Yeast protoarray-based identification of host proteins that bind to the tombusvirus p33 replication protein.
The viral replication proteins likely interact with many host proteins to facilitate viral replication. Also, some host proteins could interact with and inhibit the function of the viral replication proteins as an antiviral mechanism. To determine host proteins interacting with the viral replication proteins on a proteome-wide scale, we have utilized a protoarray-based approach with purified host proteins. Since TBSV repRNA replicates efficiently in yeast (34
), we took advantage of the yeast proteome microarray carrying 4,088 purified yeast proteins (~70% coverage of all the yeast genes). To probe the yeast proteome microarray for the identification of host proteins binding to the tombusvirus p33 replication cofactor protein, we purified recombinant p33 fused to MBP from E. coli
by use of affinity chromatography as described previously (44
). After purification, the N-terminal MBP fusion was cleaved off from p33 by use of factor Xa protease. Importantly, p33 also carried a C-terminal v5 tag to facilitate its detection. The highly purified p33 preparation was applied (in the presence of an excess amount of bovine serum albumin as a nonspecific competitor) onto the yeast protoarray to promote protein-protein interactions, followed by thorough washing. The bound p33 was detected via v5-specific antibody conjugated with Alexa Fluor dye and visualized with a microchip scanner (Fig. ). It is worth noting that the v5 tag in p33 could interfere with binding to host proteins and/or the binding of a host protein to this region of p33 could block the detection of the v5 tag on the array.
We found that 58 host proteins bound to the recombinant p33 efficiently and reliably under the in vitro conditions employed (see Materials and Methods). Among these host proteins, there are 3 protein chaperones (Gim3p, Jjj1p, and Jjj3p), 5 proteins involved in protein ubiquitination (Cdc34p, Rsp5p, Uba1p, Ubp10p, and Ubp15p), 6 translation factors involved in mRNA translation (Bfr1p, Efb1p, Hbs1p, Rpl8Ap, Tif1p, and Tif11p), and 10 proteins involved in RNA processing and metabolism (Ala1p, Bud21p, Erb1, Rib2p, Sas10p, Stm1p, Trm1p, Trz1p, Tsr2p, and Urn1p) (Table ). Additional proteins are known to be involved in various cellular processes and the functions of nine proteins are not yet defined (Table ).
To obtain further insight into p33-host protein interactions, we also performed similar binding experiments with purified recombinant p33N82 (v5 tagged), which contains only the N-terminal 82 aa of p33, as well as with p33C (Fig. ) with defined RNA-binding and protein interaction domains (43
) by use of the yeast protoarray. However, for p33C, we used biotin labeling followed by detection via streptavidin conjugated with Alexa Fluor dye and visualization with a microchip scanner. Also, because p33C is more soluble than full-length p33, which carries long hydrophobic stretches (Fig. ), we could use p33C at a concentration ~10 times higher than that used for p33 to increase the sensitivity in the protoarray experiments (see Materials and Methods). Interestingly, 29 host proteins (50% of the 58 bound to the full-length p33 on the chip) bound to p33C, and 12 (21%) bound to p33N82, whereas 8 (14%) bound to both p33C and p33N82 (Table ). The remaining nine (15%) host proteins did not bind to p33N82 and p33C, suggesting that this group of proteins might bind to the hydrophobic central domain in p33, which is missing from both p33N82 and p33C (Fig. ).
Yeast protoarray-based identification of host proteins that bind to the unique C-terminal fragment of replication protein p92pol.
To identify host proteins binding specifically to the unique, nonoverlapping portion of p92pol, we also performed protoarray-based binding experiments with purified recombinant p92C (v5 tagged), which contains the RNA-dependent RNA polymerase motifs in the unique C-terminal segment of p92pol (Fig. ). We found that 21 host proteins bound both to p33 and to p92C (Table ), whereas 11 host proteins bound only to p92C (Table ) and not to p33. The list includes an RNA helicase (Dpb3p), a methylase (Dot1p), an aminopeptidase (Map1p), an RNA-binding protein (Npl3p), and a translation factor (Tef2p) (Table ). Thus, the protoarray approach allowed the identification of host proteins that selectively bound to p33 or the nonoverlapping portion of p92pol in vitro.
The names and functions of yeast proteins bound to TBSV p92C
Interaction of recombinant p33 with selected yeast proteins in vitro.
Since the protoarray approach might be prone to identify false positives (i.e., proteins that interact with the viral p33 only on the chip under the conditions applied), we used additional approaches to confirm that the recombinant p33 could bind to the identified host proteins as found in the above-described protoarray experiments. To this end, we performed protein pulldown experiments with purified recombinant p33C and 16 host proteins chosen based on their intriguing functions in yeast. For the protein pulldown experiments, yeast lysates containing soluble proteins prepared from yeast overexpressing one of the 16 GST/His6-tagged host proteins were applied to columns containing immobilized MBP-p33C fusion protein. After elution of the bound proteins from the column, we analyzed whether the particular host protein was present in the eluted fraction by using Western blotting (Fig. ). Similarly prepared recombinant MBP bound to beads was the negative control to exclude nonspecific binders. These experiments are thus complementary to the protoarray experiments, which had the yeast proteins fixed to a solid surface, while the p33C was the probe protein present in solution.
FIG. 2. Binding of selected host proteins to p33 replication protein in vitro. MBP-tagged p33C and MBP (1 μg each) were separately immobilized on amylose beads, followed by incubation with cytosolic extracts prepared from yeast individually expressing (more ...)
As expected, GST control bound to neither MBP nor p33C (Fig. , lanes 15 and 16). In contrast, 14 host proteins, including Arp8p, Iwr1p, Tif1p, Tif11p, Gsy2p (not shown), Cdc34p, Hbs1p, Rpl8Ap, Nap1p, Spt16p, Jjj1p, Sas10p, Elf1p, and Stm1p bound to p33C-MBP much more efficiently than to MBP (Fig. ), confirming that these host proteins can interact with the C-terminal domain of p33 in vitro. The binding of Trm1p to p33C-MBP was only about twice as efficient as that to MBP (Fig. , lanes 33 and 34). Smk1p mid-sporulation-specific mitogen-activated protein kinase, which bound weakly to p33 on the protoarray (not shown), did not interact with p33C in vitro (Fig. , lanes 27 and 28). Overall, data on 16 yeast proteins from the protein pulldown experiments are mostly in agreement with the protoarray experiments, demonstrating efficient interactions between p33 and the selected yeast proteins under in vitro conditions. The major difference in the results from the protoarray and protein pulldown experiments is that Tif1p did not bind to p33C (only to the full-length p33) in the protoarray (Table ), while it did bind efficiently to p33C based on protein pulldown experiments (Fig. , lanes 5 and 6).
In vivo interaction between identified host proteins and the tombusvirus p33 based on the split-Ub two-hybrid assay.
To confirm that the identified p33-host protein interactions can also take place on subcellular membrane surfaces within yeast cells, where p33 is localized (22
), we used the split-ubiquitin yeast two-hybrid assay. This assay is based on the ability of the N-terminal (NubG) and C-terminal (Cub) halves of Ub to reconstitute a functional protein (9
). When NubG and Cub, both fused separately to interacting proteins, are brought into close proximity and reconstitute a functional Ub protein, cleavage by endogenous Ub-specific proteases (UBPs) leads to the release of an artificial transcription factor, LexA-VP16, fused to Cub. This allows the activation of LexA-driven HIS3
expression in the nucleus. In summary, the split-Ub system, unlike the original yeast two-hybrid system, does not require interacting proteins to be localized to the nucleus, allowing an analysis of protein interactions on the cytosolic surfaces of membranes, which are the natural subcellular locations of the membrane-bound p33 protein (22
The split-Ub assay with 19 of the identified host proteins revealed that Rpl8Ap, Arp8p, Ubp15p, Elf1p, Nap1p, Tif11p, Ubp10p, Gsy2p, Tif1p, and Stm1p interacted with the membrane-bound p33 when used as N-terminal fusion with NubG (Fig. ). Additional host proteins, such as Sas10p, Hbs1p, and Iwr1p, interacted with the membrane-bound p33 when used as C-terminal fusion with NubG (Fig. ). The interaction of Efb1p, Trm1p, Spt16p, and Jjj1p with p33 was detectable, but weak, in this assay. Only Uba1p, which is a Ub-activating E1 protein, and Cdc34p (not shown), which is an E2 Ub-conjugating enzyme, have not been found to interact with p33 in the split-Ub assay (Fig. ). Since Uba1p and Cdc34p are involved in ubiquitination, they might interfere with Ub cleavage in this assay (putative false negatives). Altogether, the split-ubiquitination assay confirmed that 18 of the identified host proteins interact with membrane-bound p33 in yeast.
FIG. 3. Confirmation of host protein-p33 interactions via the split-Ub two-hybrid assay. (A) The full-length sequences of 19 host proteins were fused to NubG as N-terminal (N-term) fusions. Heat shock protein 70 (SSA1) was used as a positive control because it (more ...) Effect of overexpression of selected host proteins on TBSV repRNA replication in yeast.
To test if the host proteins that interacted with p33 could affect tombusvirus RNA replication, we took advantage of the previously developed efficient tombusvirus replication system in yeast (34
) with some modifications. First, we separately overexpressed 44 of the identified host proteins from the galactose-inducible GAL1
) in yeast cells for 20 h. Subsequently, we launched TBSV repRNA replication from the CUP1
promoter in the same cells. Comparable amounts of yeast cells were harvested 24 h later, followed by Northern blotting to measure the level of TBSV repRNA produced. We used rRNA as a loading control for the normalization of data on repRNA accumulation in yeast. The accumulation level of repRNA in yeast carrying pYES plasmid, which expresses only a short peptide, was taken as 100% (Fig. ). As an additional control, we overexpressed a pseudogene (APT2
) which has no enzymatic activity when expressed (2
) and failed to interact with p33 based on the protoarray experiments (not shown). Overexpression of Apt2p led to 79% ± 9% repRNA accumulation compared with that seen for yeast carrying the pYES control (Fig. ). This suggests that protein overexpression in general might reduce the ability of yeast cells to support repRNA accumulation under the protein overexpression condition. Based on the pYES and APT2
controls, we considered the overexpression of a protein to be inhibitory if it significantly reduced repRNA accumulation below 70% and stimulatory if it increased repRNA accumulation significantly, i.e., above 130% (Fig. ).
FIG. 4. Effect of overexpression of selected host proteins interacting with p33 replication proteins on TBSV repRNA accumulation in yeast. A total of 44 zz domain-tagged yeast proteins (Open Biosystems) were expressed separately from the galactose-inducible (more ...)
These protein overexpression experiments revealed that two of the host proteins tested affected repRNA accumulation dramatically (Fig. ). These were Hbs1p translation factor and Cdc34p E2 Ub-conjugating enzyme, which increased repRNA levels by ~3.5- and ~2.2-fold, respectively. An additional five host proteins, namely, the Jjj1p cochaperone, the Pol30p proliferating cell nuclear antigen transcription factor, the Arp8p actin-related protein, the Erb1p rRNA maturation protein, and the Trz1p tRNA-processing protein, increased repRNA accumulation by 40 to 60% compared to what was seen for yeast carrying the pYES expression vector (Fig. ). In contrast, overexpression of Ddr48p, a DNA damage-responsive protein, led to a 63% ± 5% reduction in repRNA accumulation in yeast (Fig. ), which is only slightly (but significantly) less than the inhibitory effect of the Apt2p control. Overexpression of the remaining 36 host proteins affected repRNA accumulation by less than 40% (Fig. ). Overall, the above-described experiments demonstrated that ~18% of the 44 host proteins tested could affect tombusvirus repRNA accumulation when overexpressed in yeast. We could not exclude the possibility that several of the overexpressed tagged proteins are not fully functional under the expression conditions.
Effect of overexpression of selected host proteins on TBSV RNA recombination in yeast.
The effect on tombusvirus RNA recombination was tested for 44 of the identified host proteins that interacted with p33 by taking advantage of the previously developed tombusvirus recombination system in yeast (51
) with some modifications. The recombination assay was similar to the above-described replication assay, except that it was based on a highly recombinogenic repRNA termed DI-AU-FP, which contains an AU-rich recombination hot spot (51
). In this assay, recombination takes place between two molecules of DI-AU-FP RNAs (50
). The accumulation of the recombinant RNAs (recRNAs) was estimated using Northern blotting (Fig. ). Importantly, we calculated the ratio of recRNA to repRNA (DI-AU-FP), which is more informative about recombination than recRNA levels alone (50
). Among the 44 host proteins tested, 4 affected TBSV RNA recombination dramatically by enhancing the ratio of recRNA by ~8- to 12-fold (Fig. ). Overexpression of other host proteins tested had lesser effects on TBSV recRNA accumulation (not shown).
FIG. 5. Effect of overexpression of selected host proteins interacting with p33 replication proteins on recombination by a TBSV repRNA in yeast. (A) Northern blotting of total RNA obtained from yeast overexpressing zz-tagged yeast proteins as shown. The host (more ...)
Among the identified proteins, Gsy2p glycogen synthase enhanced the recRNA ratio by 12-fold, the most among the 44 host proteins tested (Fig. ). Overexpression of the Arp8p actin-related protein, the Tif11p translation initiation factor eIF1A, and Iwr1p (unknown function) increased the ratio of recRNA to repRNA by 8- to 9.5-fold (Fig. ). Because a high amount of p92pol
has been shown to increase TBSV recombination (16
), we also tested p33 and p92pol
levels in yeast overexpressing host proteins. The Western analysis showed no significant increase in p92pol
levels in comparison with p33 in the Gsy2p, Arp8p, Tif11p, and Iwr1p overexpression strains (Fig. ), suggesting that these host proteins do not affect RNA recombination through changing the p33: p92pol
ratio but via a different, yet-uncharacterized mechanism(s).
Cdc34p is present within the purified tombusvirus replicase complex.
To further test the functional relevance of the identified host proteins interacting with p33, we selected Cdc34p based on its intriguing function as an E2 Ub-conjugating enzyme. Moreover, Cdc34p might be present within the tombusvirus replicase complex, since Cdc34p matches well with an unidentified yeast protein (termed X factor) with a pI value of ~4 and a molecular mass of ~35 kDa that has been detected in the highly purified tombusvirus replicase complex via two-dimensional gel electrophoresis (49
). On the other hand, the other intriguing host protein, Hbs1p, is far less characterized and its function will be studied in the future.
To test if Cdc34p is present within the highly purified tombusvirus replicase complex, we performed Western blotting with Cdc34p-specific antibody by use of a purified preparation of the detergent-solubilized tombusvirus replicase obtained from yeast grown at 23°C (optimal for TBSV replication) or 29°C (optimal for protein expression) (Fig. ). The Flag affinity-purified tombusvirus replicase containing Flag-tagged p33 and p92pol
replication proteins (Fig. , lanes 1 and 3) contained the native Cdc34p expressed from its natural chromosomal position (Fig. , lanes 1 and 3). Copurification of Cdc34p with the Flag-tagged replication proteins indicated that Cdc34p is a component of the membrane-bound replicase complex. The preparation was RNase treated prior to loading to the affinity column, suggesting that Cdc34p was likely retained on the Flag column via its binding to the replication proteins and not via the viral repRNA bound to the replication proteins. The control Flag-purified preparations from yeast expressing His6
-tagged p33 and p92pol
(Fig. , lanes 2 and 4) lacked these replication proteins as well as Cdc34p (Fig. , lanes 2 and 4), excluding the possibility that Cdc34p is a contaminating protein bound nonspecifically to the beads. Therefore, we propose that the previously detected X factor in the tombusvirus replicase complex (49
) is Cdc34p, based on its copurification with the viral replicase (Fig. ) and its physical properties (pI of 3.96 and molecular mass of 34 kDa).
FIG. 6. Copurification of Cdc34p with the tombusvirus replicase complex. (A) FLAG- and His6-tagged p33 (p33HF) and p92pol (p92HF) replication proteins were purified after detergent-based solubilization of a membrane-enriched fraction of yeast on a FLAG affinity (more ...) Downregulation of the Cdc34p level decreases TBSV repRNA accumulation in yeast.
To test if downregulation of the Cdc34p level affects TBSV repRNA accumulation, we used doxycycline-regulatable expression of Cdc34p from its original chromosomal location (25
). Because our preliminary experiments suggested that Cdc34p has a long half-life (not shown), we downregulated Cdc34p expression 6 h prior to launching repRNA replication. Indeed, the level of Cdc34p expression was ~10 times lower in yeast grown in the presence of doxycycline than it was in its absence (Fig. , bottom, lanes 3 and 4 versus 1 and 2). The accumulation of repRNA decreased ~3-fold in yeast grown in the presence of doxycycline (Fig. , top, lanes 4 to 6), whereas the amount of p33 replication protein did not change (Fig. , bottom, lanes 1 to 4), demonstrating that Cdc34p is important for tombusvirus replication.
FIG. 7. The effect of downregulation and overexpression of WT and mutant Cdc34p on TBSV repRNA accumulation in yeast. (A) Schematic representation of the known domains in Cdc34p. The active-site mutation that abolishes the E2 Ub conjugation activity of Cdc34p (more ...) Downregulation of Cdc34p level decreases the activity of the tombusvirus replicase.
To test if a decreased amount of Cdc34p could directly affect the activity of the tombusvirus replicase, we performed in vitro replicase assays with a membrane-enriched fraction derived from yeast. The membrane-enriched fraction contains the tombusvirus replicase and is capable of performing TBSV repRNA synthesis in vitro using the copurified repRNA as a template. To compare similar amounts of replicase complexes, we adjusted the p33 content in the membrane-enriched fractions obtained from yeast cultured with or without doxycycline (17
). These in vitro experiments have shown that the tombusvirus replicase obtained from yeast with downregulated Cdc34p 22 h after launching TBSV replication was only 17% as active as the control preparation from yeast grown in the absence of doxycycline (Fig. , lanes 3 and 4 versus 1 and 2). Similarly prepared replicase preparations at earlier time points after launching TBSV replication, such as 6 and 12 h, showed only 73% and 54% decreases (Fig. ), suggesting that Cdc34p was still more readily available during the formation of the replicase complex at the early time points than at the late time point. Altogether, the in vitro assay demonstrated that Cdc34p is critical for efficient tombusvirus replicase activity.
FIG. 8. Decreased replicase activity in the presence of a low Cdc34p level. A replicase activity assay with membrane-enriched preparations obtained from yeast expressing high or low levels of Cdc34p, based on the addition of 10 μg/ml doxycycline (Dox) (more ...) The Ub conjugation function of Cdc34p is important for TBSV repRNA replication.
Cdc34p has two functional domains: the 170-aa N-terminal UBC Ub conjugation domain conserved in E2s and the unique C-terminal acidic domain involved in protein/substrate binding (18
). To test which function/domain of Cdc34p is important for tombusvirus replication, we used Cdc34p mutants in a complementation assay. We found that overexpression of Cdc34-C95
S with inactive Ub-conjugating function from a plasmid could not complement TBSV repRNA accumulation in yeast with downregulated Cdc34p expression from the chromosome, whereas the full-length wild-type (WT) Cdc34p could complement repRNA accumulation (Fig. , lanes 16 to 18 versus 10 to 12). Moreover, the N-terminal and C-terminal portions of Cdc34p rendered Cdc34p nonfunctional in a complementation assay (Fig. , lanes 13 to 20), suggesting that both the UBC and the acidic domains of Cdc34p are important during TBSV replication.
In addition, we found a normal level of replicase activity obtained from yeast expressing the WT Cdc34p from a centromeric, low-copy-number plasmid, while Cdc34p was downregulated from the chromosomal location (Fig. , lanes 7 and 8). Thus, the plasmid-borne WT Cdc34p can function in the viral replicase. On the other hand, the mutated Cdc34-C95S protein with inactive Ub-conjugating function expressed from a plasmid could not complement the downregulated Cdc34p from the chromosomal location in a replicase assay (Fig. , lanes 11 and 12). These experiments suggest that the Ub conjugation function of Cdc34p is important for the function of the tombusvirus replicase.
Ubiquitination of p33 by Cdc34p in vitro.
Since Cdc34p is part of the tombusvirus replicase complex, it also binds to p33 directly and the ubiquitination function of Cdc34p is important for the replicase function (Fig. ); therefore, we wanted to test if Cdc34p could directly ubiquitinate p33 replication protein in the absence of Skp1p/cullin/F-box (SCF) E3 ligase complex, which is involved in substrate selection during normal cellular functions (48
). To this end, we performed an in vitro ubiquitination assay with purified recombinant proteins. The in vitro assay demonstrated that Cdc34p could ubiquitinate a fraction of p33 replication protein, causing ~9-and ~18-kDa shifts in the mobility of p33 (Fig. , lane 1). This shift in mobility is expected if a single Ub or two Ubs are added to p33 (mono- and biubiquitination). The higher-molecular-mass products detected on the gels could represent p33 proteins with multi- and/or polyubiquitination (Fig. , lane 1). The ubiquitination of p33 by Cdc34p required Uba1p E1 protein, Ub, and ATP, and it did not take place if one of these components was missing in the assay (Fig. , lanes 2 to 5). A similar assay with the purified MBP did not yield ubiquitination of MBP at a detectable level (Fig. , lanes 6 and 7), suggesting that p33 is specifically ubiquitinated by Cdc34p. Moreover, the increase in the molecular mass of p33 was ~35 kDa when we used GST-tagged Ub during a standard in vitro ubiquitination assay (Fig. ), confirming that Ub was added to p33 in the in vitro ubiquitination assay. A mutated Cdc34-C95
S protein with inactive Ub-conjugating function (Fig. ), as well as the N- or C-terminally truncated Cdc34p variants (not shown), could not ubiquitinate p33, supporting the model that the ubiquitination of p33 was a specific feature of Cdc34p.
FIG. 9. Ubiquitination of p33 by Cdc34p in vitro is independent of SCF E3 ligase complex. The in vitro ubiquitination was carried out in reaction mixtures containing 50 nM GST-Uba1p E1-activating enzyme (purified from yeast), 500 nM purified recombinant GST-Cdc34p (more ...) Ubiquitination of the p33 replication protein in yeast.
To demonstrate the possible ubiquitination of p33, we FLAG affinity purified His6/FLAG-tagged p33HF from yeast coexpressing Ub tagged with c-Myc from a plasmid. In these experiments, p33HF was solubilized from the membrane fraction, followed by purification and Western blotting to detect the addition of c-Myc-Ub to p33HF via anti-c-Myc antibody. Interestingly, we found mono- and biubiquitinated p33HF based on detection by anti-c-Myc antibody as well as a shift in the molecular mass of p33HF (mono- and biubiquitination cause ~8- and 16-kDa increases, respectively) (Fig. , lane 2). The identified bands likely represent ubiquitinated p33HF because similar experiments with p33 tagged with FLAG only (termed p33F, which is about 2 kDa smaller than p33HF) resulted in slightly faster-migrating mono- and biubiquitinated p33F (Fig. , lane 1). The change in the migration pattern of the purified p33HF versus p33F proteins supports the idea that these bands represent various ubiquitinated p33 and excludes the possibility that they represent ubiquitinated contaminating host proteins in our FLAG affinity-purified samples. Moreover, the control sample containing His6/FLAG-tagged peptide (expressed from pESC-His plus YEp105 plus pYC-HF) lacked similar ubiquitinated proteins (Fig. , lane 3). We also detected putative higher-molecular-mass p33 derivatives with multiubiquitination. However, these bands could also represent high-molecular-mass ubiquitinated host proteins which could have been copurified with p33. These host proteins could be part of complexes containing p33, since similar proteins were absent from the control samples (Fig. , lanes 3 and 6).
FIG. 10. Ubiquitination of p33 replication protein in yeast. The membrane-bound p33HF (tagged with FLAG and His6) or p33F (tagged with FLAG only) was purified via FLAG affinity chromatography after solubilization from yeast coexpressing c-Myc-tagged Ub from a (more ...)
To further demonstrate that the identified bands represent ubiquitinated p33HF, we also analyzed the same FLAG-purified p33HF or p33F samples by SDS-PAGE/Western blotting with anti-FLAG antibody (49
). Accordingly, we detected p33-specific bands with ~8- and 16-kDa molecular mass increases (Fig. , lanes 4 and 5) that are consistent with mono- and biubiquitinated p33HF or p33F. The mono- and biubiquitinated p33 proteins represented a small fraction (less than 5%) of the total p33, suggesting that not all p33 is ubiquitinated or that the deubiquitination process is rapid in yeast cells.