INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Late after infection of a host cell with a cytocidal virus, macromolecular synthesis (both host and virus specific) shuts down. Often the cell is dying, while the virus has shifted from a biosynthetic phase into an assembly phase. In the case of the positive-strand RNA viruses, this entails the encapsidation of the viral genomic RNA into virions. These inevitably competing events complicate the analysis and interpretation of the kinetics of RNA synthesis at late times after infection. However, many of these complications can be circumvented by using flock house virus (FHV), a member of the Nodaviridae, as a model system for the study of RNA replication. This approach takes advantage of the nodavirus divided genome, which naturally separates the replicative and packaging functions onto two different positive-sense RNA molecules, RNA1 and RNA2, respectively (39-41).

Both genomic RNAs have cap zero structures at their 5' ends, and their 3' ends, which are not polyadenylated, are blocked either by an unusual secondary structure or by an undetermined covalent modification (3, 18, 32). RNA1 (3,107 nucleotides [nt]), encodes protein A, the catalytic subunit of the viral RNA-dependent RNA polymerase (RNA replicase), while RNA2 (1,400 nt) encodes the capsid protein precursor, protein , which is autocatalytically cleaved into the proteins ( and ) found in the mature virion (30). The genomic segments (RNA1 and RNA2) are copackaged into virions, and both are required for infectivity (35, 41, 55).

The RNA replicase catalyzes replication of both genomic RNAs via negative-strand intermediates (1, 4, 29). It is not known whether host cell factors are required for replicase activity. However, the ability of FHV RNA to replicate in insect (29), mammalian (4), plant (54), and yeast (48) cells despite a natural host range that is limited to insects (53), suggests that any such host factors must be functionally conserved in all of these disparate intracellular environments. During RNA replication, a small subgenomic RNA (RNA3) is synthesized from RNA1, probably by internal initiation on the RNA1 negative strand. RNA3 contains the 3' terminal 387 nt of RNA1 (3, 32) and encodes two small proteins (B1 and B2) of unknown function (3, 25, 26) but is not packaged into virions (25).

In general, the replication of positive-sense RNA involves at least three well-recognized mechanistic steps: parental RNA is first translated to generate active replicase and then copied by this enzyme to produce negative strands, which in turn are used as templates for the synthesis of positive-sense progeny RNAs. According to the model developed for bacteriophage Q replicase (7-11, 23), the progeny RNAs feed back both as mRNA for synthesis of more replicase and as template for further negative-strand synthesis, resulting in initial hyperbolic kinetics of RNA replication. However, the bulk of the RNA products accumulate during the linear phase of the reaction, when the replication rate is constant, because the concentrations of enzyme and template that are involved in replication have reached steady states. Later, RNA replication slows when the feedback is no longer sufficient to maintain the steady-state concentrations of functional enzyme and negative strand template. In Q-infected bacterial cells, some progeny RNA molecules are prevented from feeding back into the reaction by encapsidation into phage particles and others by the formation of double-stranded RNA (23).

Guarino and Kaesberg (33) studied the kinetics of nodavirus RNA replication in cell extracts prepared at various times postinfection (p.i.) from Drosophila melanogaster cells infected with black beetle virus (BBV), the RNA1 of which shares 99% sequence identity with RNA1 from FHV (17, 18). BBV RNA replication conformed rather well to the general pattern established for Q. After a 3-h lag period, RNA replication increased exponentially until 10 h p.i., then increased at a slower linear rate until 48 h p.i., and finally decreased by 72 h p.i. As the viral yield at each time point was not determined in these experiments, it is unclear to what extent the kinetics of BBV replication were influenced by continuous removal of positive-strand template RNAs by packaging into progeny virions (33).

Neither RNA2 nor the capsid proteins it encodes are required for replication of nodavirus RNA1, which replicates autonomously in the absence of RNA2 because it encodes the entire viral contribution to the RNA replicase (29). When transfected into susceptible cells, RNA1 reaches its peak rate of synthesis after about 17 h, replicates vigorously for at least 24 h, and attains levels comparable to those of the rRNAs (1, 4, 29). Under these conditions, RNA3 is synthesized in greater abundance than in the presence of RNA2 (29) because replication of RNA2 specifically downregulates RNA3 synthesis (63).

Nodavirus RNA replication in mammalian cells can also be initiated by the use of cDNA-based transcription systems that reconstruct many features of the natural situation (1-3, 5, 6, 34). Transcription of full-length cDNA clones of FHV RNA1 (19) by T7 RNA polymerase expressed from a recombinant vaccinia virus (VV) (27) results in abundant cytoplasmic RNA replication (1-3, 5, 38). Synthesis of primary transcripts with authentic termini was essential for optimum replication because the presence of extraneous nucleotides at either the 5' or the 3' end was detrimental (1, 5). This was accomplished by placing the promoter in the transcription plasmids precisely at the 5' end of the viral cDNA and including a cDNA copy of the antigenomic ribozyme from hepatitis delta virus (HDV) (46) immediately downstream of the FHV cDNA; the resulting transcription products self-cleaved to yield authentic 3' ends (1-3, 5). Abundant RNA1 replication was also obtained on infection of susceptible cells by using VV recombinants containing the FHV RNA1 cDNA positioned between the VV 7.5k promoter and the HDV ribozyme (6). In addition, we have achieved the replication of primary transcripts made in the nucleus of cells from a transcription plasmid that retains the general structure described above but contains the FHV RNA1 cDNA downstream of a constitutive RNA polymerase II (pol II) promoter (34).

The potential use of nodavirus RNA replication to amplify and express foreign mRNAs focused our interest on the longevity of RNA1 replication in the absence of RNA2. In the current study, we observed that RNA1 replication ceased 3 days after transfection, despite the presence of a high intracellular concentration of replication-competent RNA. This inhibition was also observed when we used a plasmid containing an inducible promoter to launch RNA1 replication from the nucleus of DNA-transfected cells. After RNA replication had stopped, it could be restored either by transferring the resulting intracellular RNA to fresh cells or by reinducing primary transcription in the original plasmid-transfected culture, indicating that replication ceased despite the competence of both the viral RNA and the cytoplasmic environment. This shutoff may represent the natural endpoint of RNA replication, functionally equivalent to the late decline in the rates of RNA replication observed in vivo with Q and BBV (23, 33).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. EcR-CHO cells, a subline of CHO cells that stably express a heterodimeric ecdysone-retinoid X receptor (EcR/RXR) were purchased from Invitrogen (Carlsbad, Calif.). The cells were grown in Ham's F-12 medium (GIBCO/BRL) supplemented with 10% fetal bovine serum (HyClone, Logan, Utah), penicillin-streptomycin (50 U/ml and 50 µg/ml, respectively), 2 mM L-glutamine, and 250 µg of Zeocin (Invitrogen) per ml. Baby hamster kidney (BHK21) cells were grown as described previously (1). VV-FHV recombinant vF1 contained full-length cDNA of FHV RNA1 between the 7.5k VV promoter and a cDNA copy of the antigenomic ribozyme of HDV (6). By autolytic cleavage of the primary transcripts, the ribozyme generated RNA that terminated at the authentic 3' nucleotide.

Plasmids. Plasmids pIND and pIND-lacZ, which contain a hormone-regulated pol II promoter, were purchased from Invitrogen, and plasmid FHV1[1,0] was as described previously (3). Plasmid pIND-FHV1[0,0] (shown schematically in Fig. 4) was constructed from pIND and FHV1[1,0] by PCR and conventional cloning techniques to put the FHV1 cDNA under the control of the inducible promoter (51). It contained a pol II promoter located downstream of a hormone response element (42), followed by the FHV RNA1 cDNA. As in previous FHV transcription plasmids (1-3, 5, 34), the promoter was resected to place the 5' terminus of the RNA1 cDNA at the transcriptional initiation site, and the 3' end of the FHV1 cDNA was juxtaposed with the cleavage site of the HDV antigenomic ribozyme (46). The nucleotide sequences of the promoter-cDNA-ribozyme junctions were confirmed experimentally and are shown in Fig. 4. The T7 transcriptional terminator (49) was also retained in pIND-FHV1[0,0], as it appeared to facilitate the ribozyme-mediated cleavage (34).

Isolation of FHV RNA1 expressed from a VV-FHV recombinant. To provide a convenient source of molecularly cloned, replication-competent FHV RNA1, BHK21 cells were infected at 28°C with the VV-FHV recombinant vF1 (6) at a multiplicity of infection of 10. Total cellular RNA was isolated at 24 h p.i., diluted 25-fold, and amplified by a second 24-h replicative passage in fresh BHK21 cells (3).

RNA transfection of cells. Monolayers containing 2 × 10⁶ EcR-CHO cells were transfected with this mixture of cellular and FHV RNAs by using Lipofectin (10 µg) and OptiMEM (1 ml) (both from GIBCO/BRL) as previously described (1). Transfected cells were incubated at 28°C for the times indicated in the figure legends. For the RNA passage experiments, total cellular RNA was isolated from transfected cells as described previously (34); for each sample, 4% of this mixture of viral and cellular RNAs was then transfected into a fresh monolayer of EcR-CHO cells as described above.

Protein labeling, extraction, and analysis. For protein labeling, transfected cells were preincubated for 30 min in modified Eagle minimum essential medium lacking methionine and cysteine (ICN). The products of protein synthesis were then metabolically labeled for 3 h by incorporation of [³⁵S]methionine-[³⁵S]cysteine (Tran³⁵S label; 50 µCi/ml; ICN) in Met-Cys-free medium. Labeled proteins were extracted, resolved by electrophoresis on a sodium dodecyl sulfate (SDS)-12.5% polyacrylamide gel, and visualized by autoradiography (1).

Plasmid transfection and hormone treatment of cells. Monolayers containing 2 × 10⁶ EcR-CHO cells were transfected with pIND-FHV1 (2.5 µg) by using Lipofectamine (14 µg) (GIBCO/BRL) and OptiMEM (1 ml) as described earlier (34). After transfection, the cells were incubated at 37°C for 6 h, the transfection mixture was replaced with complete F-12 medium, and incubation was continued at 37°C for an additional 18 h. The steroid hormone muristerone A (Invitrogen) was then added to the medium to a final concentration of 1 µM, and the cells were shifted to 28°C, the permissive temperature for FHV RNA replication. Incubation was continued at 28°C with daily replenishment of medium containing muristerone A for the times indicated in the text. In the hormone removal experiments, the medium containing muristerone A was removed, the cells were washed twice with phosphate-buffered saline, complete F-12 medium lacking muristerone A was added, and the incubation was continued at 28°C as above. Transfection efficiency was estimated to be approximately 10% at 28°C by transfection of EcR-CHO cells with reporter plasmid pIND-lacZ and subsequent induction with muristerone A, as described previously (34).

RNA labeling, extraction, and analysis. The products of RNA replication were labeled by metabolic incorporation of [³H]uridine in the presence of 20 µg of actinomycin D per ml to inhibit DNA-dependent RNA synthesis (1). After incubation at 28°C for 2 h, total cellular RNAs were extracted with guanidine thiocyanate (RNAgents; Promega Biotec) (16), as described previously (34). Samples corresponding to the RNA from 3.3 × 10⁵ cells were analyzed by electrophoresis on denaturing formaldehyde-agarose gels and visualized by fluorography (34). For quantitation of RNA replication, triplicate samples of the ³H-labeled RNA replication products from 1.7 × 10⁵ cells were precipitated with trichloroacetic acid, and the combined levels of RNAs 1 and 3 were determined by scintillation spectrometry (3). The resulting counts per minute were averaged for each triplicate set and plotted as a function of the incubation time; error bars represent the range of values obtained. Each experiment was repeated a minimum of three times, and the results of representative experiments are shown in the figures.

Primer extension. RNA samples from 2.7 × 10⁵ cells were analyzed by primer extension, using oligonucleotide primers that were complementary to nt 85 to 69 of FHV RNA1 or nt 98 to 80 of RNA3. Primer extension products were labeled by incorporation of [-³⁵S]dATP, resolved on a 6% polyacrylamide sequencing gel alongside dideoxy sequencing ladders generated with the same primers on a pIND-FHV1 template, and visualized by autoradiography as described previously (3). Where indicated, FHV RNA levels in different samples were normalized to the amount of 18S rRNA present in each sample as determined by primer extension (34). A primer specific for FHV RNA3 (complementary to nt 2819 to 2801 of RNA1) was mixed with a primer complementary to nt 82 to 65 of Rattus norvegicus 18S rRNA (57) and extended by using reverse transcriptase and [-³⁵S]dATP. The relative amounts of total RNA present in each sample were quantitated by densitometric scanning of the resulting autoradiographs and normalized to the abundance of 18S rRNA.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Kinetics of RNA replication in cells transfected with FHV RNA1. When transfected into insect (Drosophila line 1) or vertebrate (BHK21) cells, nodavirus RNA1 replicates autonomously for at least 24 h (29), reaching levels comparable to those of rRNAs (4). As part of our continuing efforts to understand and harness nodavirus RNA replication, we examined the process over longer time periods to determine the duration of replication. A subline of Chinese hamster ovary (CHO) cells (see Materials and Methods) was transfected with molecularly cloned, replication-competent FHV RNA1. At 24-h intervals, transfected cells were pulse-labeled with [³H]uridine in the presence of actinomycin D to inhibit DNA-dependent RNA synthesis. Under these conditions, uridine incorporation specifically measures the activity of the RNA replicase as it synthesizes FHV RNAs 1 and 3 (see Fig. 6B), so we could quantitate RNA replication simply by measuring uridine incorporation (Fig. 1, lower panel). The peak rate of RNA replication occurred within the first 24 h posttransfection, a finding consistent with previous observations of the replication of FHV RNA1 in Drosophila cells (29) and Nodamura virus RNA1 in BHK21 cells (1). However, on further incubation, the rate of RNA replication diminished so that after 3 days the reaction was undetectable. In view of the self-supporting nature of RNA1 replication, this cessation was surprising. Examination of the transfected cultures by phase-contrast microscopy showed no overt signs of cytopathology during the 5-day course of this experiment. Similar results were obtained upon transfection of BHK21 cells with FHV RNA1, although the kinetics of the shutoff were delayed by several days (data not shown).

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Kinetics of FHV RNA1 replication in RNA-transfected cells. (Lower panel) EcR-CHO cells were transfected with FHV RNA1 and incubated at 28°C. At 24-h intervals, cells were pulse-labeled for 2 h with [3H]uridine in the presence of actinomycin D. Total cellular RNAs were isolated. Labeled RNAs were then acid precipitated and quantitated by scintillation spectrometry. The rate of RNA replication (counts per minute per 1.7 × 105 cells) was expressed as a function of incubation time. Results were normalized to the amount of rRNA present in each sample as described in Materials and Methods. (Middle panel) The relative abundance of FHV RNA1 in the total RNA pool isolated at each time point was determined by primer extension with an RNA1-specific primer and quantitated by densitometric analysis as described in Materials and Methods. The amounts are expressed as percentages of the maximum (day 2) value. (Upper panel) A portion (4%) of the total cellular RNAs isolated at each time point shown in the lower panel were transfected into a fresh monolayer of EcR-CHO cells and incubated at 28°C. At 24 h posttransfection, RNAs were labeled, isolated, and analyzed as described for the lower panel.

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Primer extension analysis of RNA1 isolated from transfected cells. Equivalent amounts of total cellular RNAs isolated at each time point during the experiment shown in the lower panel of Fig. 1 (RNA1-transfected lanes) were analyzed by primer extension with an RNA1-specific primer as described in Materials and Methods. The RNA1 stock originally used to transfect cells in Fig. 1, lower panel, was analyzed in parallel as a control (Input RNA1 lanes). Products were separated by electrophoresis on a 6% polyacrylamide sequencing gel alongside a dideoxynucleotide sequencing ladder generated by using the same primer with plasmid pIND-FHV1[0,0] as template; an autoradiograph is shown. The sequencing lanes were labeled with the complement of the terminating dideoxynucleotide to reflect the sequence of the viral positive strand. The sequence of the 5' terminus of RNA1 is shown. The position of the primer extension product that corresponds to capped RNA1 is indicated. The FHV RNA1 sequence in vF1 lacks nt C6 so the primer extension product corresponding to capped RNA1 comigrates with nt 1 of the sequencing ladder shown in lanes 1 to 4 rather than migrating 1 nt more slowly as it does in Fig. 7, where C6 is present.

View larger version (61K): [in this window] [in a new window]   FIG. 3.   Analysis of FHV protein synthesis. Cells were transfected with authentic FHV RNA1 as indicated in the legend to Fig. 1. At 24-h intervals, the cells were labeled for 3 h by metabolic incorporation with [35S]Met-[35S]Cys. Labeled proteins were separated on an SDS-12.5% polyacrylamide gel and visualized by autoradiography. Lanes: 0, mock transfected; 1 to 5, transfected with FHV RNA1. The migration pattern of molecular mass standards is indicated at right in kilodaltons.

Development of an inducible plasmid-driven RNA replication system. In order to address the question of cause and effect and to further examine the role played by the intracellular environment in the regulation of RNA replication, we developed a system in which replication was initiated by the synthesis of DNA-templated primary transcripts. We showed previously that FHV RNA1 transcribed from a constitutive RNA pol II promoter could initiate autonomous cytoplasmic RNA replication in plasmid-transfected mammalian cells (34). Replacement of the constitutive promoter with an inducible promoter enabled us to use this system to further study replicative shutoff because it allowed temporal separation of RNA replication from primary transcription.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   Schematic diagram of plasmid designed to direct inducible FHV RNA1 replication. The transcriptional start site and self-cleavage site of the RNA transcript are indicated by vertical arrows. The nucleotide sequence of the junctions between the promoter, cDNA, and ribozyme are shown below the plasmid diagram. HRE, hormone response elements; Pr, minimal RNA pol II promoter; Rz, HDV antigenomic ribozyme; T, T7 terminator.

The inducible system reproduces the replicative shutoff. Upon transfection of EcR-CHO cells with plasmid pIND-FHV1[0,0], followed by treatment with the steroid hormone muristerone A, RNA replication was induced and then subsequently shutoff when the hormone was removed (Fig. 5, dotted line), a result reminiscent of the behavior observed after transfection with authentic RNA1. Surprisingly, the replicative shutoff was partially alleviated if the hormone was present continuously (Fig. 5, solid line), indicating that continued primary transcription prolonged the duration of RNA replication.

View larger version (19K): [in this window] [in a new window]   FIG. 5.   RNA replication in CHO cells transfected with pIND-FHV1[0,0]. Cells were transfected with plasmid pIND-FHV1[0,0] and incubated at 37°C for 24 h. The cells were shifted to 28°C and treated with muristerone A for 24 h. Thereafter, hormone was replenished daily in one set of dishes () and removed from a parallel set by washing and replacing with complete medium that lacked the hormone (). At daily intervals, cells were pulse-labeled with [3H]uridine in the presence of actinomycin D. Labeled RNAs were isolated and quantitated as described for Fig. 1. The rate of RNA replication (in counts per minute per 1.7 × 105 cells) was expressed as a function of time after hormone addition.

RNA replication can be induced repeatedly. One possible explanation for the failure to maintain long-term RNA replication after hormone removal was that the process had perturbed the intracellular environment in such a way that it was no longer able to support RNA replication. To examine this possibility, we tested whether a second round of RNA replication could be achieved by reinducing primary transcription after shutoff. Cells were either transfected with pIND-FHV1[0,0] (Fig. 6A, solid lines) or else mock transfected (Fig. 6A, dotted lines) and then treated with muristerone A as before. At daily intervals, cells in parallel dishes were pulse-labeled with [³H]uridine in the presence of actinomycin D, beginning immediately before hormone addition and continuing for 6 days. The hormone was removed after 2 days and incubation and labeling were continued as before. At 6 days after the initial hormone treatment began, the cells were passaged to minimize overcrowding and again treated with muristerone A for 2 days beginning on day 7. Day 7 was chosen to ensure that the first round of replication had been completely shut off before reinduction, allowing clear temporal distinction between the products of the two rounds. As before, sister monolayers were labeled daily with [³H]uridine in the presence of actinomycin D. Labeled replication products were quantitated by acid precipitation, and the results are shown in Fig. 6A.

View larger version (40K): [in this window] [in a new window]   FIG. 6.   Analysis of RNA replication in cells transfected with pIND-FHV1[0,0] in response to two pulses of hormone treatment. Cells were transfected with pIND-FHV1[0,0] () or mock transfected () and treated with hormone as described for Fig. 5, except that the hormone was removed from parallel dishes after 48 h, and incubation was continued in its absence for the times indicated. The cells were passaged, "(P)", onto fresh plates on day 6, and a second round of hormone addition and removal was performed. During each round of induction, cells were pulse-labeled in the presence of actinomycin D at daily intervals for 5 days, as in Fig. 1. First round, days 0 to 5 (left panels); second round, days 8 to 13 (right panels). Labeled RNAs were isolated and analyzed as follows. In panel A, RNAs were acid precipitated in triplicate, quantitated by scintillation spectrometry, and normalized to rRNA levels as described for Fig. 1. In panel B, RNAs were normalized to rRNA levels as described above and analyzed by electrophoresis on denaturing formaldehyde-agarose gels; the fluorographs are shown. Lanes 1 to 6 and 13 to 18, mock-transfected cells; 7 to 12 and 19 to 24, cells transfected with plasmid pIND-FHV1[0,0].

View larger version (54K): [in this window] [in a new window]   FIG. 7.   Primer extension analysis of RNAs from the two-round induction experiment. Equivalent amounts of total cellular RNAs isolated at each time point in Fig. 6 were analyzed by primer extension with an RNA1-specific primer as described in the legend to Fig. 2. An autoradiograph of the sequencing gel is shown. The position of the primer extension product that corresponds to capped RNA1 (which migrates as 1 nt longer than the predicted size for the product from uncapped FHV RNA1) is indicated; the sequence of the 5' terminus of RNA1 is shown.

View larger version (29K): [in this window] [in a new window]   FIG. 8.   Replication in naive cells of RNAs isolated from pIND-FHV1 transfected cells after shutoff. (Lower panel) Reproduction of Fig. 6A (right panel): solid line, cells transfected with pIND-FHV1[0,0]; dotted line, mock transfected cells. (Middle panel) The relative abundance of FHV RNA1 in the total RNA pool isolated at each time point was determined by primer extension with an RNA1-specific primer, followed by densitometric analysis, as described in Materials and Methods. The amounts are expressed as percentages of the maximum (day 12) value. (Upper panel) Portions (4%) of the total RNAs isolated at each time point in Fig. 6A (right panel) were transfected into fresh monolayers of EcR-CHO cells. After 24 h, the products of RNA replication were labeled, analyzed, and normalized as described in Fig. 1.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In general, the events involved in the replication of viral positive-strand RNA genomes via negative-strand intermediates can be divided into four phases on the basis of their kinetic properties, which have been studied extensively by using bacteriophage Q, both in vitro (7-11) and in vivo (23). During the first, or lag, phase of Q RNA replication in vivo, the input positive-sense viral RNA is used predominantly as a template for translation; the RNA concentration remains essentially constant, and the replicase concentration increases until sufficient enzyme has accumulated for the onset of RNA replication. The lag phase is followed by a hyperbolic phase characterized by rapid increases in the rates of synthesis of both RNA strands (positive and negative polarity) and of the replicase itself. At this time, the rate of replication increases so that it exceeds the translation rate, eventually saturating the machineries of translation and replication. The positive-strand RNAs can bind to ribosomes, to the replicase, or to the viral coat protein, while the negative strands bind exclusively to the replicase (23).

The reaction next enters a linear phase, when RNA replication and protein synthesis slow to constant rates due to saturation of the replicase and the ribosomes, respectively, with free viral RNA. For Q, positive-strand synthesis is stimulated at the expense of the negative strand when a factor required for the latter process becomes limiting (23, 56), so that the bulk of the positive-strand progeny are produced during this phase of the reaction. No further increase in the replicase concentration is observed, as viral protein synthesis shifts to the production of the coat protein. Late in replication only a tiny fraction of the RNA product appears to be translated or replicated. In the final phase, the rates of synthesis decline concomitantly with packaging of the progeny positive strands into the viral coat protein (23).

In the experiments presented here, we examined the transition from the early and intermediate phases of FHV RNA replication to the final stage, where the reaction slowed progressively and eventually stopped. This stage of nodavirus RNA1 replication has not been examined previously because most earlier analyses were performed over shorter time periods and it was not possible to temporally separate RNA replication from primary transcription until the development of the inducible cDNA transcription system (3, 34, 48).

Two lines of evidence suggest that the observed inhibition is specific for actively replicating viral RNAs and does not represent a general transcriptive inhibition. The first is provided by the gross pattern of cellular protein synthesis, which did not change qualitatively or quantitatively over the course of the experiment shown in Fig. 3, while viral protein synthesis was shut off. Clearly, this result demonstrates that functional levels of cellular mRNAs were maintained, because cellular protein synthesis would diminish in the absence of continued transcription. The reinduction experiment shown in Fig. 6A provides further evidence of this specificity. In that case, switching primary transcription back on at a time when replication had ceased resulted in restoration of replication, showing that de novo transcription was still active in these cells.

Strikingly, synthesis of the FHV replicase diminished with kinetics that approximated the replicative shutoff, despite the continued presence of the RNA that encodes it (Fig. 1 to 3). This observation supports the hypothesis, posed by the Q model, that replication stops when progeny RNAs can no longer feed back into the replicative cycle as templates for translation or replication. However, we observed a late decline in the rate of RNA replication with an RNA that does not encode virion structural proteins, suggesting that for FHV the block to feedback of progeny RNAs can be separated from packaging into virions.

What other reasons could account for the diminishing feedback at this point in the replicative cycle? The results presented above allowed us to eliminate several plausible explanations. For example, the accumulated RNA1 maintained its intrinsic message and template activities, as demonstrated by its ability to replicate anew when transferred to naive cells (Fig. 1 and 8). This refuted the idea that the replicative shutoff was due to the accumulation of deleterious mutations (error catastrophe, as defined in reference 44) which might result from error-prone RNA replication. These results exonerated both the quantity and the quality of the RNA1 from being responsible for mediating shutoff, and they focused our attention instead on the ability of the intracellular environment to support replication.

In cells where it had previously been shut off, replication resumed only when cDNA transcription was reinduced (Fig. 6 and 7), while any changes in cellular metabolism due to cell passaging failed to restore replication. These results are not consistent with the presence of an inhibitor of replication or depletion of a necessary cellular host factor. Instead, they established that the cytoplasmic environment remained hospitable after replication had ceased and was therefore not responsible for the shutoff.

This conclusion was further supported by the observation that the shutoff could be partially alleviated when FHV RNA1 was replenished continuously by transcription of a nuclear cDNA (Fig. 5). Could this be due to some intrinsic difference between the primary transcripts exported from the nucleus and the progeny of cytoplasmic RNA replication? None of the predicted covalent differences (cap structure, 3' end) are expected to alter the translation or replication of the RNA. However, the fates of the two RNAs are markedly different. The pathway by which pol II transcripts are transported from the nucleus evolved to export spliced mRNA and deliver it to the cytoplasmic translation machinery (14, 15, 36). Therefore, the nuclear transcripts most likely are exported and translated to yield fresh replicase that can initiate a new replication cycle in spite of the block to translational and/or replicative feedback that is imposed on the products of cytoplasmic replication.

The question still remains how the progeny RNA can be prevented from feeding back into the reaction. One possibility is that at the later stages of replication, the RNA products are sequestered away from ribosomes and/or the replicase, perhaps in membranous compartments. The FHV replicase is known to be membrane associated, and intact membranes are required for RNA replication (33, 52, 60, 61). Perhaps the later products of RNA replication are released into a separate membrane-bound compartment, where they are no longer accessible as templates for either translation or replication. The observation that most of the FHV capsid protein is assembled into provirions within a few minutes of synthesis, yet encapsidation of newly synthesized RNA occurs very slowly (30), indicates that a substantial pool of unencapsidated progeny RNA exists. A membrane-associated progeny RNA pool has been observed in poliovirus-infected cells and may be the site of virion assembly in the cell (12, 13, 47, 58). The recent observation that poliovirus RNA replication and RNA packaging are coupled, perhaps as a result of direct interaction between the replication machinery and the capsid proteins (43), lends credence to the proposal. This model does not exclude the possibility that RNA replication could result in local depletion of a factor still abundant in the cytoplasm as a whole, resulting in replicative shutoff at the initial site. In that case, primary transcription products establishing new sites of replication might have access to a fresh supply of the limiting factor.

Alternatively, it is possible that the shutoff is the consequence of a replication-specific host response mechanism, such as the interferon response, apoptosis, or homology-dependent gene silencing. However, during shutoff, no evidence for a global inhibition of protein or RNA synthesis, such as occurs in apoptosis or sometimes occurs during an interferon response (21, 22, 28, 31, 37, 62), was detected (Fig. 3 and 6). In addition, the morphologies of cells transfected with RNA1 or FHV1 plasmids when observed by light or electron microscopy were inconsistent with the cellular alterations observed during apoptosis (data not shown). Recent reports have suggested that some animal cells may have mechanisms of gene silencing (24, 45, 50) that resemble the phenomena discovered in plants and fungi (reviewed in reference 59), although no analogous phenomenon has yet been reported for mammalian cells. Nonetheless, the intriguing observation that interference with the expression of selected genes in Caenorhabditis elegans is initiated by sequence-specific double-stranded RNA (24) raises the possibility that the shutoff of FHV RNA replication reported here might be the result of a similar silencing mechanism in which double-stranded RNA is involved. The latter possibility remains to be tested.

In summary, the work presented here indicates that FHV RNA1 replication is inhibited despite the continued presence of replication-competent RNA in a cytoplasmic environment that retains the ability to support further RNA replication. Further work will be required to determine the exact mechanism of the replicative shutoff.