Mengovirus and Encephalomyocarditis Virus Poly(C) Tract Lengths Can Affect Virus Growth in Murine Cell Culture

Abstract

Many virulent aphthoviruses and cardioviruses have long homopolymeric poly(C) tracts in the 5′ untranslated regions of their RNA genomes. A panel of genetically engineered mengo-type cardioviruses has been described which contain a variety of different poly(C) tract lengths. Studies of these viruses have shown the poly(C) tract to be dispensable for growth in HeLa cells, although the relative murine virulence of the viruses correlates directly and positively with tract length. Compared with wild-type mengovirus strain M, mutants with shortened poly(C) tracts grow poorly in mice and protectively immunize rather than kill recipient animals. In the present study, several murine cell populations were tested to determine whether, unlike HeLa cells, they allowed a differential amplification of viruses with long or short poly(C) tracts. Replication and cytopathic studies with four hematopoietically derived cell lines (CH2B, RAW 264.7, A20.J, and P815) and two murine fibroblast cell lines [L929 and L(Y)] demonstrated that several of these cell types indeed allowed differential virus replication as a function of viral poly(C) tract length. Among the most discerning of these cells, RAW 264.7 macrophages supported vigorous lytic growth of a long-tract virus, vMwt (C₄₄UC₁₀), but supported only substantially diminished and virtually nonlytic growth of vMC₂₄ (C₁₃UC₁₀) and vMC₀ short-tract viruses. The viral growth differences evident in all cell lines were apparent early and continuously during every cycle of virus amplification. The data suggest that poly(C) tract-dependent attenuation of mengovirus may be due in part to a viral replication defect manifest in similar hematopoietic-type cells shortly after murine infection. The characterized cultures should provide excellent tools for molecular study of poly(C) tract-mediated virulence.

Encephalomyocarditis virus (EMCV) and mengovirus are serotypically related cardioviruses of the picornavirus family. Among the unusual features of their positive-sense single-stranded RNA genomes are long 5′ untranslated polypyrimidine tracts [poly(C)] with sequences consisting of C₁₁₅UCUC₃UC₁₀ and C₄₄UC₁₀ for EMCV (strain R) and mengovirus (strain M), respectively (6, 13). These cardioviruses and their closely related cousins, the foot-and-mouth disease viruses of the Aphthovirus genus, are the only known eukaryotic or prokaryotic genomes to contain such poly(C) tracts, and the specific function of the homopolymer region remains a biological enigma.

We have reported the construction of multiple cDNA-derived EMCV and mengovirus strains which differ from each other and from wild-type parental strains only in the lengths of their 5′ poly(C) tracts (8, 9, 13). The mengovirus panel included nine viruses with poly(C) lengths that ranged from C₄₄UC₁₀ (vMwt) down to a precise deletion of all the cytidine residues (vMC₀). The EMCV panel (vEC₂₀, vEC₉, and vEC₄) was less extensive, although it contained representative analogues for the best-characterized mengovirus strains. As a definitive phenotype, deletion of the mengovirus poly(C) tract clearly correlates with attenuation of virus virulence in animals. vMC₀, for example, has a median 50% lethal dose (LD₅₀) of >2 × 10⁹ PFU after intracerebral inoculation of mice. In contrast, the LD₅₀ of vMwt by equivalent inoculation is only 10 PFU (7). All intermediate-tract mengoviruses show correspondingly diminished virulence. For viruses with poly(C) tracts between 25 and 35 bases, the LD₅₀ in mice increases about 1 log₁₀ PFU for every 3 C's that are removed from the tract (13, 20). For EMCV strains, the correlation between tract length and murine virulence is weaker, however, and the poly(C) needs to be truncated substantially (i.e., <C₉) before the attenuation becomes measurable (e.g., the LD₅₀ of vEC₄ is 3 × 10³ PFU compared to 1 PFU for EMCV-R) (9).

Our EMCV and mengovirus recombinant isolates have been extensively characterized for growth in HeLa cells, and all were found to plaque with equivalent plating efficiencies regardless of the poly(C) tract length. While some strains do exhibit subtle changes in plaque size that correlate with incremental tract deletion (9, 13), none of the isolates show tract-dependent variations in replication kinetics or end-point titers when measured directly in single-step growth experiments. The poly(C) tracts also have no apparent influence on genome translation, virion stability, or growth temperature sensitivity, and it is clear, at least for growth in HeLa cells, that the major vegetative life cycle requirements for EMCV and mengovirus are not strongly vested in this region of the viral RNA.

However, in infected animals, especially those receiving mengoviruses, there must be some cellular or tissue determinant that rapidly detects subtle differences in poly(C) genotype and reacts in a manner that clearly means life or death for inoculated individuals. The poly(C)-dependent phenotypes are apparent shortly after murine infection. Viremic differences, for example, are measurable between vMwt- and vMC₀-inoculated animals by 8 h postinoculation (p.i.) (19). This implies that viruses with long or short poly(C) tracts are discriminated against very early after inoculation, clearly within the first rounds of cellular infection. Within 1 day after intracerebral injection, the titers of all mengovirus strains in the brain show evidence of some replication. However, thereafter the titers diverge rapidly, and the vMC₀ virus load drops precipitously relative to the vMwt load (7, 21). This again suggests that the short-tract viruses are somehow disadvantaged soon after they enter the mouse. Presumably at each infectious cycle (4 to 6 h in HeLa cell culture) the short-tract viruses continue to lose ground relative to long-tract viruses until the host immune system moderates the infection and ultimately clears the virus (1 to 2 weeks p.i.). Growth of the long-tract viruses, in contrast, is not similarly impaired, and they continue to replicate until the animal dies. The first infected cells within the murine host are presumed to be critical to the manifestation of these phenotypic differences, and we now report the identification of several murine cell lines, including a line of macrophage origin, that allow greater amplification of long-tract over short-tract mengovirus and EMCV strains. We believe that these cell lines hold excellent promise for the molecular examination of the poly(C) phenomenon.

MATERIALS AND METHODS

Viruses.

Recombinant mengoviruses vMwt, vMC₂₄ and vMC₀ have been described previously (8, 13). They have identical genotypes except for 5′ poly(C) tracts of C₄₄UC₁₀, C₁₃UC₁₀, and C₀, respectively. Similarly, recombinant vEC₂₀ and vEC₄ have poly(C) tracts of C₂₀ and C₄, respectively, and differ from EMCV-R (C₁₁₅UCUC₃UC₁₀) only in this region (9). All stocks were amplified in HeLa cells and concentrated by centrifugation through 30% sucrose cushions. Titers (PFU per milliliter) were determined by triplicate plating on HeLa cell monolayers (22) before each experiment and for all viruses propagated in the various cell types.

Cell lines.

All cells except HeLa were maintained in RPMI 1640 medium (Sigma, St. Louis, Mo.) supplemented with 100 U of penicillin per ml, 100 μg of streptomycin per ml, 2 mM l-glutamine, 50 μM 2-mercaptoethanol, and 10% fetal calf serum (RPMI 1640–10% FCS). P815 (ATCC TIB-64) is a major histocompatibility complex (MHC) class I+ murine mastocytoma cell line. CH2B (a gift from Daniel Muller, University of Wisconsin—Madison) is an MHC class I+/II+ murine (H-02^b) B-lymphoma cell line (16). RAW 264.7 (ATCC TIB-71) is an MHC class I+ murine monocyte-macrophage cell line (a gift from Donna Paulnock, University of Wisconsin—Madison). A20.J (ATCC TIB-208) is an MHC class I+/II+ murine (H-2^d) B-lymphoma cell line. L(Y) (a gift from Philip Marcus, University of Connecticut) is a murine fibroblast line. L929 is a murine fibrosarcoma cell line (ATCC CRL-2148). HeLa (ATCC CCL-2) is a human cervical carcinoma line that was maintained in minimal essential medium with 10% FCS, as previously described (22).

Single-step growth kinetics.

RAW 264.7 macrophages (10⁷ cells) were inoculated with 5 × 10⁸ PFU (multiplicity of infection [MOI] = 50 PFU/cell) of mengovirus in RPMI 1640 medium (1.5 ml). Virus attachment was carried out at 25°C for 30 min at a rotating platform (74 rpm). Unattached virus was removed by two washes with RPMI 1640, and the cells were reconstituted to 5 × 10⁵ cells/ml in RPMI 1640–10% FCS. Aliquots (1 ml) were added to 30-mm tissue culture plates (Nunc, Rochester, N.Y.) and incubated at 37°C under 5% CO₂. At the indicated intervals, duplicate plates were transferred to −20°C to halt the infection. When all samples had been collected, the monolayers were thawed, scraped, transferred to microcentrifuge tubes, and frozen-thawed two more times in an ethanol–dry-ice bath. Cellular debris was removed by low-speed centrifugation, and supernatant virus titers were determined by a standard plaque assay on HeLa cells (22).

Relative virus growth.

Cell monolayers at ∼80% confluence (∼5 × 10⁶ cells/30-mm plate) were rinsed with phosphate-buffered saline (PBS) and inoculated with virus at an MOI of 10 PFU/cell (in 250 μl of PBS). Virus attachment proceeded at 25°C for 30 min, after which unattached virus was removed by two washes with RPMI 1640. The cells were overlaid with 5 ml of fresh RPMI 1640–10% FCS and transferred to 37°C and 5% CO₂. At intervals during the next 5 days, culture supernatants were aspirated and collected, and (remaining) monolayers were rinsed twice with RPMI 1640 before being overlaid with 5 ml of fresh RPMI 1640–10% FCS. Cytopathic effect (CPE) was scored for each culture at each time point. The aspirated supernatants were made cell free by centrifugation through 0.22-μm-pore-size microfiltration tubes (Costar, Acton, Mass.), and aliquots (0.5 ml) were removed and stored at 4°C before virus titer determination by a plaque assay. The remainder (2.5 ml) was frozen for subsequent interferon (IFN) titer determinations.

Infectious center assay.

Cells infected as described above were serially diluted and seeded along with 5 × 10⁶ HeLa cells into 60-mm tissue culture plates (Falcon). The cultures were left undisturbed for 2 h at 37°C under 5% CO₂ and then rinsed three times with PBS before an agar overlay was added, as for a normal plaque assay. The plaques were counted after an additional 31 h at 37°C.

Indirect immunostaining.

RAW 264.7 and HeLa cell monolayers were infected at an MOI of 10 PFU/ml in six-well plates. At 4 h p.i., the monolayers were rinsed three times with PBS and fixed for 1 h with 2% formaldehyde at 25°C. The monolayers were again washed three times with PBS and then incubated for 10 min with 0.2% Nonidet P-40. Finally, the cells were washed five times with PBS, treated with 6% glycine for 2 h, and washed five more times with PBS. Incubation with 3% bovine serum albumin for 1 h at 37°C was followed by incubation with the primary antibody (rabbit polyclonal anti-mengovirus serum [13]), at a 1:1,000 dilution in 3% bovine serum albumin, for 2 h at 37°C and five subsequent washes with PBS. The secondary antibody (goat anti-rabbit immunoglobulin G conjugated to horseradish peroxidase) was diluted 1:1,000 in 3% BSA and incubated with the plates for 1 h at 37°C. Afterwards, the cells were again washed five times with PBS. Staining was then carried out with DAB solution (1.4 ml of 3,3′-diaminobenzidine in dimethyl sulfoxide at 10 mg/ml, 7.6 ml of 50 mM Tris [pH 7.6], 1.0 ml of 0.3% NiCl₂, 10 μl of H₂O₂), and was stopped by rinsing with PBS.

IFN bioassays.

Alpha/beta interferon (IFN-α/β) levels were determined by a bioassay on mouse L(Y) cells as described previously (23). The data were recorded as units of IFN-α/β per milliliter of supernatant and were the average of three parallel determinations. Murine IFN (Biosource, Camarillo, Calif.) was used as both a standard and internal control. In these assays, 1 measured unit was equivalent to 1 standard NIH unit of IFN-α/β.

RESULTS

Relative virus amplification.

Recombinant mengoviruses grow very well in HeLa cells regardless of their poly(C) tract lengths. Viruses with long (vMwt), short (vMC₂₄), and deleted (vMC₀) tracts were infected into murine A20.J (B-lymphoma), P815 (mastocytoma), CHB2 (B-lymphoma), RAW 264.7 (macrophage), and L(Y) (fibroblast) cells, in parallel with HeLa cells, and the titers of amplified viruses were determined at intervals over the next 5 days. Figure 1A shows representative profiles of virus synthesis after infection of three of these lines. Each point records the released titer since the previous sample. The RAW 264.7 and P815 cells typically gave peak titers for all three viruses within 24 h p.i., as did the CHB2, L(Y), and HeLa cells (results not shown). After this time, most cells in the vMwt-infected cultures were dead, and by 48 h there were no further monolayers to wash or measure. The A20.J line was somewhat different in that it gave maximum yields only after 24 to 48 h p.i. By 48 h, however, these cells too had all been killed by vMwt. At 24 h p.i., the HeLa and L(Y) cells were completely lysed during the vMC₂₄ and vMC₀ infections, but the parallel infections in the other cell lines induced negligible if any CPE in the same period (data not shown). Clearly, virus infections of the hematopoietically derived cultures yielded different phenotypes for the attenuated viruses, in terms of both virus output and CPE.

FIG. 1.

Virus replication assays. (A) Cell monolayers (A20.J, P815, and RAW 264.7 cell lines) were infected in parallel with vMwt, vMC₂₄, and vMC₀ as described in Materials and Methods. Each point represents the amount of virus released into the supernatant since the previous point, as determined in triplicate by standard plaque assays on HeLa cells. Open symbols denote points at which >80% of the original cell monolayer was destroyed. (B) Comparison of maximum titers attained during virus replication in A20.J, P815, RAW 264.7, CHB2, L(Y), and HeLa cell lines. Each line was infected with vMwt, vMC₂₄, and vMC₀ as described in Materials and Methods. Supernatant aliquots were taken at 24, 48, 84, and 132 h p.i., and virus titers in each sample were determined by plaque assay (triplicate samples) on HeLa cells. The maximum titers (24-h time point for most cell lines; 48-h time point for A20.J) are normalized relative to those achieved by vMwt and are shown as a percentage. Typical replicate platings or repeat experiments varied in titer by <1% for any of the viruses or infectious passages.

For comparative purposes, the 24-h titers for all viruses and cell lines (48 h for A20.J infections) are replotted in Fig. 1B. In every case, vMwt propagated more efficiently than vMC₂₄, which in turn grew better than vMC₀. The poly(C)-dependent differences were always stronger in the A20.J, P815, CHB2, RAW 264.7, and the L(Y) cells than in the HeLa cells. In particular, these lines consistently allowed an 8- to 15-fold-greater amplification of vMwt than of vMC₀. Although all tested cells amplified vMC₂₄ to intermediate titers, the A20.J and P815 cells were more pronounced in their discrimination against vMC₂₄ and reproducibly gave lower relative titers compared to those achieved with CHB2, RAW 264.7, L(Y), and HeLa cells. Most notably, RAW 264.7 cells were the most refractory to CPE induced by vMC₂₄ and vMC₀ even at 120 h p.i. (Fig. 1A). Among these cell lines, RAW 264.7 and L(Y) were chosen for further analysis and comparison to HeLa cells because they formed monolayers stable enough for plaque assays.

Relative plating efficiencies.

Multiple rounds of infection must occur before a plaque can be observed during a viral plaque assay. This allows a greater opportunity to manifest genetic defects or cellular responses that may influence viral growth. The relative plaque sizes and plating efficiencies of our poly(C) viruses were assessed in HeLa, RAW 264.7, and L(Y) cells and additionally in L929 cells, a murine fibrosarcoma line (Table 1). Mengovirus titers determined with HeLa cells are always the highest of our cell lines; therefore, HeLa cells are used as benchmarks for both virus titer and plaquing efficiency relative to those attained with other cell lines. In HeLa cells, the derived virus titers were equivalent for all three viruses and were defined as 100%. When the same viral stocks were tested in parallel on the other three cell monolayers, however, there was a 5- to 25-fold decrease in plaquing efficiency. Both of the L cell lines yielded plaques at only about 10 to 20% the HeLa cell titer, although there were no significant differences between the efficiencies of vMwt and vMC₀. In general, plaque sizes of 1.8, 2.1, and 2.7 mm in diameter for vMC₀, vMC₂₄, and vMwt, respectively, have been reported for infections of HeLa cells (13). Similar poly(C)-dependent plaque size differences were also seen in the L929 cells, with values corresponding to those in the HeLa cells (∼2 to 3 mm). However, interestingly, vMwt and vMC₀ plaqued to an equivalent size in L(Y) cells at an indistinguishable ∼2.5 mm in diameter, even though vMC₀ was not amplified to similar titers to those of vMwt in L(Y) cells (Fig. 1B). In contrast to these fibroblast lines, the RAW 264.7 cells developed no visible plaques with either vMC₂₄ or vMC₀, although parallel infection with vMwt yielded plaques of at least 2 to 3 mm in diameter in these monolayers. This suggests that RAW 264.7 cells are productively but not lytically infected by vMC₂₄ or vMC₀.

TABLE 1.

Plaque formation of vMC₀, vMC₂₄, and vMwt in selected cell lines

Virus	Plaquing efficiency (%) ona:				Relative plaque size onb:
	HeLa	RAW 264.7	L(Y)	L929	HeLa	RAW 264.7	L(Y)	L929
vMwt	100	3.6	9.4	16.6	+++++	+++++	+++++	+++++
vMC24	100	0	NDc	ND	++++	None	ND	ND
vMC0	100	0	12.0	20.0	+++	None	+++++	+++

^aMonolayers of each cell type were inoculated with serial dilutions of virus stocks, and a standard plaque assay was performed. The resultant plaques were counted for each virus dilution, and respective plaquing efficiencies were calculated relative to the yield for HeLa cell monolayers [(PFU per milliliter on the tested cell)/(PFU per milliliter on HeLa) × 100]. 

^bPlaque sizes were measured as described in Materials and Methods, and the relative sizes are represented by plus signs according to the following key: +++++, ∼2.7 mm; ++++, ∼2.1 mm; +++, 1.8 mm. 

^cND, not done. 

Productively infected cells.

The members of the panel of poly(C) mengoviruses have identical capsid sequences and are antigenically indistinguishable. Still, it was possible that the diminished growth of vMC₀ and vMC₂₄, relative to vMwt, on RAW 264.7 cells was a simple matter of infection frequency that resulted from differential attachment or penetration. As a control, the percentage of productively infected cells was quantified by infectious-center assays (Table 2). When infected in suspension culture, about 25% of all HeLa cells subsequently were able to initiate a plaque, a level that was relatively constant among all the tested viruses. The RAW 264.7 cells formed infectious centers at about half this frequency, but again, there was no distinction among the different virus genotypes. Consistent with these data, immunocytochemical staining of infected HeLa monolayers at 4 h p.i. showed that productive infections could be detected in about 30% of cells, regardless of the poly(C) tract size of the infecting virus (Table 2). Again, in all cases, only half as many RAW 264.7 cells showed positive infections. We conclude that poly(C)-dependent growth differences among our viruses take place at some intracellular phase after virus attachment and entry into the host cell.

TABLE 2.

.Determination of productively infected cells

Virus	% of cells forming infectious centersa		% of infectivity that was unattachedb on:		% of cells infected by immunocytochemical assayc
	HeLa	RAW 264.7	HeLa	RAW 264.7	HeLa	RAW 264.7
vMwt	24.4	9.7	0.3	>99.9	33.1	13.3
vMC24	29.2	10.2	0.2	>99.9	29.6	14.9
vMC0	25.2	10.3	0.1	>99.9	32.2	12.9

^aAliquots were taken from infected-cell cultures used for the single-step growth experiments (Fig. 2) and assayed for the percentage of total cells producing infectious centers in HeLa monolayers. 

^bUnattached infectivity was aspirated after infection of cell monolayers, and viral titers were determined relative to the input virus titer. 

^cHeLa or RAW 264.7 cell monolayers were infected and stained as in Materials and Methods. Rabbit anti-mengovirus polyclonal serum was used to detect antigen-positive cells in the virus growth studies at 4 h p.i. Mock-infected, primary-antibody-only, secondary-antibody-only, and DAB-only treated cells served as controls for nonspecific staining, and UV-inactivated virus controlled for nonspecific binding of virus to cells. The percentage of positively stained cells is indicated. 

Interestingly, these experiments exposed a key difference between the infection of HeLa and RAW 264.7 cells. During a control step in the infectious-center assays, the respective cells were washed after a defined attachment period and the material collected in this wash was assayed for infectivity (Table 2). With the HeLa experiments, less than 1% of the input virus was recovered at this stage, meaning that >99% had become associated with the cells or had become otherwise inactivated. For the RAW 264.7 cells, the opposite was true and >99.9% of the input infectivity was recovered in the wash. Although the number of infectious centers differed between these cells only by a factor of ∼2 to 3, the differences in viral binding may indicate that HeLa cells have denser, more effective receptor populations or perhaps more rapid endocytosis pathways than do the RAW 264.7 cells.

Single-step growth kinetics.

Single-step growth experiments can give a sensitive and comparative measure of virus amplification rates since the infection is synchronous and limited to only one infection cycle. The kinetics of vMC₀, vMC₂₄, and vMwt growth in HeLa cells have been reported (13) and emphasize that these particular cells are not particularly sensitive to variation in poly(C) tract length (Fig. 2, inset). When the same viruses were assayed in a single-step infection of RAW 264.7 cells, all emerged from eclipse at similar times p.i. but their respective growth rates during the logarithmic expansion phase correlated strongly with the poly(C) tract length (Fig. 2). From 4 h p.i. onward, the virus titers differed in a highly significant fashion, as is obvious from the graphs and also as measured statistically by both the Student t test and the nonparametric Wilcoxon test (P < 0.0001 in both analyses). End-point titers of vMwt and vMC₀ again differed by about 1 log₁₀ unit, which is similar to the observed gap in the continuous-infection experiments (Fig. 1A and B). The intermediate-tract virus, vMC₂₄, had a comparable, intermediate growth rate.

FIG. 2.

Single-step growth in RAW 267.4 cells. Cells were infected as described in Materials and Methods. The data are plotted as mean PFU/cell, with standard deviation. The inset depicts single-step growth results for the same three viruses after infection of HeLa cells.

IFN induction and CPE.

There are many factors which singly or in concert can influence the ability of a virus to replicate in cell culture. Clearly, the short-poly(C)-tract viruses were not impaired in their fundamental ability to infect HeLa or RAW 264.7 cells compared to vMwt. Rather, they seemed to encounter some replicative block that was present or more pronounced in the cells of hematopoietic origin. Mengoviruses, and indeed all cardioviruses, are known to be exquisitely sensitive to the antiviral effects of IFN-α/β, a cytokine that is readily induced in all hematopoietically derived cells. Therefore, we tested whether viruses with different poly(C) sizes could differentially upregulate this activity. The virus amplification experiment in Fig. 2 was repeated with RAW 264.7 cells, except that this time the supernatant samples were tested not only for virus titer (Fig. 3A) but also for IFN (Fig. 3B). The cells were also scored for comparative CPE (Fig. 3C). Upon infection with vMwt, almost all cells underwent necrotic death by 16 h p.i. Although this was enough time to amplify high titers of virus, the cells lysed before they would have had time to synthesize or release detectable amounts of IFN. In contrast, cells infected with vMC₂₄ or vMC₀ showed marginal or no CPE. Occasionally, a few of the vMC₂₄- and vMC₀-infected cells became slightly rounded or nonadherent as they underwent an apparently mild degree of stress in response to virus infection. However, these cells had limited mortality throughout the course of the experiment and eventually responded to infection with a measurable IFN release.

FIG. 3.

Comparison of vMwt, vMC₂₄, and vMC₀ growth in RAW 264.7 cells. (A) RAW 264.7 cells were infected with the indicated viruses as described in Materials and Methods. At 24, 48, and 72 h p.i., supernatant aliquots were removed, the virus titer was determined, and parallel aliquots were frozen until used in IFN bioassays. (B) Bioactive IFN-α/β titers were determined by the methods of Sekellick and Marcus (23) and calibrated relative to a control, IFN standard (Biosource). Data are presented as NIH units of IFN per milliliter. (C) The CPE in the infected RAW 264.7 monolayers was scored empirically on a relative scale compared to a mock-infected, control monolayer (+++++, >99% dead; +, < 10% dead). The vMwt IFN titers and CPE scores were limited to the 0- and 24-h time points, because all cells were dead after this time.

Infection with EMCV.

A panel of EMCV strains with short poly(C) tracts has also been engineered and tested in mice for poly(C)-dependent attenuation (9). In contrast to mengovirus, EMCV is not attenuated for virulence unless the poly(C) tract is very short (e.g., C₄). It was therefore of interest to determine whether virulence potential or the specific poly(C) lengths correlated more directly with growth in a cell line known to allow differential growth of mengovirus poly(C) mutants. EMCV-R, vEC₂₀, and vEC₄ were introduced into RAW 264.7 cells and progeny virus titers were determined as in the mengovirus experiments (Fig. 4). Like vMwt, EMCV-R killed virtually all cells within 24 h p.i. and replicated to titers significantly higher than those of either of the short-tract EMCV strains. vEC₂₀ also amplified well, but like its mengovirus counterpart, vMC₂₄, this EMCV strain did not induce visible CPE, nor did vEC₄, which gave even lower titers. Thus, cellular discrimination of poly(C) sequences is inherently dependent upon the tract length itself, and the consequent virulence or lack thereof must be a subsequent event linked to this activity.

FIG. 4.

Relative growth of EMCV-R, vEC₂₀, and vEC₄ in RAW 264.7 cells. Cells were infected as described in Materials and Methods. Maximum titers were reached at 24 h p.i. and are shown as a percentage normalized to the value for EMCV-R.

DISCUSSION

The mengovirus 5′ poly(C) tract length is strongly associated with the virulence of the virus. Recombinant mengoviruses with shortened poly(C) tracts are highly attenuated and have proven to be excellent, genetically stable, live vaccines against all serologically related EMCV strains. In many types of animals, including mice, pigs, baboons, and macaques, recombinant strains like vMC₂₄ and vMC₀ induce robust humoral and cellular responses that clear the infecting agent and provide long-term protective immunity (7, 17, 21) with negligible, if any, histopathology (1, 20). The mengovirus poly(C) phenomenon has been additionally exploited for the development of live chimeric vectors that can safely deliver immunologically relevant epitopes from diverse heterologous pathogens such as human immunodeficiency virus type 1, pseudorabies virus, and lymphocytic choriomeningitis virus (1, 2).

However, the molecular mechanism of this attenuation is still unknown. A better understanding of the poly(C) phenomenon has been difficult to achieve without an amenable experimental system. Indeed, our initial characterization of short-tract mengoviruses in cultured HeLa cells failed to reveal any differences in the relative growth of long-tract and short-tract viruses (8). It was not until we constructed an extensive panel of poly(C) mutants, including one with a complete tract deletion, that we recognized any degree of growth aberrance in tissue culture, and even then, the HeLa cell plaque size differences were subtle and difficult to quantitate (9, 13). Therefore, our most sensitive system for the analysis of poly(C) phenotypes was in vivo within mice. Identification of cell lines that discriminate against mengoviruses containing long or short poly(C) tracts now provides a more humane in vitro system to study the poly(C) phenomenon.

Since macrophages and other cells of monocyte origin are known to be linked to virulent mengovirus and EMCV infections (4, 15), these cells seemed a reasonable first choice for these experiments. In particular, RAW 264.7 cells reproducibly behaved as a faithful biological medium for the differential and progressive amplification of mengoviruses according to poly(C) tract length, as evidenced by bulk viral infections, CPE, single-step growth kinetics, plaque formation, and IFN induction after infection. The resultant progeny viruses had the same poly(C) tract lengths as did their input parental viruses (data not shown), and moreover, these particular cells passaged easily in culture and adhered well to plates, properties that also allowed direct comparison of infection frequencies among the virus panel. According to infectious-center experiments and in situ immunostaining, the long- and short-tract poly(C) isolates were equally adept at targeting these cells. Interestingly, though, the infection frequencies were only about half that of HeLa cells, perhaps indicating a lower or altered display of the VCAM-1 receptor and other factors required for viral entry (10) or indicating that only a subset of RAW 264.7 cells is capable of being infected at any time point.

Other cell lines in addition to RAW 264.7 were refractory to infection by the viruses with truncated poly(C) tracts. These include A20.J and CHB2, both of the B-cell lymphoma lineage but of different haplotypes (H-2^d and H-2^b, respectively); P815, a murine mastocytoma cell line; and J774, a monocyte line (data not shown). An experimental constant shared by these cells was their rapid lysis by vMwt but marginal CPE in response to vMC₂₄ or vMC₀. In contrast, L(Y), L929, and HeLa cells were quickly lysed by every virus. As we have previously reported with HeLa cells, the murine L lines did seem to recognize poly(C) genotypes, but their general response was not as dramatic or as sustained as with the hematopoietically derived lines. For this reason, we believe that the natural discriminatory poly(C) mechanism may rely more on the specifics of the infected cell or its developmental origin than on the particular type of host (e.g., murine versus human). Wild-type mengovirus can infect and kill many types of mammals, including primates, yet the short-poly(C)-tract viruses exhibit attenuation in all tested species (3, 20). The poly(C) mechanism therefore must be common in all these animals. Preliminary experiments with explanted murine peritoneal cells are beginning to give data similar to those obtained with the RAW 264.7 cells, in that adherent populations of these cultures seem to differentially amplify viruses of different poly(C) tract lengths (data not shown). In turn, this suggests that a natural cell or cell population within infected animals, and perhaps of hematopoietic origin, may promote or allow differential growth of these viruses. Undoubtedly, multiple rounds of subsequent replication serve to amplify the effects of the initial infectious cycle and ultimately lead to the phenotypically observed differences in virulence. According to their impaired replication potential in certain of these cells, short-tract mengoviruses are apparently disadvantaged from the moment they infect the host.

The infectious-center assay of infected RAW 264.7 cells clearly showed that all cells were equally permissive to infection, regardless of the poly(C) tract length. Discrimination then becomes manifest only after the penetration and uncoating steps. The shortest-tract mengovirus and EMCV strains failed to replicate inside these cells to the same high titer as did their long-tract counterparts. We still do not know whether the short-tract viruses are simply more effective at tripping some important protective element within infected cells or whether they are defective in some critical replication step that is not an obvious requirement for growth in HeLa cells. Perhaps it may be even more accurate to suggest that the long-poly(C)-tract viruses have enhanced growth properties rather than to conclude that the short-tract viruses have an impaired replicative capacity. Either way, the newly characterized cell types described here may provide the essential tools for dissecting the molecular mechanism of poly(C)-mediated virulence.

All cardioviruses are notoriously susceptible to the antiviral effects of IFN. The exact mechanism by which these effects are manifest during infection is unclear, but at least one cellular response to IFN treatment is the upregulation of a double-stranded RNA-activated protein kinase, PKR. This enzyme is constitutively expressed at low levels in most mammalian cells, where it acts as an intracellular sentinel against viral infection. Among other activities, PKR can help downregulate host and viral protein translation and can mediate further cycles of IFN induction (5, 12, 24). Most double-stranded RNA sequences will activate PKR, but among the best natural substrates are homopolymers of poly(C-G) that are at least 30 bp in length (14). Since PKR is present within the target cells for cardioviruses, it seems illogical that viruses with long poly(C) tracts should be more pathogenic than short-tract viruses, because the long poly(C) tracts should activate PKR more effectively. We think that this inconsistency might be a key to the poly(C) mechanism. One possibility is that the wild-type poly(C) tract [or a poly(C-G) replication intermediate] is actually designed as a molecular decoy that could bind and inactivate PKR in a manner similar to adenovirus VA_I RNA (11, 18). If the antiviral functions of activated PKR were prevented in such a manner, the long-tract viruses would have a significant replicative advantage over the short-tract viruses, just as we observed in the hematopoietic cell lines.

However, in these cells, vMC₂₄ induced higher levels of IFN than vMC₀ did. Perhaps this differential induction only reflects higher levels or multiple rounds of viral replication by vMC₂₄ in cell culture. Alternatively, this observation could indicate that a specific activation of PKR by the longer-poly(C) tracts is the real key to cardiovirus pathogenesis. Obviously, any link between poly(C) tract length and a PKR-dependent pathogenesis mechanism is premature at present, but the tissue culture systems described here can now provide excellent poly(C)-sensitive experimental tools that may enable us to solve this complex puzzle.