Lettuce Big‐Vein Associated Virus ORF3 Encodes a Functional 30K Movement Protein

ABSTRACT

Movement proteins (MPs) modulate the size exclusion limit of plasmodesmata—membrane‐lined channels connecting plant cells—thereby allowing cell‐to‐cell movement and systemic spread of plant viruses. The largest and arguably best‐studied group of MPs is the 30K superfamily. Its family members share little sequence similarity, with only a handful of residues being well conserved. Yet, all family members appear to adopt the same jelly‐roll protein fold structure. Lettuce big‐vein associated virus (LBVaV), a member of the Rhabdoviridae family, is closely associated with lettuce big‐vein disease (LBVD). It appears to facilitate the long‐distance movement of Mirafiori lettuce big‐vein virus (MiLBVV) in plants through an unknown mechanism. Notably, enhanced MiLBVV spread correlates with severe LBVD symptoms. Despite LBVaV having been known for decades, its proteins have not been studied in detail thus far. By using a combination of Alphafold2 structure modelling and FoldSeek structure‐based homology searches, we managed to annotate all LBVaV open reading frames (ORFs), with ORF3 clustering with the 30K superfamily. While ORF3 is the most conserved protein sequence among the LBVaV‐encoded ORFs, it shares only 5%–11% protein sequence identity with related MPs in the same genus. Microscopy studies confirmed that ORF3 locates at plasmodesmata, and in planta expression of ORF3 allowed cell‐to‐cell movement of two movement‐impaired plant viruses. Thus, the Alphafold2‐FoldSeek strategy allowed successful annotation of a plant viral genome even when viral proteins show little sequence similarity.

AlphaFold2 + FoldSeek structure guided search predicts the movement protein of lettuce big‐vein associated virus despite low sequence similarity.

1. Introduction

The viral family Rhabdoviridae consists of viruses with a single‐stranded, negative‐sense RNA genome. The family comprises more than 580 species grouped in 62 genera to this date. Rhabdoviruses have a broad host range covering both plants and animals from arthropods, fish, reptiles to mammals (Walker et al. 2018). Despite the fact that the genomes of both animal‐ and plant‐infecting rhabdoviruses have been characterised, functional studies on the different viral open reading frames (ORFs) remain limited. In particular, cell‐to‐cell movement is poorly understood for the plant‐infecting rhabdoviruses (Ammar et al. 2009; Kumar et al. 2015; Whitfield et al. 2018).

Once a virus replicates in a single infected cell, it depends on cell‐to‐cell movement to gain access to the vasculature for systemic spread, typically via the phloem but in certain cases also through the xylem (Sun et al. 2022). Cell‐to‐cell movement relies on the ability to pass through plasmodesmata (PD) (Cheval and Faulkner 2018; Lucas et al. 2009). PD are narrow membrane‐lined structures that span neighbouring plant cell walls, thereby connecting two adjacent cells. These microscopically sized channels facilitate transport of small molecules, including sugars, ions, proteins, and hormones. PD are critical for intercellular communication and plant physiology (Kumar et al. 2015; Kumar and Dasgupta 2021). The ability of compounds to diffuse freely through PD is in part determined by their size exclusion limit (SEL), which defines the maximum size of molecules that can pass the PD (Lucas and Lee 2004; Peters et al. 2021). Plant cells dynamically modulate the SEL via different processes including deposition and degradation of callose in proximity to the PD neck (Wu et al. 2018). This dynamic regulation of callose deposition is mediated by (i) the callose synthases producing the callose that narrows the channel and (ii) β‐1,3‐glucanase that degrades the callose thereby widening the channel and increasing the SEL (Liu et al. 2021; Zavaliev et al. 2011). As the size of most plant viruses is too large to freely pass the PD, all plant viruses express one or more viral proteins to manipulate the SEL of PD, to facilitate cell‐to‐cell movement of viral components (Hong and Ju 2017; Lucas 2006), although an exception to this rule was recently reported (Ying et al. 2024).

Besides manipulating PD function, viral movement proteins (MPs) also promote intracellular transport of viral components along the cytoskeleton (Harries et al. 2010). This facilitates the movement of viral particles from their intracellular replication sites to the cell periphery and PD (Navarro et al. 2019; Niehl et al. 2013). Once localised at the PD, the viral MPs can manipulate the SEL, thereby facilitating movement of viral particles from cell to cell (Ueki et al. 2010; Wolf et al. 1989). While many viruses employ a strategy that depends on a single viral protein that organises this intracellular transport and manipulation of the SEL (e.g., the 30K‐like superfamily) (Kumar and Dasgupta 2021; Melcher 2000), others employ a more complex strategy where these viral functions are carried by two or more viral proteins (e.g., the Triple Gene Block from potato virus X) (Kumar and Dasgupta 2021; Verchot‐Lubicz et al. 2010). Members of the 30K superfamily are named after the founding member of this family, that is, the 30 kDa movement protein of tobacco mosaic virus (TMV) (Melcher 2000). The members of this family share a conserved protein domain called the jelly‐roll fold that facilitates both the intracellular trafficking of viral components and PD manipulation to increase the SEL (Mushegian and Elena 2015; Mushegian and Koonin 1993).

Lettuce big‐vein disease (LBVD) is a disease complex that poses a major challenge in lettuce cultivation. LBVD symptoms appear under lower temperature conditions encountered during the winter and early spring growing seasons. A typical manifestation of LBVD is chlorosis along the veins (vein banding) combined with plant stunting and distorted lettuce head growth (Huijberts et al. 1990; Jagger and Chandler 1934; Verbeek et al. 2013). Although LBVD was first reported in 1934, the causal agent remained ambiguous for many decades. LBVD is transmitted by Olpidium virulentus, a persistent soil‐borne chytrid fungus that produces resting spores that remain viable in the soil for up to two decades (Campbell 1985; Hartwright et al. 2010; Lot et al. 2002). These persistent resting spores make LBVD disease management difficult once fields are infested. LBVD is associated with at least two distinct plant RNA viruses, that is, Mirafiori lettuce big‐vein virus (MiLBVV; Ophiovirus mirafioriense) and lettuce big‐vein associated virus (LBVaV; Varicosavirus lactucae). Additionally, a recent report found a third lettuce‐infecting virus, Lactuca big‐vein associated phlebovirus (LBVaPV; Olpivirus lactucae), to be present as infectious virus in soil samples collected in LBVD‐infected plants (Kuwata et al. 1983; Roggero et al. 2000; Schravesande et al. 2023; van der Wilk et al. 2002). A role for this latter virus for LBVD remains to be discovered. Initially, LBVaV was reported to be the causal agent in LBVD. Later reports suggested that both viruses together were causal for the disease, but also MiLBVV alone has been suggested to be the causal agent (Alemzadeh and Izadpanah 2012; Araya et al. 2011; Hernandez et al. 2020; Navarro et al. 2004; Verbeek et al. 2013).

Our work with a collection of LBVD‐infested soil samples has shown that the spread of MiLBVV into the lettuce heads correlates with strong symptom development (Schravesande et al. 2025). Nevertheless, LBVaV was present in all samples tested and its root‐to‐shoot movement preceded in all cases the movement of MiLBVV. This suggests a facilitative role of LBVaV on MiLBVV, possibly by promoting viral systemic spread. One explanation could be LBVaV manipulating PD such that it facilitates subsequent MiLBVV spread (Schravesande et al. 2025). It is well known that PD manipulation by viral MPs allows, as a side effect, the movement of unrelated plant viruses (Dorokhov et al. 2020). As the viral ORF responsible for LBVaV movement is unknown, we here screened for this viral protein. By leveraging recent advancements in protein structure modelling, in particular AlphaFold2 (AF2) (Jumper et al. 2021) and FoldSeek (Van Kempen et al. 2023), we were able to group ORF3 into the 30K superfamily of MPs and demonstrate that it can indeed promote the spread of two movement‐deficient viruses.

2. Results

2.1. ORF3 Is a Highly Conserved Viral Protein Under Purifying Selection

LBVaV is a bipartite negative‐sense RNA virus with a segmented genome called RNA1 and RNA2. The only two proteins with described functions are (i) the RNA‐dependent RNA polymerase (RdRp) encoded on RNA1 by the largest ORF and (ii) the nucleocapsid protein encoded on RNA2 by ORF1. Whereas the RdRp is essential for viral replication, the nucleocapsid protein is responsible for genome packaging and virion stability. However, the function of the other viral proteins encoded by ORF2, ORF3, ORF4 and ORF5 in the viral life cycle remains unknown.

Previously, we sequenced a collection of LBVaV genomes that originate from a diverse set of soil samples collected in different geographical areas and years in LBVD‐infested fields (Schravesande et al. 2025). Interestingly, all protein sequences showed various levels of sequence variation except for ORF3. In fact, ORF3 differed only at the protein level at two positions (residues 15 and 17) at its N‐terminus. This suggested that ORF3 would be under purifying selection. To test this notion, we calculated the dN/dS (ω) ratios for all LBVaV ORFs (Jeffares et al. 2015) using SLAC (Single‐Likelihood Ancestor Counting) and BUSTED (Branch‐site Unrestricted Statistical Test for Episodic Diversification) (Kosakovsky Pond and Frost 2005; Lucaci et al. 2023). The analysis revealed subtle differences in selection pressure for the LBVaV‐encoded proteins on RNA2 (ORF1, ORF2, ORF4 and ORF5), while ORF3 evolved under strong purifying selection (Figure 1B), as supported by many synonymous mutations observed across its coding sequence (Figure 1C).

FIGURE 1.

ORF3 is the most conserved protein sequence of lettuce big‐vein associated virus (LBVaV). (a) A schematic representation of the viral genome of LBVaV. (b) Gene‐wide ω‐ratio for the individual open reading frames (ORFs) of LBVaV. (c) An overview of the non‐synonymous (dN) and synonymous (dS) substitution rates per residue plotted for the protein‐coding ORFs.

2.2. ORF3 Predicted Protein Structure Resembles the 30K‐Like Superfamily of Movement Proteins

As BLAST homology searches of the LBVaV proteins encoded by ORF2–ORF5 did not provide clear hints regarding their putative protein function, we employed an alternative strategy using protein structure homology to identify functional similarities based on predicted protein folds. To this end, structure predictions were performed for all ORFs using AF2, which predicts the 3D protein conformation using a multiple sequence alignment‐informed analysis strategy (Jumper et al. 2021) (Figure 2). The obtained AF2 protein models were subsequently used as input for FoldSeek—an algorithm that enables fast and sensitive searches for structure homology across large structure databases—using three public databases and one custom‐constructed database with protein (model) structures (Van Kempen et al. 2023). Specifically, the search included (i) the AFDB proteome dataset—a comprehensive subset of the AlphaFold Protein Structure Database that provides predicted structures for complete proteomes including model organisms, pathogens, crops, and a wide range of taxa, covering all reference proteomes in UniProt—and (ii) AFDB50, a clustered core version of the AFDB in which proteins sharing at least 50% sequence identity are grouped together to reduce redundancy while preserving key structural features. Furthermore, we queried (iii) PDB100, a dataset derived from the Protein Data Bank containing unique experimental structures. Because these public resources encompass few plant viral proteins, we also constructed (iv) a custom AF2 database that comprised protein models for all sequences that derive from plant‐infecting viruses in the NCBI RefSeq database. In total, 8191 viral protein sequences were retrieved and for the entire set we predicted the protein structure using AF2. The obtained models were filtered based on a pLDDT score of 50, resulting in a final set of 7545 protein models. The AF2‐FoldSeek approach enabled us to assess the structural homology of ORF2–ORF5 despite the under‐representation of plant viral proteins in the AF2 and PDB structure databases queried. Each LBVaV ORF showed some level of structure homology to a protein group in our custom virus‐specific AF2 database (Table S1). For example, ORF4 showed homology with HC‐Pro helper proteins from potyviruses and the coat protein exhibited structural similarity with nucleocapsid proteins of related rhabdoviruses, as expected (Table S1). In contrast, ORF2 and ORF5 displayed some resemblance with phosphoproteins of rhabdoviruses (Leyrat et al. 2011). Most notably, our structure homology search revealed that ORF3 grouped with proteins of the 30K superfamily in our custom virus database. In contrast, the three public databases (AFDB proteome, AFDB50 and PDB100) mainly gave hits with viral coat proteins, albeit with a lower confidence score than the hits in the custom virus database (Figure 2A, Tables S1 and S2). Inspection of the top 100 hits (Figure 2B) confirmed that the high‐confidence matches in our custom virus database were annotated as 30K MPs, whereas the hits in the public databases predominantly corresponded to viral coat proteins.

FIGURE 2.

ORF3 model adopts the fold of 30K superfamily of movement proteins (MPs). (a) Structure homology search results using the ORF3 model as a query. Searches were performed against four protein structure databases: AFDB proteome (AlphaFold2‐predicted structures for complete proteomes), AFDB50 (clustered AlphaFold2 models at 50% sequence identity), PDB100 (experimentally determined structures from the Protein Data Bank), and a custom virus database (AlphaFold2‐predicted structures of proteins from all plant‐infecting viruses in the NCBI RefSeq database). Colour depicts database source of each hit, with density plots shown per database. (b) Bar graph depicting the top 100 hits of the structure homology search divided over five classes: 30K MP, coat protein (CP), other, unknown, and CCHC‐domain containing protein. (c) Protein structure alignment of the conserved jelly‐roll domain of the 30K MP family. Different MPs from different viral families are shown. Individual β‐strands are shown with the colour marking the respective β‐sheet. Tree is constructed based on protein structure. (d) Protein model of a representative full‐length MP of tobacco mosaic virus (TMV) and ORF3 of lettuce big‐vein associated virus (LBVaV). The β‐strands that make up the jelly‐roll domain are coloured to highlight the two β‐sheets (same colours were used as Figure 2C). N‐ and C‐terminal extensions outside the jelly‐roll are visualised in grey. (e) Percentage identity heatmap of 30K MPs per viral family.

The 30K superfamily represents a class of highly divergent viral proteins that share a jelly‐roll fold protein domain and the studied members are involved in viral movement in and between plant cells (Melcher 2000). To confirm that ORF3 adopted a true jelly‐roll fold, the predicted ORF3 protein model was superimposed on the AF2 protein models of 19 viral proteins belonging to the 30K superfamily. These proteins originate from a diverse set of plant‐infecting virus families of different realms comprising both monocot‐ and dicot‐infecting viruses. A tree based on the predicted protein structures of members of the 30K superfamily is shown alongside a schematic representation of the β‐strands that collectively form the conserved jelly‐roll fold (Figure 2C). This superimposition confirmed that ORF3 adopted a jelly‐roll fold composed of seven or eight β‐strands arranged in a sandwich of two β‐sheets (Figure 2C,D). Notably, a multiple sequence alignment of protein sequences of 30K superfamily members originating from a wide variety of virus families revealed that these proteins share little protein sequence identity between plant‐infecting viral families. Similarly, the protein sequence identity of ORF3 is low with 30K MPs of the Rhabdoviridae family (Figure 2E).

2.3. ORF3 Localises at the Cell Periphery at Plasmodesmata

Given that a common feature of viral MPs, including those of the 30K superfamily, is their ability to localise to PD, we assessed the subcellular localisation of ORF3. To this end, we fused the green fluorescent protein (GFP) to the C‐terminus of ORF3 (ORF3‐GFP). Upon transient expression in Nicotiana benthamiana leaves, ORF3‐GFP was found to localise to the cell periphery in punctate structures (Figure 3). To assess whether these punctate structures coincided with PD, a known PD marker protein, Plasmodesmata Callose Binding Protein 1 (PDCB1), was co‐expressed as a fusion protein to mCherry (Grangeon et al. 2013). Indeed, ORF3 localised in punctae together with the PD‐marker protein PDCB1‐mCherry.

FIGURE 3.

ORF3 of lettuce big‐vein associated virus (LBVaV) colocalises with the plasmodesmata (PD) marker PDCB1. ORF3‐eGFP shows localisation at the cell periphery and accumulates in punctate structures along the cell membrane. The GFP‐positive punctate structures colocalise with PDCB1‐mCherry, a well‐known marker for PD labelling.

2.4. ORF3 Restores Cell‐to‐Cell Movement of Potato Virus X and Tomato Mosaic Virus

Despite differences in movement strategies between viral MPs, it is well‐known that unrelated MPs from phylogenetically distinct plant viruses can cross‐complement viral spread of movement‐deficient infectious viral clones (Hiraguri et al. 2012; Ishikawa et al. 2013; Niehl et al. 2014; Yu et al. 2013). Correspondingly, we made use of an established MP cross‐complementation assay that allows detection and quantification of this cross‐complementation on the basis of the appearance of GFP fluorescence in non‐transformed cells (Blekemolen et al. 2018). In short, a movement‐deficient infectious viral clone that expresses GFP is co‐delivered with a cassette that drives the expression of an endoplasmic reticulum (ER)‐anchored mScarlet3 fluorescent protein; the latter protein is movement‐deficient and labels specifically the primary transformed cells (Gadella et al. 2023). To create a mosaic of transformed and non‐transformed epidermal cells, a dilute bacterial mixture is then infiltrated into the leaves. This is followed by agroinfiltration to transiently deliver the individual LBVaV ORFs to assess whether their expression allows GFP accumulation in mScarlet3‐negative (non‐transformed) cells next to mScarlet3‐positive cells.

In our setup, the viral ORFs were co‐expressed with two infectious clones of plant viruses that exploit different cell‐to‐cell movement strategies, tomato mosaic virus (ToMV) and potato virus X (PVX). Both viral clones were mutated such that their MP ORF is disrupted. ToMV is a single‐stranded RNA virus belonging to the Tobamovirus genus that achieves cell‐to‐cell movement using a 30K superfamily MP (Weber et al. 1993). ToMVΔMP, a movement‐impaired GFP‐tagged ToMV clone, and individual plasmids for the expression of the different LBVaV ORFs were delivered in N. benthamiana leaves using agroinfiltration (Zhou et al. 2019). Seven days post‐inoculation (DPI), cell‐to‐cell movement of ToMV was examined. We observed GFP fluorescence in cells neighbouring mScarlet3‐positive cells when the positive control ToMV MP was co‐expressed. Also, the expression of ORF3 gave GFP fluorescence in the neighbouring cells, while none of the other LBVaV ORFs gave a detectable GFP signal beyond the initially transformed (mScarlet3‐positive) cells (Figure 4A, Figure S1). To assess if ORF3 can also restore the movement of a second virus with a different movement strategy than the 30K superfamily, a potexvirus was used that depends on the triple gene block for viral movement. The triple gene block is a set of three small overlapping ORFs found in a subset of plant viruses like potexviruses (Beck et al. 1991). These genes encode proteins crucial for the viral movement within the plant, facilitating both cell‐to‐cell movement and long‐distance transport of the virus (Verchot‐Lubicz et al. 2010). In this case, we used a movement‐deficient clone of PVX (Yang et al. 2000). We delivered the plasmid containing pPVXdf and ER‐anchored mScarlet3 with a second plasmid that allows expression of the different LBVaV ORFs in N. benthamiana leaves. Similar to ToMV, we observed cell‐to‐cell movement of PVX only in the samples where the positive control TMV MP or ORF3 was delivered (Figure 4B, Figure S2).

FIGURE 4.

ORF3 restores cell‐to‐cell movement of tomato mosaic virus (ToMV) and potato virus X (PVX). (a) ToMVΔMP, a movement‐impaired clone of ToMV expressing GFP complemented by different open reading frames (ORFs) of lettuce big‐vein associated virus (LBVaV) in Nicotiana benthamiana. ToMV movement protein (MP) and β‐glucuronidase (GUS) were used as positive and negative controls, respectively. mSc3‐HDEL marks the primary transformed cell. (b) Similar to (a) except using a movement‐impaired clone of PVX that expresses GFP (pPVXdf). The data for the other ORFs of LBVaV that lack movement properties is shown in Figures S1 and S2.

To quantify the complementation efficiency, the number of GFP‐positive cells around a primary transformed cell (mScarlet3‐positive cell) was counted for the different LBVaV ORFs and the controls. At least 30 single cell transformation events were counted per treatment (n ≥ 30) and data was classified into four ordinal groups, ranging from ‘no movement’ to ‘movement to five or more cells’. Although the MPs of ToMV and TMV showed the highest number of GFP‐positive cells, our data consistently showed only GFP spread for ORF3 and not for any other viral ORFs coded for by LBVaV (Figure 5).

FIGURE 5.

Cross‐complementation of cell‐to‐cell movement of tomato mosaic virus (ToMV) and potato virus X (PVX) is facilitated by ORF3. The number of GFP‐positive cells surrounding each primary transformed cell (mScarlet3‐positive cell) was counted for a minimum of n ≥ 30 individual single‐cell transformation events per treatment. Data were binned into four ordinal categories: (1) GFP signal only in the initially transformed cell (no movement), (2) GFP movement to one neighbouring cell, (3–4) GFP movement to two or three neighbouring cells, and (≥ 5) GFP movement to four or more neighbouring cells. The x‐axis displays the different lettuce big‐vein associated virus (LBVaV) open reading frames (ORFs) and two positive controls (tobacco mosaic virus [TMV] movement protein [MP] and ToMV MP). Statistical differences were tested using Dunn's test (p < 0.01) and shown as different lowercase letters above the bars.

3. Discussion

In a related study involving time‐course infections of lettuce plants (Schravesande et al. 2025), we demonstrated that LBVaV consistently spreads from the root to shoot prior to MiLBVV. This difference in timing suggests that LBVaV may alter PD in a way that facilitates the systemic spread of the co‐infecting virus MiLBVV. This observation prompted us to identify the viral protein(s) responsible for LBVaV movement. This work identified LBVaV ORF3 as a viral protein under purifying selection. Structural predictions revealed that ORF3 adopts a jelly‐roll protein fold characteristic of the 30K‐like MP superfamily (Melcher 2000; Mushegian and Elena 2015; Mushegian and Koonin 1993). Besides ORF3, the other viral ORFs of LBVaV were also predicted to adopt protein topologies of ‘viral proteins’ albeit with varying levels of confidence. Furthermore, cellular localisation studies confirmed that ORF3 localises to PD. Complementation studies with movement‐deficient viral clones ToMV and PVX provide evidence that only ORF3 was able to cross‐complement viral movement, confirming that ORF3 is capable of promoting viral movement in general. None of the other LBVaV ORFs showed this feature.

Our results corroborate observations by others that structural conservation within the 30K MP superfamily is driven by functional requirements of viral movement rather than protein sequence similarity (Margaria et al. 2016; Melcher 2000). Moreover, the structure conservation of the 30K superfamily is strictly limited to the jelly‐roll protein domain itself, with the N‐ and C‐terminal parts showing clear differences in both protein length and sequence. We found that two amino acid changes permitted in the ORF3 protein sequence were located at positions 15 and 17, both outside the jelly‐roll domain. In fact, AF2 had more difficulty in predicting the structure of the parts outside the jelly‐roll domain. Apparently, the jelly‐roll structure itself and not its sequence is critical for the viral function shared by the superfamily, while the more variable regions likely play a role that is species/genus specific (Hak et al. 2023; Melcher 2000; Rossmann 2013). A recent study showed that the core domain of 30K‐like MPs is homologous to the jelly‐roll domain present in certain coat proteins of plant‐infecting RNA and DNA viruses, suggesting that the 30K superfamily and these viral coat proteins share a common ancestor (Butkovic et al. 2023). Intriguingly, despite the overall lack of protein sequence conservation between the 30K superfamily of MPs, the LxDx peptide motif stands out as the only universally conserved motif (Melcher 1990). Previous studies have demonstrated that this motif is indispensable for MP function. For example, site‐directed mutagenesis experiments that targeted the conserved aspartate within the LxDx motif showed that this change abolished the ability of the MP to facilitate cell‐to‐cell movement of viral nucleic acids (Mann et al. 2016; Zhou et al. 2019). However, an intriguing question remains, that is, what is the cause of the remarkable sequence conservation of ORF3 at the LBVaV species level given the absence of any significant protein sequence similarity with other rhabdovirus MPs or other MPs belonging to the 30K superfamily? Interestingly, sequence conservation within the 30K superfamily of MPs is not unique to ORF3 of LBVaV. A similar phenomenon has been observed for the MP of ToMV, where a conserved residue (C68) is critical for both viral movement and host immune recognition (Hak et al. 2023). Structural analysis revealed that C68 is embedded within the jelly‐roll fold and plays a key role in ER association and PD targeting (Spiegelman and Dinesh‐Kumar 2023). Functional constraints imposed by these structural features likely contribute to the sequence conservation observed despite the overall sequence divergence among viral species. The remarkable sequence conservation of ORF3 at the LBVaV species level suggests the presence of a second unknown viral function. Interestingly, the MP of the bipartite varicosavirus LBVaV is encoded on RNA2, in a genomic position analogous to the MP genes of the monopartite cyto‐ and nucleorhabdoviruses. This suggests that the genomic arrangement of the MPs is apparently retained within the Rhabdoviridae family despite differences in their genome architecture.

The localisation of ORF3 to both the PD and cytosol, as reported here, is consistent with data on other viral MPs and it supports the idea that ORF3 has a multifunctional role in LBVaV infection cycle (Powers et al. 2008; Yaegashi et al. 2007). Studies with TMV show that the 30K MP of TMV localises at the cytosolic face of the ER membrane and exhibits a capacity to recruit ER membranes to microtubules (Ferralli et al. 2006; Peiró et al. 2014). Furthermore, the reliance of viral MPs on the host cell's ER‐actin network is highlighted by studies on the MP of cucumber mosaic virus (Su et al. 2010). This MP actively disrupts F‐actin connections and inhibits actin polymerisation in N. benthamiana, demonstrating that interference with the ER‐actin network can be a viral strategy to modify host cell architecture and facilitate viral spread. Further interaction partners, such as myosins and Plasmodesmata‐located protein 1 (PDLP1), have also been identified. In N. benthamiana, myosins belonging to classes VIII and XI were found to interact with the MP of TMV (Amari et al. 2014), whereas PDLP1 was shown to interact with the MP of grapevine fanleaf virus (GFLV) (Amari et al. 2010). Together, these studies indicate that these structurally similar viral MPs use different approaches to modify PD and facilitate cell‐to‐cell movement, raising the question of whether they utilise a shared set of interaction partners in this process.

In conclusion, our findings identify ORF3 as a MP that likely facilitates cell‐to‐cell viral movement and spread of LBVaV. At the same time, we demonstrate that protein structure tools like AF2 (and now AF3 (Abramson et al. 2024)) can support viral annotation by predicting structure similarities for viral proteins that share low sequence similarity between each other.

4. Experimental Procedures

4.1. Custom Virus Database Construction

A custom‐made protein sequence database was generated that comprised all protein sequences that derive from plant‐infecting viruses present in the NCBI RefSeq database (31‐10‐2023). In total, 8191 protein sequences were retrieved and used as query sequences for protein structure predictions. Colabfold v. 1.5.2 (using localcolabfold), which is based upon AlphaFold v. 2.3.1 (Jumper et al. 2021), was used for protein model prediction (‐‐random‐seed 101 ‐‐num‐seeds 3 ‐‐use‐dropout ‐‐num‐models 1 ‐‐num‐recycle 8 ‐‐recycle‐early‐stop‐tolerance 0.5). No templates were used during the protein model prediction. Model predictions were performed on a NVIDIA A100 80GB GPU. The uniref30_2302 and colabfold_envdb_202108 databases were used to generate the multiple sequence alignments (https://colabfold.mmseqs.com/) (Mirdita et al. 2022), on an AMD EPYC 9454 processor with 192 cores and 1.5TB HPC CPU node. The predicted structures were filtered based on their predicted Local Distance Difference Test (pLDDT) value, resulting in a set of 7545 protein structures with a pLDDT ≥ 50 (Mariani et al. 2013). The dataset is available online (https://doi.org/10.21942/uva.28417079). Using this custom target database and the publicly available databases, a protein structure search was performed using FoldSeek (https://github.com/steineggerlab/foldseek) (Van Kempen et al. 2023).

4.2. Plant Material

Nicotiana benthamiana plants were sown on potting soil and grown at 23°C for 3 weeks. The plants were transplanted into 12 cm pots and transferred to a growth chamber under long‐day conditions with 16 h of light and 8 h of darkness at a temperature of 19°C to create optimal conditions for transformation by Rhizobium radiobacter, previously known as Agrobacterium tumefaciens, strain C58C1. Infected plant material was sampled from infected Lactuca sativa 'Iglo' plants. In short, L. sativa plants were sown on soil blocks and grown at 14°C for 2 weeks. The seedlings were transplanted into a hydroponic gutter system in which virus‐containing O. virulentus spores circulate. The isolate was collected in‐house from infected lettuce plants collected in De Lier, Netherlands. After 8–12 weeks, the disease symptoms are visible. Infected leaf material was directly flash frozen with the use of liquid nitrogen, for later processing.

4.3. Plant Expression Vector Construction

Total RNA was isolated from flash‐frozen infected L. sativa material. Unless otherwise stated, manufacturers protocols were followed. Leaf material was ground using a mortar and pestle to a fine powder and used as input material for total RNA isolation using the RNeasy Plant Mini Kit (Qiagen). The isolated total RNA was used for cDNA synthesis using the Maxima H Minus First Strand cDNA Synthesis Kit using random hexamer primers (Thermo Scientific). The ORFs of LBVaV were cloned into Gateway‐ready donor vectors. All PCR amplifications were performed using Phusion High‐Fidelity DNA polymerase (NEB) in a T100 thermal cycler (Bio‐Rad). The following PCR programme was used unless stated otherwise: initial denaturation at 98°C for 30 s; followed by 40 cycles of 98°C for 10 s, 58°C for 10 s, 72°C for 30 s per kb; final extension at 72°C for 5 min. PCR products were excised from a 1% agarose gel and purified using the Monarch DNA GEL Extraction Kit (NEB). Primer sequences used for the amplification of the PCR products are stated in Table S3. The purified attB‐flanked ORFs were cloned into pDONR207 using BP clonase II (Thermo Scientific) and transformed into electrocompetent Escherichia coli TOP10 (Thermo Scientific). Positive recombinants were identified using gentamicin selection, verified through colony PCR and Sanger sequenced using appropriate primers. Overnight cultures were used to generate plasmid preps using the Monarch Plasmid Miniprep Kit (NEB). Donor vectors were recombined with a destination vector containing the desired fluorescence or epitope tag (pGWB405 and pGWB406 for C‐ and N‐terminal eGFP tagging, respectively, and pGWB414 for HA‐tagging) using LR clonase II (Thermo Scientific) and transformed into electrocompetent E. coli . Destination vectors were obtained from the laboratory of Nakagawa (Nakagawa et al. 2007). Transformants were identified using spectinomycin selection and verified through colony PCR and transformed into R. radiobacter C58C1.

The pPVXdf plasmid (Komarova et al. 2006) contains an infectious clone of PVX carrying the eGFP gene fragment that is cloned into the binary vector pBIN19 and is incapable of cell‐to‐cell movement. The mScarlet3‐HDEL cassette was synthesised flanked by Sac I and Sal I restriction sites (Integrated DNA Technologies). The product was digested with SacI‐HF and SalI‐HF in the Cutsmart buffer (NEB) and cleaned up using Ampure XP magnetic beads (Beckman Coulter). The digestion product was then ligated into a SacI‐HF‐ and SalI‐HF‐digested pPVXdf plasmid to generate PVXdf; mScarlet3‐HDEL, hereafter referred to as pWSH307. As a positive control, we constructed a plasmid containing TMV_MP‐HA by amplifying the movement protein from a TMV infectious clone, described by Peart et al. (2002). We inserted the TMV_MP PCR product into pGWB414 using the Gateway system.

The pToMVΔMP construct (Zhou et al. 2019) contains an MP‐defective clone of ToMV carrying a GFP reporter. The mScarlet3‐HDEL cassette was amplified from vector pWSH307, using primers containing overlap for Gibson cloning (Table S3). The HDEL motif, a well‐established ER retention signal, ensures that the mScarlet3 fluorescent reporter is confined to the ER. The amplified mScarlet3‐HDEL cassette was inserted into the pToMVΔMP by linearising the plasmid with PmeI in Cutsmart buffer (NEB) and inserting the cassette using NEBuilder HiFi DNA Assembly Master Mix (NEB) to generate ToMVΔMP; mScarlet3‐HDEL, hereafter referred to as pWSH306. For complementation, pToMV_MP was used (Zhou et al. 2019). All plasmids are given in Table S4.

4.4. Agroinfiltration

Rhizobium radiobacter cultures were grown overnight at 28°C with 200 rpm agitation in liquid Luria Bertani medium containing the appropriate antibiotics depending on the bacterial selection marker and 50 μg/mL rifampicin. Cells were collected by centrifugation at 4000 g for 10 min, and the pellet was resuspended in Induction Medium consisting of 5× M9 salts (Sigma‐Aldrich), 1% wt/vol glucose, and 50 μM acetosyringone. The bacterial suspension was incubated at 28°C and 200 rpm agitation for 4 h and then centrifuged at 4000 g for 10 min. The cell pellet was resuspended in Infiltration Medium consisting of 0.5× Murashige & Skoog salts, 10 mM MES pH 6, 0.5% wt/vol sucrose, 0.5% wt/vol fructose, and 150 μM acetosyringone. Each suspension was adjusted to a final OD₆₀₀ of 0.5 (unless stated otherwise). Infiltrations were performed using a blunt syringe, infiltrating the abaxial side of 4‐ to 5‐week‐old N. benthamiana leaves.

4.5. Subcellular Localisation Assay

Four‐week‐old N. benthamiana plants were co‐infiltrated with the LBVaV ORFs cloned in pGWB405, the PD marker PDCB1‐mCherry, originally described by Grangeon et al. (2013), and the pGD‐VSRs silencing suppressor construct (Wang et al. 2015). Plants were maintained for 2 days at 19°C and a day/night cycle of 18/6 h. Confocal microscopy was performed using a Leica SP8 confocal microscope with a HC Plan Apo 63×, NA1.2 (water immersion) objective lens. eGFP was visualised by excitation at 488 nm with an argon ion laser, and emission was recorded in the spectrum of 497–517 nm. mCherry was visualised by excitation at 594 nm with a HeNe laser; emission was recorded at 600–620 nm.

4.6. Cell‐To‐Cell Movement Complementation Experiments

Suspensions of R. radiobacter C58C1 strains carrying either pWSH306 or pWSH307 (OD₆₀₀ 1.5) were diluted 5000× and mixed in a 1:1:1 ratio with R. radiobacter carrying the pGD‐VSRs and each of the LBVaV ORFs cloned in pGWB414. These mixtures were infiltrated into the abaxial leaf side of 4‐week‐old N. benthamiana plants. Confocal microscopy was performed using a Leica SP8 confocal microscope with a HC Plan Apo 63×, NA1.2 (water) objective lens. GFP was visualised by excitation at 488 nm with an argon ion laser, and emission was recorded in the spectrum of 497–517 nm. mScarlet3 was visualised by excitation at 594 nm with a HeNe laser; emission was recorded at 600–620 nm. Foci size scoring was performed at 7 days post‐inoculation with a fluorescence microscope (Zeiss Axio Scope A1).

Author Contributions

Willem E. W. Schravesande: conceptualization (equal), formal analysis (lead), investigation (lead), methodology (lead), visualization (lead), writing – original draft preparation (lead). Machiel V. Cligge: formal analysis (supporting), investigation (supporting). Raoul Frijters: software (lead), formal analysis (supporting), investigation (supporting). Adriaan Verhage: conceptualization (equal), supervision (equal), writing – review and editing (equal). Harrold A. van den Burg: conceptualization (equal), supervision (equal), writing – review and editing (equal).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: ORF3 restores cell‐to‐cell movement of ToMV. pToMVΔMP, a movement‐impaired clone of ToMV expressing GFP, complemented by the different ORFs of LBVaV, in Nicotiana benthamiana. ToMV MP, the MP of ToMV and GUS, facilitating no cell‐to‐cell movement capabilities, were taken along as positive and negative controls. mSc3‐HDEL functions as a marker to detect single‐cell transformation events.

Figure S2: ORF3 restores cell‐to‐cell movement of PVX. pPVXdf, a movement‐impaired clone of PVX expressing GFP, complemented by the different ORFs of LBVaV, in Nicotiana benthamiana. TMV MP, the MP of TMV and GUS, facilitating no cell‐to‐cell movement capabilities, were taken along as positive and negative controls. mSc3‐HDEL functions as a marker to detect single‐cell transformation events.

Table S1: Foldseek analysis of LBVaV ORFs against VirusDB. Each row represents a query‐target pair, where the LBVaV ORF is compared against its best‐matching viral protein from VirusDB. The target proteins are identified by their accession numbers and annotated with predicted functions and corresponding viral species.

Table S2: Foldseek results LBVaV ORF3 against AFDB50, AFDB proteome, PDB100, and custom VirusDB. Each entry lists the source database, target identifier, protein annotation and associated scientific name, followed by the e‐value and bitscore.

Table S3: List of primers used in study.

Table S4: List of plasmids used in study.

Schravesande, W. E. W. , Cligge M. V., Frijters R., Verhage A., and van den Burg H. A.. 2025. “Lettuce Big‐Vein Associated Virus ORF3 Encodes a Functional 30K Movement Protein.” Molecular Plant Pathology 26, no. 9: e70153. 10.1111/mpp.70153.

Data Availability Statement

The data that support the findings of this study are openly available in FigShare at https://doi.org/10.21942/uva.28417079.