INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

In a poliovirus (PV)-infected cell, the first virus-specific synthesis is the translation of input plus-strand viral RNA into a polyprotein which is co- and posttranslationally cleaved into functional proteins. While several of these proteins are known to be involved in the replication of the viral RNA, RNA synthesis per se is effected by the viral polymerase 3D^pol (5, 22, 45). This enzyme first copies input plus-strand RNA into minus-strand RNA, which, in turn, serves as a template for the synthesis of progeny plus strands in the multistranded replicative intermediate (RI) (23).

Data from in vitro experiments indicate that minus-strand RNA synthesis can proceed without cellular structural prerequisites (30, 49). However, membranes may be involved in the initiation step of minus-strand synthesis by presenting membrane-bound 3AB as a precursor of VPg, which is thought to be involved in priming minus-strand synthesis (25, 35).

In contrast, plus-strand RNA synthesis shows distinct structural requirements. In vivo, plus-strand synthesis is associated with specific cellular membranous structures (9, 12, 15) and takes place in a replication complex on the surface of cytoplasmic vesicles induced by the viral protein 2BC (2, 13, 17). In vitro transcription systems, derived from infected cells (11, 14, 21, 41), comprise the replication complex in a rosette-like arrangement of several vesicles surrounding the actual RNA replicating structure (11, 14). Analysis of such systems showed that the membranes of the vesicles are necessary for initiation of RNA synthesis but dispensable for elongation of RNA (20). However, the exact role of membranes in cell-free systems, derived from uninfected cells and replicating PV de novo, is not yet established (6, 7, 27, 28, 43).

The role of membranes in viral plus-strand RNA synthesis in infected cells or subcellular fractions, as summarized above, was investigated mostly at late times postinfection (p.i.), i.e., when vesicle formation and RNA synthesis were at their peaks. Only little information on membrane association of viral plus-strand RNA synthesis at early times of infection is available. Likewise, the location of viral minus-strand RNA synthesis is largely unknown. Therefore, in the present investigation we determined the intracellular location of minus-strand RNA in comparison with that of plus-strand RNA over time. This would allow for a better understanding of the interdependence of and possible differences in the mechanisms governing plus- and minus-strand RNA synthesis.

To address these questions, the locations of plus- and minus-strand RNAs were determined by fluorescent in situ hybridization (FISH) with confocal microscopy. The use of plus- or minus-strand-specific riboprobes, each labelled with a different fluorochrome, allowed for the simultaneous detection of plus- and minus-strand RNA. Association of viral RNA with membranes was determined by combining FISH and immunofluorescence (IF) detection of the viral 2B- and 2C-containing P2 proteins, which served as a marker for virus-induced membranes (9, 10, 39).

Minus-strand RNA could be detected in distinct, regularly sized, round membranous structures also containing plus-strand RNA and remaining constant in amount and size over the entire viral growth cycle. Plus-strand RNA was additionally found at early times, i.e., before peak RNA synthesis, in small clusters of vesicles, as judged from its colocalization with P2 proteins. Around peak synthesis, all RNA-associated membranous structures migrate to the center of the cell to form a characteristic juxtanuclear area of vesicles, with the distinct plus- and minus-strand-containing compartments still being clearly delineated. Interestingly, the two compartments could be distinguished only by FISH, according to their RNA content, and not on an ultrastructural level.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and virus. HEp-2 cell suspension cultures were infected with PV type 1 Mahoney at a multiplicity of infection of 30 PFU. The virus was allowed to adsorb at 4°C for 30 min in serum-free medium (Joklik minimal essential medium) (Seromed, Berlin, Germany). The cells were then washed once in cold serum-free medium and incubated at 36°C in medium containing 10% calf serum.

Isolation of RNA from infected cells. At various times p.i., 10⁶ to 10⁷ cells were washed twice in cold serum-free medium. The RNA was isolated from the cell pellet by using Trizol reagent (Gibco BRL, Gaithersburg, Md.) according to the instructions of the manufacturer.

In vitro transcription of RNA for RNase protection assays (RPAs). All PV-specific RNAs were derived from plasmid pT7PV, containing the full-length genome of PV type 1 Mahoney (16). The [-³⁵S]UTP (Amersham International, Little Chalfont, United Kingdom)-labelled plus-polarity riboprobe consisting of nucleotides (nt) 1 to 220 (riboprobe nt 1-220) was transcribed with T7 RNA polymerase (Boehringer, Mannheim, Germany) from a gel-purified BamHI fragment of pT7PV. The ³⁵S-labelled minus-polarity riboprobe nt 6065-6276 (numbering as for the positive-sense sequence) was prepared by transcription with T7 polymerase of PCR-amplified cDNA nt 6012-6276-T7, trimmed with HindIII and gel purified. Unlabelled RNAs nt 1-595 (plus polarity), nt 1-460, and nt 6012-6736 (both plus and minus polarity) were transcribed from corresponding cDNA fragments which were PCR amplified from EcoRI-linearized pT7PV DNA by using appropriate primers of 18 to 20 nt in length preceded by SP6 or T7 promoter sequences (40, 50).

RPA for detection of plus-strand RNA. Ten picomoles of ³⁵S-labelled riboprobe nt 6065-6276 of minus polarity was added to RNA extracted from infected cells. Hybridization was done in 20 µl of 80% deionized formamide-100 mM sodium citrate (pH 6.4)-300 mM sodium acetate (pH 6.4)-1 mM EDTA at 48°C overnight. RNase digestion was done at 37°C by adding 200 µl of RNase buffer (10 mM Tris HCl [pH 7.5], 1 mM EDTA, 200 mM NaCl, 100 mM LiCl) containing 0.35 U of RNase A per ml and 13.5 U of RNase T₁ per ml (Boehringer). After proteinase K digestion (final concentration, 0.13 mg/ml; Boehringer) and ethanol precipitation, RNA corresponding to 1.25 × 10⁵ (late times p.i.) to 1 × 10⁷ (early times p.i.) cells was separated on a 2.5% NuSieve (FMC, Rockland, Maine) denaturing agarose gel, blotted as described previously (20), visualized by autoradiography, and quantitated on a CDS200 densitometer (Beckman, Palo Alto, Calif.).

Two-cycle RPA for detection of minus-strand RNA. Before the first hybridization step of the two-cycle RPA (31), an excess of 1 µg of unlabelled RNA nt 1-595 of plus polarity was added to the RNA from 5 × 10⁶ to 1 × 10⁷ Trizol-extracted cells to ensure full protection of all minus strands. Hybridization was done in 40 µl of 4 M guanidine thiocyanate in 25 mM sodium citrate at 37°C overnight. RNase digestion (to eliminate excess single-stranded plus-strand RNA) was done in 400 µl of buffer (50 mM Tris HCl [pH 7.5], 10 mM EDTA, 400 mM NaCl) containing 5.7 U of RNase A per ml and 0.3 U of RNase T₁ per ml (Boehringer) at 37°C for 30 min. After proteinase K digestion and phenol extraction, the samples were put through the same RPA as described for plus-strand detection, using 10 pmol of the ³⁵S-labelled riboprobe nt 1-220 of plus polarity.

Standards for quantitative RNA determinations. Dilution series of known amounts of unlabelled standard RNA hybrids were subjected to the same RPA as used for RNA from infected cells. As a standard for the quantitation of plus-polarity viral RNA, double-stranded RNA of nt 6012 to 6736 and a ³⁵S-labelled probe (nt 6056 to 6276) of minus polarity were used. For minus-strand viral RNA, double-stranded RNA of nt 1 to 460 and a ³⁵S-labelled probe (nt 1 to 220) of plus polarity were used.

Preparation of cells for confocal microscopy. After infection with PV type 1 Mahoney, 1.5 × 10⁶ HEp-2 cells were dispersed on poly-L-lysine-coated coverslips at various times p.i. After adhesion to the coverslips for 4 min, cells were fixed for 10 min at room temperature by gentle immersion in 4% paraformaldehyde in PBS and permeabilized with 0.2 to 0.3% Triton X-100 in 4% paraformaldehyde for 10 min. After a phosphate-buffered saline (PBS) wash, aldehydes were quenched in 0.5 M ammonium chloride in PBS for 7 min. The coverslips were washed twice in PBS and processed for FISH and IF.

FISH. For plus-strand RNA detection, two riboprobes of minus polarity were prepared from appropriate DNA fragments that had the T7 promoter sequence added by PCR. The probe nt 1-7441 was labelled with fluorescein isothiocyanate (FITC)-UTP (Boehringer) during in vitro transcription, and the probe nt 6875-7441, used in double-FISH experiments, was labelled with Texas red-UTP (Molecular Probes, Eugene, Oreg.). Unincorporated nucleoside triphosphates were removed with a Micro Bio-Spin 6 column (Bio-Rad, Hercules, Calif.). To ensure good penetration of the probes into the fixed and permeabilized cells, the probes were subjected to alkaline hydrolysis (18) to generate fragments of approximately 100 nt in length. Occasionally, a second column (Bio-Spin 30) was used to remove oligonucleotides of less than 20 nt in length. The riboprobe was dissolved in hybridization buffer (50% formamide, 10 mM Tris-HCl [pH 7.4], 600 mM NaCl, 10% dextran sulfate, 10 mM dithiothreitol, 0.05% bovine serum albumin, 0.1% sodium dodecyl sulfate, 200 µg of salmon sperm DNA per µl, 100 µg of yeast tRNA per µl) and hybridized to the cells at 40°C overnight. After four washes in 0.1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 10 min, the slides were mounted with 33.3% (vol/vol) glycerol containing 16.6% (wt/vol) Mowiol (Hoechst, Frankfurt, Germany) and 2.5% (wt/vol) 1,4-diazabicyclo[2.2.2]octane (DABCO) (Sigma, Buchs, Switzerland).

IF. The preparation and specificities of monoclonal antibodies (MAb) directed against proteins 2C and 2B were described earlier (20, 33). The MAb were used for an indirect IF assay with goat anti-mouse immunoglobulin G (IgG) coupled to Texas red (Molecular Probes). For colocalization studies of Golgi complexes and protein 2C, an anti-Golgi 58-kDa protein mouse MAb (Sigma) and a rabbit anti-2C polyclonal Ab (17) were used. IF detection was done with goat anti-mouse IgG coupled to Texas red and goat anti-rabbit IgG coupled to FITC (Sigma). Images were recorded on a Leica TCS4D confocal microscope equipped with appropriate filter sets. The photomultiplier settings were adjusted so that no Texas red signal could be detected in the FITC channel and vice versa. Raw images were corrected for contrast, intensity, and background fluorescence by using Adobe Photoshop software.

Electron microscopy (EM). Cells were fixed in glutaraldehyde and OsO4 and embedded in Poly/Bed 812 (Polysciences, Warrington, Pa.) by standard protocols (8).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Kinetics of plus- and minus-strand RNA synthesis. To correlate the results of the FISH experiments reported below with those of the time course experiments, particularly the peaks of plus- and minus-strand viral RNA synthesis, we monitored the replication of PV plus- and minus-strand RNAs separately by RPA. For plus-strand detection, a ³⁵S-labelled probe of minus polarity spanning nt 6065 to 6276 (numbering is as for the positive-sense RNA of PV type 1 Mahoney) was used, and for minus-strand detection, a two-cycle RPA was performed as described previously (31) with a ³⁵S-labelled probe of plus polarity (nt 1 to 220).

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Kinetics of plus- and minus-strand RNA synthesis in PV-infected HEp-2 cells as determined by RPA. (a) Time p.i. versus amounts (logarithmic scale) of plus ()- and minus ()-strand RNA synthesized. (b) Peak synthesis of plus- and minus-strand RNA is between 3 and 3.5 h p.i.

FISH of plus- and minus-strand viral RNA. To visualize plus- and minus-strand PV RNAs separately, we performed FISH experiments with nonoverlapping plus- and minus-strand-specific RNA probes. The probes, labelled with different fluorochromes during in vitro transcription, were used either separately or, mixed together, simultaneously on the same slide. Plus-strand RNA was detectable without pretreatment of the slides (8), whereas detection of minus-strand RNA was possible only after thermal denaturation of the specimen at 65°C for 5 to 10 min, similar to earlier findings with in situ hybridization on sections (44).

View larger version (83K): [in this window] [in a new window]   FIG. 2.   Evaluation of background staining in FISH with uninfected (left) and PV-infected (right) cells in the same microscopic field with a Texas red-labelled plus-strand-specific probe (a) and an FITC-labelled minus-strand-specific probe (b). For an explanation of structural details, see the legend to Fig. 3.

View larger version (81K): [in this window] [in a new window]   FIG. 3.   Viral RNA and protein detected in PV-infected HEp-2 cells by fluorescent confocal microscopy. Bar, 5 µm. (A) Simultaneous visualization of viral plus- and minus-strand RNA in the same cell by double-labelling FISH. Plus-strand RNA (panels a and d), detected with a Texas red-labelled riboprobe of minus polarity, is found in several small granules, at first dispersed and later in infection concentrated in a juxtanuclear area. Minus-strand RNA (panels b and e), detected with an FITC-labelled riboprobe of plus polarity, is present in larger, distinct granules. Superimposed plus- and minus-strand detection shows colocalization of both signals in yellow (panels c and f). Panels a to c, 2.5 h p.i.; panels d to f, 3.5 h p.i. (B) Minus-strand RNA, detected with an FITC-labelled riboprobe as for panel A, remains in distinct granules throughout the replication cycle. Panel a, 2.25 h p.i.; panel b, 2.5 h p.i.; panel c, 3.0 h p.i.; panel d, 3.5 h p.i. (C) Simultaneous visualization of viral plus-strand RNA and viral protein 2C by combined FISH and IF. Plus-strand RNA (panels a and d) is detected with an FITC-labelled riboprobe. Protein 2C and 2C-containing precursors (panels b and e) are detected with an anti-2C MAb and Texas red-labelled anti-mouse IgG. Superimposed FISH and IF show that early in infection, plus-strand RNA and 2C are colocalized (yellow in panel c); at later times some dissociation of 2C from RNA is visible (panel f). Panels a to c, 2.0 h p.i.; panels d to f, 3.5 h p.i. (D) IF with a MAb against Golgi protein p58 detected with Texas red-labelled anti-mouse IgG, combined with IF with a polyclonal rabbit anti-2C Ab detected with an FITC-labelled anti-rabbit IgG. Panel a, intact Golgi complexes at 2 h p.i. Panel b, the same cells as in panel a, showing IF with anti-2C Ab superimposed. Little colocalization (yellow) of the two targets is found. Panel c, superimposed pictures of a cell double labelled by IF with anti-p58 and anti-2C Ab at 2.5 h p.i. The Golgi marker and viral protein 2C colocalize partially. Panel d, superimposed pictures as in panels b and c at 3 h p.i. The Golgi marker is distributed in the cytoplasm and is colocalized with protein 2C mainly at the border of the juxtanuclear area of vesicles.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   FISH for minus-strand detection in a series of optical sections, spaced 0.4 µm apart, through an infected cell at 2.75 h p.i. Minus-strand RNA is found in single spherical granules of approximately 0.5 µm in diameter in a symmetrical array around an unlabelled center. Bar, 5 µm.

View larger version (145K): [in this window] [in a new window]   FIG. 5.   Electron micrograph of a PV-infected HEp-2 cell at 2 h p.i., showing individual small clusters of virus-induced vesicles (arrowheads). Bar, 1 µm.

Combined localization of viral RNA and P2 proteins. The viral P2 proteins 2B, 2C, and 2BC are known to be markers for virus-induced membranes and have been found to be associated with the RI-containing replication complex in the juxtanuclear area of vesicles (9-11, 14). We wanted to know whether the early plus-strand-RNA-containing granules observed with the confocal microscope are composed of 2B- and 2C-containing membranes and thus presumably carry conventional replication complexes. To test this hypothesis, we combined FISH for plus-strand detection and IF for P2 protein localization with MAb specific for 2C, thus recognizing proteins 2BC and 2C. Figure 3C, panel c, shows an almost-perfect colocalization (yellow) of the P2 signals (Texas red) (Fig. 3C, panel b) with plus-strand RNA (FITC) (Fig. 3C, panel a) at early times p.i. After 3 h p.i., i.e., when a compact juxtanuclear area of vesicles is formed, the protein signal (red) dissociates partially from that of the plus-strand RNA (green) (compare panels d and e in Fig. 3C) and tends to accumulate more toward the border of the juxtanuclear area, leading to a partially nonhomogeneous appearance of this area in superimposed IF-FISH images (Fig. 3C, panel f). Combining FISH and IF analysis with an anti-2B MAb gave virtually the same result (not shown).

Early in infection, the Golgi complex is not the unique origin of the virus-induced vesicles. The onset of viral plus-strand RNA synthesis in peripheral, separated clusters of vesicles led us to investigate whether these early clusters originated from Golgi-derived vesicles, as suspected previously for morphological reasons (13) and as inferred from immunocytochemical studies implicating 2B as a Golgi-targeted viral protein (37). The presence of Golgi components in virus-induced vesicles has also been shown recently by immunoprecipitation, albeit for vesicles late in the infection (39). Since Golgi markers proved to be too delicate as antigens for the FISH conditions, even without thermal denaturation, we compared instead the location of Golgi markers with the location of the vesicle- and replication complex-associated 2C-containing P2 proteins (9, 14, 20, 39). Figure 3D, panel a, shows that at 2 h p.i. Golgi complexes (Texas red) are still intact, and only very few 2C (FITC) and Golgi epitopes colocalize (yellow) (Fig. 3D, panel b). At 2.5 h p.i., the Golgi complex starts to disintegrate, with the corresponding marker becoming distributed throughout the cytoplasm (Fig. 3D, panel c). There is some overlap between the Golgi marker- and the 2C-specific signals. At 3 h p.i. (Fig. 3D, panel d), P2 protein-carrying vesicles become condensed in the juxtanuclear vesicular area, whereas Golgi membranes remain spread out in the cytoplasm. Some colocalization (yellow) of both markers is found, particularly at the border of the juxtanuclear area. These observations argue against an origin of the vesicular clusters from Golgi complexes early in infection but are compatible with a contribution of Golgi membranes or membrane components to virus-induced vesicles in the juxtanuclear area (39).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In our virus-cell system, plus- and minus-strand viral RNA syntheses peak at the same time, i.e., at 3 to 3.5 h p.i. (Fig. 1). Concomitantly, a rapid and drastic change in the array of virus-induced membrane structures occurs, in that all replication-associated vesicles migrate toward a central area of the cell where they accumulate (Fig. 3A, panels d to f). We attempted to compare the viral RNA locations before and after this breakpoint and particularly to monitor the appearance of viral minus-strand RNA. Detection of viral RNA by RPA as well as localization of plus-strand RNA by FISH with the confocal microscope gave positive results as early as 1.5 h p.i.; minus-strand RNA was detected at 2 h p.i. by FISH. Virus-specific RNA synthesis must occur still earlier, but at present this remains beyond the detection limit.

The earliest morphological changes detectable in a PVirus-infected cell by EM were found at around 1.5 to 2 h p.i. The changes consist of small clusters of vesicles scattered through the cytoplasm. It has been suggested that such early vesicles might be Golgi derived (13) and that the Golgi complex represents an early target for PV replication (37). Therefore, we tested by double IF whether the P2 proteins, indicative of virus-induced vesicles, colocalized to the Golgi. The results obtained indicate that the membranes of the emerging virus-induced vesicular clusters are not, or are only in minute amounts, Golgi derived and thus that viral RNA synthesis employs, at its beginning, few if any Golgi-derived vesicles. However, later in infection and after virus-induced disruption of Golgi complexes, Golgi-derived membranes were found redistributed throughout the cytoplasm (37, 39) and colocalized partially to the virus-induced vesicles (Fig. 3D, panel d). This is compatible with the finding that Golgi marker proteins are present in isolated P2-containing virus-induced vesicles (39). It is not known whether the vesicles bearing Golgi markers originate in toto from the Golgi complex or whether they arise by fusions between Golgi- and endoplasmic reticulum-derived membranes.

The in situ detection of viral minus-strand RNA became feasible only with the use of strand-specific riboprobes directly labelled with the fluorochrome, which obviated any further, e.g., immunological, detection of the hybridized probe. Viral minus-strand RNA was found to be contained in relatively large, round bodies of regular size, which kept the same appearance throughout the replication cycle. Interestingly, no ultrastructural equivalent of the large, minus-strand-RNA-containing dots could be identified. Thus, these dots are defined by RNA and protein contents as determined by confocal microscopy, and they are considered to correspond, on the EM level, to morphologically nondelineated clusters or aggregates of virus-induced vesicles.

Colocalization experiments visualizing plus- and minus-strand RNA in the same cell showed viral RNA to be present in two different structures: the large round bodies, which harbor plus-strand RNA in addition to minus-strand RNA, and small granules which carry plus-strand RNA and lack a recognizable minus-strand-specific signal. From the observation that both structures carry P2 proteins, which are known to be exclusively membrane bound (9, 14, 20, 39), and from a comparison with EM data, we conclude that both structures correspond to aggregates of virus-induced vesicles.

The intracellular dispersed location of both structures changed around peak RNA synthesis, when they migrated centripetally into a juxtanuclear area. There, the smaller clusters, containing only plus-strands, coalesce, whereas the larger, plus- and minus-strand-RNA-containing dots remain as recognizable entities. The highly ordered redistribution of the individual vesicular clusters is indicative of a distinct and elaborate spatial organization of the RNA synthetic activities within the infected cell.

The presence of two distinct compartments as defined by appearance as well as RNA content suggests that they also differ in certain functions in RNA replication. Previously, we have shown by EM autoradiography that not only the compact juxtanuclear vesiculated area but also clusters of vesicles, as observed early in the infectious cycle, are involved in viral RNA synthesis (9). A comparison of the EM autoradiography and the FISH data suggests that viral plus- and minus-strand RNA synthesis is vesicle associated from the beginning. However, a reevaluation of our high-resolution EM autoradiography data did not allow us to define structures or areas corresponding to the larger, plus- and minus-strand-RNA-containing distinct bodies.

The pattern of the autoradiography signal suggests that the large plus- and minus-strand-containing bodies are not the exclusive structures of plus-strand RNA synthesis in the RI. This is also made likely by the observation that the FISH signal of a probe recognizing the 5' end of the plus strand does not preferentially label the minus-strand-containing bodies, clearly not more than the 3' probe used in this paper (not shown). This argues against predominantly nascent plus strands, i.e., RIs, in these bodies. We propose rather that these structures represent the site of viral minus-strand RNA synthesis and that this activity might be followed closely by plus-strand initiation. The resulting newly induced RI might then soon leave this compartment to continue further plus-strand synthesis on the plus-strand-specific vesicular structures (Fig. 3A) (11), where mature plus strands would accumulate (44). Thus, we propose that the plus- and minus-strand-RNA-containing bodies represent the starting points, or germ centers, of viral RNA replication.

The P2 proteins 2B and 2C and their precursor 2BC have all been found to be directly or indirectly involved in viral RNA replication (7, 26, 34, 42, 46-48), although the exact roles of these proteins are only partially understood. Our data show the P2 proteins to be colocalized primarily with all of the plus-strand FISH signals and not preferentially with the plus- and minus-strand-RNA-containing bodies. This indicates an involvement of these proteins in plus-strand synthesis (4), in addition to their proposed role in minus-strand synthesis, as concluded from data obtained with a cell-free translation-transcription system (7). Since the juxtanuclear area of vesicles acquires new vesicles at its periphery, the margination of the P2 proteins in this area at later times is still compatible with a role of 2C in the organization of the plus-strand replication complex (14) and with a role of protein 2BC in the induction of its vesicles (1, 2, 13, 17).

It was found that continuous lipid synthesis, and thus vesicle formation, is necessary for ongoing viral plus-strand RNA synthesis (24). It is tempting to speculate that the large, presumably minus-strand-synthesizing bodies stay at their constant size because membrane synthesis, and thus an increase in the number of vesicles, is not required for minus-strand RNA synthesis.

For many RNA virus families, including picornaviruses, flaviviruses, togaviruses, and a series of picornavirus-like plant viruses, membrane-dependent replication complexes have been described (reviewed in references 36 and 38). The detailed structures of these complexes as well as the exact biochemical contributions of membranes to virus replication are still largely unknown. Despite the functionality of a single vesicle in initiating and sustaining plus-strand RNA synthesis (20), PV-induced vesicles show a high tendency to accumulate into higher-order rosette-like structures (14), presumably thereby increasing the efficiency of plus-strand RNA synthesis. Even though minus-strand synthesis per se might not require membranes (29, 49), sequestration of plus-strand templates for minus-strand synthesis in the large minus-strand-containing bodies described in this paper might still be advantageous in order to avoid competition for plus strands by the translational machinery.

The observation (Fig. 1) (32) that the viral plus- and minus-strand RNA syntheses run in parallel at a constant ratio seems to indicate that the syntheses might be quantitatively interdependent, e.g., by mutually supplying the respective RNA templates. At present, the mechanism controlling this linkage is elusive. Generally, very little is known about regulation of PV replication in the sense of feedback mechanisms governing the quantitative and qualitative synthesis of viral molecules. Even the hallmark of picornavirus replication, i.e., the proteolytic cleavage of the polyprotein, implies that the entire set of gene products is present during the whole viral replication cycle and consequently that there are no early and late functions. Likewise, our findings that plus- and minus-strand RNAs are found in distinct compartments during the entire viral replication cycle could mean that PV replication depends primarily on mechanisms where a (most likely structural) constraint(s) regulates the number of molecules entering a given functional pathway.