Hepatitis a Virus Translation Is Rate-Limiting for Virus Replication in MRC-5 Cells

Virology 254, 268–278 (1999) Article ID viro.1998.9548, available online at http://www.idealibrary.com on

Hepatitis A Virus Translation Is Rate-Limiting for Virus Replication in MRC-5 Cells

Ann W. Funkhouser,*,1 Derk E. Schultz,†,2 Stanley M. Lemon,† Robert H. Purcell,* and Suzanne U. Emerson*

*Hepatitis Viruses Section, Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, Bethesda, Maryland 20892; and †Department of Microbiology and Immunology, The University of Texas Medical Branch at Galveston, Galveston, Texas 77555-1019 Received July 13, 1998; returned to author for revision August 12, 1998; accepted November 30, 1998

Translation of hepatitis A virus (HAV) RNA is controlled by an internal ribosome entry site (IRES) located within the 5Ј untranslated region (UTR). In some cell types, the characteristically slow growth of HAV may be due to inefficient viral translation. We investigated whether this is true in MRC-5 cells, which are used for vaccine production. We measured the impact of two clusters of mutations in the 5Ј UTR on virus translation and replication: the AG group was selected during passage in African green monkey kidney cells, and the MR group was selected during subsequent passage in MRC-5 cells. The efficiency of cap-independent translation was assessed by inserting cDNA encoding an HAV IRES upstream of the chloramphenicol acetyl transferase gene and transcription was driven in vivo by a hybrid T7/vaccinia virus system. A luciferase gene was inserted upstream of the IRES to serve as an internal control. Each HAV UTR was also inserted into an infectious cDNA clone; the average rate of viral RNA accumulation was determined for each mutant virus. In MRC-5 cells, the rate of virus replication was highly correlated with the efficiency of cap-independent translation (P ϭ 0.006). The MR but not the AG mutations significantly increased both translation and viral RNA accumulation. Reversion of just one MR mutation (687 G to A) eliminated all of the replication-stimulating and translation-enhancing effects of the MR mutations. In the control BS-C-1 cells, there was no discernible correlation between the rate of virus replication and the efficiency of cap-independent translation (P ϭ 0.136): the AG and MR groups combined had a small impact on translation, but no detectable impact on virus replication. We conclude that in MRC-5 cells viral translation is rate-limiting for HAV replication. © 1999 Academic Press

INTRODUCTION these mutations caused a maximum one-log increase in replication in BS-C-1 cells (Day et al., 1992). Schultz et al. Hepatitis A virus (HAV) is the only member of the (1996b) showed that these same mutations increase genus Hepatovirus in the family Picornaviridae. For rea- translation in BS-C-1 cells. Together, these published sons unknown, HAV replicates very inefficiently in cell data suggest a correlation between translation and HAV culture (Emerson et al., 1991). HAV also has an inefficient replication in BS-C-1 cells. In fetal rhesus kidney-4 internal ribosome entry site (IRES) compared with other (FRhK-4) cells, however, there are data suggesting that picornaviruses such as encephalomyocarditis virus (EMCV) (Whetter et al., 1994) and poliovirus (Schultz et HAV replication is not limited by the efficiency of virus al., 1996b). In vitro, replacement of the HAV 5Ј UTR with translation. Jia et al. (1996) found that insertion of an the more efficient EMCV IRES increases initiation at the efficient IRES (from EMCV) in front of the open reading correct polyprotein start site (Jia et al., 1991). However, it frame of HAV does not increase replication in FRhK-4 is unclear whether the extremely slow replication of HAV cells. In addition, Schultz et al. (1996b) found that muta- in cell culture is due to inefficient translation, to ineffi- tions that increase translation efficiency in BS-C-1 cells ciencies in other stages of the virus cycle, or to both. have no effect on translation efficiency in FRhK-4 cells. Viral translation limits the rate of viral replication in Slow growth and relatively low virus yields hamper many picornavirus systems (Simoes and Sarnow, 1991; production of HAV vaccine stocks in MRC-5 cells. Thus, Pilipenko et al., 1992; Todd et al., 1997). In the case of it is important to determine the restrictions to growth of hepatitis A virus, the importance of translation may re- HAV in this particular cell type. In one vaccine strain of flect the particular cells used. Day et al. (1992) showed HAV, we have discovered four naturally occurring muta- Ј that mutations at bases 152 and 203–4, and at base 687 tions in the 5 UTR clustered on a putative stem-loop in the 5Ј UTR, increase the radioimmunofocus size of before the translation start codon (A to G at position 591, HAV in BS-C-1 cells. Under one-step growth conditions, G to A at position 646, C to U at position 669, and U to G at position 687, Fig. 1). These mutations (the MR group) occurred when the HM175 strain of HAV was serially passaged in MRC-5 cells (human fetal lung fibroblasts) 1 To whom reprint requests should be addressed at present address: Section of Infectious Diseases, Department of Pediatrics, University of and are necessary for efficient replication of HM175 in Chicago, 5841 S. Maryland Avenue, Room C638B, Chicago, IL 60637. those cells (Funkhouser et al., 1994). The MR mutations 2 Present address: Qiagen Inc., Santa Clarita, CA 91355. have no discernible impact on virus replication in FRhK-4

FIG. 1. Diagram of putative secondary structure of the 5Ј untranslated region of hepatitis A virus (Brown et al., 1991). Mutations designated by black circles and nucleotide location; domains designated by Roman numerals; clusters of AG and MR mutations designated by boxes. cells (Funkhouser et al., 1996). Another group of muta- merase was provided by infection of cells with a recom- tions (U to C at position 124, deletion of bases 131 binant modified vaccinia virus Ankara (MVA/T7). The through 134 and 203, and A to G at position 152, Cohen translation efficiency of each mutant IRES was compared et al., 1987, Fig. 1) is also present in the 5Ј UTR of the to that of the wild-type in MRC-5 cells. BS-C-1 cells were MRC-5-adapted virus. These mutations (the AG group) also assayed because of the previous data showing that occurred during passage in African green monkey kid- HAV translation is correlated with radioimmunofocus ney (AGMK) cells and do not alter the rate of replication size in these cells (Day et al., 1992; Schultz et al., 1996b). in either MRC-5 or FRhK-4 cells (Funkhouser et al., 1994). Each IRES was incorporated into an infectious cDNA Since the MR mutations cause increased virus repli- clone derived from the MRC-5-adapted virus (pMR8, Fig. cation in MRC-5 cells, and all four of the MR mutations 2) and the effect of the mutated IRES on virus replication are included within the IRES, a likely hypothesis is that in either MRC-5 or BS-C-1 cells was determined. the MR mutations promote efficient replication of HAV in these cells, at least in part, by enhancing HAV transla- RESULTS tion. Because the MR mutations have such a marked Influence of MVA infection on CAT/LUC ratios cell-specific effect on replication of HAV in cultured cells (Funkhouser et al., 1996), study of these mutations pro- We confirmed that the differences noted in cap-inde- vides a unique opportunity to determine whether there is pendent translation between clones were not artifacts of a correlation between viral translation and replication. vaccinia infection. The impact of the MR mutations on We tested the hypothesis that the MR, but not the AG, cap-independent translation was tested in BT7-H cells in mutations increased viral translation in MRC-5 cells. the presence or absence of wild-type MVA. For each MRC-5 or BS-C-1 cells were transfected with bicistronic plasmid tested and in every experiment, the absolute plasmids to assay the efficiency of translation. The lucif- values of both CAT and LUC expression increased 500- erase (LUC) gene, driven by a T7 promoter, served as an to 1000-fold in the presence of vaccinia virus (data not internal control for translation of uncapped transcripts of shown). The cap-independent translations as measured the chloramphenicol acetyltransferase (CAT) gene. The by the CAT/LUC ratios of the plasmids with the most CAT gene was controlled by a HAV IRES containing the (pAGMR) and with the fewest (pWT) mutations were various naturally occurring mutations (Fig. 2). T7 poly- compared. Whereas the CAT/LUC ratio of the pAGMR 270 FUNKHOUSER ET AL.

FIG. 2. Diagram of the bicistronic plasmids and HAV infectious cDNA clones. Designation of the bicistronic construct and infectious clone containing the same 5Ј UTR.

IRES did not change significantly when vaccinia was RNA quantification added, that of pWT significantly increased in the pres- To be sure that any differences in CAT expression ence of vaccinia (Fig. 3). Therefore, vaccinia infection between different mutant RNAs reflected the level of decreased the range between mutant and wild-type cap-independent protein expression and not the level of clones, so that differences were underestimated. We transcription (Borman et al., 1994) directed by T7 ex- included MVA/T7 in our assay, nonetheless, since MRC-5 pressed from MVA, we assessed bicistronic RNA levels cells have a limited population doubling capacity and a by Northern analysis. In several experiments, slightly MRC-5 cell line that stably expresses T7 polymerase was different quantities of RNA transcript were produced not available. from either the pWT or the pAGMR plasmid (Fig. 4). However, after normalizing samples in all experiments for ribosomal RNA (see Materials and Methods), the

FIG. 4. Northern blot of T7-driven RNA transcripts probed with a 32P-labeled negative-strand RNA probe of the CAT gene. Lanes 1, 2, 7, FIG. 3. Geometric mean of nine observations with 95% confidence and 8, RNA from the pAGMR plasmid; lanes 3, 4, 9, and 10, RNA from interval of CAT/LUC ratio in the presence or absence of MVA virus after the pWT plasmid. Lanes 5, 6, 11, and 12, negative controls. Lanes 1–6, transfection of BT7-H cells with pAGMR and pWT. MRC-5 cells; lanes 7–12, BS-C-1 cells. HEPATITIS A VIRUS IN MRC-5 CELLS 271

TABLE 1 Viral Translation and Growth

Picornavirus Viral RNA 5Ј UTR GM LUCx 102a GM CATx104b Translation accumulation Virus growth (mutations) (light units) (cpm) GM CAT/LUCx102c GM rate of rised titer(log)e

MRC-5 cells HAV (AG) 0.9 2.9 2.8 1.3 3.8 HAV (AGMR) 1.2 5.6 4.7* 1.8* 5.8 HAV (MR) 1.0 3.5 3.7* 1.7* 5.9 HAV (AGϩ687) 1.0 3.5 3.4* 1.5* 4.6 HAV (AGMR-687) 2.1 4.0 1.9 1.2 3.6 HAV (WT) 1.1 2.4 2.3 1.1 2.9 EMCV 0.9 132.4 150.6* NA NA Mock 0 0 NA NA NA

BS-C-1 cells HAV (AG) 75.1 890.3 11.9 1.6 5.5 HAV (AGMR) 90.2 1,289.2 14.3* 1.6 6.1 HAV (MR) 60.2 743.6 12.4 1.5 6.0 HAV (AGϩ687) 43.8 454.1 11.9 1.5 6.0 HAV (AGMR-687) 58.5 568.3 9.7 1.6 6.0 HAV (WT) 81.6 702.2 8.6 1.4 6.3 EMCV 96.1 29,654.2 308.5* NA NA Mock 0 0 NA NA NA

Note. *Statistical significance at the P Ͻ .05 level (mutant versus wild-type 5Ј noncoding sequence) as measured by a one-way analysis of variance followed by a Tukey b test. NA, not applicable. a Geometric mean luciferase activity. b Geometric mean chloramphenicol acetyltransferase activity. c Geometric mean (chloramphenicol acetyltransferase activity)/(luciferase activity). d Slot blot quantified by phosphoimager; geometric mean integrated intensity from day x-(x-1). e Virus titer on day 7 of growth curve, as measured by radioimmunofocus assay in FRhK-4-derived cells. average intensity of the bands representing the full- Jansen et al., 1988). It increases radioimmunofocus size length bicistronic RNA transcript from the pAGMR plas- in BS-C-1 cells (Day et al., 1990) and increases IRES mid was not significantly different from the average rep- efficiency in BT7-H cells (cloned BS-C-1 cells expressing resenting the pWT plasmid (Student’s t tests, P ϭ 0.28 T7 polymerase), but not in FRhK-4 cells (Schultz et al., and P ϭ 0.52 in MRC-5 and BS-C-1 cells, respectively). In 1996b). Therefore, we were especially interested in the both cell types, both clones produced an RNase- impact of the isolated position 687 mutation on cell- sensitive species that migrated more rapidly than the specific HAV translation and virus replication in MRC-5 full-length bicistronic transcript (Fig. 4). This extra band cells. could result from RNA degradation or from aberrant in- We transfected both MRC-5 and BS-C-1 cells with ternal initiation of transcription. Its proportion relative to bicistronic plasmids after infecting with MVA/T7. LUC the full-length product was similar for both the pAGMR and CAT assays were performed. Although the data are and the pWT clones (Student’s t test, P ϭ 0.17 in MRC-5 presented as comparisons between clones from as and BS-C-1 cells). An additional finding was that, after many as 13 individual experiments (Table 1, Figs. 5A–5D), normalizing for ribosomal RNA, the average signal gen- because of the design of the statistical testing, usually erated by the full-length RNA was twofold greater in one experiment was sufficient to detect the statistical BS-C-1 cells than that in MRC-5 cells (Student’s t test, significance of the difference between clones (data not ϭ P 0.01). shown). In MRC-5 cells, all bicistronic constructs that included the U to G mutation at position 687 (pAGMR, Effect of mutations on translation pMR, and pAGϩ687, Fig. 2) exhibited an IRES efficiency The MR group of mutations is critical for efficient significantly greater than that of wild-type (Table 1, Fig. replication of the HM175 strain of HAV in MRC-5 cells 5A). In contrast, the IRES lacking the 687 mutation, but (Funkhouser et al., 1994, 1996). The U to G mutation at containing the three other MR mutations and the four AG position 687 in the MR group is also present in multiple mutations (pAGMR-687), did not drive translation more strains of HAV (Brown et al., 1991; Graff et al., 1994; efficiently than did the wild-type IRES (Table 1, Fig. 5A). 272 FUNKHOUSER ET AL.

FIG. 5. Geometric mean with 95% confidence interval of CAT/LUC ratio of all HAV bicistronic constructs (A, MRC-5 cells; C, BS-C-1 cells) or rate of RNA accumulation (B, MRC-5 cells; D, BS-C-1 cells). *Statistical significance at the P Ͻ 0.05 level (mutant versus wild-type 5Ј noncoding sequence) as measured by a one-way analysis of variance followed by a Tukey b test. Number of observations noted in parentheses. (E) Regression lines summarizing the relationship between mean log viral RNA accumulation and mean log CAT/LUC ratio of viruses and bicistronic plasmids with the same 5Ј UTR in MRC-5 cells (closed circles) and BS-C-1 cells (open circles).

The AG mutations alone did not result in a statistically Effect of mutations on RNA accumulation and viral significant increase in IRES efficiency in MRC-5 cells replication (Table 1, Fig. 5A). When the bicistronic constructs were tested in BS-C-1 If the rate of viral replication were limited by IRES cells, trends in translation were similar to those in MRC-5 function, the translation efficiency of an IRES in particular cells (Table 1, Figs. 5A and 5C), although the magnitude cells should predict the rate of virus replication in those of the differences was reduced. The pAGMR IRES was same cells. In previous studies of replication in MRC-5 the only one whose translation efficiency was signifi- cells, the 5Ј mutations were analyzed in diverse genomic cantly greater than that of pWT in BS-C-1 cells (Table 1, backgrounds and certain mutants were not tested. Fig. 5C). Therefore, we reanalyzed each of the 5Ј HAV UTRs after HEPATITIS A VIRUS IN MRC-5 CELLS 273

FIG. 6. Viral RNA accumulation. Integrated intensity of all the image in a designated region of a slot blot measured by phosphoimager. Virus titer at day 7 measured by radioimmunofocus assay in a FRhK-4-derived cell line. A, MRC-5 cells; B, BS-C-1 cells. substitution into an identical genetic background, that of replication efficiency was generally similar to that in the pMR8 cDNA clone (Fig. 2). RNA transcribed in vitro MRC-5 cells (Fig. 6); the MR8(5Ј AGMR) virus replicated from full-length infectious cDNA clones was transfected most efficiently and MR8(5Ј WT) least efficiently. How- into the permissive FRhK-4 cells, and mutant viruses ever, there was some variability from experiment to ex- were rescued. MRC-5 and BS-C-1 cells were infected periment, the differences in RNA accumulation among with rescued viruses. We measured the accumulation of the mutants did not reach statistical significance (Fig. viral RNA over time (Table 1, Figs. 5B, 5D, and 6) and 5D), and the endpoint virus titer of all mutants was determined endpoint virus titers (Fig. 6) to verify that RNA approximately the same (Fig. 6). accumulation reflected the accumulation of infectious virus particles. Viral translation as a determinant of virus replication In all experiments in MRC-5 cells, viruses with less than the full complement of MR mutations [MR8(5Ј To determine whether viral translation correlated with AGϩ687), MR8(5Ј AG), MR8(5Ј AGMR-687), and MR8(5Ј viral RNA accumulation in either of the two cell lines WT)] replicated significantly less efficiently than those tested, we compared the extent of translation (CAT/LUC with all four MR mutations [MR8(5Ј AGMR) and MR8(5Ј values from bicistronic transcripts) with accumulation of MR)] (Fig. 6). In all experiments, the rank among the less viral RNA by viruses containing the same 5ЈUTR. In efficiently replicating subset was the same: MR8(5Ј MRC-5 cells (Table 1, Figs. 5A and 5B), the viruses that AGϩ687) was the most efficient, followed by MR8(5Ј AG), replicated more efficiently were those with the more followed by MR8(5Ј AGMR-687), and finally by MR8(5Ј efficient IRESs and those that replicated less efficiently WT). were those with the less efficient IRESs. In BS-C-1 cells The extent of virus replication varied less in BS-C-1 (Table 1, Figs. 5C and 5D), the differences between mu- cells than in MRC-5 cells (Fig. 6). The hierarchy of virus tants in efficiency of translation or in viral RNA accumu- 274 FUNKHOUSER ET AL. lation were considerably smaller; therefore correlation available for translation, unlike the situation in a viral between the two variables was more difficult to demon- infection where infectious particles continually produce strate. new template. For these reasons, translation may be Linear regression was performed to evaluate the rela- linear and viral RNA accumulation logarithmic even tionship between the mean log CAT/LUC ratio (translation) when translation efficiency is determining the rate of and the mean log rate of rise of viral RNA (virus replication) RNA accumulation. However, translation efficiency is (Fig. 5E). The ratio of CAT to LUC in all bicistronic constructs clearly not the sole determinant of HAV replication effi- was significantly greater in BS-C-1 cells than in MRC-5 cells ciency in cell culture since areas of the genome besides (Fig. 5E). In BS-C-1 cells, the slope of the regression line the 5Ј UTR are critical for growth (Emerson et al., 1991, was 0.47 and was not significantly greater than zero (P ϭ 1992, 1993; Zhang et al., 1995). 0.136). This finding supported the conclusion that under our Four naturally occurring mutations (MR) that are piv- experimental conditions, a significant correlation between otal to replication of HAV in MRC-5 cells significantly the efficiency of cap-independent translation and virus rep- increased translation, whereas four other naturally oc- lication could not be detected in these cells. In MRC-5 cells, curring mutations (AG) not required for replication in the slope of the regression line was 0.87 [significantly MRC-5 cells had no significant effect on translation in greater than zero (P ϭ 0.006)], which supports the conclu- those cells. The reversion of MR mutation 687 abolished sion that the rate of virus replication was controlled to a all translation-enhancing and replication-enhancing ef- significant degree by the efficiency of translation in those fects of the other three MR mutations, but the 687 muta- cells. tion only slightly enhanced translation and virus replication when associated with the AG mutations [pAGϩ687, Ј ϩ Ј DISCUSSION pMR8(5 AG 687) versus pAG, pMR8(5 AG)]. Therefore, the 687 mutation was critical for IRES function in MRC-5 In this study we showed a strong correlation between cells, but only in combination with one or more of the IRES efficiency and virus replication in MRC-5 cells. other MR mutations. However, the impact of mutations on viral translation The A to G substitution at base 152 in the AG group of appeared linear and that on virus multiplication logarith- mutations and a 2-base deletion at bases 203–204, sim- mic (the AGMR mutations increased translation 2- to ilar to the single deletion at base 203 in the AG muta- 3-fold, Fig. 5A, and replication 1000-fold, Fig. 6). Since tions, were previously shown to increase slightly IRES viral RNA accumulation paralleled infectious particle for- function in BS-C-1 cells (Day and Lemon, 1990; Schultz et mation (virus titer) in MRC-5 cells, the efficiency of trans- al., 1996b). In our study, the AG mutations had no signif- lation appeared to determine the extent of infectious icant impact on IRES efficiency or virus replication in virus production as well as viral RNA accumulation. BS-C-1 cells. In addition to the base 152 and base 203 Given the previous data demonstrating that the MR mu- mutations, the AG group of mutations includesaUtoC tations are the critical ones for adaptation of the HM175 substitution at base 124 and a UUUG deletion at bases virus to grow in MRC-5 cells (Funkhouser et al., 1996) 131–134. The discrepancy between the results of this and the good correlation between translation and virus study and the previous study may reflect (1) the single vs replication, it is possible that adaptation of at least this double U deletion at nucleotides 203–204, (2) the addi- strain of HAV to growth in MRC-5 cells occurred because tional mutations in the AG group present in this study but of the changes affecting translation. not the previous study, (3) differences in clonality of the There are several possible reasons for the effect of the BS-C-1 cells used in the two studies, and/or (4) the use MR mutations on virus replication being logarithmic and of the vaccinia T7 hybrid expression system in the their effect on translation being linear. Infection with present study (note the greater magnitude of the differ- vaccinia virus probably decreased the apparent differences between mutant and wild-type in the vaccinia- ences between mutants in the translation system (Fig. 3) infected versus noninfected BT7-H cells, Fig. 3). but not in the replication assays since they did not We do not know how infection of cells with the avian- contain vaccinia. Although all the mutant IRESs were in adapted vaccinia virus may have affected IRES function the same bicistronic background and the assays were in the bicistronic system, though vaccinia virus infection therefore comparable, the reporter gene itself could have does dramatically inhibit cellular protein synthesis (Es- influenced the efficiency of IRES function (Kaminski and teban and Metz, 1973; Moss, 1968). When we compared Jackson, 1998), and the extent of the impact of the MR the CAT/LUC ratios of pWT and pAGMR in a T7 poly- mutations on viral translation might have been underes- merase-expressing cell line (BT7-H) with and without timated. In addition, virions accumulate at an exponential vaccinia, the difference in the CAT/LUC ratio between rate during the maturation phase of virus replication pWT and pAGMR was decreased in the presence of (Roizman and Palese, 1996). However, our translation vaccinia. Accounting for this decrease was the increased assay was performed at one time point in each experi- efficiency of the pWT IRES in the presence of vaccinia ment. Therefore, a fixed amount of RNA template was virus (Fig. 3). One possible explanation is that a compet- HEPATITIS A VIRUS IN MRC-5 CELLS 275 itive advantage for cellular translation machinery under the growth curve assay in this study could have resulted vaccinia-infected conditions gave more of a boost to the in an underestimation of the differences in replication less efficiently translating clone (pWT) (Bernard Moss, efficiency between mutant viruses in BS-C-1 cells. How- personal communication). Since there was approxi- ever, radioimmunofocus assay was performed at the mately a 1000-fold increase in protein expression with growth curve endpoint, and the titers of the mutant vi- the addition of wild-type vaccinia, vaccinia infection ruses were very similar. Differences between mutant could have increased the rate of T7 transcription or IRESs in translation efficiency may have been underes- increased the rate of both cap-dependent and cap-inde- timated in BS-C-1 cells due to vaccinia infection (Fig. 3) pendent translation (since both LUC and CAT expression or to the effect of the reporter gene (Kaminski and Jack- increased in the presence of vaccinia). The 2.5 ␮g per son, 1998). However, these factors also influenced the well of DNA inoculated did not result in maximal expres- translation assay in MRC-5 cells. Since there were very sion of the reporter proteins in either MRC-5 or BS-C-1 small differences in both translation efficiency and rep- cells, but it is not clear whether the increased level of lication efficiency in BS-C-1 cells, the impact of transla- expression observed with higher quantities of DNA (e.g., tion on virus replication may have been more difficult to 5 ␮g/well) was due to greater levels of transcription (and accurately assess. subsequent translation) within individual cells, to a In FRhK-4 cells, the relationship between viral trans- higher proportion of cells that were successfully translation and replication has been approached differently. fected, or to both. Therefore, saturation of cellular trans- Jia et al. (1996) found that insertion of an efficient IRES lation machinery could have occurred in some cells. (from EMCV) in front of the open reading frame of HAV Whatever the mechanism for the effect of vaccinia infec- does not increase replication in FRhK-4 cells. Although it tion on the magnitude of differences between bicistronic is possible that interactions between nonstructural pro- constructs, differences in translation efficiency were teins and the 5Ј UTR (Andino et al., 1990) of the chimeric most likely underestimated when vaccinia was added to virus were not optimized, these data agree with data we the assay. This could well render the smaller differences have obtained in FRhK-4 cells in which we were unable in translation between mutants and wild-type more diffi- to detect a correlation between virus translation and cult to detect in BS-C-1 cells. replication efficiency (Funkhouser, unpublished data). In general, clusters of mutations rather than single It is unclear why viral translation would be more lim- mutations have had the greatest impact on HAV replica- iting to the rate of virus replication in MRC-5 cells than in tion (Cohen et al., 1989; Day et al., 1992; Emerson et al., other cells tested. MRC-5 cells differ from FRhK-4 and 1991, 1992, 1993; Funkhouser et al., 1994, 1996; Graff et BS-C-1 cells in that they are diploid fibroblasts rather al., 1997; Zhang et al., 1995); single nucleotide changes than epithelial cells and are mortal, with the capability of have been associated with smaller, incremental changes only 42 to 46 passage doublings before the onset of in virus phenotype (Day and Lemon, 1990; Day et al., senescence. Cellular proteins whose activities differ in 1992; Emerson et al., 1992). It is therefore of great inter- senescent and nonsenescent cells could interact to dif- est that the solitary back-mutation of base 687 had such fering degrees with viral components. There are exam- a dramatic impact on both translation and replication in ples of this among DNA viruses (Apt et al., 1996). This MRC-5 cells. The MR mutations change the predicted study could represent the first reported example among secondary structure of the RNA within the putative stem- RNA viruses. loop before the translation start codon (Fig. 1), perhaps In summary, our data show that MRC-5 cells differ from changing tertiary structure as well. These structural BS-C-1 cells in the extent to which IRES efficiency de- changes could facilitate interactions between viral RNA termines HAV replication. In MRC-5 cells, HAV translation and cell-specific or viral factors. In the presence of the is rate-limiting for virus replication. other three MR mutations, the 687 mutation seems to have a dramatic effect on IRES function in MRC-5 cells. MATERIALS AND METHODS Possible explanations for the impact of the 687 mutation Bicistronic plasmids include a change in the stability of the predicted RNA stem-loop V (Fig. 1) (Schultz et al., 1996a) and/or changes The bicistronic plasmids (Fig. 2) containing mutant in its interaction with the downstream domain VI (Fig. 1) HAV 5Ј UTRs were all constructed from the pWT plasmid, in the presence of the 687 mutation, causing changes in previously described as pLUC-WT-CAT (Schultz et al., RNA–protein or RNA–RNA interactions that are essential 1996b). The pWT plasmid, driven by a T7 promoter, con- to translation. tains cDNA from base 0 to 749 of the wild-type HM175 Both translation and viral replication varied less in hepatitis A virus (preceded by 10 linker nucleotides in BS-C-1 cells than in MRC-5 cells. Differences in replica- pWT and all derivative clones) in the intercistronic space tion between mutant viruses in BS-C-1 cells have been between the firefly LUC gene and the CAT gene (Fig. 2). less pronounced when assayed by growth curve than by The pAGMR, pAG, and pMR cDNA clones (Fig. 2) were radioimmunofocus assay (Day et al., 1992). The use of constructed by replacement of the HindIII/XbaI fragment 276 FUNKHOUSER ET AL.

of pWT with that from previously described full-length mutant were assayed. Ten micrograms of plasmid DNA infectious cDNA clones (Funkhouser et al., 1996). Oligo- was transcribed in vitro with Sp6 polymerase to produce nucleotide mutagenesis and PCR mutagenesis of the RNA for transfection (Funkhouser et al., 1996). The entire HindIII/XbaI fragment were used to engineer mutations 100-␮l transcription reaction was precipitated with etha- in the pAGϩ687 and pAGMR-687 clones, respectively. nol and resuspended in 2 ml of nonserum DMEM with 12 The entire nucleotide sequence of the HAV 5Ј UTR and ␮l DMRIE-C reagent (Life Technologies, Gaithersburg, the CAT gene of all mutant bicistronic constructs was MD). This mixture was immediately inoculated onto confirmed using the Applied Biosystems 373A auto- FRhK-4 cells previously washed once with DMEM. The mated DNA sequencer and a modified Sanger method. cells were incubated for6hat34.5°C, washed once with DMEM, and refed with 10% DMEM. When the majority of Viruses cells were positive for HAV antigen by immunofluores- The modified vaccinia virus Ankara is a mammalian cence microscopy, the culture was harvested with tryp- host-restricted virus (Mayr et al., 1975). The MVA/T7 is a sin and freeze-thawed three times to release the virus. recombinant MVA expressing the T7 polymerase gene Virus titers were determined by RIFA in 11-1 cells (Funk- (Wyatt et al., 1995). All HAV strains used in this study houser et al., 1994). were rescued after transfection of FRhK-4 cells with transcripts from full-length cDNA clones (see below). Transient expression assay

Cells The bicistronic constructs were analyzed for their cell- specific expression of CAT and LUC using a transient MRC-5 cells were obtained from SmithKline Beecham expression system as described previously (Schultz et Pharmaceuticals, further passaged at DynCorp (Rock- al., 1996b) with several modifications. Monolayers of ville, MD), and used at passage doubling level 35 to 38. cells in 35-mm wells were infected with MVA/T7 or MVA. The BS-C-1 cells were originally obtained from the Amer- The virus was diluted in 0.5 ml of nonserum DMEM and ican Type Culture Collection (Rockville, MD). The 11-1 incubated for1hatamultiplicity of infection (m.o.i.) of 5 cell line was previously described (Funkhouser et al., PFU per cell. The cells were then washed with nonserum 1994) and is a robust subclone of FRhK-4 cells used here DMEM and DNA transfection was performed in triplicate only for virus quantification by radioimmunofocus assay as per manufacturer’s instructions with 2.5 ␮g DNA and (RIFA). The BT7-H cells used in transient expression 15 ␮l Lipofectin (Life Technologies) per well. After 12 to assays were a cell line subcloned from BS-C-1 cells 15 h, the transfection mixture was removed and the cells stably transformed by a plasmid expressing the T7 poly- were fed with 10% DMEM. Six hours later, the cells were merase gene and geneticin resistance (Whetter et al., harvested and CAT and LUC activities assayed, as pre- 1994). Cells were maintained in Dulbecco’s modified viously described (Schultz et al., 1996b). BT7-H cells Eagle’s medium (DMEM) supplemented with 10% fetal were assayed similarly for their transient expression ␮ calf serum, nonessential amino acids, glutamine, 50 g capability in the absence of MVA infection, but experi- ␮ of gentamicin sulfate per milliliter, and 2.5 g of ampho- ments that included the virus also were performed to tericin B per milliliter (10% DMEM). The BT7-H cell cul- determine the effect of the vaccinia virus on the results. ␮ tures also contained 500 g/ml of geneticin. The 11-1 The conditions described above for the transient and BS-C-1 cells were split 1:10 and the MRC-5 and transfection assay were obtained by optimizing the sys- BT7-H cells were split 1:4 for cell line maintenance. tem for the amount of vaccinia virus inoculated, the quantity of DNA inoculated, and the time of sample Full-length infectious cDNA clones, in vitro harvest. Both CAT and LUC activities plateaued at a transcription, and transfection m.o.i. of 5. Since greater well-to-well variability of protein The wild-type HM175 virus grows very slowly in cell expression occurred at m.o.i.s of less than 5 compared to culture even if the AG and/or MR mutations are inserted m.o.i.s of 5 or greater, 5 was chosen as the m.o.i. We into the 5Ј UTR [rescued virus detected by immunofluo- checked for reproducibility of results and saturation of rescence in the majority of cells 50 to 80 days after the cellular translation machinery by comparing levels of transfection of FRhK-4 cells with infectious cDNA LUC and CAT expression of the pAGMR and pWT plas- (Raychaudhuri, unpublished data)]. It was therefore not mids in each cell type after transfection with increasing feasible to use this genetic background to test the im- amounts of DNA (0.25, 0.5, 1.0, 2.5, and 5.0 ␮g per well). pact of the 5Ј mutations on virus replication. Instead, In both MRC-5 and BS-C-1 cells, the (CAT/LUC pAGMR)/ mutations were incorporated into an infectious clone (CAT/LUC pWT) ratio was the same for all quantities of from which rescued virus grows efficiently in MRC-5 DNA tested since CAT and LUC expression varied in cells (pMR8, Funkhouser et al., 1996). The 5Ј UTRs from parallel. In both cell types, CAT and LUC expression was various bicistronic plasmids were substituted into the efficient at 2.5 ␮g of DNA per well but was lower than pMR8 background (Fig. 2). Two sister clones of each that at 5.0 ␮g DNA per well. Therefore, since 2.5 ␮gof HEPATITIS A VIRUS IN MRC-5 CELLS 277

DNA per well appeared to result in efficient transfection, as previously described (Funkhouser et al., 1996). Vi- transcription, and translation without saturation, this ruses rescued from two sister clones of every mutant amount was transfected in all experiments. In order to cDNA clone were assayed in each cell type. Virus was determine the optimal time of sample harvest, we mea- inoculated at an m.o.i. of 3 radioimmunofocus-forming sured the CAT and LUC expression from the two plas- units per cell and harvested daily for 7 days. The cell mids with the most (pAGMR) and fewest (pWT) muta- culture supernatant and trypsinized cells were com- tions. Between 0 and 36 h posttransfection, both the bined. Volumes of harvested samples were equalized CAT/LUC ratio from expression of pAGMR and the (CAT/ with 10% DMEM before viral RNA was extracted to com- LUC pAGMR)/(CAT/LUC pWT) ratio were highest (3.2) at pensate for the evaporation that occurred over time in approximately 18 h posttransfection. Therefore, in all the 96-well plates used for culture. Viral RNA was ex- experiments, samples were harvested 18 h after trans- tracted with Trizol reagent (Life Technologies). Slot blot fection. hybridization was performed as described (Funkhouser et al., 1994) using a 32P-labeled negative-strand RNA Quantification of RNA transcripts probe. Blotted nitrocellulose membranes were exposed to Phosphoimmager 445 SI (Molecular Dynamics, Sun- To compare the relative quantities of mutant and wild- nydale, CA), and recorded data were quantified using type bicistronic RNA, the pAGMR and pWT plasmids NIH Image, Version 1.59 software. To standardize the were transfected as described above (using DMEM with measurements of RNA replication from experiment to lipofectin as a negative control). Cells were disrupted experiment, we measured the average rate of viral RNA 24 h after transfection. Total RNA was isolated with Trizol accumulation for a given virus in a given cell type. The reagent (Life Technologies) as per the manufacturer’s log integrated intensity of the area on an autoradiograph instructions. After treatment with RNase-free DNase 1 representing the RNA from a given day of virus harvest (Boehringer Mannheim Corp., Indianapolis, IN), the total (Fig. 6) minus that on the previous day was averaged RNA was mixed with ethidium bromide and electropho- over all samples of a given virus in a given cell type. resed on a 1% agarose–formaldehyde gel. The gel was Virus titers of harvested samples from day 7 (endpoint photographed, and the RNA was transferred to a posi- infectivity titers) were determined by RIFA in 11-1 cells. tively charged nylon membrane (Amersham Life Science, 32 Inc., Arlington Heights, IL). A P-labeled negative-strand Statistical analysis RNA probe was synthesized from the full-length CAT gene (plasmid was the kind gift of Dr. R. E. Rynbrand, Both the translation and the virus replication data were University of Texas Medical Branch, Galveston, TX) using tested for normality using the Shapiro–Wilk test. Since T3 polymerase (Promega Corp., Madison, WI) as per the the CAT (and CAT/LUC) data and the replication varied in manufacturer’s instructions. Prehybridization, hybridiza- a log normal fashion, all data analysis was performed in tion, and washing conditions were as described by Bra- the log scale (Table 1, Fig. 5). zas and Ganem (1996), except that the temperature for prehybridization and hybridization was 55°C. Autoradio- ACKNOWLEDGMENTS graphs and photographs of agarose–formaldehyde gels We thank Linda Wyatt and Bernard Moss for their kind gift of the MVA were scanned using the Hewlett Packard Scan Jet IICT/T, and MVA/T7 viruses. We thank Judith Graff for assistance with the and recorded data were quantified using NIH Image, Northern blot assay, and Judith Graff, Gopa Raychaudhuri, Linda Wyatt, Version 1.59 software. We compared samples by normal- and Bernard Moss for thoughtful discussions about the data. Thanks izing the densitometry values obtained from scanning are due to David Alling for assistance with both statistical analysis and data interpretation. Thanks are also extended to Ellie Ehrenfeld for her Northern blots. The values from scans of the autoradio- thoughtful review of the manuscript. graphs of T7-driven RNA transcripts were divided by values from scans of the photographs of cellular (ribo- REFERENCES somal) RNA. Andino, R., Rieckhof, G. E., and Baltimore, D. (1990). A functional ribonucleoprotein complex forms around the 5Ј end of poliovirus Virus growth curves and virus quantification RNA. Cell 63, 369–380. Because vaccinia virus changes the internal environ- Apt, D., Watts, R. M., Suske, G., and Bernard, H-U. (1996). High Sp1/Sp3 ratios in epithelial cells during epithelial differentiation and cellular ment of cells, it would be preferable to assess replication transformation correlate with the activation of the HPV-16 promoter. under the same conditions used to assess translation in Virology 224, 281–291. vaccinia-infected cells. However, the HAV replication Borman, A. M., Deliat, F. G., and Kean, K. M. (1994). Sequences within curve required 7 days and by 24 to 48 h after infection the poliovirus internal ribosome entry segment control viral RNA with the MVA/T7 virus, many of the cells were no longer synthesis. EMBO J. 13, 3149–3157. Brazas, R., and Ganem, D. (1996). A cellular homolog of hepatitis delta viable. Therefore, vaccinia virus was not included in the antigen: Implications of viral replication and evolution. Science 274, assays of viral replication. Viral RNA replication was 90–94. assessed by plotting viral RNA accumulation over time, Brown, E. A., Day, S. P., Jansen, R. W., and Lemon, S. M. (1991). The 5Ј 278 FUNKHOUSER ET AL.

nontranslated region of hepatitis A virus RNA: Secondary structure Jia, X-Y., Scheper, G., Brown, D., Updike, W., Harmon, S., Richards, O., and elements required for translation in vitro. J. Virol. 65, 5828–5838. Summers, D., and Ehrenfeld, E. (1991). Translation of hepatitis A virus Cohen, J. I., Rosenblum, B., Feinstone, S. M., Ticehurst, J., and Purcell, RNA in vitro: Aberrent internal initiations influenced by 5Ј noncoding R. H. (1989). Attenuation and cell culture adaptation of hepatitis A region. Virology 182, 712–722. virus (HAV): A genetic analysis with HAV cDNA. J. Virol. 63, 5364– Jia, X-Y., Tesar, M., Summers, D. F., and Ehrenfeld, E. (1996). Replication 5370. of hepatitis A viruses with chimeric 5Ј nontranslated regions. J. Virol. Cohen, J. I., Rosenblum, B., Ticehurst, J. R., Daemer, R. J., Feinstone, 70, 2861–2868. S. M., and Purcell, R. H. (1987). Complete nucleotide sequence of an Kaminski, A., and Jackson, R. J. (1998). The polypyrimidine tract binding attenuated hepatitis A virus: Comparison with wild-type virus. Proc. protein (PTB) requirement for internal initiation of translation of car- Natl. Acad. Sci. USA 84, 2497–2501. diovirus RNAs is conditional rather than absolute. RNA 4, 626–638. Ј Day, S. P., and Lemon, S. M. (1990). A single base mutation in the 5 Mayr, A., Hochstein-Mintzel, H., and Stickl, H. (1975). Abstammung, noncoding region of HAV enhances replication of virus in vitro. In Eigenschaften und Verwendung des attenuierten Vaccinia-Stammes “Vaccines 90: Modern Approaches to New Vaccines including Pre- MVA. Infection 3, 6–14. vention of AIDS” (F. Brown, R. M. Chanock, H. S. Ginsberg, and R. A. Moss, B. (1968). Inhibition of HeLa cell protein synthesis by the vaccinia Lerner, Eds.), pp. 175–178. Cold Spring Harbor Laboratory Press, virion. J. Virol. 2, 1028–1037. Cold Spring Harbor, NY. Pilipenko, E. V., Gmyl, A. P., Maslova, S. V., Svitkin, Y. V., Sinyakov, A. N., Day, S. P., Murphy, P., Brown, E. A., and Lemon, S. M. (1992). Mutations and Agol, V. I. (1992). Prokaryotic-like cis elements in the cap- Ј within the 5 nontranslated region of hepatitis A virus RNA which independent internal initiation of translation on picornavirus RNA. enhance replication in BS-C-1 cells. J. Virol. 66, 6533–6540. Cell 68, 119–131. Emerson, S. U., Huang, Y. K., Purcell, R. H. (1993). 2B and 2C mutations Roizman, B., and Palese, P. (1996). Multiplication of viruses: An over- are essential but mutations throughout the genome of HAV contrib- view. In “Fields Virology” (B. N. Fields, D. M. Knipe, and P. M. Howley, ute to adaptation to cell culture. Virology 194, 475–480. Eds.), 3rd ed., pp. 101–111. Lippincott-Raven, Philadelphia, PA. Emerson, S. U., Huang, Y. K., McRill, C., Lewis, M., and Purcell, R. H. Schultz, D. E., Hardin, C. C., and Lemon, S. M. (1996a). Specific inter- (1992). Mutations in both the 2B and 2C genes of hepatitis A virus are action of glyceraldehyde 3-phosphate dehydrogenase with the 5Ј- involved in adaptation to replication in cell culture. J. Virol. 66, nontranslated RNA of hepatitis A virus. J. Biol. Chem. 271, 14134– 650–654. 14142. Emerson, S. U., McRill, C., Rosenblum, B., Feinstone, S., and Purcell, Schultz, D. E., Honda, M., Whetter, L. E., McKnight, K. L., and Lemon, R. H. (1991). Mutations responsible for adaptation of hepatitis A virus S. M. (1996b). Mutations within the 5Ј nontranslated RNA of cell to efficient replication in cell culture. J. Virol. 65, 4882–4886. Esteban, M., and Metz, D. H. (1973). Early virus protein synthesis in culture-adapted hepatitis A virus which enhance cap-independent vaccinia virus-infected cells. J. Gen. Virol. 19, 201–216. translation in cultured African green monkey kidney cells. J. Virol. 70, Funkhouser, A. W., Purcell, R. H., D’Hondt, E., and Emerson, S. U. (1994). 1041–1049. Ј Attenuated hepatitis A virus: Genetic determinants of adaptation to Simoes, E. A., and Sarnow, P. (1991). An RNA hairpin at the extreme 5 replication in MRC-5 cells. J. Virol. 68, 148–157. end of the poliovirus RNA genome modulates viral translation in Funkhouser, A. W., Raychaudhuri, G., Purcell, R. H., Govindarajan, S., human cells. J. Virol. 65, 913–921. Elkins, R., and Emerson, S. U. (1996). Progress toward the develop- Todd, S., Towner, J. S., and Semler, B. L. (1997). Translation and repli- ment of a genetically engineered attenuated hepatitis A virus vac- cation properties of rhinovirus genome in vivo and in vitro. Virology cine. J. Virol. 70, 7948–7957. 229, 90–97. Graff, J., Kasang, C., Normann, A., Pfisterer-Hunt, M., Feinstone, S. M., Whetter, L. E., Day, S. P., Elroy-Stein, O., Brown, E. A., and Lemon, S. M. and Flehmig, B. (1994). Mutational events in consecutive passages of (1994). Low efficiency of the 5Ј nontranslated region of hepatitis A hepatitis A virus strain GBM during cell culture adaptation. Virology virus RNA in directing cap-independent translation in permissive 204, 60–68. monkey kidney cells. J. Virol. 68, 5253–5263. Graff, J., Normann, A., and Flehmig, B. (1997). Influence of the 5Ј Wyatt, L. S., Moss, B., and Rozenblatt, S. (1995). Replication-deficient noncoding region of hepatitis A virus strain GBM on its replication in vaccinia virus encoding bacteriophage T7 RNA polymerase for tran- different cell lines. J. Gen. Virol. 78, 1841–1849. sient gene expression in mammalian cells. Virology 210, 202–205. Jansen, R. W., Newbold, J. E., and Lemon, S. M. (1988). Complete Zhang, H., Chao, S-F., Ping, L-H., Grace, K., Clarke, B., and Lemon, S. M. nucleotide sequence of a cell culture-adapted variant of hepatitis A (1995). An infectious cDNA clone of a cytopathic hepatitis A virus: virus: Comparison with wild-type virus with restricted capacity for in Genomic regions associated with rapid replication and cytopathic vitro replication. Virology 163, 299–307. effect. Virology 212, 686–697.