Cleavage of MDA-5 was observed in cells infected with poliovirus, rhinovirus type 1a, and EMCV but not in cells infected with echovirus type 1 or rhinovirus type 16. Putative MDA-5 cleavage products of ~90 and ~70 kDa were observed in cells infected with poliovirus. Poliovirus-induced degradation of HA-MDA-5 produced from an adenovirus vector was also observed, although cleavage products were not detected.
Likely candidates for cleavage of MDA-5 during poliovirus infection are the two viral proteinases, 2Apro
. These proteinases not only process the viral polyprotein but have been shown to cleave a variety of cell proteins, including eIF4G (24
), PABP (25
), the catalytic subunit of DNA-dependent protein kinase (13
) (cleaved by 2Apro
), Nup62 and Nup153 (14
), and proteins involved in RNA polymerase transcription such as TATA-binding protein, CREB, Oct-1, p53, IIICα, and SL-1 (31
) (cleaved by 3Cpro
). To determine whether the viral proteinases participate in MDA-5 cleavage, we used mutant polioviruses with single amino acid changes in either enzyme that have been previously shown to block cleavage of host cell proteins. Degradation of MDA-5 proceeded in cells infected with these mutant viruses, suggesting that the viral proteinases do not directly cleave MDA-5. To address the possibility that the mutant proteinases are still able to effect MDA-5 cleavage, we determined whether the purified viral proteinases can directly cleave MDA-5 protein in vitro. The results indicate that neither poliovirus 3CDpro
nor coxsackievirus B3 2Apro
cleaves MDA-5 in a rabbit reticulocyte lysate. It was necessary to use the coxsackievirus 2Apro
for this experiment because it has not been possible to purify the poliovirus enzyme. The 2Apro
enzymes of poliovirus and coxsackievirus have similar substrate specificities and cleave similar sites within the viral polyprotein as well as cellular proteins such as eIF4G and PABP (17
). It would therefore be expected that the poliovirus and coxsackievirus 2Apro
proteinases would both have the ability to cleave MDA-5.
The results presented here, together with previous findings, are consistent with the hypothesis that poliovirus infection activates the apoptotic pathway, which then leads to MDA-5 cleavage. The mouse ortholog of human MDA-5 undergoes caspase-dependent cleavage in cells that are induced to enter apoptosis (23
). It was shown previously that poliovirus infection induces apoptosis, causing mitochondrial damage, release of cytochrome c
, and activation of caspase-9 and caspase-3 (4
). We have shown that poliovirus-induced cleavage of MDA-5 is blocked by the caspase inhibitor Z-VAD-FMK. Our results also indicate that cleavage of PARP occurs in poliovirus-infected cells, an event that correlates with MDA-5 cleavage. Pretreatment of cells with poly(IC) leads to accelerated cleavage of both PARP and MDA-5 in poliovirus-infected cells. Treatment of uninfected HeLa cells with puromycin, a known inducer of apoptosis, leads to cleavage of both PARP and MDA-5. The putative MDA-5 cleavage product observed in cells treated with puromycin is ~90 kDa, similar in size to the putative cleavage product observed in poliovirus-infected cells. Infection of HeLa cells with rhinovirus type 16 does not lead to apoptosis, as judged by the very low level of cleaved PARP observed by 8 h postinfection, and MDA-5 cleavage is not observed.
The main triggers of apoptosis in poliovirus-infected cells appear to be the two viral proteinases, as production of either poliovirus 2Apro
in cells is sufficient to induce apoptosis (3
). It would therefore be expected that amino acid changes in either viral proteinase would not be sufficient to block virus-induced apoptosis. Our observation that MDA-5 cleavage is not reduced in cells infected with polioviruses containing a single amino acid change in either viral proteinase is in accord with this expectation. Furthermore, MDA-5 cleavage does not occur in cells infected with a poliovirus mutant containing single amino acid changes in both viral proteinases 2Apro
. These findings provide additional evidence that induction of apoptosis by 2Apro
is central to induction of MDA-5 degradation.
Sites of caspase cleavage in murine MDA-5 have been identified at amino acids 208, 216, 251, and 260 to 280 (23
), and similar sequences are present in the human MDA-5 protein. Cleavage at some of these sites could produce the less-than-full-length MDA-5 proteins observed in poliovirus-infected cells. For example, cleavage at amino acid 216 of MDA-5 could produce the ~90-kDa protein, and cleavage at a putative caspase cleavage site at amino acid 673 could produce the ~70-kDa protein (Fig. ). It is not clear why cleavage products were not detected when HA-MDA-5 was degraded during poliovirus infection (Fig. ). It will be necessary to determine amino acid sequences from the ~90- and ~70-kDa proteins to determine whether these are bona fide cleavage products.
The inhibition of poliovirus-induced cleavage of MDA-5 by MG132 and epoxomicin indicates that the proteasome is also involved in MDA-5 degradation. The proteasome has been implicated in the regulation of apoptosis by modulating the levels of pro- and antiapoptotic molecules (10
). MDA-5 is a proapoptotic protein (1
), and therefore its cleavage by the proteasome is not unexpected. Recent studies investigating potential signal transduction pathways involved in regulating mda-5
-induced apoptosis in mammalian cells demonstrate the importance of Raf/Ras/
MEK/EFK signaling pathways in apoptosis induction by mda-5
). The results of these studies demonstrate that rodent and human cells containing an activated Raf/Ras/MEK/ERK are resistant to mda-5
-induced killing. Accordingly, this resistance is antagonized by inhibiting this important signal transduction cascade either by directly inhibiting ras
activity using an antisense strategy or by targeting ras
-downstream factors, such as MEK1/2, with the pharmacological inhibitor PD98059. The role of this pathway in poliovirus-induced MDA-5 cleavage remains to be defined.
The observation that inhibitors of either the proteasome or caspases can block cleavage of MDA-5 during poliovirus infection shows that both cellular proteases are required for degradation. The simplest explanation for this finding is that caspases and the proteasome act in a linear manner on MDA-5. For example, cleavage of MDA-5 by caspases might be required before the protein can be a substrate for the proteasome. Because inhibitors of either protease block MDA-5 degradation, it is not possible to determine whether caspases or the proteasome are first in this pathway.
Cleavage of MDA-5 during poliovirus infection might be a mechanism to evade the innate immune response by attenuating the production of IFN, thereby allowing higher levels of viral replication. If this hypothesis were correct, we would expect to observe higher poliovirus yields when cleavage of MDA-5 is blocked by MG132 or epoxomicin. The presence of MG132 or epoxomicin had no effect on poliovirus yields during infection at a high MOI (Fig. and unpublished data). Furthermore, induction of IFN-β RNA synthesis during poliovirus infection was not stimulated by conditions which block MDA-5 cleavage, such as infection with a viral mutant with amino acid changes in both proteinases, or the presence of Z-VAD-FMK. Taken together, these observations suggest that virus-induced cleavage of MDA-5 has no beneficial effect on poliovirus replication. To provide a definitive answer to the question of whether poliovirus-induced cleavage of MDA-5 leads to enhanced viral replication, it will be necessary to produce a variant of MDA-5 that is resistant to virus-induced cleavage and to determine whether the production of the altered protein in cells reduces poliovirus yields.
MDA-5, not RIG-I, is believed to be crucial for sensing picornavirus infection. This conclusion derives from studies in which mice lacking the genes encoding MDA-5 or RIG-I were infected with EMCV (11
). Mice lacking the mda-5
gene failed to produce IFN when infected with EMCV, and the animals were more susceptible to infection. However, absence of the gene encoding RIG-I had no effect on IFN production after EMCV infection of mice. While these results are compelling, it seems premature to conclude that all picornaviruses are sensed by MDA-5, because only cardioviruses have been studied. Other unanswered questions include the effect of absence of MDA-5 or RIG-I on picornavirus yields in different cell types and whether the effects of these sensor molecules on viral replication differ in cultured cells and on IFN production in animals. For example, overproduction of RIG-I in cultured cells reduces yields of vesicular stomatitis virus and EMCV (33
), but mice lacking the gene for RIG-I have no impairment in IFN production in response to EMCV infection (21
). Additional experiments with cells and with mice lacking RIG-I and mda-5
are required to assess the physiological role of these proteins in poliovirus infection.
Forced expression of mda-5
, but not RIG-I or Toll-like receptor 3, leads to enhanced IFN-β promoter activity in measles virus-infected A549 non-small-cell lung carcinoma cells (5
). In this context IFN-induced mda-5
is involved in measles virus-induced expression of antiviral cytokines. These findings provide yet another example of the potential subtle and unique roles of MDA-5 and RIG-I in sensing and responding to specific viral infections. Further studies on MDA-5 and RIG-I in cells infected with different viruses is clearly important in understanding and potentially developing strategies for selectively and effectively intervening in viral pathogenesis.