Abstract
Due to limited free diffusion in the cytoplasm, viruses must use active transport mechanisms to move intracellularly. Nevertheless, how the plant ssDNA begomovirus hijacks the host intracytoplasmic transport machinery to move from the nucleus to the plasmodesma remains enigmatic. Here, we identified nuclear shuttle protein (NSP)-interacting proteins by probing a protein microarray and demonstrated that viral NSP, a facilitator of the nucleocytoplasmic trafficking of viral (v)DNA, interacts with a new endosomal vesicle-localized plant-specific syntaxin-6 protein, designated NISP in planta. We also showed that begomovirus infection requires the NISP-NSP interaction; NISP displays a pro-viral function, but not the syntaxin-6 paralog AT2G18860 that failed to interact with NSP. Consistent with these findings, nisp-1 mutant plants were less susceptible to begomovirus infection, a phenotype reversed by NISP complementation, whereas overexpressing lines accumulated higher levels of viral DNA than wild-type. Furthermore, NISP interacts with NIG, another pro-viral factor that accessorizes the NSP-vDNA nucleocytoplasmic translocation. Additionally, the NISP-NIG interaction is enhanced by NSP. We also showed that NISP associates with vDNA and might assemble a NISP-NIG-NSP-vDNA- complex. NISP may function as a docking site for recruiting NIG and NSP into trafficking vesicles, providing a mechanism for the intracytoplasmic translocation of the NSP-vDNA complex towards the cell periphery.
Author Summary As viruses must use an active and directed intracellular movement, they hijack the intracellular host transport system for their benefit. Therefore, the identification of interactions between host proteins and begomovirus movement proteins should target the intracellular transport machinery. This work focused on the identification of these protein-protein interactions; it addressed the molecular bases for the intracellular transport of begomoviruses. We used a protein microarray to identify cellular partners for the movement protein (MP) and the viral nuclear shuttle protein (NSP), which is a facilitator of the nucleocytoplasmic trafficking of viral (v)DNA. We identified relevant protein-protein interaction (PPI) hubs connecting host and viral proteins. We revealed a novel NSP-interacting protein, which functions in the intracytoplasmic transport of proteins and DNA from begomoviruses and was designated NSP-interacting syntaxin domain-containing protein (NISP). Our data suggest an intracellular route connecting the release of newly-synthesized begomoviral DNA in the cytosol with the cell surface. Resolving viral DNA-host protein complexes led to the identification of a novel class of components of the cell machinery and a representative member, NISP, that functions as a susceptibility gene against begomoviruses. As geminiviruses pose a severe threat to agriculture and food security, this recessive gene can now be exploited as a target for engineering resistance by gene editing in crops.
Introduction
As obligate intracellular parasites, viruses must enter the host cells and hijack the cell machinery to complete their life cycle. Independent on the molecular mechanisms for virus entry into cells, which differ entirely between animal and plant viruses, once in the cytoplasm, both plant and animal viral genomes move intracellularly to specific cellular sub-compartments for replication, transcription or encapsidation [1, 2]. Due to the limitation of free diffusion in the cytoplasm, viruses have invoked the endogenous host movement mechanisms for an active and directed intracytoplasmic translocation. In fact, almost all viruses are capable of subverting the microtubule transport system to some extent to replicate and spread. Plant viruses must also use an active mechanism to move the viral genome from the replication site to the plasmodesmata, which are intercellular cytoplasmic bridges between plant cells [2, 3]. To promote this typical cell-to-cell movement, plant viruses have evolved movement proteins that increase the size exclusion limit of the plasmodesmata and actively promote the viral nucleic acid translocation into uninfected, adjacent cells [4]. In the case of viruses that replicate in the nuclei of infected cells, they have to be directed to the nuclear periphery, followed by nuclear entry. Likewise, the newly-synthesized viral proteins and viral genomes must translocate throughout the cytoplasm during exit. In animal cells, the retroviruses and DNA viruses deploy the microtubule motor system and microtubule cytoskeleton to move viral proteins and genomes into the nucleus and back to the plasma membrane [1, 5]. Geminiviruses are typical examples of plant DNA viruses that replicate in the nuclei of infected cells, but the underlying mechanisms for the intracytoplasmic movement of viral proteins and viral DNA (vDNA) are far less understood.
Geminiviruses are circular single-stranded (ss) DNA viruses, which replicate via double-stranded (ds) DNA intermediates in the nucleus of infected cells, infect a broad spectrum of economically important crops and cause significant yield losses worldwide [6]. The Geminiviridae family includes nine genera, but approximately 75% of the identified species belong to the genus Begomovirus, which are whitefly-transmitted, infect dicotyledonous species, and display a genome configuration that can be either monopartite or bipartite [7, 8]. The genomic components of the bipartite begomoviruses are designated DNA-A and DNA-B. The DNA-A encodes proteins required for viral replication (AC1 or Rep, REn or AC3), transactivation of viral genes (AC2 or TRaP), encapsidation of viral genome (AV1 or CP) and suppression of host defenses (AC4 and TrAP). DNA-B encodes the nuclear shuttle protein (NSP or BV1) and the movement protein (MP or BC1) involved in intracellular and cell-to-cell movement of vDNA.
Although the role of MP-NSP complex formation in mediating intracellular and intercellular transport of begomoviruses is widely accepted, the molecular mechanisms underlying the action of these viral proteins at individual stages of the intracellular and intercellular movement are still debated. Some studies indicated that NSP from Bean dwarf mosaic virus (BDMV) and Abutilon mosaic virus (AbMV) complexes with ss-vDNA and ds-vDNA presumably in the nuclei of infected cells, whereas NSP from Squash leaf curl virus (SLCV) binds only ss-vDNA in vitro [9, 10, 11]. NSP also interacts with MP in the cytoplasm [12], which promotes the directionality of virus translocation to the cell surface [12, 13]. The mechanism for vDNA exit from the nucleus remains unsolved. It has been conceptually accepted that NSP facilitates the nuclear exit of newly-replicated vDNA via nuclear pores [10, 14–17]. However, the interactions with the host nuclear transport machinery have not been documented. A piece of this puzzle was obtained with the identification of an NSP-interacting GTPase (NIG), which interacts with NSP at the cytosolic side of the nuclear pore complex and facilitates the release of NSP-vDNA complex from the nuclear envelope into the cytosol [12, 18]. Consistent with this transport function, the cytosolic NIG (i) accumulates around the nuclear envelope, (ii) promotes the translocation of NSP-vDNA from the nucleus to the cytoplasm, (iii) functions as a pro-viral factor during begomovirus infection and (iv) shares structural features and transport properties with the human Rev-Interacting Protein (hRIP), which has been shown to be involved in the release of HIV-1 RNAs from the nuclear periphery to the cytoplasm [12, 18, 19]. Histone H3 has been shown to bind NSP and MP from BDMV as an additional component of the vDNA-protein complex that traffics intracellularly and intercellularly [20]. A nuclear acetyltransferase (AtNSI) also interacts with the Cabbage leaf curl virus (CaLCuV) NSP and may regulate the nuclear export of ss-vDNA-NSP via acetylation of CP and histones [21, 22]. Other NSP-interacting partners include the receptor-like kinases (RLKs) NSP-interacting kinase (NIK1) and NSP-associated kinase (NsAK), which are not involved in intracellular movement of vDNA but rather are implicated in antiviral immunity against begomoviruses [23–26].
For the cell-to-cell transport of vDNA, two distinct mechanisms, which basically differ in the nature of the vDNA-protein complex translocated into adjacent cells, have been suggested [11, 27]. In the first mechanism, designated “relay race” model, NSP facilitates the translocation of viral DNA from the nucleus to the cytoplasm and then is replaced by MP, which promotes the cell-to-cell movement of vDNA via plasmodesmata [10, 14]. The second mechanism represented by the “couple-skating” model preconizes that MP facilitates the vDNA-NSP nucleocytoplasmic trafficking and promotes the translocation of vDNA-NSP complex into the adjacent cells via endoplasmic-reticulum-derived tubules induced by the viral infection [11–13, 18, 28–32]. More recently, an alternative route for intracellular translocation and cell-to-cell spread of vDNA via chloroplasts and stromules has been proposed, which was based on experimental data of AbMV MP interactions (33,34). AbMV infection has been shown to induce the formation of a stromule network interconnecting different chloroplasts, chloroplasts with nuclei, and expanding to the cell periphery presumably at plasmodesmata [34]. This stromule network is stabilized by the plastid chaperone cpHSC70, which has also been shown to interact with MP [33]. The interaction between MP and cpHSC70 may bridge viral nucleoprotein complexes with stromules. Independent on the cell-to-cell transport models, the underlying mechanisms of the intracytoplasmic retrograde and anterograde movement of vDNA complexes remain to be determined. CaLCuV and SLCV MPs have been shown to interact with synaptotagmin A (SYTA), which is located in the endosomes of plant cells and is required for the cell-to-cell trafficking of different MPs [35]. These findings raised the hypothesis that distinct viral MPs promote an intracytoplasmic transport of vDNA-complexes to plasmodesmata via an endocytic recycling pathway. Likewise, the AbMV MP has been shown to interfere with microtubule assembly and stability when co-expressed in plant tissues with a viral inducer of cell progression, which may suggest a microtubule-associated transit of MP or begomoviruses to the plasmodesmata [36]. However, the host components of the vDNA complex competent for intracytoplasmic translocation have not been identified, and the underlying mechanism for intracytoplasmic trafficking of vDNA remains to be elucidated.
In this investigation, we used a protein microarray assay to identify MP- and NSP-host protein-protein interactions (PPIs). From the NSP-MP-host PPI network map, we selected an intracellular transport-associated protein, designated NSP-interacting syntaxin domain-containing protein (NISP), as an NSP specific target, to gain further insight into the intracytoplasmic transport of begomoviruses. We showed that NISP, which was located in motile vesicle-like structures, many of which were targeted to the endosomes, interacted with NSP in vivo and displayed a pro-viral function. Furthermore, NISP also interacted with NIG, and the complex formed was enhanced by the presence of viral NSP. In addition, vDNA was recruited into the NISP complex. Our data indicate that NISP may be part of the vDNA complex that traffics intracellularly via endosomes.
Results
Identification of Arabidopsis proteins interacting with in vitro transcribed and translated MP and NSP
To search for the begomovirus movement proteins-Arabidopsis protein-protein interactions (PPIs), we used a nucleic acid programmable protein array (NAPPA)-based approach [37], in which we interrogated 4600 in vitro synthesized ORFs from Arabidopsis for their capacity to be targeted by MP and/or NSP from CaLCuV. The in situ synthesized NAPPA protein microarray has previously generated a transcription factor (TF)-NAPPA dataset, which has been validated by pull-down assays and bimolecular fluorescence complementation (BiFC) assays in tobacco leaves using a subset of newly identified interactions [37]. NSP and MP were in vitro transcribed and translated to probe this protein microarray (S1a and S1b Fig). A total of 40 proteins were identified as associated with MP, and 40 interacted with NSP. The majority (35 proteins of 40) were found to interact with both proteins, which was not surprising because MP itself interacts with NSP and may be part of common multiprotein complexes as both participate in viral DNA movement (S2 Fig). We found five new NSP-specific protein interactions (dark orange) and five MP-specific protein interactions (light yellow), including the stomatal cytokinesis defective 2 (SCD2) protein (AT3G48860) and a peptidyl-prolyl cis-trans isomerase family protein (AT1G73655) belonging to the same protein family as Pin4, which have been shown previously to interact with MP (S1 Table) [36]. As previously identified NSP-interacting proteins were absent in the protein array, the identification of known MP-specific interacting proteins contributed to validate our current data.
The protein-protein interactions between the viral proteins and the Arabidopsis proteins were integrated into the protein-protein interactome experimentally determined for Arabidopsis thaliana (BioGRID database and the Arabidopsis interactome database), using Cytoscape software. This procedure identified the PPI network containing MP, NSP and all, directly and indirectly, interacting host proteins (Fig 1). AT1G68185, an ubiquitin-like protein, which formed a large hub (hub 20, degree 148), may represent a convergent targeting node among begomovirus viral proteins as it was found to be a direct target of NSP and MP independently (S1 Table). Furthermore, the ubiquitin-proteasome pathway (UPS) is targeted by many viruses to maintain suitable levels of viral proteins and to induce, inhibit or modify ubiquitin (Ub)-related host proteins [38]. Other MP-interacting proteins, including FKBP-like peptidyl-prolyl cis-trans isomerase family protein (this work), chloroplast heat shock protein 70-1 [33] and synaptotagmin A [35], which have been shown to assist the MP movement function, are connected to the AT1G68185 hub (hub 20), further supporting the notion that this hub may be biologically significant during begomovirus infection.
To gain further insights into the cellular processes affected by NSP or MP, we performed functional enrichment analyses of their direct and indirect interactors. Significantly enriched GO terms (with p-value < 0.05) were identified in all three categories, Biological Process, Molecular Function, or Cellular Component ontology (S2 Table). Some over-represented GO terms in this set of viral protein-interacting host proteins are consistent with NSP and MP function or localization. Consistent with their function as viral movement proteins, a subset of host interactors was significantly over-represented under the transport activity term and protein binding term in Molecular Function (S2 Table). Under the cellular component ontology, we observed an over-representation of proteins under membrane-bounded organelle term, intracellular vesicle term, and SNARE complex term, which may suggest an intracytoplasmic route for begomovirus proteins and vDNA trafficking (S2 Table). Accordingly, two NSP-interacting proteins, AT5G23550 and AT4G30240, form convergent hubs (17 and 19) enriched for membrane-bound proteins and transport activity function (Fig 1, S1 Table). The NSP-interacting protein AT5G23550 (hub 17) is annotated as a Got1/Sft2-like vesicle transport protein, which may be located to a late-Golgi compartment and may be involved in the fusion of retrograde transport vesicles derived from an endocytic compartment with the Golgi complex [39]. Likewise, AT4G30240 (hub 19) is a soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE)-like protein harboring a syntaxin-6 domain typical of proteins found in the endosomal transport vesicles [40].
Under the biological process ontology, the response to auxin term represented a significantly enriched GO term, which has not been previously associated with begomovirus infection (S2 Table). The NSP-MP-host PPI network uncovered two relevant hubs enriched for auxin response-related proteins, which are represented by an NSP-MP general interaction-derived hub (Fig 1, hub 9) and an NSP-specific host interaction-derived hub (hub 11). Remarkably, these two viral protein hubs are interconnected to each other via interactions with proteins (in gray), which are also over-represented under the auxin response term, forming a large auxin affected hub target by begomovirus (S1 Table). This hub may represent an undocumented biological process affected by begomovirus infection. The resulting PPI network map may provide a framework for future studies in detailing the begomovirus complex life cycle. One such example is described here, as following.
A SNARE-like protein designated NSP-Interacting Syntaxin domain-containing Protein (NISP), but not its paralog, interacts with NSP via the syntaxin domain
The NSP-MP-Host PPI network uncovered enriched GO terms linked to intracellular vesicle and membrane-bound proteins, which may be associated with a mechanism for intracytoplasmic translocation of viral proteins and vDNA. To pursue further with this possible connection and to validate further the microarray data, we selected a new specific NSP-interacting protein, the SNARE-like protein AT4G30240, which harbors a syntaxin-6 domain at the N-terminus and a transmembrane segment at the C-terminus, for further analyses (Fig 2a). SNARE-complexes drive the fusion of membrane-bound vesicles with target membranes required for intracellular transport via the endosome system. We used the syntaxin domain superfamily as the prototype for the identification of homologs in the genomes of Arabidopsis thaliana, Oryza sativa, Drosophila melanogaster, Homo sapiens, and Saccharomyces cerevisiae. We then selected the most related proteins to construct phylogenetic trees using Bayesian inference method (S3a Fig). AT4G30240 was clustered in pair with AT2G18860 and formed a clade of syntaxin-6 domain-containing proteins, which also included a third Arabidopsis paralog AT1G27700. This AT4G30240-derived clade (in red), also designated Syntaxin-6-like subgroup, clustered only plant proteins, carrying an N-terminal syntaxin-6-like domain and a C-terminal transmembrane segment. AT4G30240 and AT218860 may be a result of genome duplication, as these syntaxin-6-like proteins clustered in pair even in phylogenetic analyses that include proteins from other organisms, share a high degree of sequence conservation and display similar intron/genomic DNA configuration (https://www.arabidopsis.org/).
The remaining syntaxin-6-like proteins clustered together in a distinct clade that included proteins from human, rice and the Arabidopsis AT1G28490, also named SYP61, a characterized trans-Golgi network (TGN)/endosome localized syntaxin [41–43]. This SYP61 second clade (in yellow) was designated target (t)-syntaxin-6/SNARE-like subgroup, as it included structurally similar proteins carrying the N-terminal syntaxin-6 domain and also a tSNARE coiled-coil homology domain at the C-terminus.
Due to the lack of a coiled-coil tSNARE domain at C-terminus, the syntaxin-6-like subgroup may represent a new, functionally uncharacterized, plant-specific class of the syntaxin-like family of the t-SNARE superfamily. This hypothesis was further confirmed in a phylogenetic analysis of syntaxin-6 domain-containing proteins from Arabidopsis thaliana, Oriza sativa, Glycine max, Zea mays, Zostera marina, Homo sapiens, Mus musculus, Bos Taurus, Equus caballus, Danio rerio, and Drosophila melanogaster using the full-length sequence of AT4G30240 as prototype (S3b Fig). This phylogenetic analysis further confirmed that the syntaxin-6 domain-containing proteins form two distinct clades with high internal stability (100%). The AT4G30240-derived clade, syntaxin-6-like clade, includes only plant syntaxin-6-like proteins, and the tSNARE/syntaxin clade includes proteins from all species tested, except yeast.
To complement the protein microarray data, we used the yeast two-hybrid to examine further the NSP-AT4G30240 interaction. We also included the paralog AT2G18860 in our assays, which was not present in the protein microarray. The second most closely related paralog, AT1G28490, was present in the protein microarray but was not found to interact with NSP. Co-expression of the full-length AT4G30240 (NISP, NSP-Interacting Syntaxin domain-containing Protein) ORF fused to the activation domain (AD), and NSP fused to the binding domain (BD) of GAL4 promoted histidine prototrophy (Fig 2b). The HIS3 marker gene was not activated in yeast cells cotransformed with the controls. In contrast, AT2G18860 failed to interact with NSP in yeast, as neither AD-AT2G18860 nor BD-AT2G18860 promoted his prototrophy when cotransformed with the respective cognate BD-NSP or AD-NSP (Fig 3c).
To map the NSP-interacting domain on NISP, we generated NISP deletion fragments and assay for protein interactions by two-hybrid assay. The N-terminal NISP fragment extending up to amino acid position 107 and containing an intact syntaxin-6 domain (NISP1-107, Fig 2a) was sufficient to bind NSP in yeast (Fig 2d). The interaction of the truncated NISP1-107 with NSP was specific because the HIS marker gene was not activated in yeast cells co-transformed with AD- or BD-truncated NISP1-107 and the empty vectors (Fig 2d). In contrast, the NISP truncated versions, comprising the NISP internal region, delimited by positions 105-200 (NISP105-200; Fig 2e) or the C-terminal portion (NISP200-300; Fig 2f), were not capable of recognizing NSP, confirming that NISP binds NSP via its syntaxin-6 domain. Although the syntaxin-6 domains of NISP and AT2G18860 are conserved (78.57% identity), NSP interacts specifically with NISP in yeast (Fig 2b). Sequence alignment revealed an 8-amino acid deletion at the syntaxin-6 domain of AT2G18860 as the most striking feature that may account for the differences in protein binding specificity (S4 Fig).
Although NISP lacks a typical coiled-coil tSNARE domain, it harbors a syntaxin-6 domain at the N-terminus and a predicted transmembrane segment at the C-terminus, which suggests an endomembrane-bound localization (Fig 2a). Mammalian syntaxin-6 domain proteins are localized in the trans-Golgi network (TGN) membranes and endosomal vesicles [44]. To examine the subcellular localization of NISP, we transiently expressed AT4G30240 fused to a green fluorescent protein (GFP) under the control of the native and constitutive promoters, in Nicotiana benthamiana leaves and analyzed the localization of the fluorescent signal by confocal microscopy. NISP-GFP was predominantly localized in motile spherical structure, resembling vesicles (Fig 3a, 3b; S1 Video). We also examined the localization of the paralog AT2G18860 fused to GFP, which concentrated predominantly in membrane-bound vesicles and plasma membrane (S5 Fig).
To examine more precisely the NISP localization, we performed co-localization assays, using SYTA-mCherry as an endosome-associated marker [35], the plasma membrane-associated molecular marker FLS2-mCherry [45], ER-, TGN- and tonoplast-associated molecular markers constructed by Nelson et al. [46]. These co-localization studies showed that NISP-GFP did not co-localize with ER-mCherry- (Fig 3c, asterisks indicate the field of the insets) and tonoplast (vac-mCherry)- (Fig 3e) associated molecular markers. Likewise, NISP-containing vesicles did not co-localize with the plasma membrane FLS2-mCherry (Fig 3f). In contrast, we observed a small population of NISP-GFP-containing vesicles that co-localized with the TGN-associated molecular marker (Fig 3d) and a large fraction of NISP-GFP-containing vesicles precisely co-localizing with the endosomal protein SYTA-mCherry (Fig 3g). Because many of the NISP-labeled vesicles were associated with the plasma membrane, we examined whether the NISP-associated vesicle membrane was derived from the plasma membrane by staining NISP-GFP-expressing cells with FM4-64, a marker for plasma membrane internalization that labels endosomes [47]. In addition to labeling plasma membrane, dye-labeled vesicles were observed that also showed discrete fluorescence with NISP-GFP (Fig 3H). Collectively, these results indicate that NISP localizes in TGN/early endosome vesicles as it co-localized with Golgi-associated markers and endosomal SYTA. Furthermore, the membranes of the NISP-associated vesicles were labeled by FM4-64, a marker for vesicle components of the plant endocytic pathway.
NISP interacts with NSP in intracellular vesicles
We next examined whether NISP interacts with NSP in vivo using the bimolecular fluorescence complementation (BiFC) assay. The formation of NISP/NSP complexes occurred in vesicles of tobacco epidermal cells independent of the orientation of the NISP fusions (N-terminus or C-terminus of YFP; Fig 4a), and the reconstituted fluorescent signal was much higher than that of the background (control panels with combinations of the protein fusions with empty vectors). The reconstituted fluorescent signal was co-labeled by FM4-64, indicating that the complex was formed in endosomal vesicles derived from the plasma membrane (Fig 4b, see arrows). We further demonstrated that NISP and NSP interacted in vivo using co-immunoprecipitation assays. Both NISP-GFP and NSP-6HA, were co-immunoprecipitated from co-infiltrated extracts using anti-GFP antibody, demonstrating a specific association of NISP-GFP with NSP-6HA fusions (Fig 4c). Switching the protein tags did not alter the results because NISP-6HA also associated with NSP-GFP in co-infiltrated extracts (Fig 4d). The Co-IP results also demonstrated that GFP alone did not interact with either NSP-6HA or NISP-6HA, demonstrating the specificity of the interactions between the viral protein and NISP.
NISP, but not its paralog AT2G18860, displays a pro-viral function
The biological relevance of the NSP-NISP interactions in begomovirus infection was examined using several different approaches. We first analyzed the expression of NISP from a released microarray data, which showed that NISP was induced by begomovirus infection. This result was further confirmed by quantitative RT-PCR, which also showed that AT2G18860, the NISP paralog that failed to interact with NSP, was not induced by begomovirus infection (S6 Fig). We then identified a T-DNA insertion mutant in the NISP gene (nisp-1) and the AT2G18860 gene (at2g18860-1). The accumulation of NISP and AT2G18860 transcripts in the respective T-DNA insertion mutant was much lower than in Col-0, confirming that nisp-1 and at2g18860-1 were loss-of-function mutant lines (Fig 5a, 5b). These mutant lines were complemented by expression of NISP-GFP and AT2G18860-GFP in the own knockout background in three independently transformed lines (Fig 5a, 5b, nisp-1/NISP lines, and at2g18860-1/AT2G18860 lines). We also prepared transgenic lines overexpressing NISP and AT2G18860, under the control of a constitutive promoter. The accumulation of NISP and AT2G18860 at transcript (Fig 5a, 5c) and protein (Fig 5b, 5d) levels was higher than in Col-0. The knockout and overexpressing lines were phenotypically indistinguishable from Col-0 (S7 Fig).
We next examined whether NISP and AT2G18860 participate in begomovirus infection. The knockout lines (nisp-1 and at2g18860-1), complemented lines (nisp-1/NISP and at2g18860-1/AT2G18860), NISP-overexpressing lines and Col-O were inoculated with infectious clones of CaLCuV and the course of infection was monitored by visualization of symptom development and PCR diagnosis. The accumulation of viral DNA was quantified by qPCR (Fig 5).
The nisp-1 knockout line displayed attenuated symptoms (Fig 5e), delayed course of infection (Fig 5f), and lower viral DNA as compared to Col-0 (Fig 5g). Expression of NISP-GFP in nisp-1 restored the wild type susceptibility to begomovirus infection in all three independently transformed lines (nisp-1/NISP), confirming that the attenuated susceptibility displayed by the nisp-1 line was indeed due to loss of NISP function. In contrast, NISP overexpression conferred enhanced susceptibility to begomovirus infection; the progress of infection was accelerated (Fig 5f), and viral DNA load was significantly higher in the NISP-overexpressing lines than in Col-0 and nisp-1 lines (Fig 5g). We predicted that if NISP-NSP binding were the bases for the NISP pro-viral function, the closest paralog of NISP, AT2G18860, which harbors the syntaxin-6-like domain but does not interact with NSP, would not affect begomovirus infection. Accordingly, the at2g18860-1 knockout line, at2g18860-1/AT2G18860 complemented lines, and an AT2G18860- overexpressing line displayed similar symptoms, course of infection and viral DNA load as Col-0, indicating that AT2G18860 does not interfere with begomovirus infection (Fig 5e, 5f, 5g). These results further substantiated the notion that the formation of NSP-NISP complex may be the determinant for the NISP pro-viral function rather than the presence of a syntaxin-6 like domain per se.
NISP also interacts with NIG and complex formation is enhanced by viral NSP
There is a gap in our knowledge of how the vDNA-NSP complex, which binds to NIG in the cytosolic side of the nuclear pores, traffics to the cell periphery. From the PPI network map, we hypothesized that NIG could serve as a bridge to attach vDNA to the intracellular transport system of plant cells. The integration of our data with the Arabidopsis interactome uncovered a possible interaction between NIG and NISP, which was also shown here to bind specifically to viral NSP. The interaction between NIG and NISP was first demonstrated by two-hybrid assay (S8 Fig 8a). The co-expression of the NISP ORF fused to BD and NIG ORF fused to AD from GAL4 promoted histidine prototrophy in medium supplemented with 2.5 mM 3AT, whereas co-expression with empty vectors did not.
To examine whether NIG and NISP interact in vivo, we used BiFC and co-immunoprecipitation assays. NIG has been shown to be unevenly distributed in the cytosol, concentrating around the nuclear envelope (S8b and S8c Fig) [25, 48]. The BiFC data demonstrated that nYFP-NIG is relocated to the vesicle-like compartments via interaction with cYFP-NISP (and vice-versa), as the reconstituted fluorescent signal, which concentrated in membrane-derived vesicles and was barely detected in the controls (Fig 6a), was co-stained by FM4-64 (Fig 6b). The inclusion of NSP-6HA in the BiFC assay increased the NISP-NIG complex formation, as judged by the enhanced intensity of the reconstituted fluorescent signal, showing the fluorescent vesicles associated with the plasma membrane (S9 Fig). To provide further evidence of the NSP-mediated enhancement of NIG-NISP interaction, we used a quantitative co-immunoprecipitation assay (Fig 6, S8 Fig). We first confirmed that NIG interacts with NISP in vivo (Fig 6c, S8d). The Co-IP results indicated that NISP-GFP, but not GFP alone, interacted with NIG-HA. Likewise, NIG-GFP, but not GFP alone, interacted with NISP-HA. NSP promoted a stronger association of NISP-GFP with NIG-6HA, as judged by the amount of co-immunoprecipitated NIG by anti-GFP in the presence of co-expressed GST-NSP (Fig 6d, 6e). Likewise, GST-NSP enhanced NISP-6HA-NIG-GFP complex formation (S8e and S8f Fig). These results indicate that NSP/NIG and NISP might form a multiprotein complex assembled in intracytoplasmic transporting vesicles. Syntaxin-6 domain-containing proteins have been shown to participate in the endosome-plasma membrane route of protein cargos.
NISP complexes viral DNA in vivo
The current model of vDNA intracellular translocation informs that NSP binds to nascent viral DNA in the nucleus, facilitates its translocation to the cytosol via nuclear pores and moves vDNA intracytoplasmically to the cell periphery, where MP mediates the intercellular movement of vDNA-NSP or vDNA-MP via plasmodesmata. Because NSP also binds to NISP in vesicle-like structures, we asked whether the vDNA would be recruited into a NISP multi-protein complex using chromatin immunoprecipitation (ChIP) assay (Fig 7). The untransformed line (Col-0) and NISP-GFP-overexpressing (AtNISP-1, AtNISP-3, and AtNISP-3, Fig 7a) lines were infected with CaLCuV infectious clone and the ChIP assay was performed using an anti-GFP antibody. We confirmed first that the inoculated plants were infected (IN) by performing PCR from input DNA using CaLCuV-specific primers, which amplify a 770-bp fragment from CaLCuV DNA-B (Fig 7b and S10b Fig, primers F1/R1) and a 389-bp fragment from DNA-A (Fig 7e and S10a Fig, primers F1/R1). This viral DNA-B and DNA-A fragments were also amplified from ChIP-DNA of infected (IN) AtNISP-overexpressing lines but not of infected (IN) or uninfected (UN) Col-0, which lacks GFP-tagged NISP (Fig 7a, 7e). Using two sets of DNA-B-specific primers (F2/R2 and F3/R3, S10b Fig), ChIP-qPCR further demonstrated significant enrichment of DNA-B-specific fragments in infected AtNISP-GFP-overexpressing samples, which had been precipitated with anti-GFP serum, but not in pulled down samples from infected Col-0 lines (Fig 7c and 7d). Likewise, the use of F2/R2 set of DNA-A primers (S10a Fig 10a) also resulted in significant enrichment of the DNA-A-specific fragment in ChIPed DNA from infected AtNISP-GFP-overexpressing samples but not from infected Col-0 (Fig 7f). Collectively, our results indicate that NISP is a functional component of the vDNA-complex that traffics along the cytoplasm to the cell periphery for the MP-mediated cell-to-cell movement of vDNA into adjacent cells.
Discussion
Viruses are obligate intracellular parasites and, once in the host cell cytoplasm, they must move intracellularly to and from the replication site to the plasma membrane for spreading. The cytoplasmic limitations on free diffusion have forced viruses to evolve efficient mechanisms to manipulate the host transport system for their benefit. Understanding how viruses hijack the host intracellular transport system using a limited repertoire of proteins provides an opportunity to uncover the molecular bases of the host transport machinery. Bipartite begomoviruses, plant ssDNA viruses, use two movement proteins, MP and NSP, to move intracellularly from the site of replication in the nucleus to the cell surface [17]. However, the host components of the vDNA complex competent for intracytoplasmic translocation have not been identified, and the underlying mechanism for intracytoplasmic trafficking of vDNA remains to be elucidated. Here, we used a previously fabricated, in situ synthesized protein microarray containing 4600 Arabidopsis ORFs [37] to identify NSP- and MP-Arabidopsis protein-protein interactions (PPIs). Consistent with the movement function of the viral proteins, the identified NSP-MP-PPI network uncovered direct and indirect interactions, over-represented under transport activity, protein binding, membrane-bound organelles, intracellular vesicle, and SNARE complex ontology terms, which may define an intracellular route for vDNA trafficking. Precedents in the literature have implicated a microtubule-associated intracellular movement of viral proteins and vDNA in addition to an endocytic recycling pathway for the MP-mediated cell-to-cell movement of vDNA protein complexes (35; 36]. Accordingly, transport function-related proteins, including kinesin motor protein, plant syntaxin 121, Rab GTPase, Ras, Got1/Sft2-like vesicle transport protein, target SNARE coiled-coil domain protein, syntaxin-6 plant proteins, prenylated Rab acceptor, Heat shock protein 70 (Hsp 70), nuclear pore anchor, synaptotagmin A, were over-represented in the MP-NSP-PPI network. These studies thus provided a critical framework for future investigations to elucidate the molecular mechanisms for intracytoplasmic transport of begomoviruses.
The NSP-interacting syntaxin domain-containing protein (NISP) is identified as a new functional cellular partner of begomovirus NSP
In the present investigation, we characterized an NSP-specifically interacting syntaxin-6 domain-containing protein, designated NISP and provided several lines of evidence indicating that NISP might be involved in the intracytoplasmic movement of NSP-vDNA. First, co-localization studies using organelle-associated molecular markers, and the FM4-64 marker demonstrated that NISP is associated with TGN/early endosomal vesicles. Second, we confirmed that NISP interacted with NSP in planta and showed that the complex formation occurred in vesicle-like structures, which resemble the NISP-associated TGN/endosomal vesicles, the typical subcellular localization of syntaxin-6-like proteins from Arabidopsis and humans [49]. Third, we showed that NISP exhibits a pro-viral function, and begomovirus infection required NISP-NSP interaction. The mutant nisp-1 was resistant to begomovirus as the knockout lines displayed attenuated symptoms, a delayed course of infection, and accumulated much lower viral DNA as compared to Col-0. In contrast, the overexpressing lines were hypersensitive to begomovirus. We took advantage of a highly conserved NISP paralog, AT2G18860, to show that NISP-NSP interaction was the molecular basis for the NISP pro-viral function. AT2G18860, which shares with NISP a 78,6% identical syntaxin-6 domain, did not interact with NSP and thus did not interfere with begomovirus infection. Consistent with these data, NISP, but not AT2G18860, is induced by begomovirus infection. Fourth, NISP also interacted with NIG, which facilitates the release of the NSP-DNA complex from the nuclear pores to the cytosol [12, 18], and the presence of NSP enhanced the NISP-NIG complex formation, indicating that NSP, which interacts with NIG [12] and NISP (Fig 4), may be recruited by the complex. We used ChIP assay to demonstrate that NISP was also associated with vDNA in infected cells, which may be assembled into a NISP-NIG-NSP multiprotein complex. Finally, the NISP interactions relocated the dispersed cytosolic NIG and viral NSP to trafficking vesicle resembling the NISP-containing spherical structure associated with TGN/endosome. The NISP-mediated translocation of NSP-vDNA to endosomal vesicles favors the association of the viral complex with MP, which is brought to the endosome via interaction with the endosome-associated SYTA for the MP-mediated cell-to-cell movement of vDNA [35].
NISP belongs to a new, previously unidentified plant-specific subfamily of syntaxin-6-like proteins
The phylogenetic reconstruction of syntaxin-6 domain-containing proteins from different eukaryotic organisms divided this family into two subclades, one containing proteins from all tested organisms, including SYP61 from Arabidopsis, and a new, previously unidentified plant-specific subclade, which includes NISP and AT2G18860 paralogs clustered in pair. SYP61 has been shown to form functional SNARE complexes involved in the vacuolar-TGN recycling pathway and in exocytotic trafficking to the plasma membrane [43, 50]. The process of membrane fusion is the final step of vesicle transport, which is generally driven by a specific pairing of 1 v-SNAREs with the 3 or 3 cognate t-SNARE molecules into a trans-SNARE complex [49, 51]. The proteins from the SYP61 subclade contain the typical structural configuration of t-SNARES, including the N-terminal syntaxin-6 domain, followed by a C-terminal t-SNARE coiled-coil homology domain and a transmembrane segment at the C-terminus. The proteins from the NISP subclade harbor the N-terminal syntaxin-6 domain and the transmembrane segment but lack a C-terminal SNARE domain, which may impair the participation of these new syntaxin-6-like proteins in the assembly of a functional trans-SNARE complex for membrane fusion. Nevertheless, we have demonstrated that a representative of this new syntaxin-6 subclade, NISP, is functionally expressed in plant cells. In addition to being induced by begomovirus infection, NISP functions as a susceptibility factor to begomoviruses, forms complexes with viral and host proteins and is located in TGN/endosome-associated vesicles. We also mapped the NSP-binding domain on NISP to the syntaxin-6 domain. Therefore, we propose that NISP may function as an intracellular vesicle docking site to which the cargo proteins are specifically recruited through interactions with the syntaxin-6-like domain.
We also showed that members of the new syntaxin-6 subclade might diverge functionally concerning protein-protein binding specificity. The NISP paralog, AT218860, which shares with NISP a conserved syntaxin-6 domain (78.6% identity), fails to interact with viral NSP and hence does not affect begomovirus infection. Therefore, specificity of protein interactions may rely on marginal features of the syntaxin-6 domain. A striking difference between NISP and AT2G18860 is an eight-amino acid deletion toward the N-terminus of the syntaxin-6 domain, which may account for the NISP specificity to NSP (S4 Fig). Complementary studies are necessary to precisely define the determinants of a functional syntaxin-6 domain for NSP interaction, which will be useful for engineering resistance to begomoviruses.
NISP may participate in the intracytoplasmic transport of NSP-vDNA
Based on the current data and others previously reported data, we propose a model for the NISP integration into an intracytoplasmic transport route used for the begomovirus anterograde movement. Begomoviruses replicate in the nuclei of infected cells, where NSP interacts with newly-synthesized vDNA and facilitates the translocation of vDNA to the cytoplasm via nuclear pores [13, 32]). This nuclear exportation process may be mediated by an exportin-like protein yet to be identified. At the cytosolic side of the nuclear envelope, NIG associates with the NSP-vDNA complex to accessorize the release of NSP-vDNA from the nuclear pores into the cytosol [12, 18]. The pro-viral function of NIG is associated with its cytosolic localization because the WWP1-mediated recruitment of NIG into nuclear bodies impairs begomovirus infection [48]. At the cytosol, we showed here that NISP may recruit the NIG-NSP-vDNA complex to TGN/endosome-associated transport vesicles. The nature of the NISP-associated vesicle was identified using cell biology molecular markers and FM4-64, which demonstrated that NISP is predominantly localized at TGN and in the early endosome (Fig 3), which was consistent with the localization of syntaxin-6 proteins in most eukaryotic cells [40]. As MP has been demonstrated to associate with the endosomal SYTA, a possible NISP sequestration of NSP-DNA into TGN/endosome-associated vesicles may provide the means for the predicted interactions between endosomally localized MP-SYTA and NSP-DNA, which would favor the MP-mediated intercellular transport of vDNA via an endocytic recycling pathway [35]. In summary, our data support a model in which direct interactions between the host factor NISP and NIG/NSP modulate vDNA intracellular trafficking to endosomes where MP, recruited by the endosomal SYTA, opportunistically interacts with NSP-vDNA to mediate the cell-to-cell movement of vDNA via plasmodesmata.
Material and Methods
Phylogenetic analysis
The amino acid full-length sequences of plant proteins were retrieved from TAIR (http://arabidopsis.org/), Phytozome v12 database (https://phytozome.jgi.doe.gov/) and amino acid full-length sequences of other species from the Ensemble database (https://www.ensembl.org/index.html). The syntaxin domain superfamily was predicted using the software HMMER v3.2.1 (http://hmmer.org/) using Pfam v.30 database (https://pfam.xfam.org/). The syntaxin domain-containing protein sequences were selected to perform the phylogenetic reconstruction. The amino acid sequences were aligned using the MUSCLE algorithm [51]. Phylogenetic trees were constructed using Bayesian Inference with MrBayes v3.2.2 [52] and mix model of amino acid substitution (Wag). The analyses were performed running 10.000.000 generations and excluding the first 2.500.000 generations as burn-in and visualized with Figtree software (http://tree.bio.ed.ac.uk/software/figtree/). Plant NISP homologous proteins were identified using the BLAST algorithm implemented in the Ensemble and Phytozome databases. The amino acid sequences of the NISP homologs were selected for phylogenetic reconstruction of N-terminal Syntaxin-6 family using the same tools and criteria applied for the superfamily syntaxin. However, the amino acid substitution model Jones was used for the N-terminal Syntaxin-6 phylogenetic tree.
PPI network analysis
NSP- and MP-interacting proteins from Arabidopsis were used as a query term to identify their respective interactions described in the BAR database (Bio-Analytic Resource for Plant Biology, http://bar.utoronto.ca/interactions/). The Biogrid and IntAct databases were selected for searching. The protein-protein interactions (PPI) were retrieved and imported into the Cytoscape software (https://cytoscape.org/) to visualize the topology of the PPI network and to calculate the network centrality metrics for each protein. The measured network centralities were betweenness, closeness, eccentricity, and degree. Briefly, the betweenness centrality in the PPI network of the graph G = (V,E) was calculated by observing the number of times a node v (protein) contributes as a link along the shorter paths among all nodes. The betweenness centrality of this node can be analyzed as: where V is the total set of nodes, σ(s,t) is the number of shorter paths of (s,t), and σ(s,t|v) is the number of paths crossing v [53]. The betweenness centrality can indicate the relevance of a protein as functionally capable of keeping together interacting proteins. Closeness centrality (of a node v is the sum of the shortest path distances from w to all other nodes and calculated as described by Freeman [54]: A high closeness value can be determined by the fact that all nodes are usually close to node v. The eccentricity centrality of a node v is the maximum distance from v to all other nodes in graph G and can be analyzed as described by Hage and Harary [55]: In addition, the degree centrality of node v was calculated by , where d represents each adjacent node and K, the total number of adjacent nodes. Nodes with a high degree can be designated hubs and hold together different nodes with a lower degree.
GO analysis of the PPI network
The gene set enrichment analysis (GSEA) of the proteins that compose the PPI network was performed using the Bioconductor GOstats package [56]. GSEA analysis computes hypergeometric test p values for over- or underrepresentation for each GO term, and the annotation of genes is based on the Bioconductor annotation package org.At.tair.
Plasmid constructions
All recombinant plasmids were generated by the GATEWAY cloning system (Invitrogen, http://www.invitrogen.com/). The primers used for gene amplification are listed in S3 Table. The coding regions of NISP (AT4G30240) and AT2G18860, with or without a translational stop codon, were amplified from leaf cDNA of Arabidopsis thaliana ecotype Columbia (Col-0) and inserted into the entry vectors pDONR201 and pDONR207. The resulting clones were designated At4g30240-St-pDONR201 (pUFV2134), At4g30240-Ns-pDONR201 (pUFV2135), At2g18860-St-pDONR201 (pUFV3193), At2g18860-Ns-pDONR201 (pUFV3194), At4g30240-St-pDONR207 (pUFV2236), At4g30240-Ns-pDONR207 (pUFV2237), At2g18860-St-pDONR207 (pUFV3195), At2g18860-Ns-pDONR207 (pUFV3196). Then, the DNA inserts were transferred by recombination from the entry vector to different yeast and plant expression vectors.
For the yeast two-hybrid assay, the NISP and AT2G18860 coding regions were transferred from the respective entry vectors (At4g30240-St-pDONR201 and At2g18860-St-pDONR201) to the yeast expression vectors pDEST32 and pDEST22, generated the clones At4g30240-St-pDEST32 (pUFV2141), At4g30240-St-pDEST22 (pUFV2142), At2g18860-St-pDEST32 (pUFV3197) and At2g18860-St-pDEST22 (pUFV3198), fused to binding domain (BD) (pDEST32) or activation domain (AD) (pDEST22) of GAL4. The NSP- and NIG-derived clones used in the yeast two-hybrid assay have been previously described and designated NSP-St-pDEST32, NSP-Cl-pDEST22 (Fontes et al., 2004), and NIG-pDEST22 (pUFV2547; Calil et al., 2018).
For the subcellular localization experiments, the NISP and AT2G18860 coding region were fused to GFP and expressed in planta under the control of the CaMV 35S promoter in the vector pK7FWG2. For the co-immunoprecipitation assay, in addition to the GFP fusion, NISP ORF was also fused to hemagglutinin (HA) tag, using the binary destination vector pK7m34GWfor three fragment recombination. The resulting DNA constructions were At4g30240-pK7FWG2 (pUFV2576), At2g18860-pK7FWG2 (pUFV3199) and 2×35S::At4g30240-Ns-6HA-pH7m34GW (pUFV2586).
For the subcellular co-localization assays, the ER-associated molecular marker ER-mCherry (CD-959), the TGN-associated molecular marker G-mCherry (CD-967) and the tonoplast-associated molecular marker vac-mCherry (CD-975) were obtained from the Arabidopsis Biological Resource Center. The plasma membrane-associated molecular marker, FLS2-mCherry was obtained by amplifying the FLS2 ORF from Arabidopsis cDNA with the primers AtFLS2-Fwd and AtFLS2-NS-Rvs (S3 Table), inserting the amplified fragment by recombination into pDON207 and transferring it by triple recombination to the destination vector pH7m34GW along with FLS2-Ns-pDON207, pDONR-P4-P1R+2×35S, and pDONR-P2R-P3+mCherry, resulting in the clone FLS2-mCherry (pUFV3034). Likewise, the SYTA ORF was amplified from leaf cDNA with the primers AtSYTA-Fwd and AtSYTA-NS-Rvs (S3 Table), cloned into the entry vector pDONR-207 and transferred via triple recombination to the destination vector pH7m34GW along with SYTA-NspDON207, pDONR-P4-P1R+2×35S, and pDONR-P2R-P3+mCherry, generating SYTA-mCherry (pUFV3296).
The NISP coding region was also fused to GFP and expressed in plants under the control of its endogenous promoter. For this, about 2000 bp of 5’ flanking sequence of At4g30240 was PCR-amplified using the primers described in S3 Table and cloned into pDONR-P4-P1R, yielding proAt4g30240-pDONR-P4-P1R (pUFV2436). Then, the 5’ flanking sequence was transferred by triple recombination to the destination vector pK7Fm34GW along with At4g30240-Ns-pDON201, proAt4g30240-pDONR-P4-P1R, and eGFP-pDONR-P2R-P3, resulting in the clone At4g30240::At4g30240-GFP (pUFV2849). The NSP- and NIG-derived clones used for co-immunoprecipitation assays have been previously described and designated as 2×35S-NSP-Cl-6HA-pH7m34GW (pUFV2424), NSP-Cl-pK7FWG2(pUFV2536), 2×35S-NIG-Ns-6HA-pH7m34GW (pUFV1946) and NIG-pK7FWG2 (pUFV1177) [48].
The SPYNE-GW and SPYCE-GW vectors, which contain the N-terminal (nYFP) and C-terminal (cYFP) of YFP, respectively, were used for the BiFC experiments. At4g30240-Ns-pDONR207 was transferred to SPYNE-GW and SPYCE-GW, yielding At4g30240-Ns-SPYNE (pUFV2238) and At4g30240-Ns-SPYCE (pUFV2239). The NSP- and NIG-derived clones used in the BiFC assays have been previously described and designated as NSP-Cl-Ns-SPYNE (pUFV1654), NSP-Cl-Ns-SPYCE (pUFV1655), NIG-Ns-SPYNE (pUFV1646), and NIG-Ns-SPYCE (pUFV1647) [48].
NISP truncated versions were obtained from At4g30240-St-pDONR201 by PCR using primers listed in S3 Table. The generated fragments were cloned into the entry vector pDONR201 and designated as At4g30240(1-107)-St-pDONR201 (pUFV2806), At4g30240(105-200)-St-pDONR201 (pUFV2807) and At4g30240(200-300)-St-pDONR201 (pUFV2808). The respective DNA inserts were then transferred to the yeast expression vectors, yielding At4g30240(1-107)-St-pDEST32 (pUFV2822), At4g30240(1-107)-St-pDEST22 (pUFV2819), At4g30240(105-200)-St-pDEST32 (pUFV2823), At4g30240(105-200)-St-pDEST22 (pUFV2820), At4g30240(200-200)-St-pDEST32 (pUFV2824) and At4g30240(200-300)-St-pDEST22 (pUFV2821).
Detecting MP- and NSP-host protein interactions on high-density halo tag-NAPPAs
The identification of NSP and MP partners was carried out through a previously fabricated high-density HaloTag nucleic acid programmable protein array (HaloTag-NAPPA), as described by Yazaki et al. [37]. The protein microarray, which consisted of a set of in situ synthesized proteins from 4600 Arabidopsis thaliana ORFs, has been designated AtNAPPA01. 3×HA-fused CaLCuV MP and NSP proteins were independently expressed using the TNT system (Promega) according to the manufacturer’s recommendations. The in vitro synthesized HA-fused MP, and HA-fused NSP were used to probe individually AtNAPPA01. The identification of MP- and NSP-interacting proteins with the NAPPA protein-protein interaction method was from a single experiment using duplicated spots on one glass slide. Interactions were considered positive only when both replicated spots displayed normalized signal intensity above background. Experiments, data scanning, quantification, and data processing were carried out as described by Yazaki et al., 2016.
Plant growth conditions and transformation
Arabidopsis thaliana, ecotype Columbia (Col-0), was used as the wild type. Homozygous seeds from the T-DNA insertional mutants nisp-1 (Salk_117751) and at2g18860-1 (Salk_099699) were obtained from the Arabidopsis Biological Resource Center. Arabidopsis seeds were surface-sterilized and cold treated at 4°C for 72 h in the dark and then exposed to white light. Seedlings were grown at 22°C on plates containing Murashige-Skoog (MS) medium for 2 weeks and then transferred to soil. Plants were grown in a growth chamber at 22 °C under short-day conditions (12h light/12h dark). Transgenic lines were obtained via A. tumefaciens-mediated transformation using the floral dip method [57]. Transformed plants with the indicated DNA constructions were selected in MS medium supplemented with the appropriate antibiotic (kanamycin, pK7FWG2). Selected transformants were confirmed by PCR, transferred to soil, and grown in a growth chamber at 22 °C under long-day conditions (16h light/8h dark) to generate seeds. For segregation analysis, seeds were germinated on MS medium containing 50 µgmL-1 kanamycin. Homozygous T1 lines for the T-DNA loci were selected by determining the frequency of their kanamycin-resistant T2 seeds after self-pollination. Accumulation of NISP-GFP and AT2g18860-GFP transcript and protein levels were monitored in each generation by qRT-PCR using gene-specific primers (S3 Table) and immunoblotting using anti-GFP antibody. From the independently transformed lines (8-15 lines of each construct and genotype transformed), we selected three independently transformed lines which appeared to have an integrated T-DNA locus on a single chromosome, since 75% of their T1 segregating seedlings were resistant to kanamycin (S4 Table). The selected lines are presented in S4 Table. The nomenclature nisp-1/NISP-1-3 indicates that the genotype nisp-1 was transformed with At4g30240-pK7FWG2, harboring the NISP ORF fused to the N-terminus of GFP under the control of the 35S promoter to generate NISP-complementing lines. The designations AtNISP-1, AtNISP-2 and AtNISP-3 indicate three independently transformed lines, which were generated by transformation of Col-0 with At4g30240-pK7FWG2 to obtain NISP-GFP-overexpressing lines. The same nomenclature was applied for the transformation of at2g18860-1 mutant and Col-0 with At2g18860-pK7FWG2, containing the At2g18860 ORF fused to the N-terminus of GFP under the control of 35S promoter (S4 Table).
RNA isolation and qRT-PCR analysis
Primers were designed using the Primer Express 3.0 (Life Technologies). Total RNA was extracted from Arabidopsis leaves using TRIzol reagent (Thermo Fisher Scientific) and quantified with a NanoDrop spectrophotometer. Total RNA (2 μg) was treated with RNase-free DNase I and reverse transcribed using oligo (dT) (0,5 μM) and 1U of M-MLV (Thermo). Real-time RT-PCR was carried out using SYBR Green PCR Master Mix (Life Technologies), DNA template, and primers in 10 μL of reaction and an ABI7500 real-time PCR system (Life Technologies). The expression of each gene was normalized to the expression of Actin and quantified using the 2-ΔCt method. The primers used to detect specific transcripts with real-time RT-PCR are listed in S3 Table.
Yeast two-hybrid assay
Competent cells of Saccharomyces cerevisiae, strain AH109 (MATa, trp1-901, leu2-3, 112, ura3-52, his3-200, gal4Δ, gal80Δ, LYS2::GAL1UAS-GAL1TATA-HIS3, GAL2UAS-GAL2TATA-ADE2, URA3::MEL1UAS-MEL1 TATA-lacZ), were cotransformed with the yeast expression vectors pDEST22 (Gal4 AD) and pDEST32 (Gal4 BD), carrying the coding region of the tested proteins, along with 100 μg of salmon sperm carrier DNA (ssDNA), using the lithium acetate/polyethylene glycol (PEG) method (Gietz and Woods, 2002). For the selection of double transformants, the cells were plated on medium deficient in leucine and tryptophan (SD, Synthetic Dropout, -Leu, -Trp) and cultured for 4 days at 28°C. Interactions were monitored by the ability of the reporter strain to grow on media lacking leucine, tryptophan, and histidine but supplemented with 2.5 mM 3-AT (3-Amino-1,2,4-triazole; Sigma) for 4 days at 28°C.
Co-immunoprecipitation assay
The in vivo interactions between NISP and NSP or NIG were monitored by co-immunoprecipitation using the μMACS™ Epitope Tag Protein Isolation (MACS/Miltenyil Biotec) kit, according to manufacturer’s recommendations. Agrobacterium tumefaciens (strain GV310)-mediated transient expression in N. benthamiana leaves was carried out as described (Carvalho et al. 2008b). Total protein extracts were obtained from infiltrated leaves expressing the recombinant proteins and incubated for 2 h with anti-green fluorescent protein (GFP) magnetic beads (MACS / Miltenyi Biotec) at 4 °C under rotation. After incubation, the extracts were applied to a MACS column for the indicated protein tag, washed five times before elution with a pre-heated buffer at 95 °C. The immunoprecipitated proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) 10% (p/v), immunoblotted with the monoclonal antibodies anti-human influenza hemagglutinin (anti-HA) epitope tag, amino acids 98-106 (Miltenyi Biotec, 130-091-972) or anti-GFP (Miltenyi Biotec, 130-091-833), and detected using Signal West Pico Chemiluminescent Substrate (Thermo Scientific) according to the recommendations of the manufacturer.
Bimolecular fluorescence complementation (BiFC) assay
Different combinations of A. tumefaciens strain GV3101 expressing the indicated proteins fused to N-YFP/C-YFP were co-infiltrated into the abaxial surface of N. benthamiana leaves at OD600nm of 1:1 proportion. Three days after infiltration, fluorescence was examined in epidermal cells using a Zeiss inverted 510 META laser scanning microscope equipped with an argon laser and a helium laser as excitation sources. YFP was excited at 514 nm using the argon laser, and YFP emission was detected using a 560–615-nm filter. The images were processed using LSM Image Browser 4 (ZEISS) software.
Subcellular localization
The subcellular localization of proteins was examined in epidermal cells of N. benthamiana leaves transiently expressing GFP fusions. Briefly, N. benthamiana leaves were agro-infiltrated with A. tumefaciens strain GV3101 carrying proAt4g30240-pDONR-P4-P1R (pUFV2436), At4g30240-pK7FWG2 (pUFV2576) or At2g18860-pK7FWG2 (pUFV3199) (Method S1) and fluorescence of the GFP fusions were visualized in epidermal cells 3 days after infiltration by confocal microscopy using a Zeiss inverted 510 META laser scanning microscope equipped with an argon laser and a helium laser as excitation sources. GFP was excited by the argon/helium-neon laser system in 488 nm wavelength, and emission was collected in 500–530 nm band-pass filter. Images were captured and processed with LSM Image Browser 4 (Carl-Zeiss) software.
For co-localization assays, NISP-GFP was co-expressed with ER-, Golgi-, tonoplast-, plasma membrane- and endosome-associated molecular markers fused to mCherry (see in plasmid constructions) and fluorescence was examined by confocal microscopy, using a Zeiss inverted LSM510 META laser scanning microscope, on whose system, the default scanning mode is sequential scanning (multitrack mode). For imaging GFP, the 488 nm excitation line and the 500 to 530 nm band pass filter were used. Excitation of mCherry was at 540 nm and emission of 608-680 nm. For the FM4-64 staining, 48 h after agro-infiltration with the DNA construct expressing NISP-GFP, the leaves were infiltrated with a solution of 50 µM FM4-64 (Molecular Probes, Eugene, OR) and examined by confocal microscopy 1h post-infiltration.
Infectivity assay
The generation of transgenic lines overexpressing NISP (AtNISp) and At2g18860 (AT2G18860), knockout lines (nisp-1 and at2g18860-1), NISP complementing lines are described in Method S1. Transgenic lines and Columbia at the seven-leaf stage were inoculated with plasmids containing partial tandem repeats of CaLCuV DNA-A and DNA-B by biolistic delivery using a microprojectile bombardment model PDS-1000/He accelerator at 900 psi. (24, 26, 58). The inoculated plants were transferred to a growth chamber and monitored for symptoms development and progress of infection [23, 59]. Total DNA was extracted from systemically infected leaves [60], and viral DNA accumulation was detected by PCR using universal DNA-B specific primers PBL1v2040 and PCRC1 [61] (S3 Table). In each experiment, 15 plants of each line were inoculated with 2 μg of tandemly repeated DNA-A plus DNA-B per plant. The course of infection was examined using data from three independent experiments.
Quantification of viral DNA in infected plants
The accumulation of viral DNA in infected plants was quantified by qPCR [26]. The reactions were prepared in 10 μL using SYBR Green Master Mix (Life Technologies), 100 ng of total DNA from systemically infected leaves, and CaLCuV-DNA-B-specific primers (S3 Table), and analyzed on a real-time PCR system. The genomic copies of CaLCuV were normalized against an internal control (18S rRNA) to account for template input variation between tubes. For viral DNA quantification, standard curves were prepared using serial dilutions of DNA-B of CaLCuV (100 to 106 copies of viral genome per reaction).
Chromatin immunoprecipitation (ChIP) assay
The in vivo interaction between NISP and viral DNA was evaluated by ChIP. NISP-overexpressing transgenic lines (AtNIS) and Columbia (Col-0) at the six-leaf stage were inoculated by biolistic delivery with CaLCuV DNA-A and DNA-B infectious clones. Tissue harvest, isolation of nuclei, and chromatin breaks from 45 days-old infected plants were conducted according to Alves et al. [62] and Yamaguchi et al. [63]. The chromatin immunoprecipitation was carried out with 1 µl of polyclonal anti-GFP serum (Life Technologies A11122; 3x diluted).
ChIP-qPCR
NSP primers (S3 Table) were initially used to amplify viral DNA-B from the immunoprecipitated DNA and input DNA to ensure DNA-B-specific amplification in samples. Likewise, DNA-A-specific primers were used to examine initial amplification of the DNA-A component. Each reaction contained 1 µL of immunoprecipitated DNA or 20-fold diluted input DNA, 0,2 mM CaLCuV-DNA-B-specific primers (S3 Table), and 5 µL of SYBR Green PCR Master Mix (Life Technologies) in 10 µL total volume. The thermal cycling conditions of the reactions consisted of an initial step at 95°C for 10 min, followed by 40 cycles at 94°C for 15 s and 58°C for 45 s. The cycle threshold (Ct) values from Col-0 and AtNISP were normalized with the Ct values of the respective inputs, and the results were expressed as a percentage of the input.
Author Contribution
B.C.G.M performed most of the experiments; L.G.C.M. performed the co-localization experiments and protoplast protein localization assays; M.D.-B. performed protein microarray assay; B.C.G.M. and J.P.B.M. performed two hybrid assays; J.C.F.S. performed in silico and statistical analyses; B.C.G.M. and A.A.S. performed BiFC assays; E.P.B.F. and J.R.E. conceived experiments; B.C.G.M. and E.P.B.F. designed all experiments, analyzed the data and wrote the manuscript.
Funding
We are grateful to the Brazilian funding agencies CAPES, CNPq, FAPEMIG, and the National Institute of Science and Technology in Plant-Pest interactions for supporting this study.
Competing interests
The authors have declared that no competing interests exist.
Supporting information
S1 Table- Protein-protein interactions between the viral proteins and the Arabidopsis proteins.
S2 Table- Enriched GO terms in three categories, Biological Process, Molecular Function, or Cellular Component ontology
S3 Table. List of primers used for cloning, PCR, RT-qPCR and qPCR
S4 Table. Expression of kanamycin resistance in the T1 generation of transgenic Arabidopsis plants.
S1 Video/Movie -Vesicle-associated localization of NISP
S1 Fig. Identification of MP- and NSP-interacting proteins.
(A) In vitro synthesized 3xHA-NSP and 3xHA-MP. In vitro translated HA-NSP and HA-MP were electrophoresed by SDS-PAGE and immunoblotted using an anti-HA antibody. (B) Protein interaction between in situ synthesized Arabidopsis hallo-proteins and 3xHA-MP or 3xHA-NSP. The AtNAPPA02 high-density array containing 4600 Arabidopsis ORFS spotted in duplicates was probed with 3xHA-MP or 3xHA-NSP, and candidate interactors were detected with an anti-HA antibody. The bright spots are signals from candidate interactors.
S2 Fig. A network of NSP- and MP-directly interacting Arabidopsis proteins.
The network was assembled by the Cytoscape software. The viral proteins are indicated in red, MP-specifically interacting proteins in yellow, NSP-specifically interacting proteins in orange and proteins that associate with both MP and NSP are indicated in green.
S3 Fig. In silico analyses of NISP.
(A) Phylogenetic tree of syntaxin superfamily domain-containing proteins from the species indicated. The phylogenetic tree was constructed using Bayesian inference performed with MrBayes v3.2.2 and the mixed amino acid substitution model Wag. The syntaxin-6 domain-containing proteins from Arabidopsis are indicated in blue, and NISP (AT4G30240) was clustered with plant-specific syntaxin-6 proteins forming the red clade. (B) Phylogenetic tree of N-terminal Syntaxin-6 domain-containing proteins. NISP (AT4G30240) clusters together with plant-specific N-terminal syntaxin-6 proteins, which do not harbor a typical tSnare domain at the C-terminus. The syntaxin-6 domain proteins from Arabidopsis are in blue. The phylogenetic tree was constructed using Bayesian inference performed with MrBayes v3.2.2 and the mixed amino acid substitution model Jones.
S4 Fig. Sequence alignment of NSP and AT2G18860
The N-terminal syntaxin-6 domain is indicated in red and the transmembrane segment in black. The sequence alignment was performed using CLUSTAL OMEGA. Identical amino acids are indicated with asterisks, highly conserved residues with (:) and lower conservation with (.).
S5 Fig. Subcellular localization of AT2G18860
AT2G18860-GFP distribution in plasma membrane-associated vesicles. N. benthamiana leaves were infiltrated with A. tumefaciens carrying a DNA construct expressing AT2G18860-GFP under the control of 35S promoter. Confocal images were taken 36 h post-infiltration. Approximately 200 cells were examined. Scale bars, 10 μm.
S6 Fig. NISP, but not AT2G18860, expression is induced by CaLCuV
NISP and AT2G18860 transcript levels were quantified by qRT-PCR in uninfected Col-0 leaves (Col-0), tungsten-inoculated Col-0 leaves (Col-0 T), and systemically infected Col-0 leaves with CaLCuV (Col-0 IN). Gene expression was calculated using the 2-ΔCt method, and actin was used as an endogenous control. Error bars indicate 95% confidence intervals based on replicated samples (n = 3) from three independent experiments.
S7 Fig. The transgenic lines are visibly indistinguishable from the wild-type plants.
Columbia (Col-0) ecotype of Arabidopsis thaliana was used as the wild-type control for phenotype comparison, and the Col-0 ecotype was used for generating almost all the transgenic plants except for the complementation test in nik1-1 and at2g18860-1 mutants. The genotypes are the same as in Fig. 5. Plants were grown in a growth chamber at 22°C under long-day conditions (16 h light/8 h dark). (A) Developmental phenotypes associated with inactivation of NISP gene and overexpression of NISP-GFP in the R3 generation of transgenic lines at 45 days after germination. (B) Developmental phenotypes associated with inactivation of NISP gene and overexpression of NISP-GFP in the R3 generation of transgenic lines at 60 days after germination. (C) Developmental phenotypes associated with inactivation of AT2G18860 gene and overexpression of AT2G18860-GFP in the R2 generation of transgenic lines at 45 days after germination. (D) Developmental phenotypes associated with inactivation of AT2G18860 gene and overexpression of AT2G18860-GFP in the R2 generation of transgenic lines at 60 days after germination
S8 Fig 8. NSP enhances NISP-NIG complex formation.
(A) NISP interacts with NIG in yeast. AD-NIG and BD-NISP fusions were expressed in yeast and interactions between the recombinant proteins were monitored by His prototrophy in selective medium (-Leu,-Trp,-His) and supplemented with 2.5 mM 3-AT. The co-expression of PRNIG fused to BD and CSN5A fused to AD was used as a positive control. (B) Cytosolic and perinuclear distribution of NIG-GFP. N. benthamiana leaves were infiltrated with A. tumefaciens carrying NIG-GFP constructs with expression driven by the 35S constitutive promoter. NIG-GFP was imaged by confocal microscopy 36 h after infiltration. (C) Cytosolic localization of NIG in transgenic lines. Confocal fluorescence image of Arabidopsis root cells stably transformed with 35S:NIG-GFP. (D) NISP interacts with NIG in planta. Total protein extracts from N. benthamiana expressing NISP-6HA and NIG-GFP were used for co-immunoprecipitation assays using anti-GFP. Input and IP show the levels of the expressed proteins NISP-6HA and NIG-GFP. Anti-HA was used to detect NISP-6HA from the immunoprecipitated complex. GFP was used as an unrelated protein. The experiment was repeated three times. (E) NISP-NIG complex formation in the presence of viral NSP. The Co-IP assay was performed as described in A, except that co-expressed NSP-GST was included in the assay. NSP-GST was detected by immunoprecipitating it from transfected leaves and immunoblotting with anti-GST. The experiment was repeated three times. (F) The interaction of NISP and NIG is increased by the presence of viral NSP. NIG-GFP levels in the immunoprecipitated complex in the presence and absence of viral NSP were quantified using the Band Analysis tools of the ImageLab software (Bio-Rad). The signal values were normalized using the IP NIG-GFP band. A.U. denotes arbitrary unit.
S9 Fig. NISP interacts with NIG in vivo and NSP enhances NISP-NIG complex formation.
(A) BIFC assay showing the interaction between NISP and NIG in vesicles of N. benthamiana leaf cells. N. benthamiana leaves transiently expressing NISP and NIG fused to the C-terminus (cYFP) or N-terminus (nYFP) of YFP, were examined by confocal microscopy 3 days after infiltration. Scale bars, 10 μm and 20 μm. (B) Co-expression of NSP-6HA strengthens interaction between NIG and NISP. The BiFC assay was performed by agro-infiltration of N. benthamiana leaves with nYFP-NIG, cYFP-NISP and vice-versus along with the expression of NSP-6HA. The reconstituted fluorescence signal was observed by confocal microscopy, 3 days after agroinfiltration. The figure displays representative samples from three independent biological repeats. Scale bars, 10 μm. (C) Confocal fluorescent image of NISP and NIG fused to the C-terminus (cYFP) or N-terminus (nYFP) of YFP in combination with empty vectors, as indicated in the figure.
S10 Fig. Schematic representation of CaLCuV DNA-A, DNA-B and ChIP-DNA primers.
(A) Schematic representation of DNA-A. The number 1 corresponds to the 5’ end of the nick site (TAATATT/AC) within the conserved nonanucleotide sequence located in the intergenic region. The ORFs are indicated with arrows. The positions of the primers used for PCR (F1/R1), and qRT-PCR (F2/R2) are indicated. (B) Schematic representation of DNA-B. The numbering scheme is the same as in a. The ORFs are indicated with arrows. The positions of the primers used for PCR (F1/R1), and qRT-PCR (F2/R2 and F3/R3) are indicated.
Acknowledgments
The authors are grateful to the microscopy facility at Universidade Federal de Viçosa.