Screening of the Yeast yTHC Collection Identifies Essential Host Factors Affecting Tombusvirus RNA Recombination

Abstract

RNA recombination is a major process in promoting rapid virus evolution in an infected host. A previous genome-wide screen with the yeast single-gene deletion library of 4,848 strains, representing ∼80% of all genes of yeast, led to the identification of 11 host genes affecting RNA recombination in Tomato bushy stunt virus (TBSV), a small model plant virus (E. Serviene, N. Shapka, C. P. Cheng, T. Panavas, B. Phuangrat, J. Baker, and P. D. Nagy, Proc. Natl. Acad. Sci. USA 102:10545-10550, 2005). To further test the role of host genes in viral RNA recombination, in this paper, we extended the screening to 800 essential yeast genes present in the yeast Tet-promoters Hughes Collection (yTHC). In total, we identified 16 new host genes that either increased or decreased the ratio of TBSV recombinants to the nonrecombined TBSV RNA. The identified essential yeast genes are involved in RNA transcription/metabolism, in protein metabolism/transport, or unknown cellular processes. Detailed analysis of the effect of the identified yeast genes revealed that they might affect RNA recombination by altering (i) the ratio of the two viral replication proteins, (ii) the stability of the viral RNA, and/or (iii) the replicability of the recombinant RNAs. Overall, this and previous works firmly establish that a set of essential and nonessential host genes could affect TBSV recombination and evolution.

RNA viruses are successful pathogens because they are capable of rapid evolution that helps them to overcome host resistance and other antiviral strategies (13, 14, 17, 27, 54, 55, 64). RNA recombination, the joining of two noncontiguous RNA segments together, is an especially powerful tool for viruses to create new resistance-breaking or drug-resistant strains and/or viruses (27, 64). Accordingly, the generation of novel recombinant RNAs (recRNAs) has been described for many human, animal, and plant viruses as well as RNA bacteriophages (1, 4, 5, 11, 16, 21, 23, 24, 27, 32, 42, 59, 64, 65).

Progress in our understanding of viral RNA recombination has been slowed down by the difficulty of detection of new recRNAs, the adverse selection pressure on some recRNAs, and the poor predictability of recombination events. Development of powerful model RNA recombination systems, however, has revealed many unique features of viral RNA recombination. For example, sequencing of numerous recRNAs in Brome mosaic virus (BMV) (3, 33, 36, 53), Turnip crinkle virus (TCV) (6, 7, 37, 38, 40), and tombusviruses (61, 62) established that recombination does not occur randomly within the viral RNA genome but rather, there are recombination “hot spots”. These include AU-rich sequences (31, 34, 58), inter- or intramolecular secondary structures (19, 35, 62), and cis-acting RNA elements with high affinity toward the viral replicase (8, 10, 40). Mutagenesis of the replicase proteins has led to altered recombination frequencies or altered the sites of recombination (15, 30, 47), suggesting that many recombination events are due to template switching (replicase jumping) by the viral replicase (22, 27, 39). In vitro template-switching experiments confirmed the abilities of purified viral RNA-dependent RNA polymerases or partially purified viral replicases to switch templates (8, 25, 52). Nonreplicative RNA recombination events have also been demonstrated for a small group of RNA viruses (11, 18).

In spite of our growing understanding of RNA recombination, the role of the host in the recombination process is not yet understood. However, recombination is known to be more frequent in some host species than in other host species (12, 64), indicating that host genes likely affect the RNA recombination process, similar to the significant influence of host genes on viral RNA replication (2, 26, 45). Identification of all the host genes affecting viral RNA recombination would be a major advance toward understanding the mechanism of RNA recombination and the selection pressure favoring or selecting against the new emerging recombinant viruses.

Tombusviruses, such as Tomato bushy stunt virus (TBSV) and Cucumber necrosis virus (CNV), are simple, single-component RNA viruses that code for five proteins. The two N-terminally overlapping proteins, termed p33 and p92, are essential for replication. The p92 is the viral RNA-dependent RNA polymerases, whereas p33 replication cofactor, an RNA-binding protein (46, 49, 50), is involved in template selection and recruitment of viral RNA into replication (29, 43, 49). These proteins interact with each other and the viral RNA in cells (43, 51), which leads to replication complexes formed on peroxisomal membranes (41, 43). Recent development of yeast as a model host for tombusvirus replication facilitated the identification of 96 host genes whose separate deletions affected the replication of a tombusvirus replicon RNA in yeast (45). These observations support the significant role of the host in virus replication (45).

One of the most intriguing features of tombusviruses is their ability to frequently participate in RNA recombination and generate defective interfering (DI) RNAs. DI RNAs are noncoding deletion derivatives of genomic RNA (gRNA) that have been used as model templates for replication and recombination studies (20, 63). In addition to the in vitro (8, 9) and in planta (47, 58, 60-62) recombination assays, a novel yeast-based RNA recombination assay has been developed recently (44). The tombusvirus recRNAs obtained in yeast were dimer-sized RNAs which were also present in plant cells, suggesting that yeast could be a suitable host for RNA recombination studies (44). Indeed, based on the available yeast single-gene knockout (YKO) library containing ∼4,800 strains (∼80% of all yeast genes in yeast), 11 nonessential host genes were identified which either increased or suppressed the accumulation of tombusvirus recRNAs (57). In this paper, we extended the screening for genes affecting tombusvirus recombination to the essential yeast genes. Among the 800 essential host genes present in the yeast Tet-promoters Hughes Collection (yTHC) (out of ∼1,100 predicted essential yeast genes) (28), we found that 16 genes affected the accumulation of tombusvirus recRNAs. These essential yeast genes, which either increased or decreased the accumulation of tombusvirus recRNAs, are involved in RNA transcription/metabolism, protein metabolism/transport, or unknown cellular processes. Detailed analysis of the effect of a selected group of yeast genes revealed that they could affect (i) the amount of viral replication proteins, (ii) the stability of the viral RNA, and/or (iii) the replicability of the new recRNAs. Overall, this and previous genetic screening of yeast led to the identification of 27 host genes (out of ∼5,600 genes that represent ∼ 95% of all predicted yeast genes) that affected the accumulation of TBSV recRNAs.

MATERIALS AND METHODS

Yeast strains and expression plasmids.

The titratable yTHC of yeast strains was obtained from Open Biosystems. yTHC was provided in the haploid strain R1158 background (URA3::CMV-tTA MATahis3-1 leu2-0 met15-0). This strain was created from strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) by a one-step integration of the tTA transactivator, under the control of the cytomegalovirus (CMV) promoter, at the URA3 locus. The kanR-tetO7-TATA cassette on a plasmid was then integrated into the genome, replacing the endogenous promoter for each gene (28).

The expression plasmid pGAD-His92 (containing CNV p92 gene and LEU2 marker) has been previously described (48). For construction of the pYC/DI-AU-FP plasmid, pDI-73-AU-FP (58) was digested with XbaI and BstXI, followed by gel purification of the 330-bp fragment (containing AU-FP sequence and portion of DI-72 RII) and ligation with two DI-72-derived fragments (the 170-bp HindIII-XbaI fragment, containing RI, and the 450-bp BstXI-SacI, including portion of RII, the entire RIII/IV, and the ribozyme sequences [44]). After ligation, the DI-AU-FP region was PCR-amplified with primers 542 (GCCCGAAGCTTGGAAATTCTCCAGGATTTC) and 1069 (CCGGTCGAGCTCTACCAGGTAATATACCACAACGTGTGT). The obtained PCR product was treated with HindIII and SacI and ligated into pYC/DI-72 (44, 48).

For construction of pGBK-His33/DI-AU-FP, we PCR amplified the GAL1-DI-AU-FP region from plasmid pYC/DI-AU-FP using primers 1546 (CCGCAATTCACGGATTAGAAGCCGCCGAGCGGGT) and 1069 (CCGGTCGAGCTCTACCAGGTAATATACCACAACGTGTGT). The PCR product was treated with EcoRI and SacI, followed by ligation into pHisGBK-His33-DI-72 (T. Panavas and P. D. Nagy, unpublished data) treated with EcoRI and SacI.

For construction of pGBK-His33/recΔ170RII/Δ70RII, we PCR amplified three separate regions of pYC/DI-AU-FP: first, GAL1 with primers 1546 and 1667 (GGACAAGCTTAATATTCCCTATAGTG), followed by treatment with EcoRI and HindIII; second, the Δ170RII/RIII/RIV part of DI-72 with primers 1668 (GGACAAGCTTGGAGAGTCTGCATATCACACCTG) and 1638, followed by treatment with HindIII and XmaI; and third, the Δ70RII/RIII/RIV part of DI-72 with primers 1640 (GGACCCGGGAAAGCGGTTTGTGAGAAG) and 1069, followed by treatment with XmaI and SacI. Then the GAL1 and Δ170RII/RIII/RIV PCR products were ligated together, followed by a new round of PCR amplification with primers 1546 and 1638 (GGACGGTACCCCGGGCTGCATTTCTGCAATG). The obtained 780-bp PCR product was treated with EcoRI and XmaI, followed by ligation with Δ70RII/RIII/RIV and cloning into pHisGBKHis33-DI-72 (T. Panavas and P. D. Nagy, unpublished data) treated with EcoRI and SacI.

Yeast transformation and cultivation.

The parental strain (BY4741) and the strains in the yTHC collection were cotransformed with different combinations of plasmids using the LiAc/ssDNA/PEG method (17a), and transformants were selected by complementation of auxotrophic markers.

For analysis of DI-AU-FP RNA accumulation and generation of new recRNAs, each transformed yTHC strain was inoculated into SC-LH⁻ medium containing 2% galactose and supplemented with Geneticin G418 (200 mg/liter) and cultured for 24 to 48 h at 29°C until an optical density at 600 nm of ∼0.8 to 1.0 was reached. For the maximum level of essential gene expression, yeast was grown in the absence of doxycycline, whereas to reduce the expression levels of the essential genes, yeast was grown in the same medium in the presence of 10 mg/liter doxycycline (28). Our preliminary experiments showed that the use of 10 mg/liter doxycycline was as good as 25 or 50 mg/liter doxycycline and better than 3.3 mg/liter to affect TBSV recombination (not shown). Therefore, we used 10 mg/liter doxycycline throughout the experiments to turn the particular gene off.

In the RNA stability experiments with DI-AU-FP RNA (the replicon RNA) or recΔ170RII/Δ70RII RNA (a representative recRNA), yeast strains (lacking plasmid pGAD-His92) were grown in SC-H⁻ medium containing 2% galactose for 24 h at 29°C. Cells were harvested by centrifugation at 1,100 × g for 5 min, followed by resuspension of cells in SC-H⁻ medium containing 2% glucose and by culturing at 29°C. Samples for total RNA extraction were collected every 30 min up to 4 h.

RNA analysis.

Total RNA isolation and Northern blot analysis were performed as described previously (44, 48). Briefly, for extraction of total RNA, yeast cells were broken by shaking for 1 to 2 min at room temperature with equal volumes of RNA extraction buffer (50 mM NaOAc [pH 5.2], 10 mM EDTA, 1% sodium dodecyl sulfate [SDS]) and water-saturated phenol and then incubated for 4 min at 65°C, followed by ethanol precipitation. The obtained RNA samples were separated on a 1.5% agarose gel and transferred to a Hybond-XL membrane (Amersham) before hybridization with a DI-72 RNA-specific probe. For detection of plus-strand replicon RNA and the recRNAs, we prepared ³²P-labeled RIII/IV(−) probe with T7 transcription from PCR product obtained with primers 1165 (AGCGAGTAAGACAGACTCTTCA) and 22 (GTAATACGACTCACTATAGGGCTGCATTTCTGCAATGTTCC) on DI-72 templates.

Protein analysis.

For protein analysis, yeast strains were cultivated as described above for RNA analysis. A total of 50 ml yeast culture was harvested, the pelleted cells were resuspended in 200 μl cold TG buffer (50 mM Tris-HCl [pH 7.5], 10% glycerol, 15 mM MgCl₂, 10 mM KCl), and 250 μl of glass beads was added to each sample. The cells were broken with a Genogrinder for 2 min at 1,500 rpm. Each sample was further mixed with 600 μl prechilled TG buffer, and unbroken cells were removed by centrifugation at 100 × g for 5 min. Enriched-membrane fractions were collected by centrifugation at 21,000 × g for 10 min, resuspended in SDS-polyacrylamide gel electrophoresis loading buffer, and incubated at 85°C for 15 min. The supernatant was used for SDS-polyacrylamide gel electrophoresis and Western blot analysis as described previously (44, 48). The primary antibody was anti-His₆ (Invitrogen), and the secondary antibody was alkaline phosphatase-conjugated anti-mouse immunoglobulin G antibody (Sigma).

5′ Rapid amplification of complementary ends (RACE) and RT-PCR analysis of the junction sites in the recRNAs.

We used both total yeast RNA extracts and gel-isolated recRNAs for reverse transcription (RT)-PCRs to specifically amplify regions covering the junction sites (57). First, the RT reaction included primer 14 (GTAATACGACTCACTATAGGGTTCTCTGCTTTTACGAAG) for cDNA synthesis, followed by PCR with primers 168 (TCGTCTTATTGGACGAATTCCTGTTTACGAAAG) and 270 (TTGGAAATTCTCCTTCAGTCTGAGTTTGTGGA). The PCR products were cloned into pGEM-T Easy vector (Promega) and sequenced using M13 reverse primer.

The 5′ and 3′ sequences of recRNAs were determined using 5′ RACE and 3′ RACE (57). The resulting products were cloned and sequenced (8).

RESULTS AND DISCUSSION

Rationale.

To test the effect of essential host genes on tombusvirus recombination, we took advantage of (i) the development of recombination assays for tombusviruses in yeast (Fig. 1A) (44, 57) and (ii) the availability of the yTHC collection for 800 essential yeast genes (28). In the yTHC collection, the expression of a given essential yeast gene is under the Tet-titratable promoter in the genome (Open Biosystems). The expression of the essential gene can be switched off by the addition of doxycycline to the yeast growth medium (28). Therefore, tombusvirus recombination was examined when the expression of the particular host gene was either on (no doxycycline) (Fig. 1B) or off (plus doxycycline) (Fig. 1C) (28). Then the ratio (Fig. 1B) between tombusvirus recRNAs versus the original (nonrecombinant) replicon RNA was calculated (based on their accumulation levels) (Fig. 1D) to identify those host proteins that affected the recombination for each strain present in the yTHC collection.

FIG. 1.

Replication and recombination of TBSV replicon RNA in yeast from the yTHC collection. (A) Schematic representation of the DI-AU-FP replicon RNA with four noncontiguous regions (RI to RIV) derived from TBSV genomic RNA and the artificial AU-FP region to promote recombination (58). (B and C) The scheme of launching DI-AU-FP replication and regulation of host gene expression from the Tet-titratable promoter. P33 and p92 replication proteins are expressed constitutively from the ADH1 promoter, whereas DI-AU-FP is expressed from the regulatable GAL1 promoter. Replication (and probably recombination) of DI-AU-FP takes place in the cytoplasm (on peroxisomal membrane surfaces). The expression of a particular host gene occurs in the absence of doxycycline, and it is switched off in the presence of doxycycline. (D) (Top). Ethidium bromide-stained agarose gel of total RNA extracts obtained from the parental yeast strains (grown without doxycycline [−DOX] or with doxycycline [+DOX]) showing the accumulation of DI-AU-FP replicon RNA and the recombinant RNA. The yeast cells coexpressed DI-AU-FP RNA, and p33/p92 replication proteins. (Bottom). Northern blot analysis of total RNA extracts with a ³²P-labeled RNA probe specific for RIII/IV of DI-AU-FP. The samples are the same as in the top panel. The unmarked faint bands represent degradation products (5′ truncated RNAs that migrate fast) and additional recombinants (slow migrating).

Efficient generation of tombusvirus recRNAs in yeast using DI-AU-FP replicon RNA.

To identify essential host genes affecting tombusvirus recombination, we transformed each of the yeast strains present in the yTHC collection (Open Biosystems) with two plasmids (pGAD-His92 and pGBK-His33/DI-AU-FP) (Fig. 1B and C). These plasmids expressed the p33 and p92 replication proteins of CNV from the constitutive ADH1 promoter and the highly recombinogenic DI-AU-FP RNA (Fig. 1A) (57, 58) from the galactose- or glucose-inducible or -repressible GAL1 promoter (44). DI-AU-FP RNA contains a 186-nucleotide insertion with an extended AU-rich stretch between region I (RI) and RII, which promotes DI RNA recombination in yeast and in plant cells (57, 58). Expression of p33, p92, and the DI-AU-FP replicon RNA in the parental yeast strain under standard growth conditions (see Materials and Methods) led to efficient replication of the 807-nucleotide-long replicon RNA (∼20,000 copies per yeast cells). In addition, new recRNAs (∼15% of the level of the original replicon RNA) (Fig. 1D) also accumulated after 24 h of culturing. The addition of 10 mg/liter doxycycline to the growth medium did not alter the accumulation of the original replicon RNA, the formation of recRNAs, and the ratio between recRNAs and the replicon RNA (Fig. 1D) in the case of the parental yeast strain (which carries all yeast genes and with expression of genes from their natural promoters), suggesting that the presence of doxycycline did not affect the replication or recombination of the DI-AU-FP replicon RNA in the parental yeast strain.

To characterize the putative recRNAs that accumulated in the parental yeast strain replicating DI-AU-FP, we performed 5′ RACE, 3′ RACE, and RT-PCR to cover the junction sites in the recRNAs (Fig. 2). Sequence analysis of the recRNAs showed that they were derived by recombination possibly between two 5′ truncated replicon RNAs, as shown in Fig. 2 (C.-P. Cheng, E. Serviene, and P. D. Nagy, submitted for publication). Although the sequences present at the 5′ end and at the junction sites in the recRNAs were variable, we placed the recRNAs into two groups, long and short (Fig. 2), based on their sizes. The short recRNAs with larger 5′ deletions were more common (visible on ethidium bromide-stained agarose gels or Northern blots) (Fig. 1D). Similar to other recRNAs characterized in the in vitro replicase assays (8), in plants or plant protoplasts (61-63), and in yeast (44, 57), both the short and long recRNAs contained one to eight extra, nonviral sequences at the junction sites (Fig. 2). The presence of extra nucleotides is likely due to replicase errors during the template-switching events (8). Overall, Fig. 1 and 2 established that DI-AU-FP replicon RNA could efficiently generate recRNAs at detectable levels in the parental yeast strain, making it a suitable construct for high-throughput recombination studies (see below).

FIG. 2.

5′ Sequences and junction sequences in recRNAs generated in the parental yeast coexpressing DI-AU-FP RNA and p33/p92. (A and B) (Top) Schematic presentation of the two types of recombinants containing partially duplicated sequences from the 3′ half of DI-AU-FP replicon RNA joined in a head-to-tail fashion. The actual 5′ ends (left panel) and junction sequences are shown for a number of recRNAs. Note that most of the 3′ sequences (3′ part of RII, RIII, and RIV) and 5′ deletions (RI and 5′ part of RII) are not shown. ▵ indicates the number of deleted nucleotides, whereas virus-templated and nonviral sequences are shown in uppercase and lowercase letters, respectively. The 3′ end in RIV (at both the internal and 3′-terminal locations) contained the authentic viral sequence.

Identification of 16 essential host genes affecting tombusvirus recombination.

Based on the above findings, we performed high-throughput screening of the 800 strains present in the yTHC collection. Individual transformed yeast strains carrying the expression plasmids for p33, p92, and DI-AU-FP RNA were grown with or without added doxycycline (Fig. 1B and C, respectively), as described in the Materials and Methods. Total RNA was extracted from samples (at least six samples per yeast strain, three each from yeast grown with or without doxycycline), electrophoresed in 1.5% agarose gel, stained with ethidium bromide, transferred to a nylon membrane, and analyzed by Northern blotting specific for the 3′ end of DI-AU-FP (Fig. 3A). We found that six of the yTHC strains were not transformable with the two plasmids and 41 strains grew too slowly to obtain enough viral and host RNAs for subsequent analysis. Out of the remaining yeast strains, we found that 16 strains showed altered ratio in recRNA/replicon accumulation (we scored only those strains that showed an at least twofold alteration from the recRNA/replicon ratio seen in parental yeast) (Fig. 3A and B). We chose the recRNA/replicon ratio as the critical parameter, because this might reflect the specific effect of the host factor on recombination and/or accumulation of the recRNAs. In contrast, absolute recRNA levels could greatly depend on the level of accumulation of the replicon RNA, which was variable in these yeast strains (Fig. 3C). For example, higher replicon levels in some yTHC strains can likely increase absolute recRNA levels via providing more abundant recombination substrates. However, these host factors are primarily replication factors and they will be presented in a separate paper.

FIG. 3.

Sixteen yTHC yeast strains show an altered recRNA/replicon ratio compared to the parental strain. (A) Northern blot analysis of total RNA extracts from the shown yeast strains (four independent samples are shown for each strain [two samples were grown without and two with doxycycline {DOX}] to illustrate the reproducibility of recombinant accumulation) was performed with a radiolabeled RNA complementary to RIII/IV. Arrow points at the DI-AU-FP replicon, whereas the novel recRNAs (short and long [Fig. 2]) are depicted with arrowheads. Asterisks mark a 5′-truncated replicon RNA that might serve as a recombination substrate (C.-P. Cheng, E. Serviene, and P. D. Nagy, unpublished). rRNA is shown as a loading control. Host genes in open and closed circles showed increased or reduced accumulation of recRNAs compared to the replicon RNA in the presence of doxycycline. (B) The ratio of recRNA versus replicon RNA is shown in selected yTHC strains graphically. (C) The accumulation level of replicon RNA is shown in selected yTHC strains graphically. The accumulation level of DI-AU-FP in the parental strain was scored as 100%.

Among the 16 identified yTHC strains showing altered recRNA/replicon ratio (Fig. 3A and B), three yTHC strains (i.e., POL1, COP1, and RGR1) showed a 2- to-12-fold-increased ratio in recRNA accumulation when grown with doxycycline versus without doxycycline (Fig. 3A and B). On the contrary, six strains (i.e., RPT4, RPB11, MPS1, RRP9, ARP9, and SEN1) supported a two- to sixfold-reduced recRNA/replicon ratio when grown with doxycycline (Fig. 3A and B). Seven additional strains (NOP10, RPM2, YDR327W, ORC6, FAS2, YKL033W, and RIB7) supported a two- to threefold higher ratio of recRNA accumulation than the parental yeast, but in a doxycycline-independent manner (Fig. 3A and B).

The observation that seven of the identified host genes affected tombusvirus recombination in a doxycycline-independent manner might be explained by the use of a Tet-titratable promoter for their expression. We propose that these host genes might be expressed at lower levels from the Tet-titratable promoter than from their natural promoters. Thus, these proteins might be present at a reduced level even in the absence of doxycycline compared to the parental strain.

The 16 essential host genes identified in the screen mentioned above code for proteins with different molecular functions in various cellular processes (Yeast Genome Database, SGD [http://www.yeastgenome.org]). These include RNA binding/processing (RRP9, NOP10), RNA helicase/unwinding (SEN1), RNase (RPM2), or RNA polymerase/RNA transcription (RPB11, RGR1, ARP9) (Table 1). Others are involved in protein modification (MPS1), protein catabolism (RPT4), or protein transport (COP1). FAS2 codes for a protein reductase/synthase, which affects fatty acid biosynthesis, whereas RIB7-coded protein has deaminase activity. POL1 has DNA polymerase function, whereas ORC6 is involved in DNA replication. We also identified two genes with currently unknown functions (YDR327W and YKL033W) (Table 1). Altogether, the identified host factors could have either direct or indirect effects on tombusvirus recombination (see below).

TABLE 1.

Names and functions of the identified host genes

Gene/open reading frame name	Molecular/biological function(s)a
POL1	Alpha DNA polymerase, synthesis of RNA primer
COP1	Protein transporter, alpha subunit of COP1 vesicle coatomer complex, endoplasmic reticulum to Golgi body and retrograde transport
RGR1	Transcription mediator, component of RNA polymerase II holoenzyme
MPS1	Protein threonine/tyrosine kinase
RPT4	Endopeptidase, ATPase involved in the degradation of ubiquitinated substrates
ARP9	RNA polymerase, actin-related protein involved in transcriptional regulation
SEN1	RNA helicase required for processing of tRNAs, rRNAs, and snRNAs
RRP9	RNA binding, involved in pre-rRNA processing, associated with U3 snRNP
RPB11	RNA polymerase II subunit B12.5; part of central core
FAS2	Alpha subunit of fatty acid synthetase, which catalyzes the synthesis of long-chain saturated fatty acids
YDR327W	Unknown
NOP10	RNA binding, constituent of small nucleolar ribonucleoprotein particles containing H/ACA-type snoRNAs, which are required for pseudouridylation and processing of pre-18S rRNA
RPM2	Ribonuclease P activity, protein biosynthesis
RIB7	Deaminase, riboflavin biosynthesis
ORC6	DNA replication
YKL033W	Cytoplasmic protein with unknown function

^aBased on references annotated in the SGD genome database (http://www.yeastgenome.org/).

To exclude the possibility that some of the above host genes affected viral RNA recombination via contributing to the artifactual generation of recombinants during DNA plasmid replication and/or plasmid-based RNA transcription, we performed Northern blot analysis on yeast total RNA isolated from the 16 identified yTHC strains and the parental strain expressing only p33 and DI-72 RNA, but not p92. This experiment revealed the absence of viral recRNAs in each sample lacking viral replication (Fig. 4). Therefore, we conclude that viral recRNAs were not formed during DNA replication and/or RNA transcription, but instead, the recRNAs were generated during viral RNA replication. The presence of extra nonviral sequences at the 5′ ends and junctions of the recRNAs is also compatible with the replicase-driven recombinational model (8). This is because the purified tombusvirus replicase has been shown to add nonviral sequences at the recombinational junctions in vitro (8) and in vivo (58). Overall, these and previous findings support the replicase-driven template-switching model in tombusvirus recombination.

FIG. 4.

Northern blot analysis of DI-AU-FP RNA transcripts from total RNA extracts obtained from selected yTHC yeast strains coexpressing p33 but lacking p92. Note the lack of recombinant RNAs and the presence of the original (containing plasmid-borne 5′ and 3′ sequences) and ribozyme-cleaved DI-AU-FP RNA transcripts in all samples in the absence of TBSV replication. The ribozyme-cleaved DI-AU-FP RNA transcripts are more active in replication/recombination than the uncleaved transcripts (data not shown). DOX, doxycycline.

The effect of the identified host factors on transcription and on the amount of replication proteins.

Template switching-driven RNA recombination might depend on the amount (concentration) of viral RNA templates and the amount of replicase complexes available in the cells (27, 39). Therefore, we tested the amount of viral RNA transcripts and then the level of replication proteins present in the above identified 16 yTHC strains in comparison with the parental yeast.

Comparison of the accumulation of DI-AU-FP transcripts generated from the expression plasmids in the selected yTHC strains revealed that the amount of transcripts decreased in nine strains, increased in five strains (MPS1, ARP9, SEN1, NOP10, and ORC6), and did not change in two strains (RRP9 and YKL033W) plus the parental strain in the presence of doxycycline. Altogether, we did not find a good overall correlation between altered DI-AU-FP transcript levels and the alteration in the recRNA/replicon ratio. For example, the POL1, RPB11, and RPM2 strains showed decreased levels in DI-AU-FP transcript accumulation, yet they increased (POL1), decreased (RPB11), and left unchanged (RPM2), respectively, the recRNA/replicon ratio in the absence or presence of doxycycline (Fig. 4). However, we did observe a reverse correlation between low DI-AU-FP transcript levels and high recRNA accumulation in strains POL1, COP1, and RGR1 in the presence of doxycycline (Fig. 5). Therefore, it is possible that the initial amount of replicon RNA could affect recRNA generation/accumulation in some strains.

FIG. 5.

Western analysis of p33 and p92 replicase proteins in total protein samples using anti-His tag antibody. See further details in the legend to Fig. 3.

To test for the possible effect of the identified host genes on p33/p92 levels, we performed Western blotting on protein extracts obtained from membrane-enriched fractions (48) (Fig. 5). These experiments revealed that seven strains (COP1, RGR1, RPB11, NOP10, RPM2, RIB7, and ORC6) accumulated p92 at reduced levels, three (RPT4, ARP9, and YDR327W) at higher levels, and the remaining six yTHC strains and the parental strains at similar levels in the presence versus in the absence of doxycycline (Fig. 5). The accumulation levels were also variable in the case of the p33 protein. We found that (i) three strains (POL1, RPB11, and RIB7) accumulated p33 at reduced levels, (ii) two strains (RGR1 and ARP9) accumulated p33 at higher levels, and (iii) the remaining 11 yTHC strains and the parental strains accumulated p33 at similar levels in the absence or presence of doxycycline (Fig. 5). The changes in p33 and/or p92 levels, however, did not show good correlation with an altered recRNA/replicon ratio. For example, the POL1, RIB7, and RPB11 strains showed decreased levels of p33 accumulation, yet they increased (POL1), left unchanged (RIB7), or decreased (RPB11) the recRNA/replicon ratio in the absence versus in the presence of doxycycline (Fig. 3A and B). Therefore, these data do not support the idea that many of the host genes identified in this work would affect the ratio of recRNA versus replicon RNA by affecting p33 and/or p92 levels in host cells. However, we did observe that the ratio of p33/p92 changed drastically in POL1, COP1, and RGR1 strains in the presence of doxycycline (Fig. 3 and 4). Therefore, it is possible that the p33/p92 ratio (together with the initial replicon RNA levels) could affect recRNA generation/accumulation in some strains.

Role of the identified host factors in postrecombinational accumulation of the recRNAs.

The accumulation of viral recRNAs in the host cells depends on two major processes: the primary recombination events, followed by the secondary postrecombinational amplification/selection (27, 31). Host factors might be able to affect either of these processes. Unfortunately, it is difficult to measure the primary recombination events (i.e., the frequency of recombination) in our yeast recombination assay due to efficient postrecombinational amplification of the recRNAs. However, the role of postrecombinational selection can be tested because it likely depends on (i) the in vivo stability of the recRNA, (ii) the ability of the recRNA to replicate, and (iii) the ability of the recRNA to compete with the abundant replicon RNA in the cells. Altogether, each of the above secondary selection processes could influence the ratio of recRNA and replicon RNA in the identified host strains. Therefore, we performed tests to determine the effect of postrecombinational selection on the accumulation of recRNAs in nine yTHC strains and the parental strains.

Stability of a representative cloned recRNA (named recΔ170RII/Δ70RII) (Fig. 6A) versus the replicon RNA (i.e., DI-AU-FP) was analyzed by determining the half-life of these RNAs in the absence of replication. To do so, either the recRNA or the replicon RNA was expressed (in the presence of p33 but in the absence of p92 to prevent replication) for 24 h by growing the selected yTHC strains and parental strains in an inducing medium (containing galactose), followed by repression of RNA transcription (via growing yeast in glucose-containing medium) and time-course RNA analysis. For these experiments, we selected nine yeast strains, which showed the most pronounced effects on RNA recombination (Fig. 3). Altogether, these experiments established that the recRNA had a ∼6-fold-increased stability compared to the replicon RNA in the parental yeast (Table 2). The stability of these RNAs was not affected by doxycycline in the parental strain, suggesting that doxycycline does not indirectly affect viral RNA stability. We also found a ∼2-fold increase in recRNA stability in a doxycycline-independent manner for the FAS2 and NOP10 strains. Similarly, increased stability of recRNA over the replicon RNA was observed in the case of COP1, RGR1, and RPB11, but in a doxycycline-dependent manner (Table 2). On the contrary, the stability of recRNA versus replicon RNA decreased in the MPS1 and RRP9 strains (Table 2). Altogether, these data suggest that the stability of recRNA is a factor that could affect the accumulation of the recRNA in selected yTHC cells. However, stability of the recRNA versus the replicon RNA cannot explain the altered recRNA/replicon ratio for most of the identified yTHC strains. For example, the POL1 strain showed high RNA stability for both replicon RNA and the recRNA in a doxycycline-independent manner (Table 2) and yet recRNA accumulated efficiently only in the presence of doxycycline in the POL1 strain (Fig. 3A and B). Also, recRNA showed a ∼6-fold increase in stability in parental yeast and yet the replicon RNA accumulated to an amount ∼6-fold higher than the more stable recRNA (Fig. 3A and B). Therefore, altered stability of the recRNA in the host strains identified above is unlikely to be the only factor affecting recombinant accumulation compared with the original replicon RNA.

FIG. 6.

Accumulation of a representative recRNA in selected yTHC strains. (A) Schematic presentation of a representative “short” recRNA, termed recΔ170RII/Δ70RII RNA, that was coexpressed with p33/p92 replication proteins in yeast. (B) Northern blot analysis of total RNA extracts from the shown yeast strains was performed as described for Fig. 3. Asterisk marks the accumulation of 5′-truncated viral RNA product (Fig. 3). (C) The accumulation level of recΔ170RII/Δ70RII RNA is shown in selected yTHC strains graphically, and the level in the parental strain was scored as 100%.

TABLE 2.

Half-life^a of the replicon and the recRNA in yTHC yeast

Gene/strain	DI-AU-FP		recRNA
	−DOX	+DOX	−DOX	+DOX
Parental	25	20	140	150
POL1	210	180	210	150
COP1	50	70	120	80
RGR1	30	510	140	330
MPS1	240	150	50	45
RPT4	35	50	60	45
RRP9	240	240	130	130
RPB11	60	120	240	120
FAS2	50	55	150	120
NOP10	50	50	90	90

^aHalf-life of the RNAs is shown in minutes.

The role of post-recombinational amplification of recRNA in altering the ratio of recRNA versus replicon RNA.

The recRNAs shown in Fig. 2 contain essential cis elements for replication, and they likely replicate in yeast. To test the role of postrecombinational accumulation of the recRNA in yeast, we coexpressed recΔ170RII/Δ70RII recRNA (Fig. 6A) with p33 and p92 in yTHC cells in the presence or absence of doxycycline. Comparison of accumulation levels for recRNA in parental strains and the selected yTHC strains revealed that recRNA accumulated at different levels (more than twofold differences) in all nine yTHC strains tested compared to the parental yeast strain (Fig. 6B and C). For six strains (POL1, COP1, RGR1, RPT4, RRP9, and FAS2), the recRNA replicated two- to sixfold better in the presence of doxycycline, whereas for RPB11 and NOP10, the recRNA replicated less in the presence of doxycycline (Fig. 6B and C). Interestingly, recRNA replication was ninefold higher in MPS1 than in the parental strain, but independent of doxycycline (Fig. 6B and C).

Overall, the replication data obtained with recRNA (Fig. 6B and C) versus replicon RNA (Fig. 3C) suggest that POL1, COP1, and RGR1 (in a doxycycline-dependent way) and NOP10 (in a doxycycline-independent way) might show increased recRNA levels (Fig. 3B) because these strains accumulated the recRNAs more efficiently than the replicon RNA. Therefore, it is possible that these host proteins could affect tombusvirus recombination indirectly by facilitating the accumulation of the recRNA better than the replicon RNA.

In contrast, in spite of the efficient replication of recRNA in RPT4 and RRP9 (Fig. 6B and C), these strains supported the low recRNA/replicon ratio compared to the parental strain (Fig. 3A and B). Therefore, RPT4 and RRP9 host genes could be directly involved in tombusvirus recombination, not by affecting recRNA versus replicon RNA accumulation.

Another interesting gene is MPS1, which increased recRNA replication ninefold (Fig. 6B and C) in a doxycycline-independent manner, and yet the MPS1 strain supported a low recRNA/replicon ratio in the presence of doxycycline compared to the parental strain (Fig. 3A and B). In the case of RPB11, both the recRNA and the replicon RNA accumulated threefold less efficiently in the presence of doxycycline than in its absence (Fig. 6B), whereas the recRNA/replicon ratio decreased fourfold in the presence of doxycycline (Fig. 3B). Also, the replication data on recRNA and the replicon RNA do not correlate with the recRNA/replicon ratio for FAS2 in the absence of doxycycline, whereas the correlation exists in the presence of doxycycline. Thus, it is possible that MPS1, RPB11, and FAS2 could also be directly involved in RNA recombination.

Altogether, these replication data suggest that POL1, COP1, RGR1, and NOP10 could enhance the ability of the recRNA to replicate efficiently in these yeast strains. Thus, these host factors might have only an indirect effect on the recombination process. On the other hand, the replication data cannot explain the observed changes in the recRNA/replicon ratio in RRP9, MPS1, RPT4, and RPB11 and can only partially explain changes for the FAS2 strain, suggesting that these genes could affect tombusvirus recombination directly.

A small set of host factors affects tombusvirus recombination.

The possible effect of host genes on viral RNA recombination has been studied with Tombusviruses (57). The previous genetic screen involving ∼4,800 nonessential single gene deletions in the YKO library (57), and the current genetic screen with 800 essential yeast genes in the yTHC library identified 11 and 16 host genes, respectively, that affected viral recRNA accumulation. Altogether, the two genetic screens combined cover close to 95% of all the estimated number of genes (∼5,800) present in the yeast genome. Therefore, we estimate that less than 0.5% of host genes affect viral RNA recombination. This number is probably an underestimate, because genetic screens frequently overlook redundant genes (such as gene families) with similar functions whose deletion/regulation could be compensated by other genes. Nevertheless, the identified host genes represent the first set of host genes shown to affect viral RNA recombination and thus, they should be valuable for studies on the mechanism of RNA recombination and the role of the host in virus evolution.

Conclusions.

The identified host genes in the previous (57) and current (Table 1) works can be placed into several groups: (i) RNA-binding proteins, ribonucleases, and a helicase; (ii) intracellular transport proteins; (iii) kinases and phosphatases; (iv) protease, protein reductase, and endopeptidase; (v) transcription factors; (vi) DNA replication factors; and (vii) genes with unknown functions. Some of these proteins might have a direct effect, while other proteins could have an indirect effect, on viral RNA recombination. Further detailed studies will establish the functional roles of many of the identified proteins in tombusvirus recombination.