|Home | About | Journals | Submit | Contact Us | Français|
The rate of protein secretion in host cells is inhibited during infection with several different picornaviruses, with consequences likely to have significant effects on viral growth, spread, and pathogenesis. This Sin+ (secretion inhibition) phenotype has been documented for poliovirus, foot-and-mouth disease virus, and coxsackievirus B3 and can lead to reduced cell surface expression of major histocompatibility complex class I and tumor necrosis factor receptor as well as reduced extracellular secretion of induced cytokines such as interleukin-6 (IL-6), IL-8, and beta interferon. The inhibition of protein secretion is global, affecting the movement of all tested cargo proteins through the cellular secretion apparatus. To test the physiological significance of the Sin− and Sin+ phenotypes in animal models, Sin− mutant viruses are needed that fail to inhibit host protein secretion and also exhibit robust growth properties. To identify such Sin− mutant polioviruses, we devised a fluorescence-activated cell sorter-based screen to select virus-infected cells that nevertheless expressed newly synthesized surface proteins. After multiple rounds of selection, candidate Sin− mutant viruses were screened for genetic stability, increased secretion of cargo molecules and wild-type translation and growth properties. A newly identified Sin− mutant poliovirus that contained coding changes in nonstructural proteins 2A (N32D) and 2C (E253G) was characterized. In this virus, the 2C mutation is responsible for the Sin− phenotype and the 2A mutation suppresses a resulting growth defect by increasing the rate of cell death and therefore the rate of viral spread. The 2A-N32D suppressor mutation was not allele specific and, by increasing the rate of cellular apoptosis, affected a completely different pathway than the 2C-E253G Sin− mutation. Therefore, the 2A mutation suppresses the 2C-E253G mutant phenotype by a bypass suppression mechanism.
The ability to inhibit host cell protein secretion during infection has been reported for a number of picornaviruses. Viruses with a secretion inhibition (Sin+) phenotype thus far include poliovirus (16), foot-and-mouth disease virus (31), and coxsackievirus (12). During infection with these viruses, the secretion of newly synthesized proteins from host cells is blocked or the rate of secretion is significantly slowed. Thus far, this block to protein secretion has not been shown to be specific for any particular cargo, with several different secreted or membrane-associated proteins in several different cell lines showing similar decreases. When the secretion of exogenously expressed cargo proteins, such as plasmid-expressed varicella-zoster virus G protein or α-1 protease inhibitor (17, 18), or endogenous, constitutively expressed cargo proteins, such as tumor necrosis factor (TNF) receptor or major histocompatibility complex class I (11-13, 34), were monitored, cells infected with Sin+ viruses secreted a much lower proportion of the protein synthesized than uninfected cells. The existence of viral variants, here termed Sin− viruses, which display diminished ability to inhibit host protein secretion, has allowed comparison of the rates of protein synthesis and protein secretion in cells infected with Sin+ wild-type and Sin− mutant viruses. For example, when cells were infected with a Sin− mutant poliovirus termed 3A-2, which contains a single amino acid insertion in the 3A protein, much larger amounts of virally induced proteins such as cytokines interleukin-6 (IL-6), IL-8, and beta interferon were secreted compared to those of cells infected with wild-type Sin+ virus (15). This 3A-2 virus, however, displays a slight growth defect in tissue culture cells during single-cycle infections (3, 5, 15). It has been shown that a Sin− coxsackievirus B3 that contains the 3A-2 mutation is attenuated in mice, but whether this resulted from the known growth defect of the mutant virus or increased inflammation due to its Sin− phenotype is not known (46). Differentiating between the possible effects of the Sin− phenotype on viral growth and the suppression of the immune response has been hampered by the lack of Sin− variants that do not show intrinsic growth defects and would therefore be suitable for studies in animal models.
The mechanism of inhibition of host protein secretion varies for different picornaviruses. During poliovirus infection, secretory cargo is arrested in the endoplasmic reticulum (ER)-to-Golgi intermediate compartment and the ER (6). An interruption in transport at this step is also observed when either coxsackievirus or poliovirus 3A protein is expressed in isolation (45, 47). Poliovirus and coxsackievirus proteins 2B and 2B-containing precursors are also able to inhibit cellular protein secretion (16, 44). The expression of poliovirus 2B or 2BC in isolation results in the accumulation of cargo in the Golgi apparatus (16). For foot-and-mouth disease virus, secretory cargo has been shown to accumulate in the ER, and foot-and-mouth disease virus protein 2BC alone, but not 3A, has been shown to be sufficient to inhibit host protein secretion (31). All the aforementioned picornavirus proteins (3A, 2B, 2C, and 2BC) are membrane associated and likely to complex with each other. They are also multifunctional, playing many roles in viral RNA replication and intracellular membrane rearrangement in addition to the inhibition of host protein secretion (reviewed in references 33 and 40). When present in complexes, these proteins exhibit emergent properties different from the sums of their individual functions (19). For example, poliovirus 2BC expressed in isolation induces the formation of single-membrane vesicles (10, 41), and 3A expressed in isolation causes ER swelling (17), but only the combination of 2BC and 3A induces the formation of autophagosome-like double-membrane vesicles similar to those observed during infection (25, 41). Furthermore, the previously described Sin− poliovirus mutant 3A-2, which contains only a single amino acid insertion in the 3A protein (3, 15), shows a greatly reduced ability to inhibit host protein secretion even in the presence of the wild-type 2B and 2BC proteins. The phenotype of this Sin− poliovirus argues that, in the context of a membrane-associated RNA replication complex that contains mutant 3A protein, the independent abilities of the 2B and 2BC proteins to inhibit secretion are not fully revealed.
Two attractive, nonexclusive hypotheses for the role of protein secretion inhibition by picornaviruses are (i) that the membrane alterations resulting in the inhibition of host protein secretion facilitate viral growth or (ii) that the inhibition of host protein secretion prevents the release of inflammatory mediators that would otherwise suppress growth and spread of the virus in tissues. The slight growth defect of Sin− poliovirus mutant 3A-2 could be interpreted as being consistent with the first hypothesis that the inhibition of protein secretion by poliovirus is intrinsically related to the membrane rearrangements required for viral growth. However, that the growth defect of the 3A-2 virus was only slight could be interpreted as being consistent with the second hypothesis, in which the growth defect of the 3A-2 virus is related to one of the 3A protein's other effects on RNA replication complexes, such as the binding of GBF-1 (45, 46), a cellular protein whose recruitment is known to be defective in the 3A-2 virus (3). In this scenario, the inhibition of protein secretion would serve as a virulence factor, not necessary for growth per se but instead facilitating growth in the presence of innate and acquired immune responses.
In tissues and during multiple cycles of infection in culture, the growth and spread of viruses depends both on the intrinsic growth rate of the virus within cells and the rate of its spread from cell to cell. Picornaviruses, which are nonenveloped, are thought to spread primarily via the lysis of infected cells, although several examples of apparently nonlytic spread have been documented (25, 29, 39, 43). In poliovirus and many other viral infections, cell lysis and therefore premature viral spread are delayed by the inhibition of apoptosis. Interestingly, it has been shown that poliovirus infection induces apoptosis early in infection and suppresses apoptosis later in infection (1, 4, 42). The early induction of apoptosis requires a functional poliovirus receptor (32) but is also likely to involve the inhibition of transcription and translation by 3C and 2A proteinases, which can both induce apoptosis when expressed in isolation (2, 7, 20). Later in infection, the activation of proapoptotic caspases, which can best be observed in slow-growing mutant viruses or in infections interrupted by treatment with the RNA synthesis inhibitor guanidine (1, 4) is inhibited. The mechanism of this late inhibition of apoptosis is not yet known but is accompanied by the aberrant processing of procaspase-9 (4).
In this study, we developed a genetic selection to identify Sin− polioviruses with robust growth properties. Ironically, the only such mutant virus that proved to be genetically stable contained a mutation in the 2C protein (E253G) that conferred the Sin− phenotype and an additional mutation (N32D), in the 2A protein, that suppressed an accompanying growth defect by increasing the rate of viral spread. Herein, both the 2C-E253G and the 2A-N32D mutant viruses are characterized. Although genetic suppression is commonly interpreted to signify physical interaction, we present evidence that the suppression of the 2C-E253G mutant virus defect by the 2A-N32D mutation is via a bypass, not an interactive, genetic mechanism.
Human HeLa and osteoblastoma-derived MG63 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum and penicillin-streptomycin (100 units/ml each) at 37°C and 5% CO2. The titers of poliovirus were determined on both HeLa and MG63 cells, and the multiplicities of infection (MOIs) refer to titers for the appropriate cell line. Viral growth curves and plaque assays were carried out as described previously (14).
IL-8 amounts were determined by enzyme-linked immunosorbent assays (ELISAs) from Invitrogen (Carlsbad, CA) and GE Healthcare (Niskayuna, NY) for plaque-purified viral isolates from a fluorescence-activated cell sorter (FACS)-based screen and for the 10-8 isolate and its characterized derivatives, respectively. Cell viability was assayed using the Live/Dead fixable stain (Invitrogen) according to the manufacturer's instructions. Pan-caspase inhibitors zVAD-fmk and qVD-OPh were purchased from R&D Systems (Minneapolis, MN). The activation of caspase-9, -3, and -7 was monitored using Caspase-GLO assays (Promega, Inc., Madison, WI) according to the manufacturer's instructions.
A HeLa cell line was constructed that stably expressed a single-chain antibody (9) from the pHook-2 vector (Invitrogen). The antibody recognizes the hapten phOx (4-ethoxymethylene-2-phenyl-oxazolin-5-one) and was placed under translational control of the poliovirus internal ribosome entry site (IRES). These cells were infected at 0.1 PFU/cell. Two hours postinfection, the cells were treated with trypsin (10 mg/ml) for 10 min at 37°C and then washed three times in serum-free medium. Following a 2- to 3-h recovery in complete medium, the cells that expressed surface antibody were detected by staining with a fluorescein-bovine serum albumin (BSA)-phOx conjugate and sorted by FACS. Cells positive for antibody expression were subjected to three rounds of freezing and thawing to recover virus, which was used in the subsequent round of selection, for a total of four rounds. Following the final round of selection, putative Sin− viruses were amplified by growth in HeLa cells and then plaque purified. Secondary screens to identify viruses with stable Sin− phenotypes were performed using MG63 cells seeded onto 96-well plates. In the first round of screening (Fig. (Fig.1B),1B), each well was inoculated with a preparation of virus that represented 10% of that obtained from a single plaque but for which the titer was not determined. For the second round of screening (Fig. (Fig.1C),1C), each well was inoculated at a defined MOI of 5.0 PFU/cell.
To make infectious 10-8 poliovirus cDNA, viral RNA was purified from HeLa cells infected with 10-8 mutant virus as described previously (37), amplified by reverse transcription-PCR, digested with BglII and AvaI, and cloned into a BglII- and AvaI-cut pGEM3polio plasmid. To make the 2A-N32D infectious clone, amplified cDNA from 10-8 virus was digested with BseRI and cloned into BseRI-cut pGEM3polio. To construct the 2C-E253G infectious clone, primers were designed to introduce the mutation into the wild-type cDNA clone via a splicing-by-overlap-extension protocol (28) to construct a mutated BseRI fragment. Viral RNA was then transcribed from poliovirus cDNA plasmids using a MEGAscript T7 kit (Ambion, Austin, TX) according to the manufacturer's instructions. Viral RNA was transfected into 70% confluent HeLa cells on 10-cm dishes. Briefly, 3 μl Lipofectamine 2000 (Invitrogen) was incubated in 1 ml DMEM without serum for 5 min at room temperature. This mix was then added to 1.0 μg or 0.1 μg RNA in 1 ml DMEM, incubated for 20 min at room temperature, and added to cells in 5 ml total volume. After 4 h, the transfection mix was removed and replaced with 10 ml DMEM/agar overlay with 10% calf serum and grown for 42 h. For each virus, several plaques were picked and titers determined to ascertain the phenotype and to perform further amplification. To guard against selecting revertant viruses, the specific infectivity (number of plaques/unit transfected RNA) of mutant and viral RNA was determined; in the cases of the mutant viruses studied herein, the specific activities were found to be virtually identical to those of the wild-type virus.
HeLa cells were infected for the indicated times, washed once with phosphate-buffered saline (PBS), and then labeled with 100 μCi/ml [35S]methionine for 15 min. The cells were washed once with PBS and harvested by scraping. Following another wash, the cells were resuspended for 15 min on ice in lysis buffer (50 mM Tris [pH 8.0]), 5 mM EDTA, 150 mM NaCl, 0.5% NP-40) with protease inhibitors. Nuclei and insoluble debris were pelleted by centrifugation at 800 × g for 5 min, and the supernatants were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The gel was dried and visualized on a phosphorimager (Storm 860; GE Healthcare).
HeLa cells were infected with wild-type and mutant viruses at an MOI of 20 PFU/cell. For experiments using pan-caspase inhibitors, zVAD and qVD-OPh (quinolyl-valyl-O-methylaspartyl-[2,6-difluorophenoxy]-methyl ketone) (8), obtained from R&D Systems, Inc. (Minneapolis, MN), were added to medium at 100 μM for the entire infection. At the indicated time postinfection, the cells were harvested by scraping and washed twice in PBS. The cells were resuspended in 1 ml PBS, 1 μl fixable Live/Dead reagent was added, and the cells were incubated for 30 min at 4°C. The cells were then washed in PBS and fixed in 4% paraformaldehyde for 20 min. Following two washes, the cells were run on a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ) and analyzed using FlowJo 8.8 (Tree Star, Inc., Ashland, OR).
Our goal was to identify new Sin− mutant polioviruses that showed diminished ability to inhibit host protein secretion but also displayed robust growth properties. To this end, we designed a FACS-based isolation (Fig. (Fig.1A)1A) to collect virus-infected cells that rapidly presented cell surface proteins on the exterior of the cell. Any virus contained in those cells would, presumably, not have interfered with the transport of the cell surface protein through the secretory pathway. Specifically, HeLa cells that expressed a single-chain antibody (9) on the cell surface were infected with wild-type poliovirus at low MOIs to allow any mutant virus in the population to express its phenotype, even if that phenotype were recessive. At 2 h postinfection, the cells were treated with trypsin to remove preexpressed antibody from the surfaces of the cells. After a further 2 h of incubation, the cells were sorted for those that stained brightly with the antibody's fluorescently labeled ligand. Cells that did not recover the expression of the antibody were presumed to be infected with wild-type Sin+ poliovirus and discarded. However, cells that did reexpress antibody at the cell surface were presumed to be either uninfected or infected with Sin− virus. Virus from these cells was collected and used to infect fresh antibody-expressing cells to initiate another round of selection. Four rounds of selection were performed. For rounds 1 to 3, the most fluorescent 0.15%, 3.8% and 2.3% of cells were selected, respectively. After the fourth round, the most fluorescent 1.4% of cells were harvested and found to contain 20 times more virus per cell than the corresponding population of cells infected with virus that had not undergone previous selection. Therefore, we had enriched for viruses that allowed more cell-surface protein to be expressed. From the pool of viruses from the fourth round of selection, 80 individual isolates were collected by plaque purification at 37°C on HeLa cells. Large plaques were selected with the goal of ensuring that the Sin− isolates would also display robust growth.
To test whether virus from the 80 plaque isolates also allowed increased secretion of IL-8 during infection, consistent with the anticipated phenotype of a Sin− variant, human MG63 osteosarcoma cells, commonly used to monitor cytokine secretion, were infected in 96-well plates, and the amount of IL-8 secreted into the supernatant after 4 h of infection was determined by ELISA (Fig. (Fig.1B).1B). IL-8 is a virally induced chemokine secreted via the normal secretory pathway. It is not secreted by uninfected cells, and a small amount is secreted from wild-type Sin+ virus-infected MG63 cells. However, it is robustly secreted by cells infected with Sin− mutant poliovirus 3A-2 (15). Although we had anticipated that most of the 80 plaque isolates would display Sin− phenotypes, only a few displayed considerably higher amounts of IL-8 secretion than wild-type poliovirus (Fig. (Fig.1B).1B). Viruses from isolates 10, 17, 25, 41, and 49 were further plaque purified, and the IL-8 secretion assay was repeated with 10 new plaque isolates from each (Fig. (Fig.1C).1C). The second round of purification revealed that the Sin− phenotype was not stable in most of the isolates. However, several of the descendants of original isolate 10 showed reproducibly higher amounts of IL-8 secretion. One of these potential Sin− virus isolates, identified as 10-8, was chosen for further characterization.
To identify the mutation or mutations in the 10-8 isolate that were responsible for the Sin− phenotype, cDNA made from 10-8 viral RNA was sequenced. To our surprise, no mutations were found in the 3A coding region. Instead, two coding mutations, 2A-N32D and 2C-E253G, and three noncoding changes were identified (Fig. (Fig.2A).2A). An infectious cDNA clone (10-8 reconstructed) was then made in which a region of the 10-8 viral cDNA, including the 2A-N32D and 2C-E253G mutations as well as a silent mutation in 3C, was cloned into a wild-type poliovirus cDNA. In addition, the 2A-N32D and 2C-E253G mutations were engineered into wild-type poliovirus cDNA in isolation and in combination (Fig. (Fig.2B2B).
To test the contributions of these mutations to the Sin− phenotype, we compared the amount of IL-8 secreted from the MG63 cells infected with wild-type virus, the 10-8 mutant poliovirus, and viruses that carried the two mutations individually and in tandem. As can be seen in Fig. Fig.3A,3A, at 1.5 and 3 h postinfection, 10-8 virus-infected cells allowed the secretion of higher amounts of IL-8 than wild-type virus-infected cells, confirming the Sin− phenotype. For the 2A-N32D virus-infected cells, the amount of IL-8 secreted was virtually identical to that secreted from wild-type virus-infected cells (Fig. (Fig.3B).3B). However, virus carrying the 2C-E253G mutation alone or in combination with the 2A-N32D mutation allowed the secretion of more than twice as much IL-8. The 2C-E253G was thus identified as the mutation responsible for the Sin− phenotype and its effect on secretion shown to be independent of the 2A-N32D mutation.
We have shown that 10-8-infected cells have increased the surface expression of a membrane-bound antibody and secrete higher amounts of a virally induced cytokine than wild-type virus-infected cells. This makes it very likely that the 10-8 virus is Sin−, that is, does not perform the Sin+ function of the wild-type virus in inhibiting cargo transport nonspecifically. However, to verify that these results were not due to a global increase in protein synthesis or to a failure of the mutant poliovirus to inhibit host translation, we monitored host and viral protein synthesis during infection with the wild-type, 10-8-reconstructed, and 2A-N32D mutant poliovirus. Poliovirus 2A proteinase is known to inhibit host cap-dependent translation partially through cleavage of the host proteins eIF4GI and -II (27, 48), and the cleavage products of eIF4GI then activate poliovirus IRES-mediated translation (23, 35) and stabilize polyribosomes formed on viral RNA (26). Cells were infected with the wild-type, 10-8, or 2A-N32D virus and pulse-labeled with [32S]methionine at various times postinfection (Fig. (Fig.4).4). During wild-type poliovirus infection, the inhibition of host translation became apparent from 2 to 3 h postinfection. The cap-independent translation of individual viral proteins could be seen by this time as well. For both the mutant viruses tested in Fig. Fig.4,4, the inhibition of host protein synthesis occurred slightly more rapidly, but this increase in the rate of host translation inhibition was small and did not result in the increased synthesis of viral proteins. Thus, the increased amounts of secreted protein from 10-8 viruses are not likely to result from an altered ability of the mutant 2A proteinase to inhibit host translation or to promote viral translation.
To determine the effects of the 2C-E253G and 2A-N32D mutations on viral growth and spread, the plaque sizes of the wild-type, 10-8, 2A-N32D, and 2C-E253G viruses were analyzed after 42 h of growth under agar on HeLa monolayers. As can be seen in Fig. Fig.5A,5A, the 10-8-reconstructed virus containing both the 2A and 2C mutations displayed plaques about equal in size to those formed by the wild-type poliovirus, suggesting that this Sin− virus had growth and spread properties similar to that of the wild-type virus. However, the Sin− poliovirus that contained the single 2C-E253G mutation produced plaques smaller than the wild-type plaques, and the 2A-N32D virus produced plaques larger than those of the wild type. This suggested that although our selection had isolated a Sin− mutant poliovirus with robust growth properties, this virus contained one mutation, 2C-E253G, that conferred the Sin− phenotype and another, 2A-N32D, that suppressed a defect in the growth or spread of the 2C-E253G virus.
We then explored the effects of the 2A-N32D and 2C-E253G mutations on viral growth in single-cycle infections, as opposed to the multiple cycles required to form a plaque. Intracellular yields of the wild-type, 10-8, 2A-N32D, and 2C-E253G viruses were determined after inoculation at either high (20 PFU/cell) or low (0.1 PFU/cell) MOIs over a time course. At high MOIs (Fig. (Fig.5B),5B), no difference in the growth of the wild-type virus was observed in comparison to that of the 10-8 virus or to any of the viruses that contained subsets of the 10-8 mutations. A possible exception was 2A-N32D, which displayed a drop in intracellular virus 5 to 6 h postinfection. No difference in viral growth was observed during the low-MOI infections (Fig. (Fig.5C),5C), in which the inclusion of the 3A-2 mutant virus confirmed its slight growth defect. Therefore, little information concerning the mechanism by which the 2C-253G mutation decreased viral growth or spread or by which the 2A-N32D mutation increased viral growth or spread was provided by the single-cycle growth curves.
To determine whether the increased spread of the 2A-N32D-containing viruses resulted from premature cell death after the first cycle of infection, infected cells were harvested at 8, 10, and 12 h postinfection and labeled with a fluorescently conjugated reagent that preferentially binds dead cells with permeable membranes (fixable Live/Dead stain; Invitrogen, Inc.). This particular stain was used because it survives fixation, allowing the inactivation of virus before analysis by flow cytometry. As shown in Fig. Fig.6A,6A, the numbers of dead cells infected with the wild-type or 2C-E253G mutant virus were very similar and increased steadily from 8 to 12 h postinfection, at which time approximately 25% of the infected cells were dead. However, cells infected with viruses that contained the 2A-N32D mutation (Fig. (Fig.6B)6B) started to die at 8 h, and most cells were dead by 12 h postinfection. The presence of the 2C-E253G mutation did not alter the premature cell death caused by the 2A-N32D mutation (Fig. (Fig.6B)6B) or change the rate of cell death compared to that of the wild-type virus during longer time courses (Fig. (Fig.6C6C).
To test whether the increased cell death observed during infection with viruses that contain the 2A-N32D mutation was apoptotic, we examined the effects of two pan-caspase inhibitors, zVAD-fmk, commonly used in caspase studies, and Q-VD-OPh, which has been reported to display reduced cellular toxicity (8), on the survival of wild-type virus-and 2A-N32D- infected cells. At 10 h postinfection, both compounds significantly reduced the amount of cell death observed in 2A-N32D-infected cells (Fig. 7A and B, left panels). As has been shown previously for zVAD-fmk (1), neither compound inhibited the intracellular yield of poliovirus in a single-cycle infection (Fig. 7A and B, right panels). Therefore, the early cell death induced in 2A-N32D-infected cells occurs via caspase-dependent apoptosis.
To explore any differences in caspase activation between wild-type virus-and 2A-N32D-infected cells, we used an assay that monitors the release of luciferin from an inactive form by specific proteolysis by initiator caspase-9, or effector caspases-3 and -7, which, having similar substrates, are not distinguished from each other (see Materials and Methods). As expected from previously published results (4, 38), little activation of the caspase-9 (Fig. (Fig.7C)7C) or caspase-3 and -7 (Fig. (Fig.7D)7D) was observed at 8 h after infection with the wild-type poliovirus. However, 2A-N32D-infected cells showed marked increases in caspase-9 activity, which is comparable to that observed following treatment with cycloheximide and TNF alpha to induce apoptosis (Fig. (Fig.7C).7C). In addition, 2A-N32D-infected cells show strikingly high increases in caspase-3 and -7 activity (Fig. (Fig.7D).7D). This dramatic effect on caspase activity occurs independently of other known functions of the 2A proteinase. As was shown in Fig. Fig.4,4, other functions of 2A, such as the inhibition of host translation and the enhancement of viral translation, were found to be unaffected by the N32D mutation. Therefore, we consider it likely that a wild-type function of 2A is to inhibit, directly or indirectly, the activation of caspases, and this effect is lost with the N32D mutation.
To determine whether the increased spread of 2A-N32D-containing viruses correlated with the increased rate of apoptotic cell death, the amount of extracellular virus released from the wild-type, 2A-N32D, 2C-E253G, and 2A-N32D/2C-E253G viruses was determined over time in the presence and absence of zVAD-fmk to inhibit caspase activation (Fig. (Fig.8A).8A). Only small amounts of extracellular virus were detected in any of the infections at 6 and 8 h postinfection. By 10 h postinfection, however, both viruses that carried the 2A-N32D mutation released virus, and by 12 h postinfection, nearly 20 times more virus was present in the medium of 2A-N32D-infected cells than of wild-type virus-infected cells. In the presence of zVAD-fmk, the amount of extracellular virus remained small for all viruses. These data strongly support the hypothesis that the increased spread of 2A-N32D virus is the result of the early caspase-dependent death of infected cells that leads to early virus release and early infection of neighboring cells.
To measure the cell-to-cell spread of the wild-type and mutant viruses directly, we determined the intracellular yield during a second round of infection (Fig. (Fig.8B).8B). The cells were infected at a very low MOI (0.01 PFU/cell) to ensure that only a fraction of the cells were infected during the first round of infection and many cells remained to be infected during a second round. Previous work had determined that, under these conditions, the second round of infection began to produce virus about 14 h after infection (data not shown). For the wild-type, 3A-2, and 2C-E253G viruses, the yield of intracellular virus, beginning at 12 h postinfection, increased slowly and gradually. However, the yield of the 2A-N32D and 2A-N32D/2C-E253G viruses in the second round of infection increased faster and was larger in magnitude at the time points measured than even the wild-type virus in parallel infections (Fig. (Fig.8B).8B). Therefore, the 2A-N32D mutation resulted in higher levels of both extracellular virus release from the first round of infection and spread to neighboring cells for a second round. Given that this increase in spread is likely to have been caused by the premature cell deaths of the 2A-N32D-infected cells, an explanation is provided for the early decrease in cell-associated virus observed for the 2A-N32D-containing viruses in the first infectious cycle at high MOIs (Fig. (Fig.5B5B).
The 2A-N32D mutation was almost certainly identified as a genetic suppressor of the small-plaque phenotype of the 2C-E253G Sin− virus (Fig. (Fig.5A).5A). If this genetic interaction were the result of an altered physical interaction between the 2A and 2C mutant proteins, it would be expected to be allele specific; that is, the 2A-N32D mutation should not be a general suppressor of other defects in growth or spread. To determine whether the 2A-N32D mutation can suppress the growth defects of other small-plaque viruses, the mutation was cloned into cDNA that encodes 3D-ΔIpalm (3D-D339,S341,D349A), a virus with a small-plaque phenotype caused by three mutations on the “palm” side of Interface I (36), a contact surface for polymerase-polymerase interactions. The plaque phenotypes of the wild-type, 2C-E253G, and 3D-ΔIpalm viruses in the absence and presence of the 2A-N32D mutation were determined after growth at either 37°C or 39.5°C for 46 h. As shown in Fig. Fig.9,9, the addition of the 2A-N32D mutation increased the plaque sizes of all viruses tested, giving rise to large plaques in a wild-type background and suppressing the small-plaque phenotype of both the 2C-E253G virus and 3D-ΔIpalm virus at both temperatures. This suppression was most striking for the 3D-ΔIpalm mutant virus, which was previously reported to display a small-plaque phenotype at 37°C (36) but displayed, in addition, a strong temperature sensitivity, forming plaques too small to see at 39.5°C (Fig. (Fig.9).9). In the presence of the 2A-N32D mutation, these plaques became readily apparent. Therefore, the 2A-N32D mutation, by virtue of increasing the rate of viral spread, acts as a non-allele-specific suppressor of growth defects in multiple-cycle assays.
One of the purposes of this study was to identify and characterize Sin− viruses with wild-type viral growth properties in tissue culture, so that they could be used in a mouse model for poliovirus infection to determine the benefits that poliovirus might derive by blocking protein secretion in an infected animal. In tissue culture, poliovirus infection inhibits the secretion of IL-6, IL-8, and beta interferon (15) and the cell surface expression of major histocompatibility complex class I and TNF receptor (13, 34). The increased secretion of any one of these factors, or others that have not yet been tested directly, could adversely affect the ability of poliovirus to grow and spread in an infected host with a competent immune system, possibilities that are now more amenable to experimental determination.
Using a FACS-based approach, we selected and characterized a Sin− poliovirus with wild-type virus-like growth properties. The 10-8 virus was termed a Sin− poliovirus because infected cells showed both increased cell-surface expression of a single-chain antibody and increased secretion of IL-8. In the 10-8 virus, the mutation that conferred the Sin− phenotype was an E253G change in the 2C protein. The 2C sequences are present both as processed 2C and as a stable uncleaved form, 2BC. Proteins 2BC, processed 2C, 2B, and 3A are all known to interact within the RNA replication complex. We hypothesize that the E253G mutant protein in 2C, 2BC, or a larger precursor interferes with the ability of the 3A and 2B moieties to block secretion.
Virus carrying the 2C-E253G mutation alone also displayed a defect in growth or spread that was difficult to discern in single-cycle infections but resulted in smaller plaques. Like the 2C-E253G virus, previously described Sin− virus 3A-2 also displayed a slight growth defect. We identified no Sin− mutant viruses in this screen that displayed robust growth; in fact, several putative isolates proved to be genetically unstable. Although this is a negative result, it is possible that the inhibition of protein secretion per se may play a direct role in viral growth, spread, or both. Further studies will be required to determine whether the inhibition of host protein secretion per se affects viral growth and spread, perhaps as a side effect of the membrane rearrangements that accompany poliovirus infection.
Coisolated with the 2C-E253G Sin− mutation was a 2A-N32D mutation that independently caused cells to die by apoptosis much earlier than wild-type virus-or 2C-E253G-infected cells. This is consistent with the hypothesis that a wild-type function of the 2A protein is to inhibit apoptosis late in infection, perhaps by directly or indirectly leading to the aberrant cleavage of procaspase-9, and this activity is abrogated by the 2A-N32D mutation. However, it remains formally possible that the 2A-N32D mutation is a gain-of-function mutation, providing a novel antiapoptotic function. The identification of additional substrates of the 2A and 2A-N32D proteinases will help to resolve this issue.
The 2A-N32D mutation in the original Sin− virus isolate was most likely selected for its ability to suppress the defective growth and spread phenotype caused by the 2C-E253G mutation. Classically, the extragenic suppression of a defective mutant phenotype can arise via the following three different classes of mechanism: informational, interactional, and bypass suppression (reviewed in references 21 and 22). Informational suppression is best exemplified by recoding strategies; for example, the suppression of phenotypes caused by nonsense mutations is how suppressor tRNAs were named. During interactive suppression, the defect in a cellular product is rendered functional via direct physical interaction with another product that is altered by the suppressor mutation. Although this is the most commonly sought class of genetic suppressor, additional evidence is needed to make the argument that the genetic interaction is the result of a physical interaction. One of the most diagnostic criteria for interactive suppressors is that they are highly allele specific. The third class, bypass suppression, occurs when the suppressing mutation is found in a different biochemical pathway than the original mutation that caused the growth defect or is downstream (epistatic) in the same pathway.
It seems likely that 2A-N32D is a bypass suppressor of the growth or spread defect of the 2C-E253G virus. First, the 2C-E253G mutation does not affect the rate of cell death and the 2A-N32D mutation does not affect the rate of host protein secretion. In addition, the 2A-N32D mutation is not allele specific: it can suppress the small-plaque phenotypes of both the 2C-E253G virus and another mutant virus, 3D-ΔIpalm. As can be seen in Fig. Fig.8,8, the presence of the 2A-N32D mutation led to the early release of virus, which correlated with increased viral spread. Therefore, we hypothesize that by failing to inhibit apoptosis, the 2A-N32D mutation can increase the rate of spread of any virus that contains it; this property allowed the selection of 10-8, a Sin− virus that displayed wild-type growth during the multiple-cycle infections used in the plaque purification of the isolated viruses. Bypass suppression is a common genetic strategy that is important to consider during the analysis of viral selection and evolution in cultured cells and in natural environments.
We thank Nicole Cobb for suggesting the allele specificity assay, David Kamm for assistance with the IL-8 assays, and Peter Sarnow, Ernesto Mendez, and Michel Brahic for critical readings of the manuscript.
This work was supported by NIH training grant AI-007328 (T.B.B.), the UNCF/Merck Graduate Science Research program (J.A.J.), ACS postdoctoral fellowship PF-04-040-01-GMC (J.F.S.), and NIH grant AI-25166 and an NIH Director's Pioneer award to K.K.
Published ahead of print on 22 July 2009.