mRNA levels of iNOS (Fig. (Fig.5E).5E). It is thus reasonable to assume that these treatments modify HuR posttranslationally and/or change its association with its protein ligands. Furthermore, it is well accepted that the beneficial effect of physical activity is related to AMPK activation, which reduces the risk of developing insulin resistance and type 2 diabetes (25). Therefore, it is possible that AMPK activation ameliorates the symptoms of NO-induced muscle wasting in patients. This strategy is already applied to combat type 2 diabetes by use of the AMPK activators metformin (61) and thiazolidinediones (16).
The discovery that HuR, via the activation of the iNOS-NO pathway, contributes to NF-κB-mediated muscle loss suggests that proteins involved in these posttranscriptional processes should also be considered as targets for the design of anti-cachectic drugs. The importance of identifying novel drugs is undermined by the fact that high doses of sodium salicylate, which inhibits IKKβ and thus NF-κB activity (10), cause unwanted secondary effects and are not well tolerated by patients. Furthermore, it was demonstrated that low doses of this drug induce iNOS expression in muscle cells independently of NF-κB (4, 51). Therefore, further studies defining the effects of cytokine treatments on HuR's association with its mRNA targets, as well as protein ligands, will uncover better ways to specifically target its function during muscle wasting.
We are grateful to Jerry Pelletier, Denis Guttridge, Maria Hatzoglou, and Paul Anderson as well as K. Van der Giessen and J. Behrmann for helpful discussions and comments on the manuscript. We thank A. B. Lassar (Harvard Medical School, Boston, Mass.) for MyoD, myogenin, and p21Cip1 plasmids.
This work was supported by a CIHR Cancer Consortium Training Grant Fellowship Award to S.D.M., an NCIC TFF Research Fellowship to R.M., an NIH operating grant (NIH/NHGRI HG003679) to S.A.T., and CIRH operating grant Mop-57680 and an FRSQ “Subvention d'etablissement de jeune chercheur” (Frsq 23516-2760) to I.-E.G.
†Supplemental material for this article may be found at http://mcb.asm.org/.