Co-opted CHIP E3 Ub-ligase provides protein quality control by regulating molecular properties of condensates formed by a positive-strand RNA virus

The TBSV p33 replication protein drives the formation of a specialized vir-condensate, which is associated with the membranous VROs due to the two membrane-spanning domains of p33 ( Fig. 1A ) ( 35 , 36 ). The vir-condensate provides essential proviral functions during tombusvirus replication. To inhibit the function of vir-condensate, we took advantage of a micropeptide, called killswitch, which promotes specific and stable self-interaction ( 44 ). The fusion of killswitch micropeptide to proteins could promote changes in condensates structures and protein mobility by contributing enhanced protein interactions ( 44 ). ( A ) Schematic representation of the strategy of used Killswitch peptide-fused ALFA nanobody to specifically recognize the ALFA-tagged p33 replication protein of TBSV. The scheme shows the key domains and characterized sites of TBSV p33 replication protein, including transmembrane domains (yellow boxes), mono-(bi)-ubiquitination (“ub”) and phosphorylation (“P”), sites arginine/proline-rich (RPR) region (gray box), and p33-p33 interaction domains (purple boxes). aa, amino acids. ( B ) Confocal images show RFP-tagged nbALFA-Killswitch induced irregular aggregate formation of ALFA-tagged eGFP-p33C in the presence of 2.5% PEG-8000 in vitro. Note that purified eGFP-p33C-ALFA forms round shaped droplets, whereas RFP-nbALFA-Killswitch shows little to no condensate formation under the same conditions. For the assay, 5 μM eGFP-p33C-ALFA and 2 μM RFP-nbALFA-Killswitch were used. The right panel shows similar results in the presence of 5% PEG-8000. Scale bars, 10 μm. ( C ) Right panels: Expression of nbALFA-Killswitch induces large aggregates/inclusions (>1 μm, marked with yellow arrows) within VROs formed by ALFA-tagged p33 in N. benthamiana cells. Left panels: VROs (marked with white arrows) formed by eGFP-p33-ALFA, RFP-p33-ALFA, and eGFP-p33 (control) show ring-like structures around the clustered peroxisomes that lack large aggregates. pGD empty vector (EV) is used as control. Scale bars, 5 μm. ( D ) Top: Northern blot analysis demonstrates decreased TBSV accumulation after coexpression of eGFP-p33-ALFA and nbALFA-Killswitch. Second panel: Ribosomal RNA is shown as loading control in an ethidium bromide-stained agarose gel. Bottom: The accumulation of eGFP-p33-ALFA and p33 was analyzed by Western blot using anti-p33 antibody, and anti-Myc antibody was used to detect nbALFA-Killswitch. Total protein loading was visualized by Ponceau S staining. ( E ) Fluorescence recovery after photobleaching (FRAP) analysis shows slower fluorescence recovery of co-opted BFP-NbADH1 within a single VRO induced by RFP-p33-ALFA in a N. benthamiana cell coexpressing nbALFA-Killswitch (top) relative to the empty vector (EV) control (bottom). An arrow points at the large VRO consisting of clustered peroxisomes. Note that the signal of membrane-bound p33 was not recovered. Scale bars, 5 μm. ( F ) The graph shows the extent of fluorescence recovery of BFP-NbADH1 within VROs at a given time point. Six to eight VROs from three independent plants were tested. Error bars show SE. ( G ) ATP production is reduced by nbALFA-Killswitch within the VROs formed by p33-ALFA and p33 fused with an ATP biosensor (ATeam YEMK ) in N. benthamiana cells. The more intense fluorescence resonance energy transfer (FRET) signals are green and red (between 0.5 and 1.0 ratio), whereas the low FRET signals (0.5 and below) are light blue and purple. We show the quantitative FRET values (obtained with ImageJ) for a number of samples in the graph. Error bars show SE. Statistical analysis was performed using the Student’s t test (**** P  < 0.0001). Scale bars, 10 μm. CFP, cyan fluorescent protein; YFP, yellow fluorescent protein. The C-terminal portion of TBSV p33 induces gel-like droplets in vitro due to liquid-liquid phase separation ( 35 ). To test the effect of the killswitch micropeptide on p33C-induced droplet formation, we fused killswitch to a nanobody, which specifically recognizes the “ALFA”-tag sequence [nanobody ALFA (nbALFA); Fig. 1A ] ( 45 , 46 ). His 6 –enhanced green fluorescent protein (eGFP)–p33C–ALFA was purified from Escherichia coli , and we found that it induced mostly round-shaped droplets in the presence of 2.5 or 5% PEG-8000 (polyethylene glycol, molecular weight 8000) crowding agent ( Fig. 1B ). His 6 -eGFP-p33C-ALFA induced droplets comparably to the previously characterized TBSV eGFP-p33C ( 35 ), suggesting that the ALFA tag has not changed p33 characteristics. However, mixing purified His 6 –red fluorescent protein (RFP)–nbALFA–Killswitch with His 6 -eGFP-p33C-ALFA resulted in irregular-shaped aggregations ( Fig. 1B and fig. S1, A and B). These in vitro data suggest that His 6 -RFP-nbALFA-Killswitch bound to His 6 -eGFP-p33C-ALFA and changed His 6 -eGFP-p33C-ALFA droplets into irregular aggregates. Mixing purified His 6 -RFP-nbALFA with His 6 -eGFP-p33C-ALFA formed codroplets (fig. S1A), even at high concentration, without forming aggregates (fig. S1B). Mixing purified His 6 -RFP-nbALFA-Killswitch with His 6 -eGFP-p33C (lacking the ALFA tag) did not result in irregular p33C aggregates (fig. S1A). Together, these control experiments confirmed that only the nbALFA-targeted killswitch micropeptide could induce aggregate formation by p33C-ALFA (i.e., His 6 -eGFP-p33C-ALFA) in vitro. Based on the above in vitro experiments, we propose that the nbALFA-targeted killswitch micropeptide could change the molecular properties of p33-ALFA–induced droplets/condensates. To test whether the killswitch micropeptide could also change the properties of the vir-condensate induced by p33 in vivo, we coexpressed TBSV eGFP-p33-ALFA protein and nbALFA-Killswitch in Nicotiana benthamiana cells to induce VRO formation ( Fig. 1C ). The p33-induced VROs form from co-opted clustered peroxisomes and vir-condensate ( Fig. 1C , left panels) ( 35 , 36 ). The expression of nbALFA-Killswitch in N. benthamiana cells also expressing TBSV eGFP-p33-ALFA protein or RFP-p33-ALFA protein led to VRO formation that contained large puncta ( Fig. 1C , right panels). As expected, the expression level of nbALFA-Killswitch in N. benthamiana cells affected large puncta formation by TBSV eGFP-p33-ALFA in VROs (fig. S1, C and D). After a 2-day expression, all VROs contained large puncta structures when nbALFA-Killswitch was expressed to a high level in N. benthamiana cells (fig. S1D). The large puncta by eGFP-p33-ALFA were formed despite of the reduction of eGFP-p33-ALFA amount when nbALFA-Killswitch was expressed at the highest level [fig. S1, C and D; Agrobacterium culture, optical density at 600 nm (OD 600 ) of 0.4]. The formation of puncta-like structures is rare in N. benthamiana cells only expressing TBSV eGFP-p33-ALFA or RFP-p33-ALFA proteins ( Fig. 1C , left panels) or eGFP-p33 in the absence ( Fig. 1C , left panel) or presence of nbALFA-Killswitch ( Fig. 1C , right panel). In control experiments, the coexpression of increasing amounts of nbALFA (in the absence of killswitch) with eGFP-p33-ALFA (fig. S2, A and B) or nbALFA-Killswitch with eGFP-p33 (lacking the ALFA tag) did not inhibit VRO formation and did not lead to large puncta formation in VROs (fig. S3, A and B). Highest nbALFA expression increased the accumulation of eGFP-p33-ALFA, but only mildly affected puncta formation in VROs (fig. S2, A and B). These data indicate that the killswitch micropeptide targeted to VROs via nbALFA caused the aggregation/puncta formation of eGFP-p33-ALFA and RFP-p33-ALFA proteins via the killswitch-killswitch interactions. To demonstrate that the coexpression of nbALFA-Killswitch could inhibit TBSV p33 functions in N. benthamiana cells, we measured TBSV replication when nbALFA-Killswitch and eGFP-p33-ALFA were coexpressed in TBSV-infected cells. Northern blot revealed ∼4-fold inhibition of TBSV replication ( Fig. 1D , lanes 13 to 16 versus 9 to 12). In contrast, expression of nbALFA-Killswitch in the absence of eGFP-p33-ALFA expression had only minor inhibition on TBSV replication ( Fig. 1D , lanes 5 to 8 versus 1 to 4). Expression of nbALFA in the presence of eGFP-p33-ALFA (fig. S2C) or nbALFA-Killswitch in the presence of eGFP-p33 (fig. S3C) had no effect on TBSV replication. These data suggest that both targeting to VROs via nbALFA-ALFA tag interaction and specific binding of nbALFA-Killswitch to eGFP-p33-ALFA are needed for the killswitch micropeptide to inhibit TBSV replication. The VRO-targeted killswitch micropeptide likely inhibits p33 function to induce functional vir-condensate. The p33-induced vir-condensate consists of a gel-like structure allowing slow movement of co-opted (client) cytosolic proteins, mostly within or into the vir-condensate ( 35 , 36 ). To test whether the expression of nbALFA-Killswitch “hardened” the vir-condensate induced by eGFP-p33-ALFA in N. benthamiana cells, we performed fluorescence recovery after photobleaching (FRAP) analysis of the co-opted alcohol dehydrogenase (BFP-ADH1) fermentation enzyme, which is co-opted and a client component of the vir-condensate in association with VROs ( 47 ). The recovery of the BFP-ADH1 fluorescent signal in RFP-p33-ALFA–induced VROs was considerably slowed by the coexpression of nbALFA-Killswitch [ Fig. 1, E (top panels) and F ] when compared with cells not expressing nbALFA-Killswitch ( Fig. 1E , bottom panels). In the control experiment, the recovery of the BFP-ADH1 fluorescent signal in RFP-p33-ALFA–induced VROs was comparable when nbALFA was coexpressed or not (fig. S4). To obtain additional data on the altered molecular properties of vir-condensate caused by the killswitch micropeptide, we performed FRAP analysis of BFP-RPN11, which is co-opted and a client component of the vir-condensate ( 35 , 41 , 42 ). BFP-RPN11 showed reduced level of fluorescent signal recovery in VROs induced by RFP-p33-ALFA when nbALFA-Killswitch was coexpressed (fig. S5, A and B). In the control experiment, the recovery of the BFP-RPN11 fluorescent signal in RFP-p33-ALFA–induced VROs was not affected by coexpression of nbALFA (fig. S5, C and D). These findings are consistent with the model that expression of nbALFA-Killswitch leads to inhibition of client protein mobility, thus resulting in less dynamic and less functional vir-condensates associated with TBSV VROs. In another test for the effect of the killswitch micropeptide on the molecular properties of vir-condensate, we measured ATP production within VROs. One of the functions of TBSV vir-condensate is to produce ATP locally via co-opted glycolytic and fermentation enzymes, which are needed for rapid tombusvirus replication ( 30 , 47 , 48 ). To test whether VROs containing nbALFA-Killswitch (in the presence of p33-ALFA) could inhibit ATP generation, we used an ATP-biosensor module fused to TBSV p33 (i.e., p33-ATeam), targeting the fusion proteins to VROs ( 30 , 38 ). The fluorescence resonance energy transfer–based assay showed that the expression of nbALFA-Killswitch significantly inhibited relative ATP levels in VROs formed by p33-ALFA ( Fig. 1G ). In the control experiment, the ATP production in p33-Flag-induced VROs is not affected by the expression of nbALFA-Killswitch, likely due to lack of targeting to VROs (fig. S6A). In addition, ATP production in p33-ALFA–induced VROs was not affected by coexpression of nbALFA (fig. S6B), suggesting that the killswitch micropeptide must be targeted to p33 in VROs to reduce ATP production, likely due to the altered mobility of co-opted glycolytic and fermentation enzymes in the hardened vir-condensates. Overall, these experiments revealed that making the vir-condensate more solid-like or less dynamic via the recruitment of the killswitch micropeptide markedly interferes with the viral functions of vir-condensate and VRO. The vir-condensate, induced by p33 replication protein, forms during TBSV replication, and it surrounds the co-opted clustered peroxisomes ( 35 ). The p33 replication protein is embedded in the peroxisomal membrane, decorating the membranes in a mostly evenly distributed “smooth” pattern when expressed either as eGFP-p33 or the C-terminally tagged p33-eGFP or RFP-p33 ( Fig. 2A ). However, expression of p33-BFP in N. benthamiana cells leads to formation of several small punctate structures within the large VROs ( Fig. 2, A and B ) consisting of the clustered peroxisomes marked with the GFP-SKL luminal marker protein. The punctate structures within p33-BFP-induced VROs might represent p33 aggregates similar to the above aggregates induced by the targeted killswitch micropeptide, except the formation of smaller-sized puncta. This suggests that p33-BFP might be aggregation-prone, which could possibly alter its ability to induce “proper” vir-condensates. Expression of p33-BFP in N. benthamiana cells replicating TBSV led to ∼4-fold inhibition of viral RNA accumulation ( Fig. 2C , lanes 5 to 8 versus 1 to 4), suggesting that p33-BFP has dominant negative effect on the function of wild-type (WT) p33 expressed from the TBSV genomic RNA (gRNA), likely by forming nonfunctional aggregations of p33-BFP–WT p33 oligomers. Self-oligomerization of p33 is critical for TBSV replication ( 33 , 49 , 50 ). In contrast, expression of eGFP-p33-ALFA in N. benthamiana cells replicating TBSV did not inhibit viral RNA accumulation ( Fig. 1D , lanes 9 to 12 versus 1 to 4). ( A ) Confocal images show VRO-like structure induced by different fluorescently labeled p33. Note that eGFP-p33–, p33-eGFP–, and RFP-p33–induced VROs show ring-like structures around the clustered peroxisomes that lack large aggregates. ( B ) Confocal images of p33-BFP-induced VRO show many small puncta, which represent aggregations/inclusions. The clustered peroxisomes within the VRO are marked by GFP-SKL. Scale bars, 10 μm. ( C ) Northern blot analysis reveals inhibitory effect of p33-BFP expression on TBSV replication. Bottom panel: Ribosomal RNA is shown as loading control in an ethidium bromide-stained agarose gel. ( D ) Top three panels: FRAP analysis shows highly reduced fluorescence recovery of co-opted RFP-NbADH1 (top panel) within a single VRO induced by p33-eGFP (middle panel) in a N. benthamiana cell coexpressing p33-BFP (bottom panel). Note that p33-eGFP and p33-BFP signals in VRO were not recovered because of the lack of mobility by the membrane-bound p33. The photobleached area in the cell is circled. Bottom two panels: FRAP analysis of fluorescence recovery of RFP-NbADH1 (top panel) within a single VRO induced by p33-eGFP in the absence of p33-BFP expression (control). The graph shows the extent of fluorescence recovery of co-opted RFP-NbADH1 within VROs at a given time point. Six to eight VROs from three independent plants were tested. Error bars show SE. Scale bars, 5 μm. ( E ) FRAP analysis of co-opted RFP-NbPK within a single VRO induced by p33-eGFP in the presence (top three panels) or absence of p33-BFP (bottom two panels, as control) in a N. benthamiana cell. See additional details in (D) above. ( F ) ATP production is reduced by coexpression of p33-BFP within the VROs formed by p33 fused with an ATP biosensor (ATeam YEMK ) in N. benthamiana cells. See Fig. 1G for further details. Scale bars, 10 μm. CFP, cyan fluorescent protein; YFP, yellow fluorescent protein. The puncta-forming p33-BFP affected the molecular properties of vir-condensate by rendering the vir-condensate less dynamic, based on FRAP experiments showing reduced rate of fluorescent recovery by the co-opted RFP-ADH1 or RFP-PK (RFP-fused pyruvate kinase) glycolytic enzymes in VROs in comparison with cells expressing only p33-eGFP ( Fig. 2, D and E ). As expected, the less dynamic vir-condensate due to the presence of p33-BFP resulted in poor accumulation of ATP within VROs ( Fig. 2F ). Together, the data obtained with p33-BFP show similar outcome to those obtained with the VRO-targeted killswitch micropeptide approach above. Namely, protein aggregation and the reduced dynamic nature of vir-condensate within VROs lead to reduced level of ATP production and inhibition of TBSV replication. We decided to use p33-BFP in additional experiments due to its excellent fluorescent features and high aggregation property affecting the vir-condensate function in plant cells. Mutations within the TBSV p33 replication protein cause protein aggregation in vitro ( 35 ), suggesting that the p33 replication protein containing two IDRs and posttranslational modifications, such as sumoylation, could be prone to forming nonfunctional protein aggregations under some conditions ( 36 ). Yet, the lack of punctate structures formed by the expression of eGFP-tagged or RFP-tagged p33 ( Fig. 2A ) or during TBSV replication ( 35 ) might indicate a putative activity of a co-opted host protein that inhibits protein aggregation by the TBSV p33 replication protein under standard virus infections. Because TBSV p33 and CIRV p36 replication proteins interact with heat shock protein 70 (Hsp70) chaperone ( 51 , 52 ) and several TPR-containing co-chaperones, such as Cyp40 cyclophilin (Cpr7p in yeast), HOP (Sti1 in yeast), and the Ttc4 oncogene-like Cns1 (based on earlier yeast screens) ( 53 – 55 ), we decided to test whether CHIP E3 Ub-ligase that has three TPR domains, binds to Hsp70, and acts as a co-chaperone and functions in PQC ( 56 , 57 ) could affect TBSV replication and vir-condensate formation. When we coexpressed RFP-CHIP with nbALFA-Killswitch and eGFP-p33-ALFA in N. benthamiana cells, then we observed only a few small puncta within eGFP-p33-ALFA VROs in N. benthamiana cells ( Fig. 3A , right panels). This is in contrary to the large puncta formation caused by nbALFA-Killswitch within eGFP-p33-ALFA–induced VROs when CHIP was not overexpressed ( Fig. 3A , left panels). In addition, transient expression of CHIP led to recovery of TBSV replication in the presence of eGFP-p33-ALFA and nbALFA-Killswitch ( Fig. 3B , lanes 5 to 8 versus 1 to 4). ( A ) Confocal images show that eGFP-p33-ALFA aggregates induced by nbALFA-Killswitch expression are reduced in size in VROs upon coexpression of RFP-NbCHIP (right) compared with the empty vector control (left). We show two independent representative image sets for each experiment. Scale bars, 5 μm. ( B ) Top: Northern blot analysis shows that expression of RFP-NbCHIP enhances TBSV replication when eGFP-p33-ALFA and nbALFA-Killswitch are coexpressed in comparison with the empty vector (EV) control. Second panel: Ribosomal RNA is shown as loading control in an ethidium bromide-stained agarose gel. Bottom: The accumulation of RFP-NbCHIP, eGFP-p33-ALFA, and nbALFA-Killswitch was analyzed by Western blot using anti-RFP, anti-GFP, and anti-Myc antibodies, respectively. Total protein loading was visualized by Ponceau S staining. ( C ) Top two panels: Coexpression assay. Confocal images show that p33-BFP accumulation is inhibited by the coexpression of RFP-NbCHIP. Top: The coexpression assay shows that VRO-like structures consisting of clustered peroxisomes (marked with GFP-SKL) lack detectable p33-BFP signals, which represent most VROs (95%). Middle: Only a small fraction of VROs (∼5%) exhibits colocalization of p33-BFP and RFP-NbCHIP, and these structures do not contain visible aggregates. Scale bars, 10 μm. Bottom: Sequential expression assay. In the sequential expression assay, when p33-BFP and GFP-SKL were expressed 1 day before RFP-NbCHIP expression, colocalization of p33-BFP and RFP-NbCHIP within VROs that did not have visible aggregates were observed by confocal microscopy. Scale bars, 5 μm. ( D ) Western blot analysis shows that expression of RFP-NbCHIP reduces the accumulation of p33-BFP in N. benthamiana . p33-BFP was detected using anti-p33 antibody, and RFP-NbCHIP was detected using anti-RFP antibody. ( E ) Top: Northern blot analysis reveals the effect of expression of RFP-NbCHIP E3 ligase on TBSV replication when p33-BFP is coexpressed in comparison with control (empty vector). Bottom: Ribosomal RNA is shown as loading control in ethidium-bromide stained agarose gel. Note that the TBSV gRNA is visible in the image. ( F ) Western blot analysis shows higher accumulation of p33-BFP in NbCHIP-silenced plants compared with control plants. The middle panel shows the silencing efficiency of CHIP mRNA in N. benthamiana plants by semiquantitative RT-PCR. Note that some variation in CHIP mRNA levels was observed among the five control samples. Such variation is not unexpected as CHIP is expressed at relatively low basal levels and is responsive to environmental and physiological cues. Consequently, transcript abundance may differ among individual plants. Despite this variability, CHIP transcript levels were consistently reduced in the silenced plants compared with the controls, confirming effective gene silencing. Bottom: RT-PCR was used to show tubulin mRNA level in the same samples as control. ( G ) Larger aggregates/inclusions are observed in p33-BFP-induced VROs in NbCHIP-silenced N. benthamiana cells compared with control plants (bottom). Scale bars, 5 μm. To obtain further evidence that CHIP targets the p33-induced punctate structures (inclusions), we tested RFP-CHIP in N. benthamiana cells also coexpressing p33-BFP. The puncta-forming p33-BFP was eliminated by the transient expression of CHIP, although clustered peroxisomes were observed (marked by eGFP-SKL peroxisome luminal marker) ( Fig. 3C , top panel) ( 28 , 58 ). The inhibitory effect of RFP-CHIP on p33-BFP accumulation is likely a “late” and incomplete effect because p33-BFP expression resulted in clustered peroxisomes in many cells, the hallmark of TBSV VROs. Also, Western blot analysis confirmed the reduced accumulation of p33-BFP in N. benthamiana cells expressing RFP-CHIP ( Fig. 3D ). Note that, in rare cases, we observed p33-BFP decorated VROs in CHIP-expressing cells ( Fig. 3C , second panel). These VROs lacked punctate p33-BFP structures. Preexpression of p33-BFP 1 day before RFP-CHIP expression resulted in VRO formations lacking puncta structures ( Fig. 3C , bottom panel). All these data suggest that CHIP specifically targets the misfolded, puncta-forming fraction of p33-BFP for degradation while “shielding” a fraction of p33-BFP molecules that are not aggregated. Accordingly, coexpression of CHIP with p33-BFP in TBSV-infected cells led to partial recovery of TBSV replication ( Fig. 3E , lanes 10 to 12 versus 7 to 9), suggesting that CHIP mitigates the dominant inhibitory effect of p33-BFP on TBSV replication. Virus-induced gene silencing (VIGS)–based knockdown of CHIP expression in N. benthamiana led to ∼3-fold increased p33-BFP accumulation ( Fig. 3F ), and it resulted in large puncta formation in VROs ( Fig. 3G ) in comparison with the smaller p33-BFP puncta in control, nonsilenced plants. Transient expression of CHIP lacking the E3 ligase function (ΔC79, deletion of the U-box region, fig. S7A) did not alter the accumulation of the aggregation-prone p33-BFP (fig. S7B) and the p33-BFP inclusions from VROs (fig. S7D), whereas expression of CHIP lacking the TPR domain (ΔN112) reduced the accumulation of p33-BFP (fig. S7C) and inclusions in VROs (fig. S7D), suggesting that the E3 Ub-ligase activity of CHIP is required to reduce p33-BFP levels in cells. Similar role of the U-box domain of CHIP in removal of the aggregation-prone, mitochondria-associated CIRV p36-BFP was observed (fig. S7E), highlighting the essential role of E3 Ub-ligase activity of CHIP. Based on the obtained data, it seems that CHIP is guarding the TBSV VROs from the abundant formation of inhibitory p33 puncta structures, possibly by promoting degradation of the misfolded and aggregated p33 proteins. To gain insight how CHIP affects TBSV replication, first we tested the putative interaction of CHIP with the TBSV p33 replication protein ( Fig. 4A ). Using the membrane-based split-Ub yeast two hybrid assay (MYTH) ( 49 ), we found that both the Arabidopsis CHIP and the N. benthamiana CHIP interacted with TBSV p33 ( Fig. 4B ). Purification of the functional tombusvirus replicase from yeast or N. benthamiana showed that CHIP is present in the VRO fraction ( Fig. 4, C to E ). Purification of the CIRV replicase also confirmed the presence of copurified CHIP in mitochondria-associated VROs (fig. S8A). Confocal microscope analysis confirmed that the NbCHIP was relocalized to the TBSV VROs marked with eGFP-p33 and BFP-SKL (peroxisome luminal marker) ( Fig. 4F ). FRAP assay showed the recovery of the RFP fluorescent signal after photobleaching, suggesting that RFP-CHIP is present in the vir-condensate fraction of the VROs ( Fig. 4, G and H ). This was further supported by the reduced copurification of CHIP with p33 from membrane fraction when new VRO formation was inhibited by cycloheximide treatment (fig. S8B). This is because the co-opted host proteins, such as CHIP, could move within, into or out of the vir-condensate, whereas those co-opted host proteins, which are permanent components of the replicase complex, are anchored and trapped in spherules at the co-opted peroxisomal membrane ( 59 , 60 ). ( A ) Schematic representation of the key domains of CHIP, including the N-terminal TPR domain, which is known to interact with HSP70, the C-terminal U-box E3 ligase domain, and the central coiled-coil domain functioning as a linker region. Different truncation mutants tested are shown. aa, amino acids. ( B ) The split Ub-based MYTH assay shows the interaction between the TBSV p33 and Arabidopsis thaliana CHIP and its various truncation mutants. Note that both A. thaliana (top) and N. benthamiana CHIP homologs (bottom) were tested. The Ssa1p heat shock protein 70 (Hsp70) and the empty prey vector (NubG) were used as the positive and the negative controls, respectively. The left panel shows the interaction strength, and the right panel demonstrates that comparable amounts of yeasts were used for these experiments. ( C ) Copurification of the NbCHIP with TBSV Flag-p33 and Flag-p92 pol replication proteins from isolated subcellular membranes. Top two panels: Western blot analysis of copurified His 6 -NbCHIP (lanes 1 and 2) with Flag-affinity captured Flag-p33 and Flag-p92 pol . His 6 -tagged proteins were detected with anti-His antibody, whereas Flag-p33 was detected with anti-Flag antibody. The negative control was from yeast expressing His 6 -p33 and His 6 -p92 pol using a Flag-affinity column (lane 3). Samples were cross-linked with formaldehyde, followed by quenching with glycine before cell lysis. Bottom two panels: Western blot of total His 6 -NbCHIP, His 6 -p33, and Flag-p33 in the total yeast extracts. ( D ) Copurification of the AtCHIP and truncation mutants with TBSV Flag-p33 and Flag-p92 pol replication proteins from subcellular membranes. See (C) for further details. ( E ) Copurification of NbCHIP with TBSV p33-Flag from N. benthamiana . Top two panels: Western blot analysis of copurified 3Myc-tagged NbCHIP (lanes 1 and 2) with Flag-affinity purified p33-Flag. p33-Flag was detected with anti-Flag antibody, whereas 3Myc-NbCHIP was detected with anti-Myc antibody as shown. Bottom two panels show the accumulation of p33-Flag, and 3Myc-tagged NbCHIP in the total plant extracts. Flag-RFP-SKL and pGD empty vector were used as negative control. ( F ) Confocal images show colocalization of TBSV eGFP-p33 and RFP-NbCHIP within VROs consisting of clustered peroxisomes (marked by BFP-SKL) in N. benthamiana leaves. Bottom: Z -stack image shows that RFP-NbCHIP is distributed in the cytosol and nucleus, and no clustered peroxisomes are seen in the absence of p33. Scale bars, 10 μm. ( G ) FRAP analysis shows fast fluorescence recovery of RFP-NbCHIP within a single VRO induced by eGFP-p33 in a N. benthamiana cell. Note that the signal of the membrane-bound p33 was not recovered. Scale bars, 5 μm. ( H ) The graph shows the extent of fluorescence recovery of RFP-NbCHIP within VROs at a given time point. Six to eight VROs from three independent plants were tested. Error bars show SE. CHIP has three conserved domains, including the N-terminal TPR domain, which interacts with Hsp70 chaperone ( 56 , 57 ). The C-terminal region contains the unique U-box domain with E3 Ub-ligase function ( Fig. 4A ). These domains are connected by a linker domain (charged coiled coil region) ( 56 ). The above yeast MYTH assay revealed that the TPR domain showed strong interaction with p33, whereas the U-box domain and the linker domain also showed interactions with p33. The interaction was confirmed by the efficient copurification of full-length AtCHIP and the N-terminal region of CHIP (construct ΔC79, Fig. 4D ) and the lesser copurification of the U-box region (construct ΔN107, Fig. 4D ) with the purified tombusvirus replicase. The role of the CHIP TPR domain in interaction with the CIRV p36 was confirmed on the basis of the purified CIRV replicase (fig. S8A, lanes 4 to 5). Pulldown experiments also supported the efficient binding of the C-terminal region of TBSV p33 or the CIRV p36 replication proteins with CHIP and the TPR domain of CHIP (fig. S8, C to E). This indicates that the N-terminal TPR domain of CHIP is the major region to interact with p33/p36, with other domains of CHIP also playing roles. Overall, these data supported that CHIP interacts with the TBSV p33 or the CIRV p36 replication proteins, and it is recruited into the VROs from the cytosol. To explore the role of CHIP in tombusvirus replication, we knocked down NbCHIP mRNA levels based on a VIGS approach in N. benthamiana ( 61 , 62 ). Replication of the peroxisome-associated TBSV was reduced by ∼3-fold in the knockdown plants when compared with the nonsilenced control plants 1.5 days after inoculation ( Fig. 5A , lanes 6 to 10 versus 1 to 5). CHIP knockdown in N. benthamiana also reduced the accumulation of the related, but mitochondria-associated CIRV gRNA by ∼3-fold ( Fig. 5B ) and the related peroxisome-associated cucumber necrosis virus (CNV) by ∼2-fold ( Fig. 5C ). Knockdown of CHIP did not alter plant phenotypes when compared with the control [tobacco rattle virus (TRV)–cGFP] plants ( Fig. 5D ). Single plant cells (protoplasts) with CHIP knockdown supported ∼3-fold reduced accumulation of TBSV RNAs ( Fig. 5E ), demonstrating that CHIP has proviral function at a single cell level, in addition to organismal level. ( A ) Knockdown of NbCHIP mRNA level inhibits TBSV replication in N. benthamiana plants. Top: Accumulation of TBSV gRNA in NbCHIP-silenced N. benthamiana plants at 1.5 days postinoculation in the inoculated leaves was measured by Northern blot analysis. Second panel: Ribosomal RNA is shown as loading control in an ethidium bromide-stained agarose gel. Third panel: RT-PCR analysis of NbCHIP mRNA level in the silenced and control plants. Fourth panel: RT-PCR analysis of tubulin mRNA level in the silenced- and control plants. Each experiment was repeated three times. ( B and C ) Knockdown of NbCHIP mRNA level inhibits both CIRV and CNV replication in N. benthamiana plants. ( D ) No obvious phenotype is observed in NbCHIP-silenced N. benthamiana plants as compared with the control plants. ( E ) Knockdown of NbCHIP mRNA levels inhibits TBSV replication in N. benthamiana protoplasts, as measured by Northern blot analysis. The graph shows the silencing efficiency of NbCHIP through VIGS in protoplasts measured by RT-qPCR. ( F ) NbCHIP silencing leads to reduced ATP levels within VROs induced by p33 in N. benthamiana cells. The ATP biosensor ATeam YEMK was fused to p33 for targeting into VROs. See Fig. 1G for further details. Scale bars, 10 μm. CFP, cyan fluorescent protein; YFP, yellow fluorescent protein. ( G ) FRAP analysis shows reduced fluorescence recovery of co-opted RFP-PK within a single VRO (marked with an arrow) induced by p33-BFP in a cell of CHIP KD N. benthamiana when compared with a single VRO in a control, nonsilenced cell (bottom panels). ( H ) The graph shows the extent of fluorescence recovery of RFP-PK within VROs at a given time point. Six to eight VROs from three independent plants were tested. Error bars show SE. ( I ) Expression of NbCHIP mRNAs is highly induced by TBSV (left) and CIRV and CNV (right) in N. benthamiana plants determined through semiquantitative RT-PCR. ( J ) RT-qPCR analysis shows that NbCHIP expression is induced by TBSV and CIRV infections in N. benthamiana plants. Tubulin was used as an internal reference gene. ( K ) CHIP promoter-driven GFP expression is induced by TBSV infection and detected by Western blot analysis using anti-GFP antibody. Total protein loading was visualized by Coomassie blue staining. ** P < 0.01; **** P < 0.0001. Because CHIP is present in the vir-condensate portion of VROs ( Fig. 4G ), CHIP might affect the function of vir-condensate indirectly via eliminating aggregated, nonfunctional molecules. Measuring the ATP production in VROs formed in CHIP silenced plants showed significant reduction of ATP production in VROs ( Fig. 5F ). In addition, CHIP-silenced plants showed reduced recovery of the fluorescent signal for the co-opted RFP-PK glycolytic enzyme in VROs, suggesting less dynamic vir-condensate formation ( Fig. 5, G and H ). Together, these data strongly support a proviral function of CHIP in vir-condensate function and replication of tombusviruses. The emerging picture is that the co-opted CHIP E3 enzyme might be involved in maintaining p33 PQC within the VROs to keep the associated vir-condensate dynamic and highly functional during TBSV infection. To test whether tombusvirus infections alter the expression of CHIP, we measured the mRNA levels in infected N. benthamiana leaves. Reverse transcription polymerase chain reaction (RT-PCR) assay showed the robust induction of CHIP mRNA expression with all three tombusviruses tested ( Fig. 5I ). RT–quantitative polymerase chain reaction (qPCR) analysis of total RNA extracts from TBSV- or CIRV-infected leaves versus mock-inoculated leaves revealed ∼30- to 45-fold increased expression of CHIP in the inoculated leaves 2 days after inoculation ( Fig. 5J ). The CHIP mRNA level was low in mock inoculated plants ( Fig. 5, I and J ), suggesting that basal CHIP expression is low in N. benthamiana leaves. The promoter test with eGFP ORF placed behind DNA containing CHIP promoter region also showed increased CHIP promoter activity after TBSV infection based on measurement of eGFP level ( Fig. 5K ). All these data support a tombusvirus-induced robust expression of CHIP in N. benthamiana leaves. The above data indicated a p33 PQC function for the co-opted CHIP that resulted in removal of the aggregation-prone p33 when expressed as a BFP fusion protein or in the presence of the killswitch micropeptide. TBSV p33 becomes sumoylated, and it also contains a long IDR and a SUMO-interacting motif (SIM2), all of which drive droplet formation in vitro ( 32 , 35 , 36 ). We propose that the phase-separation–inducing features of p33 might render p33 aggregation-prone that could undergo transition to solid-like phase under some conditions and the host-coded CHIP maintains p33 functions by eliminating aggregated p33 molecules from VROs. To further explore the quality control role of CHIP in tombusvirus replication, we tested a p33 mutant that is aggregation prone. Namely, we observed that a four–amino acid alanine mutant of p33C (the cytosol-exposed segment of p33) within the SUMO-interaction motif (SIM2-m4) formed aggregated structures in vitro instead of the droplets (condensates) formation supported by WT p33C ( Fig. 6A , right panels) ( 36 ). Mixing the purified aggregation-prone eGFP-p33C-SIM2-m4 with the droplet-forming RFP-SUMO-p33C resulted in irregular-shaped aggregations in vitro in the presence of 2.5% PEG-8000 ( Fig. 6A , left panels). In contrast, mixing the purified eGFP-p33C with RFP-SUMO-p33C mostly resulted in round-shaped droplets ( Fig. 6A , middle panels). These in vitro data suggest that mutations within the SIM2 domain in p33 act dominantly by “forcing” the droplet-forming RFP-SUMO-p33C into coaggregated structures. We interpret these data as p33 mutation induces the coaggregation of p33 carrying WT sequences, likely interfering with p33 function(s). ( A ) Confocal images show irregular-shaped coaggregations of eGFP-p33C-SIM2-m4 (top, left panel) and RFP-SUMO-p33C (second, left panel) in the presence of 5% PEG-8000 in vitro. Whereas eGFP-p33C forms round-shaped codroplets with RFP-tagged SUMO-p33C (middle panel) under the same conditions. Far right panels show the control droplet formation assays with single proteins. For the assay, 1 μM RFP-SUMO*-p33C and 2.5 μM eGFP-p33C or eGFP-p33C-SIM2-m4 were used. Scale bars, 10 μm. ( B ) Northern blot analysis demonstrates decreased TBSV accumulation when p33-SIM2-m4 was coexpressed, in comparison with expression of WT p33-Flag or pGD empty vector. ( C ) Expression of p33-SIM2-m4 reduces the accumulation of WT p33 and p92 pol . The accumulation of p33-Flag, p33, SIM2-m4 and p92 pol was analyzed by Western blot using anti-p33 antibody. Total protein loading was visualized by Ponceau S staining. ( D ) TBSV p33-SIM2-m4 mutant shows reduced stability compared with WT p33. Western blot shows p33 levels after expression in N. benthamiana protoplasts at a given time point after treatment with 100 μM CHX, an inhibitor of new protein synthesis. The statistical analysis was performed using the Student’s t test. Each experiment was repeated three times. h, hours. ( E ) The expression of p33-SIM2-m4 leads to reduced ATP levels within VROs induced by p33. The ATP biosensor ATeam YEMK was fused to p33 for targeting into VROs. See Fig. 1G for further details. Scale bars, 10 μm. **** P < 0.0001. ( F and G ) Knockdown of NbCHIP mRNA level enhances the accumulation of p33-SIM2-m4 but not WT p33 in N. benthamiana . Semiquantitative RT-PCR was performed to determine the silencing efficiency of NbCHIP expression. ( H ) Overexpression of NbCHIP decreases the accumulation of p33-SIM2-m4 (left) but increases WT p33 level (right) in N. benthamiana . The accumulation of WT p33, p33-SIM2-m4, and NbCHIP was analyzed with Western blot using anti-Flag antibody, whereas anti-Myc antibody was used to detect NbCHIP. Each experiment was repeated. To test the dominant effect of p33-SIM2-m4, we expressed it in TBSV-infected plant cells, which produced the WT p33 from the gRNA. Transient expression of p33-SIM2-m4 reduced TBSV RNA accumulation by ∼6-fold when compared with the transient expression of p33-Flag ( Fig. 6B , lanes 9 to 12 versus 5 to 8). Expression of p33-SIM2-m4 also inhibited CIRV replication in plants (fig. S9A). Transient expression of p33-SIM2-m4 resulted in low p33 mutant accumulation ( Fig. 6C ). However, it also led to decreased accumulation of the WT p33 and p92 pol expressed via agroinfiltration ( Fig. 6C ). Transient expression of p33-Flag did not affect the expression of the WT p33 and p92 pol ( Fig. 6C ), and it enhanced TBSV repRNA accumulation (fig. S9B). Therefore, it is likely that the aggregation-prone p33-SIM2-m4 has a dominant negative effect on the WT p33 and p92 pol accumulation leading to their elimination from N. benthamiana cells. On the other hand, p33-SIM2-m4 expression had minor effects on the expression of the TBSV coat and movement proteins (fig. S10, A to D) or the heterologous peroxisome-targeted Flag-RFP-SKL and the cytosolic GFP proteins (fig. S10, E and F), indicating that p33-SIM2-m4 has dominant negative effect specifically on p33 and p92 pol . Protein stability assay showed that p33-SIM2-m4 protein was unstable, unlike the WT p33 in plant cells ( Fig. 6D ). Expression of the aggregation-prone p33-SIM2-m4 in the presence of VRO-forming p33-ATeam reduced the production of ATP within the VROs when compared with the control ( Fig. 6E ). This indicates that the vir-condensates containing the co-opted glycolytic enzymes are less functional in the presence of the aggregation-prone p33-SIM2-m4. To test whether CHIP is involved in elimination of p33-SIM2-m4 mutant, we knocked down CHIP expression in N. benthamiana . The CHIP KD plants supported ∼5-fold increased accumulation of p33-SIM2-m4 mutant than the control nonsilenced plants ( Fig. 6F , lanes 6 to 10 versus 1 to 5). This is in contrast to the similar accumulation of WT p33 in CHIP KD or nonsilenced plants ( Fig. 6G ). Moreover, transient expression of CHIP reduced the accumulation of the aggregation prone p33-SIM2-m4 mutant ( Fig. 6H , lanes 2 and 4 versus 1 and 3), whereas it enhanced the accumulation of WT p33-Flag ( Fig. 6H , lanes 5 to 8). We have confirmed the role of CHIP in elimination of mutant p33 by studying a second mutant, p33-Sm5, carrying five separate mutations that reduce p33 sumoylation, TBSV replication ( 36 ), and CIRV replication (fig. S9A). Expression of TBSV eGFP-p33-Sm5 frequently leads to small puncta formation in VROs (fig. S11A) and reduces ATP production in VROs (fig. S11B), thus acting as a dominant negative mutant. CHIP KD increased p33-Sm5 accumulation (fig. S11C), whereas CHIP expression reduced p33-Sm5 amount in plant cells (fig. S11D). Similarly, the accumulation of the aggregation-prone Killswitch-p33 also increased by ∼3-fold in CHIP KD plants (fig. S12). Thus, the accumulation of aggregation-prone p33 mutants (i.e., p33-SIM2-m4 mutant and p33-Sm5) and Killswitch-p33 protein is regulated by CHIP E3 ligase in plants. Unexpectedly, transient expression of CHIP reduced the accumulation of TBSV p92 pol , whereas CHIP KD enhanced it (fig. S13, A to C), indicating that p92 pol is targeted for degradation. However, coexpression of p33 replication protein restored p92 pol accumulation when CHIP was expressed (fig. S13D), suggesting that CHIP could be involved in removal of surplus p92 pol RdRp that is not in complex with p33 from cells. Note that the N-terminal segment of p92 pol RdRp overlaps with p33 (fig. S13A), and they interact in the TBSV replicase complex ( 50 ). In sum, these data suggest that the co-opted CHIP E3 ligase is involved in elimination of mutated aggregation-forming p33 from VROs. Thus, a major new emerging function of co-opted CHIP E3 Ub-ligase is to provide PQC within the TBSV VROs that is required for proper assembly and function of vir-condensate and the entire VROs.

(+)RNA viruses concentrate replication factors in VROs, which are specialized subcellular compartments containing subverted membranes and dozens of co-opted host proteins to facilitate efficient RNA replication. TBSV p33 replication protein not only is involved in the assembly of the membrane-bound viral replicase and formation of VROs but also binds to and sequesters host proteins in a unique vir-condensate substructure within the membranous VROs ( 35 ). To gain further insight into the functions of the vir-condensate, we used approaches to “harden” the condensate and reduce protein mobility that likely could interfere with its proper functions. Targeting of the killswitch micropeptide via a nanobody into VROs made the vir-condensate less dynamic, based on the formation of large punctate structures (inclusions) within VROs and the slowed recovery of fluorescent signals by the co-opted host proteins, which are clients in vir-condensate ( 35 ). In addition, the targeted killswitch micropeptide caused the aggregation of p33 that originally forms droplets via liquid-liquid phase separation in vitro. These altered features of the vir-condensate are likely due to the known property of the killswitch micropeptide to contribute additional protein (self)interactions within the condensate structures that reduce protein mobility ( 44 ). The BFP tag fused to p33 replication protein also led to several puncta formations in VROs and reduced FRAP signals of the fluorescently tagged co-opted host proteins, indicating altered protein interactions and less dynamic vir-condensate. Both the killswitch approach and the BFP-tagged p33 resulted in reduced ATP production by the co-opted glycolytic enzymes in VROs and inhibition of TBSV replication. The aggregation-prone p33-BFP has provided us a tool to demonstrate how TBSV uses CHIP E3 Ub-ligase as a regulator of protein quality in VRO and anti-p33 aggregation system during infection. Moreover, the findings with p33-BFP provide independent support for the conclusions drawn from the nbALFA-Killswitch experiments. This is important to exclude that CHIP is not directly targeting Killswitch micropeptide, but CHIP targets the aggregated/misfolded proteins in VROs for degradation. Aggregation-inducing mutations in p33 also changed droplet formation by p33 in vitro and had dominant negative effects on TBSV replication. These observations support the model that the dynamic vir-condensate structure is critical for tombusvirus replication, whereas protein aggregations are detrimental for vir-condensate functions. Moreover, this highlights the opportunity to inhibit virus infections by changing the molecular properties of the vir-condensate. Altering the molecular dynamics in cellular condensates has been shown to affect several cellular processes ( 44 , 63 ). For example, reducing protein mobility within cellular condensates has been shown to reduce proliferation of condensate-driven leukemia cells, suppress viral capsid assembly, alter ribosomal protein mobility in nucleoli, and inhibit microRNA transcription in transcription bodies ( 44 ). Drug-based hardening of condensates formed by negative-strand RNA viruses inhibited virus replication ( 64 ). Artificial molecular condensates might be effective against infecting viruses by sequestering critical host factors away from virus replication ( 65 ). Together, targeting the molecular properties of virus-induced condensates is emerging as novel antiviral strategies to combat viruses. The above model on the critical function of the dynamic vir-condensate also opened a new question: How can the dynamic nature of the vir-condensate be maintained during virus replication? The TBSV p33 replication protein self-interacts to form oligomers ( 50 , 66 ). In addition, p33 serves as a scaffold protein in vir-condensates, where it also binds to dozens of co-opted host proteins, several of which serve as client proteins in vir-condensates ( 49 , 50 , 67 – 69 ). The numerous protein-protein interactions within the VROs could favor the formation of nonfunctional protein aggregates/inclusions under certain conditions, which would limit virus replication. Accordingly, mutations in p33 replication protein caused aggregations in vitro ( Fig. 6 ) ( 36 ), suggesting that p33 is aggregation prone. Yet, we observed only very few punctate structures formed by GFP- or RFP-tagged p33 or the co-opted cytosolic proteins in VROs in plant cells, indicating the existence of a PQC mechanism that “guards” the vir-condensates to become nonfunctional aggregates/inclusions. Accordingly, in this work, we have identified the CHIP E3 Ub-ligase as the core factor that drives the elimination of the aggregated p33 replication protein from plant cells. Accordingly, CHIP expression reduced the dominant negative effect of VRO-targeted killswitch micropeptide or the aggregation-prone p33-BFP on TBSV replication. On the contrary, knocking down CHIP expression in plant cells resulted in enhanced accumulation of large puncta-forming p33-BFP or aggregation-prone p33 mutants. CHIP KD also reduced ATP production by the co-opted glycolytic enzymes and reduced the recovery of fluorescent signals of co-opted host proteins, which are client proteins in vir-condensate. All these data indicate the critical role of CHIP in regulating protein interactions and stability and maintaining dynamic vir-condensate. This function of CHIP might be relevant to remove the protein aggregations/inclusions caused by the expression of aggregation-prone mutant p33 (p33-SIM2-m4 and p33-Sm5), p33-BFP, or the killswitch micropeptide-targeted p33 in the VROs. The U-box domain with E3 Ub-ligase function is needed to remove the mutant aggregation-prone p33 from cells. Similar to the human microtubule-binding tau protein ( 70 ), TBSV p33 protein seems to bind to CHIP via the TPR domain independent of Hsp70 chaperone activity. We propose that TBSV exploits the co-opted CHIP to provide PQC within the VROs that consist of p33 decorated membranes and highly enriched co-opted proteins in the vir-condensates. The need for PQC mechanism is a previously unknown phenomenon during (+)RNA virus replication. It may be needed because the assembly of large membranous structure with vir-condensate and actin filaments ( 35 ) is a complex process involving many aggregation-prone proteins present in high concentrations. However, whether CHIP could target other co-opted host proteins in VROs will be explored in the future. TBSV-driven subversion of CHIP E3 ligase might be especially suitable for the above proviral function due to the crucial role of CHIP in protein degradation and PQC, helping the maintenance of proteostasis in the noninfected cells ( 57 , 71 ). The role of CHIP in removal of misfolded and aggregation-prone proteins has also been demonstrated for neurotoxic proteins, such as tau, alpha-synuclein, or amyloid-β, and insoluble plant proteins during heat or oxidative stress ( 12 , 13 , 71 – 73 ). Together, it is remarkable that the CHIP E3 Ub-ligase centric PQC is exploited by tombusviruses to remove aggregated viral replication proteins from VROs in a manner reminiscent of the CHIP/STUB1-mediated removal of disease-causing protein aggregates in mammals and hyperphosphorylated tau from brain cells ( 4 , 70 ). CHIP is documented to affect viral infections by degrading immune signaling components, such as RIG-I and MAVS ( 74 , 75 ). Bovine coronavirus–coded nsp14 recruits CHIP to ubiquitinylate TRAF3 and inhibit interferon production ( 76 ). CHIP also affects virus replication of plant potyvirus (potato virus A) through regulation of the accumulation of the coat protein and CP degradation ( 77 ). HIV-1 Tat regulatory protein, which has intrinsic disordered nature, and the Vif protein are degraded through CHIP-driven ubiquitination, affecting HIV-1 pathogenicity ( 78 ). Senecavirus A VP1 protein is targeted for degradation by STUB1/CHIP and this antiviral activity of CHIP is dampened by cleavage performed by the viral-coded 3C protease ( 79 ). Thus, the highly conserved CHIP seems to be at the core of regulating multiple virus infections. Another aspect of orchestrating PQC within VROs is the reduction of possible damage caused by the accumulation of misfolded p33 to the host cell. It is documented that high accumulation of misfolded proteins could be highly toxic due to proteotoxic interactions with critical cellular factors ( 9 , 71 ). For example, improper accumulation of misfolded p33 aggregates might permanently trap the PQC components, such as Ub, protein chaperones, and UPS/autophagy components that would induce detrimental pathological changes in the host cells. These factors would lead to rapid cellular death, ultimately limiting virus spread in plants. Therefore, inducing and subduing the CHIP-based PQC by TBSV is likely beneficial for virus replication by limiting the proteotoxic effects of virus infections on the host. Together, tombusviruses need to co-opt CHIP E3 ligase to prevent p33 protein aggregations and maintain dynamic vir-condensates. A major emerging function of co-opted CHIP E3 Ub-ligase is to provide PQC within the TBSV VROs that is required for proper assembly of the VROs.

The proper assembly and disassembly of large protein complexes, including intracellular biomolecular condensates, also called membraneless organelles, is regulated by protein quality control (PQC) mechanisms to maintain protein homeostasis in cells. PQC facilitates chaperone-mediated protein refolding or the degradation of misfolded proteins via the ubiquitin (Ub)-proteasome system or autophagy ( 1 , 2 ). Improper functioning of the PQC machinery could lead to misfolded protein aggregations/inclusions that might result in various disease states, such as neurodegenerative disorders, including Alzheimer’s, Huntington’s, and Parkinson’s diseases ( 1 , 3 ). PQC components are frequently enriched in condensates that sequester many proteins to compartmentalize biological reactions. However, dysregulation of condensates could lead to aggregations, leading to disease-associated processes ( 2 , 4 ). Viruses also induce the formation of condensates to facilitate viral processes, such as replication or virion assembly ( 5 , 6 ). However, the role of PQC factors in viral condensates is understudied. CHIP/STUB1 is an E3 Ub-ligase involved in PQC ( 7 ). CHIP has three conserved domains: 3× tandem tetratricopeptide repeat (TPR) domain at the N terminus, a middle charged coiled-coil domain, and a C-terminal U-box domain ( 8 ). CHIP promotes the degradation of prematurely terminated proteins, aggregation-prone mutant proteins, and a diverse range of cellular proteins ( 9 – 11 ). The role of CHIP in removal of misfolded neurotoxic proteins, such as tau, alpha-synuclein, or amyloid-β, has also been demonstrated ( 12 , 13 ). CHIP plays critical roles in many processes, including aging, autophagy, cancer, autoimmunity, and antiviral immune response ( 7 , 14 , 15 ). CHIP regulates glycolysis in human diseases (endometriosis) by ubiquitinating PFKFB4 glycolytic enzyme ( 16 ). CHIP E3 Ub-ligase has been established as a core component in the PQC machinery ( 9 – 11 ). Many plant and animal viruses have positive-strand (+)RNA genomes and are important pathogens, such as severe acute respiratory syndrome coronavirus 2 and flaviviruses, causing numerous diseases in humans and significant losses in agriculture. (+)RNA viruses have to co-opt numerous host resources, including proteins and cellular membranes and rewire cellular pathways to support their infections. A central step is virus replication, which occurs in unique viral replication organelles (VROs) ( 17 – 21 ). The VROs consist of subverted subcellular/organellar membranes that not only contribute optimal microenvironment for progeny RNA synthesis but also provide protection from recognition and destruction by the host antiviral machinery ( 17 , 19 , 21 – 25 ). VRO biogenesis is a complex and coordinated process driven by viral replication proteins in coordination with co-opted host proteins. Tomato bushy stunt virus (TBSV), which belongs to the large flavivirus-like supergroup, has extensively been studied to unravel how (+)RNA viruses reprogram host machineries to support viral infections ( 19 , 26 , 27 ). TBSV research has led to dissection of virus-induced complex rearrangements of cellular membranes and alterations in lipid and other metabolic processes during infections ( 28 – 31 ). TBSV has a ∼4.8-kb single-component (+)RNA genome that codes for five proteins. The two essential replication proteins are the abundant p33 and p92 pol RdRp protein. P92 pol is a read-through product of the p33 ORF. P33 replication protein serves as a master regulator of VRO biogenesis. In addition, p33 has RNA chaperone activity involved in selective recruitment of the viral (+)RNA for replication ( 32 , 33 ). Tombusviruses hijack either peroxisomes in case of TBSV or mitochondria by the closely related carnation Italian ringspot virus (CIRV). Similar to many (+)RNA viruses, the TBSV VRO membrane has vastly modified lipid compositions that facilitate the formation of numerous invaginated spherules, the sites of viral RNA synthesis ( 28 , 34 ). Recent work revealed that TBSV p33 and CIRV p36 replication proteins induce phase-separation sequestering several co-opted host proteins in unique condensate substructures, called virus-induced replication (vir)–condensates, which associate with membranous VROs ( 35 ). TBSV p33 has an intrinsically disordered region (IDR) ( 35 ) and becomes sumoylated by the co-opted host sumoylation machinery ( 36 ), which facilitates condensate formation in vitro and biogenesis of vir-condensate during TBSV infection of plant cells. The vir-condensate provides proviral functions ( 35 , 37 ), including in situ adenosine triphosphate (ATP) production by co-opted glycolytic and fermentation enzymes ( 30 , 38 – 40 ). The vir-condensate also contains the co-opted SUMO enzymes ( 36 ) and regulatory particle non-ATPase 11 (RPN11), a metalloprotease subunit of the 26 S proteasome lid ( 35 , 41 , 42 ), and it also traps antiviral factors, such as exportin 1 (XPO1) and its cofactors ( 37 , 43 ). Together, the C terminus of p33 acts as the scaffold factor, whereas the above co-opted host factors interacting with p33 are “client” proteins in vir-condensate, which is associated with the membranous VROs. Here, we targeted a micropeptide termed “killswitch” to VROs to enhance p33 self-association and alter the material properties of p33-induced vir-condensates. Killswitch is a hydrophobic, aromatic amino acid–rich micropeptide that self-associates, and it has been developed as a tool to selectively manipulate the microenvironment in condensates ( 44 ). When targeted to a condensate-forming protein, killswitch could probe the function of condensate by changing the dynamics of molecular interactions within the condensate. We find that the targeting of killswitch to TBSV p33 induced large aggregates/inclusions in VROs that led to significant inhibition of TBSV replication. The killswitch-driven solidification of the vir-condensates reduced the mobility of key co-opted host factors in vir-condensate, thus limiting in situ ATP accumulation within VROs. These findings support an emerging strategy of targeting virus replication–induced biomolecular condensates as a potential approach for future crop protection. We also show that mutations in p33 replication protein cause protein aggregation in vitro, suggesting that p33 could be prone to forming nonfunctional protein aggregations/inclusions under some conditions. However, we discovered that TBSV co-opts CHIP/STUB1 E3 Ub-ligase involved in PQC to reduce the detrimental effect of protein aggregation. We propose that the phase-separation–inducing features of p33 might render p33 aggregation prone under some conditions and the host-coded CHIP maintains p33 functions by eliminating aggregated p33 molecules. The need for hijacking PQC mechanism is a a previously undocumented phenomenon during (+)RNA virus replication.

N. benthamiana is one of the most widely used model hosts in plant virology because it is highly susceptible to a broad range of plant viruses, including tombusviruses, and it is particularly amenable to Agrobacterium -mediated transient protein expression and VIGS-based gene expression knockdowns ( 61 ). Unlike mammalian cell culture systems, protein expression in N. benthamiana is typically achieved by agroinfiltration of leaf tissue, which enables rapid and efficient expression of recombinant proteins in intact plant cells. This approach allows the analysis of VROs, condensate formation, and virus-host interactions in the native cellular environment of a plant host ( 67 , 80 – 82 ). The TRV-based VIGS in N. benthamiana was conducted as previously described ( 61 ). To generate the VIGS construct TRV2-NbCHIP, a fragment of NbCHIP was PCR amplified from N. benthamiana cDNA with primers #7404 and #7405, digested with Bam HI and Xho I restriction enzymes, and subsequently inserted into the pTRV2 plasmid treated with Bam HI and Xho I. Agrobacterium strain C58C1 carrying TRV1 was mixed with agrobacteria harboring TRV2-cGFP (control) or TRV2-NbCHIP vectors (OD 600 of 0.5 for each), and the mixture was infiltrated into N. benthamiana plants. Twelve days after VIGS treatment, plant phenotypes were documented, and upper leaves were harvested for total RNA extraction. cDNA was synthesized by reverse transcription using oligo(dT) primers. Transcript levels of NbCHIP and Tubulin (used as an internal reference) were analyzed by semiquantitative PCR with primers #7402/#7403 and #8093/#8094, respectively, or by real-time qPCR using primers #9165/#9166 and #8178/#8179. The silenced leaves were subsequently inoculated with sap containing TBSV, CIRV, or CNV20Kstop (a CNV mutant carrying a stop codon that blocks p20 translation, a suppressor of gene silencing protein) to initiate viral infection. Samples from the inoculated leaves were collected at 1.5, 2, and 2.5 days postinoculation in case of TBSV, CIRV, and CNV20Kstop, respectively. Total RNA extraction and DIG-based Northern blot analysis were performed as described previously to determine virus accumulation ( 35 ). To visualize the subcellular localization of NbCHIP in plant leaves, the NbCHIP coding sequence was PCR amplified from N. benthamiana cDNA using primers #9180/#9181. The PCR product was digested with Bam HI and Xho I and ligated into the pGD-RFP vector digested with Bam HI and Sal I, generating an RFP-NbCHIP fusion construct. In addition, the constructs pGD-p33-BFP, pGD-eGFP-p33, pGD-eGFP-SKL (peroxisome marker), pGD-BFP-SKL (peroxisomal marker), pGD-p19 (gene silencing suppressor encoded by TBSV), and pGD-RFP-NbCHIP were individually transformed into Agrobacterium strain C58C1. Agrobacterium suspensions with different combinations of these constructs (OD 600 of 0.3 for each) with or without viral components were infiltrated into N. benthamiana leaves. pGD-p19 (OD 600 of 0.1) was coinfiltrated with all these combinations to enhance protein expression. At 2.5 days postagroinfiltration (dpai), the agroinfiltrated leaves were examined using confocal laser microscopy (Olympus FV3000). BFP, GFP, and RFP were excited at 405, 488, and 561 nm, respectively, and emissions were detected in the ranges of 430 to 470, 495 to 535, and 580 to 620 nm. Z -stack imaging was used to obtain a comprehensive view of the distribution of fluorescently labeled proteins within cells. More than three independent plant leaves were analyzed. Because p33-BFP was degraded in the presence of RFP-NbCHIP when coexpressed in plant cells, a sequential agroinfiltration procedure was also used. Agrobacterium cultures carrying pGD-p33-BFP and pGD-eGFP-SKL were coinfiltrated into plant leaves, followed by infiltration of pGD-RFP-NbCHIP at 1 dpai. Agroinfiltrated leaves were examined by confocal laser microscopy at 3 dpai. To investigate the negative roles of aggregates/inclusions formed within VROs, the ALFA-tag/nbALFA–killswitch system was used to promote p33 aggregation. In this system, the ALFA tag (SRLEEELRRRLTE) is specifically recognized and bound by the ALFA nanobody (nbALFA) ( 45 ), whereas the killswitch micropeptide is a hydrophobic, aromatic-rich sequence capable of self-association, thereby strengthening intermolecular interactions and altering the material properties of endogenous condensates ( 44 ). Agrobacterium cultures carrying pGD-eGFP-p33-ALFA, pGD-RFP-p33-ALFA, or pGD-eGFP-p33 (control) were coagroinfiltrated with either pGD-3myc-nbALFA-Killswitch or pGD empty vector (control) (OD 600 of 0.3 for each) into N. benthamiana leaves using a half-leaf assay. pGD-p19 (OD 600 of 0.1 for each) was coagroinfiltrated with all combinations to enhance protein expression. To test whether NbCHIP could prevent p33 aggregation, Agrobacterium cultures carrying pGD-eGFP-p33-ALFA (OD 600 of 0.2) and pGD-3myc-nbALFA-killswitch (OD 600 of 0.4) were coagroinfiltrated with either pGD-RFP-NbCHIP or pGD-EV (control) (OD 600 of 0.4) into N. benthamiana leaves using a half-leaf assay. pGD-p19 (OD 600 of 0.1) was coagroinfiltrated with all combinations to enhance protein expression. Confocal analysis was performed at 2.5 dpai. To examine the effect of different fluorescent tags on p33 behavior within viral VROs, C-terminally BFP- or eGFP-fused p33 and N-terminally eGFP- or RFP-fused p33 (OD 600 of 0.3 for each) were transiently expressed in plant leaves and analyzed by confocal microscopy at 2.5 dpai. To determine whether p33-BFP is subject to NbCHIP-mediated regulation, Agrobacterium cultures carrying pGD-p33-BFP, pGD-L-p92, and pGD-L-DI72 (OD 600 of 0.3 for each) were coagroinfiltrated after VIGS-based silencing with either TRV-MBP or TRV-NbCHIP plants. Three leaf discs from three independent leaves were examined. To evaluate the VROs formed by p33 and p33-SIM2-m4 and p33-Sm5 mutants, Agrobacterium cultures carrying pGD-eGFP-p33 or pGD-eGFP-p33-SIM2-m4 or pGD-eGFP-p33-Sm5 (OD 600 of 0.3 for each) were coagroinfiltrated with pGD-RFP-SKL and pGD-p19 (OD 600 of 0.1 for each) into plant leaves. The samples were analyzed by confocal microscopy at 2.5 dpai. To investigate the mobility of NbCHIP within vir-condensates, FRAP assays were performed as described previously ( 35 ), with minor modifications. Briefly, the entire VRO region in agroinfiltrated N. benthamiana leaves was subjected to photobleaching. Full bleaching was carried out using a 405-nm laser at 60 to 80% intensity for 3 s. Time-lapse images were acquired every 10 s for 1 min. Six to eight VROs labeled with RFP-NbCHIP and eGFP-p33 were analyzed per experiment, with at least three independent leaves examined. Images collected after FRAP were analyzed using the ROI Manager tool in ImageJ. Fluorescence intensities of photobleached regions were quantified, and recovery curves were plotted using GraphPad Prism 9. To assess the effect of p33 aggregation on condensate dynamics within VROs, the p33-ALFA/nbALFA–killswitch system and p33-BFP were included in FRAP assays. For the p33-ALFA/nbALFA–killswitch system, Agrobacterium cultures carrying pGD-RFP-p33-ALFA and pGD-BFP-NbADH1 or pGD-BFP-NbRpn11 were coagroinfiltrated with either pGD-3myc-nbALFA–killswitch or pGD-EV (control) (OD 600 of 0.3 for each) into N. benthamiana leaves through a half-leaf assay. pGD-p19 was coagroinfiltrated with all combinations to enhance protein expression. For p33-BFP assay, Agrobacterium cultures carrying pGD-p33-eGFP and pGD-RFP-NbADH1 or pGD-RFP-NbPK were coagroinfiltrated with either pGD-p33-BFP or pGD-EV (control) (OD 600 of 0.3 for each) into N. benthamiana leaves through a half-leaf assay. FRAP imaging and analysis were performed as described above. All recombinant proteins used in this study were purified from E. coli BL21(DE3) cells as described previously ( 35 ) and in the Supplementary Materials. Droplet formation assays were performed to assess aggregation of eGFP-p33C induced by the ALFA/nbALFA–killswitch system and by p33-SIM2-m4. Briefly, 5 μM eGFP-p33C-ALFA and 2 μM RFP-nbALFA–killswitch were mixed in a 40-μl reaction volume containing 30 mM tris-HCl (pH 7.5), 150 mM NaCl, and PEG-8000 at the indicated concentrations (2.5 or 5%). As a control, 5 μM eGFP-p33C-ALFA or 2 μM RFP-nbALFA–killswitch alone was diluted in the same reaction buffer. In parallel, 2.5 μM eGFP-p33C or eGFP-p33C-SIM2-m4 was mixed with 1 μM RFP-AtSUMO1*-p33C in a 40-μl reaction volume containing 30 mM tris-HCl (pH 7.5), 150 mM NaCl, and 5% PEG-8000. As a control, 2.5 μM eGFP-p33C, 2.5 μM eGFP-p33C-SIM2-m4, or 1 μM RFP-AtSUMO1*-p33C alone was diluted in the same reaction buffer. Protein mixtures were incubated in 384-well glass-bottom plates (Cellvis) at room temperature, and images were acquired randomly from the field of view 20 to 50 min after incubation. Additional methods are described in the Supplementary Materials, including the list of primers and templates used for overlapping PCR together with details of all other plasmid constructions in tables S1 to S3.