Sugarcane mosaic virus remodels multiple intracellular organelles to form genomic RNA replication sites

Positive-stranded RNA viruses usually remodel the host endomembrane system to form virus-induced intracellular vesicles for replication during infections. The genus Potyvirus of the family Potyviridae represents the largest number of positive single-stranded RNA viruses, and its members cause great damage to crop production worldwide. Although potyviruses have a wide host range, each potyvirus infects a relatively limited number of host species. Phylogenesis and host range analysis can divide potyviruses into monocot-infecting and dicot-infecting groups, suggesting that they differ in their infection mechanisms, probably during replication. Comprehensive studies on the model dicot-infecting turnip mosaic virus have shown that the 6K2-induced replication vesicles are derived from the endoplasmic reticulum (ER) and subsequently target chloroplasts for viral genome replication. However, the replication site of monocot-infecting potyviruses is unknown. In this study, we show that the precursor 6K2-VPg-Pro polyproteins of dicot-infecting potyviruses and monocot-infecting potyviruses cluster phylogenetically in two separate groups. With a typical gramineae-infecting potyvirus—sugarcane mosaic virus (SCMV)—we found that replicative double-stranded RNA (dsRNA) forms aggregates in the cytoplasm but does not associate with chloroplasts. SCMV 6K2-VPg-Pro-induced vesicles colocalize with replicative dsRNA. Moreover, SCMV 6K2-VPg-Pro-induced structures target multiple intracellular organelles, including the ER, Golgi apparatus, mitochondria, and peroxisomes, and have no evident association with chloroplasts.


Introduction
During positive-strand RNA virus replication, the host endomembrane system is usually rearranged and contributes to the formation of virus-induced intracellular membranous vesicles providing scaffold for anchoring of virus replication complexes (VRCs) including viral proteins, RNA and host factors crucial for virus replication [1][2][3]. Diverse intracellular organelles or cellular comparts are subverted and modi ed by distinct viruses infecting plants and animals [4][5][6]. In particular, endoplasmic reticulum (ER) is widely targeted by various viruses to induce invagination of its outer membrane and form spherules or vesicles [7][8][9][10][11][12][13]. Besides ER, chloroplasts, mitochondria, peroxisomes, tonoplasts, nuclear membrane, and plasma membrane can also be rearranged to form membrane-bound VRCs in infected cells, indicating that these cellular membrane compartments are replication sites of corresponding viruses [14][15][16][17][18][19]. These membranous vesicles are thought to function in concentrating viral components and con ne the virus replication in a speci c compartment to escape host defense response and thereby to support the e cient replication of viruses [6,20].
The genus Potyvirus contains agriculturally and economically important members causing great damage on many cultivated plant species including monocots and dicots [21,22]. The potyviral genome is a positive-sense single-stranded RNA of approximately 10 kb in length that encodes a large polyprotein and subsequently being processed into 10 mature proteins [22][23][24]. Additionally, a P3N-PIPO resulted from viral RNA polymerase slippage is expressed within P3 cistron [25]. Of these viral proteins, 6K2 is an integral membrane protein that induces the formation of replication vesicles [8]. The 6K2-VPg-Pro polyprotein of TuMV is demonstrated to be responsible for the induction of ER-derived cytoplasmic vesicles housing VRCs for virus replication by nding the presence of viral components of RNAdependent RNA polymerase (RdRp), 6K1, P3, CI, VPg-Pro and double-stranded viral RNAs [26][27][28][29][30]. Various host factors are recruited to 6K2-VPg-Pro-induced replication vesicles for viral infection [31,32], further suggesting that the 6K2-induced ER-derived vesicles are the sites for potyviral replication. However, potyviruses may also replicate in chloroplasts by evidences that viral RNAs of tobacco etch virus (TEV) and potato virus Y (PVY) were found in chloroplasts [33,34]. Further studies precisely and clearly present evidence that 6K2-induced vesicles accumulate on ER membrane and migrate to Golgi apparatus, then target chloroplast envelope for TuMV genome replication [35,36], indicating that both ER and chloroplasts are utilized for TuMV replication.
Previously, phylogenetic analysis of polyproteins encoded by potyviral genome sequences showed that monocot-infecting potyviruses clustered separately from that of dicot-infecting potyviruses [37,38], suggesting a correlation between potyviral genomic sequences and host ranges. Considering this phylogenetic difference, the replication mechanisms of dicot-and monocot-infecting potyviruses may be distinct. Plenty of evidences have supported that both ER and chloroplasts are targeted by several dicotinfecting potyviruses for replication [7,8,[33][34][35], while the replication sites for monocot-infecting potyviruses remain unclear.
Maize (Zea mays) is one of the most important and largest staple food crops in the world [39,40]. The potyvirus sugarcane mosaic virus (SCMV) is a prevalent viral pathogen inducing maize dwarf mosaic disease worldwide [41,42]. Beside maize, SCMV can infect several important monocot crops such as sorghum (Sorghum vulgare) and sugarcane (Saccharum sinensis) [41,42]. Identi cation and characterization of the replication sites of SCMV would assist in precise understanding the virus replication process and its pathogenesis. In this study, we provide evidences that SCMV 6K2-VPg-Pro polyprotein-induced cytoplasmic vesicles are the virus replication sites that target multiple intracellular organelles.

Phylogenetic analysis
The amino acid sequences of potyviral 6K2-VPg-Pro were downloaded from the GenBank database in National Center for Biotechnology Information and then subjected to multiple sequence alignment and phylogenetic analysis using MAFFT online [43]. The phylogenetic tree was constructed with the neighborjoining method. The phylogy was tested by 500 bootstrap resampling, poisson model amino acid substitutions type.

Plasmid construction
The coding regions of 6K2, NIa-VPg, and NIa-Pro were PCR-ampli ed using SCMV-BJ infectious clone as the template [44] with primers that were designed to modify the cleavage sites at the junction of 6K2 and NIa-VPg, NIa-VPg and NIa-Pro to prevent polyprotein proteolysis (Supplementary Table 1) [45]. PCR products of 6K2, NIa-VPg and NIa-Pro were then fused together by overlap PCR to obtain 6K2-VPg-Pro, which was then inserted into pGD-mCherry and pGD-EGFP vectors [46] by Infusion enzyme (Takara, Kyoto, Japan) ligation to produce constructs pGD-6K2-VPg-Pro-mCherry and pGD-6K2-VPg-Pro-EGFP. The resulting constructs were con rmed by DNA sequencing.

Plant growth conditions and virus inoculation
Maize inbred line B73 and Nicotiana benthamiana plants were grown in growth champers under 16 h light at 24 o C/8 h dark at 22 o C conditions. The rst true leaves of 8-day-old maize B73 seedlings were rubinoculated with fresh crude extracts from plants infected by SCMV-BJ as previously described [47,48].
Plants mock-inoculated with phosphate buffer (0.01 M) were used as controls.

Maize protoplasts isolation and transfection
Maize seeds were inoculated with fresh crude extracts from SCMV-BJ-infected leaves or mock-inoculated via vascular puncture, and the germinated seedlings were kept in dark at 24 o C to obtain etiolated plants [49,50]. Maize protoplasts isolation and transfection were conducted as previously described [49,51].

Particle bombardment assay
For particle bombardment assay, mock-inoculated or SCMV-BJ-infected maize plants which were sapinoculated at 3-leaf-stage were allowed to grow to 9 leaf-stage in a greenhouse. The particle bombardment assay was performed as described previously with some modi cations [52]. The prepared microcarriers (50 μL) were mixed with 2.5 μg plasmids, 20 μL of 0.1 M spermidine and 50 μL of 2.5 M CaCl 2 . After vortex for 3 min, the resulting mixture was centrifuged at 10000 g for 20 s. The pellet was resuspended and washed with 140 μL 70% ethanol twice. Finally, the pellet was resuspended by 10 μL absolute ethanol for bombardment. The bombardment assay was performed with a PDS-1000/He system (Bio-Rad) as instructions. Tissue helium pressure at the tank was regulated at 1,500 pounds per square inch. The bombarded maize leaf tissues were kept in dark conditions at 25 o C for 18-24 h until confocal microscopy analysis.

Agrobacterium-mediated transient expression in N. benthamiana
Plasmids for agroin ltration were introduced into Agrobacterium tumefaciens strain GV3101 by freezethaw method [53]. The bacteria were then cultured overnight, centrifuged to obtain pellet, and nally resuspended with MMA buffer containing 10 mM MgCl 2 , 10 mM MES, pH 5.6, and 100 mM acetosyringone. The resuspended bacteria were then diluted to OD 600 =0.5, followed by incubation at room temperature for 2-3 h until in ltration into the abaxial face of leaves of 5-or 6-leaf-stage N. benthamiana plants. To enhance protein expression, agrobacteria harboring a plasmid expressing the RNA silencing suppressor p19 protein of tomato bushy stunt virus (TBSV) were also in ltrated simultaneously [54].

Confocal microscopy analysis
Agroin ltrated N. benthamiana leaves, transfected maize protoplasts and bombarded leaf tissues were analyzed using a confocal laser-scanning microscopy (Leica TCS SP8). The uorescence of both EGFP and YFP was excited at 488 nm, mCherry and chloroplasts auto uorescence was excited at 552 and 638 nm, respectively. A line sequential scanning mode was used to avoid uorescence across.

Western blot analysis
Total protein of maize leaf tissues and protoplasts were extracted as previously described [48]. The protein extracts were separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by transferring proteins to PVDF membranes (Millipore, Massachusetts, USA). The blotted membranes were then separately incubated with anti-SCMV CP [55] and anti-actin (CW0264, CWBIO, Beijing, China) antibodies with a dilution of 1:5000 for 1 h, and nally detected chemiluminescence with eECL Western Blot Kit according to manufacturer's protocol (CW0049, CWBIO, Beijing, China).

6K2-VPg-Pro polyproteins of monocot-infecting potyviruses phylogenetically separate from that of dicotinfecting potyviruses
To investigate whether the replication site of monocot-infecting potyviruses is same or distinct to that of dicot-infecting potyviruses, we rstly conducted a multiple sequence alignment of potyviral 6K2-VPg-Pro polyprotein amino acid sequences (Supplementary Fig. 1) and performed a phylogenetic analysis (Fig. 1) using the neighbor-joining method implemented with MAFFT online [43]. Multiple sequences alignment showed that the 6K2-VPg-Pro polyproteins share 58.55% identities among the representative 20 potyviruses, and there are 54 absolutely conserved residues for all of them ( Supplementary Fig. 1).

SCMV replication sites do not correlate with chloroplasts
Double-stranded RNA (dsRNA) is regarded as the hallmark of virus infection [56]. For positive singlestranded RNA viruses, replicative RNA intermediates, namely viral dsRNA is formed during virus replication, whose localization is reasonably thought to be the site of virus replication. To label the replication sites of SCMV in maize cells, we performed the dsRNA binding-dependent uorescence complementation (dRBFC) assay by co-expressing B2-nYFP and VP35-cYFP in SCMV-infected maize cells following a previous report [36]. In this assay, maize protoplasts isolated from mock-or SCMVsystemically infected leaves (see methods) were transfected with plasmids expressing B2-nYFP and VP35-cYFP. Confocal microscopic analysis showed that uorescent punctate structures indicating SCMV dsRNA localization were observed in cytoplasm of SCMV-infected maize protoplasts ( Fig. 2A), and the reconstituted YFP signals have no obvious association with chloroplasts auto uorescence ( Fig. 2A). No uorescent signal was observed in mock-infected maize protoplasts ( Fig. 2A). Western blotting con rmed the infection of SCMV in maize protoplasts (Supplementary Fig. 2A).
Moreover, we performed the dRBFC assay in maize leaves by particle bombardment. The upper maize leaves were collected from healthy and SCMV-systemically infected plants at 9-leaf stage, and subjected to bombardment. The dsRNA-reconstituted YFP uorescent signals were observed in cytoplasm but did not merge with the red auto uorescence of chloroplasts in SCMV-systemically infected maize leaf cells (Fig. 2B). No YFP uorescence was observed in healthy leaf cells. Western blotting con rmed the infection of SCMV in maize leaves for bombardment ( Supplementary Fig. 2B).
The 6K2-VPg-Pro-induced cytoplasmic vesicles are replication sites of SCMV Studies on dicot-infecting potyviruses, such as TEV and TuMV, demonstrated that the 6K2-VPg-Pro polyprotein can induce the formation of cytoplasmic vesicles, which were further shown to be their replication sites [8,26,35]. To determine whether SCMV-encoded 6K2-VPg-Pro could induce cytoplasmic vesicle structures to function as virus replication sites, we performed the dRBFC assay by co-expressing B2-nYFP and VP35-cYFP with 6K2-VPg-Pro in SCMV-infected maize leaves. To this end, 6K2-VPg-Pro was tagged with mCherry at C-terminal (Fig. 3A). It should be noted that the Glutamine (Q) and Glutamic acid (E) residues preceding the cleavage sites locate at 6K2-VPg and VPg-NIa-Pro junctions were both changed by histidine (H), thereby preventing protein proteolysis [45]. We then conducted a particle bombardment assay in SCMV-infected maize leaves, in which mCherry-tagged 6K2-VPg-Pro (6K2-VPg-Pro-mCherry) was co-expressed with B2-nYFP and VP35-cYFP. The results showed that granular YFP foci formed by dsRNA binding of B2-nYFP and VP35-cYFP perfectly overlapped with the irregular structures induced by 6K2-VPg-Pro-mCherry (Fig. 3B), suggesting that the 6K2-VPg-Pro-induced irregular structures are the active replication sites of SCMV in maize leaf cells.

The 6K2-VPg-Pro-induced replication vesicles reside on multiple intracellular organelles excluding on chloroplasts
Considering that the TuMV 6K2-induced replication vesicles migrate sequentially from ER to chloroplasts for viral genome replication [35], we wondered whether SCMV could also target chloroplasts for replication. For this test, we made a construct 6K2-VPg-Pro-EGFP by fusing EGFP to the C-terminal of 6K2-VPg-Pro (Fig. 4A), and delivered it into protoplasts isolated from SCMV-systemically infected maize leaves. Comparing with the nucleus and cytoplasm-localized wild EGFP, aggregate structures showing green uorescence could only be observed in cytoplasm of protoplasts transfected with 6K2-VPg-Pro-EGFP (Fig. 4B). Unexpectedly, these structures clearly had no obvious association with the red auto uorescence from chloroplasts (Fig. 4B). Moreover, in N. benthamiana epidermal cells, transient expression of either 6K2-VPg-Pro-EGFP or 6K2-VPg-Pro-mCherry also formed aggregate bodies and punctate structures in cytoplasm (Supplementary Fig. 3). These aggregate bodies and punctate structures did not merge with the chloroplasts auto uorescence (Supplementary Fig. 3), suggesting that SCMV 6K2-VPg-Pro-induced replication vesicles did not reside on chloroplasts.
Previous ndings that TuMV 6K2 vesicles accumulate at ER exit sites (ERES) on the ER membrane and subsequently migrate to Golgi apparatus [35,36] prompted us to investigate whether SCMV 6K2-VPg-Proinduced vesicles locate at ER and/or Golgi apparatus. We co-expressed 6K2-VPg-Pro-EGFP with an ER marker mCherry-HDEL or a Golgi apparatus marker mCherry-GmMan1 in maize protoplasts [57].

Discussion
In this study, we investigate the active replication sites of SCMV in maize cells through dRBFC assay. As a sensitive and speci c dsRNA reporter system for visualizing dsRNA distribution and dynamics in living cells, dRBFC has been used in vivo to visualize the subcellular distribution of dsRNA intermediates in the replication of TuMV, carnation Italian ringspot virus, barley stripe mosaic virus [36,59]. Given that dRBFC assay requires the active viral replication which generates double-stranded replicative RNA intermediates [36], we used young maize leaves already systemically infected with SCMV for protoplasts preparation and bombardment to locate the replication sites of SCMV. The systemic infection of SCMV in young leaves ensures the active replication state of SCMV in maize cells. The dRBFC assay clearly showed that SCMV replicative dsRNA localized in cytoplasm but did not reside on chloroplasts in maize protoplasts and leaf tissues.
We then demonstrated that SCMV replicative dsRNA associated with 6K2-VPg-Pro-induced vesicles. SCMV-encoded 6K2-VPg-Pro polyprotein is responsible for the formation of vesicles represented by irregular punctate and aggregated structures in cytoplasm, which is consistent with that the expression of TuMV 6K2-VPg-Pro could induce the formation of cytoplasmic vesicles [26]. The ring-like vesicle structures induced by SCMV 6K2-VPg-Pro were similar to that of vesicle structures induced by TuMV 6K2 [35]. These punctate or aggregated bodies may come from the membrane-binding and fusion effects of these vesicles [8]. Among the SCMV 6K2-VPg-Pro-induced vesicles, some perfectly colocalized to dsRNA replicative intermediates indicating virus replication, while some scattered in cytoplasm rather than colocalized with dsRNA (Fig. 3B). Since the co-localization of 6K2-VPg-Pro-induced vesicles with dsRNA binding protein (B2-YN and VP35-YC), the SCMV 6K2-VPg-Pro-induced vesicles can provide compartments for SCMV active RNA replication.
Studies on TuMV 6K2 revealed it can induce small mobile vesicles derived from ER at early infection stage and form an irregularly shaped structure juxtaposed to nucleus at late infection stages [8]. TuMV RNA replication takes place within these 6K2 vesicles associating with chloroplasts or in cytoplasm [8,27,35,36]. TuMV 6K2-induced replication vesicles form at ERES on ER membrane, and then transport to Golgi apparatus by COPI and COPII secretory pathway, and subsequently reside on chloroplasts for virus replication. Viral RNA was also detected in chloroplasts [7,8,[33][34][35]60]. Even in the absence of other viral proteins/viral replication, TuMV 6K2 can induce the biogenesis of ER-derived vesicles that target chloroplasts in N. benthamiana leaves [36,60]. However, for SCMV 6K2-derived vesicles, we only observed them in cytoplasm. We did not nd SCMV 6K2-derived vesicles targeting chloroplasts in maize protoplasts or bombarded epidermal cells.
In this study, SCMV 6K2-VPg-Pro-induced vesicles were rstly found to be associated with ER and Golgi apparatus, while had no evident co-localization with chloroplasts. We wondered whether other intracellular organelles can be employed by SCMV as replication sites. Interestingly, SCMV 6K2-VPg-Pro-EGFP-formed punctate structures also colocalized with both mitochondria and peroxisome markers in N. bethamiana leaf epidermal cells, which was similar to the replication sites of a well-studied tombusvirus TBSV [61,19]. TBSV employs peroxisomes and mitochondria to form VRCs, and can target ER for replication when peroxisome biogenesis is defective, implying its preference in targeting organelles for replication [19,35,61]. It would be interesting to explore whether SCMV is preferential in remodeling organelles to form VRCs by using organelles biogenesis-defective maize plants under infection condition.
A question is which mechanisms exist underlying the difference of replication sites between monocotinfecting and dicot-infecting potyviruses. Previous studies showed that polyproteins encoded by monocot-infecting and dicot-infecting potyviruses cluster separately in distinct subgroups [37,38]. Genetic variation and phylogenetic analysis revealed that the 6K2-VPg-Pro polyproteins of monocotinfecting potyviruses cluster separately with that of dicot-infecting potyviruses. There are 54 conserved residues in 6K2-VPg-Pro of monocot-and dicot-infecting potyviruses, suggesting their important function in potyviral replication. Intriguingly, we found the combination of three relatively conserved amino acids at positions 39, 299 and 413 between monocot-and dicot-infecting potyviruses clusters, that are W 39 , N 299 and S 413 for dicot-infecting potyviruses, while L 39 , P 299 and D 413 for monocot-infecting potyviruses. The genetic variations on 6K2-VPg-Pro between monocot-and dicot-infecting potyviruses potentially suggest their roles in determining the replication sites. TuMV 6K2-induced vesicles can tra c to chloroplasts by relying on actomyosin motility system [35]. As for monocot-infecting potyviruses like SCMV, we propose that actomyosin motility system or other chloroplast transport pathway may not be adopted by 6K2-VPg-Pro. Future work on genetic variations determining the distinct characteristics of replication vesicles will provide clues for the differences on replication sites and host range between monocot-infecting and dicot-infecting potyviruses.