Alternanthera mosaic virus - an alternative ‘model’ potexvirus of broad relevance

Introduction

The potexvirus now known as Alternanthera mosaic virus (AltMV) was first reported from Australia, infecting Alternanthera pungens (Amaranthaceae) (Geering and Thomas, 1999), which is designated as a noxious weed in both the USA and Australia. AltMV was initially described as closely related to Papaya mosaic virus (PapMV), to which AltMV is serologically very similar, as well as being the most closely related potexvirus (c.63 - 64% pairwise full genome nucleotide

identity). The next report was of an isolate affecting the ornamental moss rose, or portulaca, (Portulaca grandiflora; Portulacaceae) in Italy (Ciuffo and Turina, 2004), which was followed by a full genome sequence of an isolate from creeping phlox (Phlox stolonifera; Polemoniaceae) and identification of another portulaca isolate in the USA (Hammond et al., 2006a).

Since then, AltMV has been identified in many different hosts, primarily among ornamental crops, causing varying symptoms, largely of the mottle or mosaic type but occasionally inducing systemic necrosis under cool growing conditions. AltMV has been shown to differ in some interesting features from Potato virus X (PVX), the type member of the genus Potexvirus, and thus offers an alternative model system for understanding the behavior of potexvirus. In addition, naturally-occurring variability in the AltMV Triple Gene Block 1 (TGB1) protein been identified, affecting the efficiency of RNA silencing suppression (Lim et al., 2010a), as a result of which AltMV has been shown to have significant potential as a tool for biotechnology, with utility for both protein expression and Virus-Induced Gene Silencing (VIGS) depending upon the variant of TGB1 protein included in the viral vector construct (Lim et al., 2010c).

Here we summarize the current state of knowledge regarding AltMV, including host range, strain differentiation, host interactions, and utility as a plant viral vector for both protein expression and VIGS approaches to reverse genetics.

Host range, Transmission, and Mixed Infections

Since its initial discovery, the known host range of AltMV has expanded rapidly, including natural hosts in many taxonomically divergent plant families, and additional experimental hosts. Some caution must be applied to interpretation of host identifications not supported by sequencing of the virus isolate in question, as the primary commercially available test is utilized as a dual test for AltMV and PapMV, but PapMV is reported to have a very limited host range (Purcifull and Hiebert, 1971; Büchen-Osmond et al., 1988; Noa-Carazana et al., 2006), and AltMV has been identified by sequencing from multiple hosts, or ability to infect other hosts confirmed by inoculation with characterized AltMV isolates (e.g. Geering and Thomas, 1999; Ciuffo and Turina, 2004; Hammond et al., 2006a; Baker et al., 2006; Duarte et al., 2008; Lockhart and Daughtrey, 2008; Tang et al., 2010; Vitoreli et al., 2011; Iwabuchi et al., 2016; Hammond, unpublished).

Whereas the natural host range of PapMV is restricted to papaya (Carica papaya, Caricaceae), tree spinach (Cnidoscolus chayamansa, Euphorbiaceae), and ulluco or quecha (Ullucus tuberosus, Basellaceae) (Purcifull and Hiebert, 1971; Büchen-Osmond et al., 1988; Noa-Carazana et al., 2006), about 17 species in nine families of dicotyledonous plants have been reported to be experimentally infected (Purcifull and Hiebert, 1971). The reported natural host range of AltMV is very much larger, including species in at least 24 taxonomically diverse families (Geering and Thomas, 1999; Ciuffo and Turina, 2004; Hammond et al., 2006a; Baker et al., 2006; Duarte et al., 2008; Lockhart and Daughtrey, 2008; Tang et al., 2010; Vitoreli et al., 2011; Iwabuchi et al., 2016; Hammond, unpublished; and Agdia, personal communication to JH; see Table 1). The experimental host range encompasses at least 38 species from 14 diverse dicotyledonous plant families, including seven families not (yet) reported to be naturally infected (Hammond et al., 2006a; Lim et al., 2010c; and Table 2), meaning that at least 31 families contain species susceptible to AltMV infection. The number of families containing susceptible species is only expected to grow as more species (and especially ornamental species produced in the same locations as currently known susceptible ornamentals) are tested.

AltMV transmission may occur through soil in which infected plants have been grown, and the virus may be both exuded from the roots of infected plants, and taken up from infested soil by the roots of healthy plants, but this remains to be demonstrated. In nursery production systems, transmission is likely to occur through routine horticultural operations such as propagation, pinching or pruning, flower cutting, etc. - and if tools are not sterilized between crops, readily transmitted from one susceptible crop variety or species to another. In experimental situations, a high percentage of plants become infected following mechanical inoculation using even high dilutions of infected sap; a high degree of success is achieved with different molarity buffers from pH 7.0 to 1% K₂HPO₄ in deionized water (c. pH 8.7).

Mixed infections of AltMV with other virus types have been reported in multiple crops; in various phlox species, AltMV has been identified in combination with Phlox virus M (PhlVM; carlavirus) in annual phlox hybrids (Hammond and Reinsel, 2011), and with Angelonia flower break virus in Angelonia (Lockhart and Daughtrey, 2008). In Phlox stolonifera, AltMV has been detected in multiple mixed infection with Phlox virus S (carlavirus), Tobacco ringspot virus (nepovirus), Spiranthes mosaic virus 3 (SpiMV-3; potyvirus), and a probably distinct second potyvirus (Hammond and Reinsel, 2001); mixed infections of AltMV, an unidentified tobamovirus, and a SpiMV-3-related potyvirus were also detected in Phlox divaricata (Hammond and Reinsel, 2011). AltMV was also detected in Mazus reptans in mixed infection with Ligustrum necrotic ringspot virus (carlavirus; Henderson et al., 2014); and in Nandina domestica in mixed infection with Apple stem grooving virus (capillovirus; Tang et al., 2010). In some instances mixed infections may lead to increased symptom expression (e.g. mixed with PhlVM in annual phlox hybrids; J. Hammond, unpublished).

Infections of two distinct genotypes of AltMV were also detected in an isolate from creeping phlox (Phlox stolonifera) which had been passaged mechanically over a period of five years prior to creation of full-length infectious clones; subsequent analysis of stored frozen tissue demonstrated that both isolates were present from an early stage, and probably in the original phlox plant (Lim et al., 2010a). The two infectious clones were subsequently individually marked by insertion of either the green fluorescent protein (GFP) or the red fluorescent protein, DsRed, and shown to co-exist within the same leaf areas, with some cells dually infected; this suggests that the milder isolate ameliorated the symptoms of the more severe isolate, resulting in less severe effects on the host (Lim et al., 2010a).

AltMV Detection and Identification

AltMV was initially identified as closely related to PapMV, which remains the most closely related species. Indeed, commercially available serological tests (Enzyme linked immunosorbent assay and lateral flow serological tests) utilize antibodies which detect both PapMV and AltMV. AltMV is a good immunogen, and an excellent rabbit polyclonal antiserum has been produced (Geering and Thomas, 1999).

A number of primer sets for AltMV detection and identification have been developed for either RT-PCR, including PP1/PP4 (Hammond et al., 2006a), PP23/BNSNC (for cDNA primed with NSNC-odT; Hammond et al., 2006b) and PP12/PP15 (Lim et al., 2014); or for real-time quantitative RT-PCR, including Coat-F/Coat-R and RdRp-F/RdRp-R (Lim et al., 2010a).

Virus Propagation and Purification

AltMV is readily propagated and maintained in Nicotiana benthamiana, with transmission by mechanical transmission using sap extracts prepared in 10 - 100 mM potassium phosphate buffer, pH 7 - 7.5, or 1% K₂HPO₄, either with a pinch of Celite added to the extract as an abrasive, or dusting of the leaves to be inoculated with carborundum powder, according to individual preference. Systemically-infected leaves of N. benthamiana harvested between two to four weeks after inoculation can either be extracted fresh, or stored frozen until it is convenient to perform the purification. We have used a purification protocol first developed for Plantain virus X (Hammond and Hull, 1981), and subsequently adapted for potyviruses (Hammond and Lawson, 1988), and which also works well with multiple potexviruses, carlaviruses, and Lolium latent virus (e.g. Hammond et al., 2006a; Vaira et al., 2008; Hammond, unpublished). Yields of 8.6 - 12.5 mg AltMV/100g of infected tissue were obtained from leaves harvested from 19 - 28 days post inoculation (dpi) (Hammond et al., 2006a).

Partially purified and concentrated TGB1 inclusions could be prepared from the same tissue extract by following the potyvirus ‘cytoplasmic inclusion pellet’ fraction as described (Hammond and Lawson, 1988), centrifuging the pellet fraction over 50/60/70/80% sucrose step gradients, and collecting the material from the interfaces between each of the gradient steps; most of the TGB1 crystal-like aggregates were found at the 70%/80% sucrose interface and within the 80% layer (Seo et al., 2014).

Strain Differentiation

Alignment of coat protein (CP) sequences of AltMV isolates from multiple species reveal several specific amino acid residue differences which differentiate isolates into ‘phlox-like’ and ‘portulaca-like’ groups; the ‘phlox-like’ isolates included most from different Phlox species, plus the original isolate from Alternanthera pungens, and an isolate from Nandina domestica, whereas the ‘portulaca-like’ group included four isolates from Portulaca grandiflora, plus one each from an annual phlox hybrid and from a Pericallis hybrid (Hammond and Reinsel, 2015). Phylogenetic trees based on the Triple Gene Block 2 (TGB2) and 3 (TGB3) proteins and CP amino acid sequences separated the ‘phlox-like’ and ‘portulaca-like’ isolates into distinct clades (Hammond and Reinsel, 2015).

In TGB3, all ‘portulaca-like’ isolates have T(10)A and DY(56,57)EH compared to ‘phlox-like’ isolates (Jang et al., 2013; Hammond and Reinsel, 2015), but to date no specific biological difference has been associated with these variations.

In the CP, all of the ‘portulaca-like’ isolates have MN(13,14)ID, T(66)A, L(76)I, and T(154)A compared to ‘phlox-like’ isolates; five ‘portulaca-like’ isolates have A(77)S, and four E(78)G compared to ‘phlox-like’ isolates (Fig. 1). At least some of these differences are related to the ability of AltMV isolates to induce systemic necrosis at low temperature, as substitution of either MN(13,14)ID or LA(76,77)IS in an infectious clone was found to eliminate induction of systemic necrosis in Nicotiana benthamiana at 15℃ (Lim et al., 2014).

The recently-described isolate from Achyranthes bidentata has the most divergent sequence (Iwabuchi et al., 2016), and the CP has multiple variant amino acids, mainly in the N-terminal half, that are unique to this isolate (Fig. 1).

Within the 1,541 residue RdRp protein, a total of 68 amino acid residues differentiate the fully-sequenced isolates available before the sequence of AltMV-Ac became available; addition of the divergent AltMV-Ac to the alignment reveals a total of 190 positions at which at least one isolate differs from the consensus, with variations in AltMV-Ac particularly concentrated in the region between residues 425 - 535, between the Methyl transferase (MT) and 2-oxyglutarate/FE(II)-dependent oxygenase (2OG-F) domains. There are many additional RdRp residues at which AltMV is distinguished from the next most-closely related potexvirus, PapMV (Hammond and Reinsel, 2015), also concentrated in the region between MT and 2OG-F domains (Supplemental Fig. 1). However, the RdRp of the fully-sequenced portulaca isolate was not obviously distinguished from three phlox isolates in a phylogenetic tree (Hammond and Reinsel, 2015).

The 3’-terminal portion of the genome including the TGB3 and CP genes is not available for all isolates for which partial genome sequences have been determined. However, partial RdRP sequences of isolates from angelonia (Lockhart and Daughtrey, 2008), torenia (Duarte et al., 2008), and skullcap (Baker et al., 2006) share several variant residues differing from a nandina isolate (Tang et al., 2010) which is identical in this region to phlox isolate AltMV-PA; one of the variant residues from the Angelonia, torenia, and skullcap isolates is also shared with the full sequence of a portulaca isolate (Hammond and Reinsel, 2015). These isolates may therefore be considered ‘portulaca-like’.

Little differentiation was observed between the CP nucleotide or 3′ untranslated region (UTR) sequences of different isolates, with ‘phlox-like’ and ‘portulaca-like’ isolates all falling within a single clade; the 3′ UTR is presumed to be conserved in order to maintain replicase recognition signals (Hammond and Reinsel, 2015).

The differences in distribution of ‘phlox-like’ and ‘portulaca-like’ isolates may change as more isolates from different crops are sequenced, but it is notable that most of the various Phlox spp. and Nandina domestica are perennial crops, whereas the only phlox isolate which is ‘portulaca-like’ was isolated from an annual phlox hybrid. We have therefore speculated that the annual crop portulaca may have been the source of infection for other herbaceous and annual bedding plants, whereas phlox may be the sources for most of the perennial crops, as annual and perennial ornamentals are typically produced by separate groups of nursery producers (Hammond and Reinsel, 2015). The recent report of a ‘phlox-like’ isolate from the perennial ornamental penstemon (KT210149; Fig. 1) supports this suggestion.

Ultrastructure and Cytopathology

AltMV can reach very high concentrations in suitable hosts, such as N. benthamiana. Virus particles accumulate in the cytoplasm in ‘fingerprint whorls’ typical of other potexviruses and carlaviruses, or in banded inclusions, which are composed of layers of quasi-parallel particles, often with a pronounced curvature of the individual particles. In some cells, almost the whole of the cytoplasm may appear to be taken up by accumulated virions (Fig. 2A). Areas of banded inclusions may also have the ‘bands’ separated by large numbers of particles oriented almost perpendicular to the plane of sectioning, so that particles are seen in cross-section or at an oblique angle (Fig. 2B). In some infected cells crystalline structures presumed to be accumulations of TGB1 are detected both in the cytoplasm and in nuclei, often near cytoplasmic arrays of virions; these crystals cannot be found in every cell, and it is not yet clear whether the crystals represent active forms of TGB1 or accumulations of excess protein (Fig. 3A). Similar crystalline aggregates have been observed in TGB1-containing fractions purified from AltMV-infected plants (Seo et al., 2014). The appearance of the crystals differs depending on the angle of the section, with crystals appearing to have parallel striations of two different periodicities, or hexagonal arrays of apparent tubular construction; TGB1 aggregates in the cytoplasm often appear to have ribosomes closely associated with the outsides of the crystals (Fig. 3B). The nuclear aggregates are sometimes observed in close association with the nucleolus (Fig. 3C), which is confirmed by localization of GFP:TGB1_L88 fusions to the nucleolus in images captured by confocal microscopy (see below).

Localization of Fluorescent-labeled AltMV Proteins

TGB1

Two distinct forms of TGB1 have been identified in naturally occurring isolates of AltMV; the prevalent form has Leu at residue 88 (TGB1_L88), while a mild form present as a minor constituent of isolate AltMV-SP has a substitution of Pro for Leu (TGB1_P88) (Lim et al., 2010a). A GFP:TGB1_L88 fusion expressed in N. benthamiana by agroinfiltration was observed to localize in punctate spots at the cell periphery, at the nuclear membrane, and at the nucleolus (Fig. 4; identified by co-localization with fibrillarin and absence of staining by DAPI; Lim et al., 2010c); when cells were plasmolyzed, TGB1 remained at the cell wall (Fig. 5B). In contrast, GFP:TGB1_P88 formed larger cytoplasmic aggregates and accumulated only at the nuclear periphery without nucleolar localization (Fig. 4C; Lim et al., 2010c). When AltMV GFP:TGB1 was co-expressed with Barley stripe mosaic virus (BSMV) DsRed:TGB3, both localized in punctate spots along and either side of the cell wall (Fig. 6A), typical of the localization of BSMV TGB3 at plasmodesmata (Lim et al., 2009).

When TGB1 variants were examined for interaction with TGB1_L88 by Bimolecular fluorescence comple-mentation (BiFC), TGB1_P88, TGB1_D(81)R, and TGB1_Q(101)R were found to interact with TGB1_L88, while mutant TGB1_G(31)R and TGB1_GK(33,34)RR did not; GFP:TGB1 mutant D(81)R and Q(101)R localized at the nuclear periphery and the nucleolus, as for TGB1_L88, whereas mutants G(31)R and GK(33,34)RR were distributed throughout the nucleoplasm but not at the nucleolus, and natural variant L(88)G showed nuclear but not nucleolar localization (Nam et al., 2013). Interestingly, a balance of nuclear and cytoplasmic aggregates comparable to the distribution of TGB1 crystals observed by electron microscopy was not observed. Localization of mutants could be related to efficacy of silencing suppression (see below).

TGB2

AltMV GFP:TGB2 was found to localize to the cell periphery as small punctate aggregates in the epidermal layer, and to the ER network in mesophyll cells; when epidermal cells were plasmolyzed, the GFP:TGB2 remained attached to the plasma membrane, and separated from the cell wall (Fig. 5C). No change in TGB2 localization was observed when unfused TGB3 was co-expressed.

TGB3

AltMV GFP:TGB3 or DsRed:TGB3 was rarely observed in the epidermal layer, associated with the cell wall (Fig. 5D), but instead found almost exclusively in the mesophyll layer, in close association with chloroplasts (Lim et al., 2010b). No alteration of the localization of DsRed:TGB3 in the presence of unlabeled TGB2, TGB2 + TGB3, or TGB3 was observed, indicating a lack of AltMV TGB3 interaction with TGB2 (Lim et al., 2010b).

Studies of subcellular localization thus revealed several differences between AltMV and other potexviruses; whereas interactions between TGB2 and TGB3 of PVX is reported at the endoplasmic reticulum (ER) and in ER-associated granular vesicles (Schepetilnikov et al., 2005; Ju et al., 2007; Samuels et al., 2007), no interaction or co-localization was observed for AltMV TGB2 and TGB3, and AltMV TGB3 was instead found to be closely associated with the chloroplast membrane (Lim et al., 2010b). In addition, while DsRed-labeled PVX TGB3 (DsRed:TGB3_PVX) expressed by agroinfiltration was localized almost exclusively to the epidermal layer, the equivalent AltMV construct, DsRed:TGB3_AltMV, was found primarily distributed in the mesophyll layer, in association with chloroplasts; combined expression of GFP:TGB3_PVX and DsRed:TGB3_AltMV confirmed the separation of PVX and AltMV TGB3s between the epidermal and mesophyll layers respectively (Lim et al., 2010b). The mesophyll localization of AltMV TGB3 was also emphasized by comparison to the distribution of BSMV TGB3, which was distributed primarily in the epidermal layer (Fig. 7).

N-terminal deletions of AltMV TGB3:GFP or TGB3:DsRed, and C-terminal deletions of GFP:TGB3 or DsRed:TGB3 were used to determine that the N-terminal 18 TGB3 residues contribute to chloroplast localization, and mutation of two residues, VL(17/18)AR in the context of the full-length TGB3 was sufficient to abolish chloroplast localization and allow accumulation at the periphery of epidermal cells, whereas WT and chloroplast-localizing deletion mutants were not detected in the epidermal layer (Lim et al., 2010b). The N-terminal domain of TGB3 was found to contain a ‘zipcode’ for delivery of DsRed or GFP to the chloroplast (Fig. 8; Lim et al., 2010b).

Analysis of the AltMV TGB3 sequence revealed a putative N-terminal signal peptide (SP), corresponding to residues 1 - 19, which includes the residues VL(17,18) shown critical for chloroplast localization; this SP is not cleaved, as GFP or DsRed fusions to the TGB3 N-terminus are stably maintained and targeted to the chloroplast. The SP identified by Phobius software prediction is therefore likely to be a signal anchor chloroplast-targeting signal, as the PVX TGB3 (and also TGB3 of WClMV, PlAMV, and Tulip virus X [TVX]) is predicted to have a transmembrane domain instead of an SP; however, PapMV, Clover yellow mosaic virus (ClYMV), Zygocactus virus X (ZVX), and Foxtail mosaic virus (FMV) were also predicted to have an N-terminal SP, and BaMV yielded an inconclusive N-terminal prediction. The C-terminal domains of PVX, WClMV, TVX, and PlAMV TGB3s are predicted to have a cytoplasmic localization, whereas the TGB3 of AltMV, PapMV, ClYMV, ZVX, and FMV have C-terminal domains predicted as non-cytoplasmic. These results - and the demonstrated difference in localization between AltMV and PVX TGB3s - suggest that there are at least two subclasses of TGB3 properties in the genus Potexvirus (Lim et al., 2010b).

CP

AltMV GFP:CP localized to the periphery of epidermal cells, mainly to punctate spots, and to larger punctae in mesophyll cells, as well as associated with chloroplasts; in plasmolyzed cells the punctae remained associated with the cell wall (Fig. 5E). When co-expressed with BSMV DsRed:TGB3, AltMV GFP:CP co-localized to dual spots either side of the cell wall, indicating plasmodesmatal location (Fig. 6B).

Host Interactions

As noted above, there are sequence variants that are associated with isolates from different ornamental crop types, probably with little regard to the geographic origin of the crop. For example, the first report of a portulaca isolate was from Italy (Ciuffo and Turina, 2004), but other isolates from portulaca, or ‘portulaca-like’ isolates from other hosts, have been identified from several hosts in Brazil (Duarte et al., 2008), the USA (Baker et al., 2006; Lockhart and Daughtrey, 2008; Hammond and Reinsel, 2015), and a full sequence obtained in Russia (Ivanov et al., 2011). However, experimental inoculation of isolates from perennial phlox to an assortment of ornamental plants and virus bioassay hosts has shown that typical phlox isolates readily infect species in taxonomically diverse genera (Hammond et al., 2006a; Lim et al., 2010c; J. Hammond, unpublished), and the distribution of isolate types is probably related more to the crop production systems than to inherent biological properties of the different isolates. However, we have identified a number of variants which affect host interactions, and in some cases linked biological differences to specific viral proteins and particular amino acid residues affecting virus:host interactions. In other cases interactions between viral proteins and host proteins have been identified without identification of differences between virus isolates.

Symptom severity

Whereas detailed tests of host response have not been reported with a wide variety of either isolates or host species, ‘phlox-like’ isolates produce more severe symptoms than ‘portulaca-like’ isolates in the experimental host Nicotiana benthamiana (Hammond et al., 2006b), and severity of symptoms is also correlated with induction of necrosis at low temperatures (15℃) (Lim et al., 2010a; 2014).

When four distinct infectious clones of AltMV were derived by population cloning from phlox isolate AltMV-SP, they were found to induce symptoms ranging from very mild (clone 4-1) to very severe (clone 3-7); these four clones resulted from each possible combination of two clones of the 5′ portion of the genome with two clones representing the 3′ portion, joined at a unique restriction site (Lim et al., 2010a) near the end of the viral helicase domain of the RdRp protein. Alignment of the aa sequences revealed multiple aa differences in the 5′ portion of the genome (5 in MT; 24 between MT and 2OG-F; 4 in 2OG-F; 8 between 2OG-; and 4 in the Helicase [HEL] domain), and only five in the 3′ region of the genome (3 in the ‘core’ Polymerase [POL] domain, one in the ‘variable’ POL C-terminal domain, and one in TGB1), without any substitutions in TGB2, TGB3, or CP (Lim et al., 2010a). Clones 3-1 and 4-1 not only induced mild symptoms, but failed to induce necrosis when infected plants were grown at 15℃; exchange of portions of ‘severe’ clone 3-7 and mildest clone 4-1 to substitute either the ‘core polymerase’ residues, the ‘variable’ POL C-terminal domain and TGB1 residue, or both regions identified both independent and additive effects of these two regions in symptom severity and on RNA accumulation level (Lim et al., 2010a). The single residue difference in TGB1 was also demonstrated to have a major effect on efficiency of RNA silencing expression (Lim et al., 2010a; 2010c).

RNA replication levels of AltMV from both mild and severe infectious clones was considerably higher at 15℃ than at 25℃, although the most severe clone 4-7 killed plants prior to 30 days post-inoculation (dpi) at 15℃ (Lim et al., 2010a). Interestingly, amplification and sequencing of a central portion of the genome from the original infection of AltMV-SP, from which the infectious clones were derived, was shown to contain a preponderance of sequences identical to severe clone 4-7, a minor component identical to mild clone 3-1, and two additional sequence variants; a mixed infection of clones 3-1 and 4-7 reconstituted an infection with appearance and RNA accumulation levels very similar to AltMV-SP and clone 3-1 alone, and significantly lower accumulation than severe clone 4-7 alone. Mixing clone 3-1 labeled by insertion of the gene for the Green fluorescent protein (GFP), and clone 4-7 labeled with DsRed, resulted in some areas of tissue infected by either clone alone, and other areas in which both could be clearly identified within the same cells; at 25℃ 4-7:DsRed spread to an area about 10-fold larger than 3-1:GFP in systemically infected leaves (and c.8 : 1 ratio of 4-7:DsRed to 3-1:GFP RNA), but at 15℃ each isolate spread to a similar extent in upper leaves, with a c.6 : 4 ratio of RNA (Lim et al., 2010a).

The CP has also been shown to affect symptom expression; an isolate from Portulaca (AltMV-Po) was found to induce chlorotic rather than necrotic local lesions, and mild mosaic in N. benthamiana (Hammond et al., 2006b), without necrosis at 15℃. Substitution of the CP of AltMV-Po into severe infectious clone 3-7 (which otherwise induces severe systemic necrosis at 25℃ and plant death at 15℃) resulted in mild mosaic in N. benthamiana without necrosis, even at 15℃ and 30 dpi, conditions under which clone 4-7 routinely kills plants (Lim et al., 2014). AltMV 3-7_Po-CP was also compared to parental AltMV 3-7 in Alternanthera dentata and soybean (Glycine max). In A. dentata, AltMV 3-7_Po-CP was symptomless rather than inducing the obvious local lesions, stunting, and early anthocyanin production observed with clone 3-7; AltMV 3-7_Po-CP caused symptomless infection of the first trifoliate leaf of soybean inoculated on the primary leaf, whereas 3-7 induced severe necrosis (Lim et al., 2014).

Induction of necrosis

As noted above, some severe symptom isolates frequently induce systemic necrosis even at 25℃, and severe clones 4-7 and 3-7 routinely kill N. benthamiana grown at 15℃ by 30 dpi (Lim et al., 2010a, 2014). After demonstrating that substitution of the AltMV-Po CP into clone 3-7 caused ablation of necrosis in N. benthamiana, the ability of the AltMV-CP to act as an inducer of systemic necrosis was examined using a heterologous expression system -as an additional protein expressed from the genome of Potato virus X (PVX). Plants infected with PVX:CP_Po showed symptoms without necrosis, and similar to control PVX at either 25℃ or 15℃, whereas PVX:CP_SP developed interveinal necrosis in upper leaves at 25℃, and systemic necrosis associated with the veins from about two weeks post-infection at 15℃ (Lim et al., 2014).

There are nine amino acid differences between the CP of AltMV-SP (from which clone 3-7 is derived; CP_SP) and AltMV-Po (CP_Po), most of which are in the N-terminal half of the CP (Fig. 1). Substitution of these variant residues from CP_Po into clone 3-7 created CP variants MN(13,14)ID, N(36)S, A(69)V, LA(76,77)IS, E(78)G, T(154)A, and the combination MN(13,14)ID/LA(76-77)IS. Of these variants, all retained severe symptoms and induction of necrosis at 15℃ except for MN(13,14)ID, LA(76,77)IS, and the combination MN(13,14)ID/LA(76-77)IS. The differences in induction of necrosis were not related to RNA accumulation level, as levels of accumulation of each viral variant were similar to parental clone 3-7, and there was no obvious correlation between symptom expression and level of CP expression detected by western blot (Lim et al., 2014). Similarly, there was no obvious correlation between the predicted isoelectric point of the CP variants and symptom production; CP_SP and CP_Po have predicted pI of 5.73 and 5.716 respectively, while the variants range from 5.212 to 7.026; however, there was an obvious difference in homologous interaction between CP_SP and CP_Po as determined by Yeast two hybrid (Y2H) analysis. Whereas CP_SP showed a strong self-interaction apparent after 20 h growth, CP_Po showed a weak self-interaction not obvious until about 80 h; all of the residue substitutions except for CP_LA(13,14)IS showed strong self-interaction, while CP_LA(13,14)IS showed a weak interaction similar to that of CP_Po (Lim et al., 2014). Electron microscopy of virions of different variants revealed no obvious differences in particle structure or flexibility to explain the differences in induction of necrosis; however, in non-denaturing gel electrophoresis, subunits of CP_SP, but not CP_Po, formed dimers, again suggesting differences in self-association between these CP variants (Lim et al., 2014).

The strong self-association of CP_SP was confirmed by Bimolecular fluorescence complementation (BiFC), as was negligible self-interaction of CP_Po, or between CP_SP and CP_Po. Y2H was used to screen an Arabidopsis (Arabidopsis thaliana) cDNA library to identify CP_SP-interacting host proteins, and an A. thaliana boron transporter 1 (AtBOR1) protein identified, which showed a weaker binding to CP_Po (Lim et al., 2014). The N. benthamiana homolog of AtBOR1 (NbBOR1) was then cloned and shown by BiFC to interact with CP_SP, but not with CP_Po, with localization occurring at the cell plasma membrane (Lim et al., 2014). An association of NbBOR1 with necrosis was demonstrated by silencing of NbBOR1 using the Tobacco rattle virus (TRV) VIGS system. AltMV-challenged plants of N. benthamiana in which NbBOR1 was significantly down-regulated developed intraveinal necrosis at 25℃, whereas control plants pre-infected with wild-type (WT) TRV developed only a mosaic after AltMV 3-7 challenge; plants challenged with AltMV 3-7 at 15℃ developed necrosis regardless of pre-infection with WT TRV or TRV:NbBOR1. No necrosis was induced in any plants challenged with AltMV 3-7_Po-CP (Lim et al., 2014).

Over-expression of CP_SP by agroinfiltration was not sufficient to induce necrosis at either 25℃or at 15℃, probably because the level of CP expression achieved was significantly lower than that obtained by systemic viral infection; over-expression of AltMV CP from PVX resulted in an estimated 200-fold higher level than CP expression by agroinfiltration. It is also notable that significant necrosis was only observed in systemically-infected leaves at about two weeks after infection, whereas agroinfiltration results in only local accumulation without amplification by viral replication (Lim et al., 2014).

Systemic necrosis was therefore demonstrated by various methods to be due to a specific host response difference to CP_SP compared to CP_Po,_,and not to differences in levels of virus accumulation as shown for necrosis induced by the HEL elicitor of Plantago asiatica mosaic virus (PlAMV; Komatsu et al., 2011). The resistance gene Rx from potato was originally thought to be specific for PVX (Goulden et al., 1993), but has since has been shown capable of conferring resistance against several other potexviruses (Baurès et al., 2008; Candresse et al., 2010). AltMV 3-7 and 3-7_Po-CP each induced similar symptoms in transgenic N. benthamiana line Rx-18 to those observed in WT N. benthamiana at either 25℃ or 15℃, indicating no interaction with Rx (Lim et al., 2014). However, silencing of NbSGT1, a gene previously shown to be involved in the disease resistance pathways mediated by multiple R (resistance) genes (Azevedo et al., 2002; Peart et al., 2002), was shown to ablate induction of systemic necrosis by AltMV 3-7, whereas typical systemic necrosis occurred in controls (Lim et al., 2014). SGT1 is not the only gene required for Rx function; Hsp90, Rar1, RanGap1, and RanGap2, plus at least five other host factors mediate Rx resistance, maybe through mediating specificity or sensitivity of recognition of the PVX CP elicitor of Rx (Sturbois et al., 2012). This indicates that the necrosis induced by AltMV results from an interaction occurring at least in part through an R gene; CP is the elicitor of temperature-sensitive necrosis, and also functions as a pathogenicity determinant, and NbBOR1 may serve as a host factor mediating recognition of AltMV-CP_SP as an elicitor. It is possible that components of a mitogen-activated protein kinase (MAPK) signaling cascade are required for AltMV-induced necrosis, as noted for PlAMV (Komatsu et al., 2010; Hashimoto et al, 2012).

CP structure related to virion assembly and necrosis.

It was interesting that substitution of either of two pairs of amino acid residues- MN(13,14)ID and LA(76,77)IS - could independently ablate the temperature-dependent necrosis associated with CP_SP; whereas MN(13,14)ID are clearly predicted to be located in the CP N-terminal domain on the virion surface, LA(76,77)IS are not. A predicted structure for the most closely-related PapMV CP by Yang et al. (2012) shows PapMV CP residue F13 overlapping the adjacent subunit in the virion, and interacting with an exterior surface hydrophobic pocket, in line with a prior report that F13 is critical for PapMV virion assembly (Laliberté Gagné et al., 2008), and that the N-terminal domain of PVX CP plays a role as a scaffold (Lukashina et al., 2012). Other residues and domains of PapMV also affect RNA binding and subunit self-assembly, and mutants with substitutions for F13 still form multisubunit assemblies, so F13 is not the only major determinant of subunit-:subunit interactions (Laliberté Gagné et al., 2008; Tremblay et al., 2006).

Overlaying the AltMV CP sequence onto the structure of the PapMV CP derived by Yang et al. (2012) suggested that not only are MN(13,14)ID located on the virion surface with potential to overlap the adjacent subunit, but that LA(76,77)ID are located on the subunit surface within the same turn of the virion helix. Notably the other residues which differentiate AltMV CP_SP and CP_Po, except for E(78)G and N(36)S, are on subunit surfaces predicted to be between adjacent turns of the virion helix (Yang et al., 2012). Both MN(13,14)ID and LA(76,77)IS may affect secondary or tertiary structure of the AltMV CP so as to affect either self-interactions or interactions with host protein(s) involved in elicitor recognition and thus induction of necrosis. In the case of the PVX resistance gene Rx, it has been noted that there is relatively little amino acid sequence conservation in the elicitor domains of the various potexvirus CPs interacting with Rx, and that conservation of structural elements is more important than a specific amino acid sequence (Baurès et al., 2008; Candresse et al., 2010).

Although our results (Y2H, BiFC, and absence of dimers in non-denaturing gel electrophoresis) all suggest that CP_Po has weak self-interactions, it is notable that in contrast the CP of portulaca isolate AltMV-MU can form RNA-free virus-like particles (VLPs) in vitro, and filamentous VLPs at both pH 4.0 and pH 8.0 (Mukhamedzhanova et al., 2011), whereas the closely-related PapMV formed filamentous VLPs only at pH 4.0, and only small aggregates of 13 - 33S at pH 8.0 (Erickson et al., 1976); the only amino acid difference between AltMV-MU and AltMV-Po is residue I(110)M, which is not considered likely to explain the apparent difference in subunit:subunit interactions.

The differential interactions between CP_SP and CP_Po and both AtBOR1 and NbBOR1 is potentially directly linked to necrosis. AtBOR1 localizes to the plasma membrane, especially in roots, where it regulates xylem loading with boron to maintain boron concentration, protecting shoots from boron deficiency (Takano et al., 2002). Either excessive or deficient boron concentrations can result in severe plant damage, which often involves necrosis (e.g. Cakmak and Römheld, 1997; Nable et al., 1997), with a narrow physiologically acceptable range (Takano et al., 2005). Interaction of NbBOR1 with AltMV CP_SP may disturb boron regulation leading directly to necrosis, but effects on necrosis of silencing SGT1 may indicate a rather more complicated scenario.

Movement and Chloroplast Interactions

TGB3 interactions

Availability of infectious clones of AltMV enabled examination of the viral genes necessary for movement, and comparison to what has been reported for other potexviruses, including the type member of the genus, PVX, and others including White clover mosaic virus (WClMV) and Bamboo mosaic virus (BaMV) (e.g. Lin et al., 2006; Lough et al., 1998; Krishnamurthy et al., 2002; Mitra et al., 2003; Schepetilnikov et al., 2005; Verchot-Lubicz, 2005). A number of differences have been elucidated as a result, which suggest that different potexviruses utilize different types of interaction with their hosts, and that AltMV is an excellent additional model species to gain further understanding of differences within the genus Potexvirus.

As a basis for the examination of AltMV movement, the AltMV clone expressing GFP as a marker from a position between the Triple Gene Block and the CP under the control of as duplicated subgenomic promoter (Lim et al., 2010a) was utilized; variants were made with deletions in TGB2, TGB3, or CP, and infection and spread monitored by Laser scanning confocal microscopy (LSCM). In the case of TGB2 and TGB3 mutations, the amino acid sequence of the overlapping reading frames (TGB1 for ΔTGB2; TGB2 for ΔTGB3) was maintained (Lim et al., 2010b). Infection of WT AltMV-eGFP progressed from initially infected cells to large areas of infected cells on the inoculated adaxial leaf surface, and within 5 dpi within the mesophyll and to infection of similar areas visible on the abaxial leaf surface; systemic infection was apparent with a few more days. Deletion of the TGB2 start codon resulted in restriction of infection to single cells on the inoculated adaxial surface of leaves inoculated with infectious RNA transcripts, indicating that AltMV-ΔTGB2-eGFP was not competent to spread beyond the initially infected cell; no diffusion of free GFP to cells beyond the initially infected cells was detected; as for AltMV-ΔTGB2-eGFP, AltMV-ΔCP-eGFP was also limited to single cells. Results for both AltMV-ΔTGB2-eGFP and AltMV-ΔCP-eGFP parallel what has been reported for WClMV (Beck et al., 1991) and PVX (Lough et al., 2000).

In contrast, in leaves inoculated with AltMV-ΔTGB3-eGFP, small groups of fluorescent cells were consistently observed at the inoculated adaxial service, spreading somewhat over the course of 7 dpi, but never reaching the size of clusters observed with WT AltMV-eGFP, and never spreading beyond the epidermal layer. In parallel experiments experiments examining the subcellular localization of TGB3, deletion mutants identified TGB3 residues 17/18VL as critical to localization (see below), so mutant TGB3_VL(17/18)AR was created for both localization and movement studies. AltMV-TGB3 _VL(17/18)AR-eGFP was also able to spread from cell-to-cell in the inoculated adaxial epidermis, but not to enter the mesophyll or reach the abaxial epidermis. The ability of AltMV-ΔTGB3-eGFP and AltMV-TGB3 _VL(17/18)AR-eGFP to spread to multiple cells within the epidermis was unlike the restriction of TGB3 deletions of WClMV, PVX, and BaMV to initially inoculated cells (Lough et al., 1998, 2000; Lin et al., 2006), suggesting a difference in movement properties of AltMV compared to these three previously characterized viruses.

While both AltMV-ΔTGB3-eGFP and AltMV-TGB3 _VL(17/18)AR-eGFP infectious clones were able to spread to multiple cells in the inoculated adaxial epidermal layer, there was no spread to the mesophyll or the abaxial epidermis; however, agroinfiltration of free TGB3 expressed from the 35S promoter at 4 dpi was able to complement movement of AltMV-ΔTGB3-eGFP in trans, allowing detection of GFP expression throughout the mesophyll and abaxial epidermis in the agroinfiltrated region, but without local movement outside this region. Two further constructs added either WT TGB3, creating AltMV-ΔTGB3 (TGB3+), or TGB3:GFP, creating AltMV-ΔTGB3 (TGB3-GFP+), in order to examine complementation in cis. AltMV-ΔTGB3 (TGB3+) was able to systemically infect N. benthamiana, inducing mild vein-associated symptoms and demonstrating that expression of functional TGB3 separated from the context of the Triple Gene Block could complement both local and systemic movement. However, while AltMV-ΔTGB3 (TGB3-GFP+) was able to spread within the inoculated leaf, within both adaxial and abaxial epidermal layers and the mesophyll, no symptoms could be detected in non-inoculated leaves, and only trace levels of CP were detected by western blotting, indicating that TGB3-GFP could complement local movement, but not effective long-distance movement; these results and the chloroplast localization of both TGB3-GFP (or DsRed) and GFP-TGB3 (or DsRed-TGB3) fusions suggests that movement requires the N-terminal TGB3 domain including residues 17 and 18, and chloroplast targeting (Lim et al., 2010b). It is possible that full systemic movement may also require the C-terminal domain to be free to exit the vascular system, as expression of TGB3+, but not TGB3-GFP+, complemented systemic movement.

In addition to the close association between TGB3 and chloroplasts revealed by confocal microscopy, Fluorescent in situ hybridization revealed concentration of AltMV RNA in the mesophyll layer surrounding the chloroplasts, suggesting the possibility of replication in association with the chloroplast membrane (Lim et al., 2010b). Overexpression of AltMV TGB3 either as an added gene from the AltMV genome, or from the PVX genome, resulted in an increase in symptom severity including veinal necrosis (Fig. 9), whereas overexpression of PVX TGB3 from the AltMV genome caused no increase in symptom severity; the leaf lamina around necrotic veins became bleached, with significantly fewer chloroplasts than in plants infected with WT PVX (Lim et al., 2010b). Electron microscopy of leaf tissue infected with either AltMV or PVX overexpressing AltMV TGB3 revealed ‘necked’ vesicular invaginations of the chloroplast membrane not observed in WT infections (Lim et al., 2010b; Jang et al., 2013), and similar in appearance to chloroplast invaginations reported in Turnip yellow mosaic virus (TYMV; tymovirus) infected plants (Ushiyama and Matthews, 1970), and to mitochondrial or peroxisomal vesiculation induced by different tombusviruses and associated with replication (Rubino and Russo, 1998). It is therefore possible that AltMV TGB3 forms vesicles at the chloroplast to serve as protected sites for viral replication, and to recruit other components of the AltMV replication complex, in a similar way that the TYMV 66K and 140K replication proteins are targeted to the chloroplast, where replication occurs at chloroplast invaginations (Prod’homme et al., 2003), and the Barley stripe mosaic virus (BSMV; hordeivirus) TGB2 and γb proteins localize to chloroplasts and induce vesicles for a replication-related function (Torrance et al., 2006).

The chloroplast interactions of AltMV TGB3 were further investigated through identification of host protein interactions, and the effects of chloroplast damage under dark conditions. Seven TGB3 amino acid sequence variants were identified among a collection of AltMV isolates from various hosts; these were separately expressed as additional genes from the PVX genome, and infected plants grown under either normal light (16h light/8h dark), or incubated for 6 days in constant darkness starting at 7 dpi. Control plants infected with AltMV showed somewhat more severe symptoms under ‘dark’ conditions, while WT PVX-infected plants showed milder symptoms than AltMV under both ‘light’ and ‘dark’ conditions respectively. Plants infected with PVX (AltMV TGB3+) under ‘light’ conditions developed symptoms more severe than WT PVX, including veinal-associated necrosis, and fewer chloroplasts were observed by confocal microscopy. However, the symptom severity was significantly increased under ‘dark’ conditions, with all PVX (AltMV TGB3+) plants showing some degree of collapse, but without obvious correlation with particular amino acid variations (Jang et al., 2013). The chloroplast destruction observed is similar to that reported under normal lighting conditions in plants infected with TYMV (Chalcroft and Matthews, 1966).

Comparison of the TGB3 sequence variants showed that two isolates had an alanine (A) substitution at residue 17 (Jang et al., 2013), one of the two residues mutated in TGB3_VL(17,18)AR, shown to lose chloroplast localization in our earlier studies (Lim et al., 2010b). This implies that it is residue L(18)R which is critical in abolishing chloroplast localization (Jang et al., 2013).

In addition to the ‘necked’ small chloroplast vesicular invaginations previously observed in plants infected with either AltMV or PVX overexpressing AltMV TGB3, we also observed large cytoplasmic invaginations into some chloroplasts in plants infected with AltMV (TGB3+), including some cytoplasm with significant aggregates of virus particles (Jang et al., 2013). Some stromule-like extensions of chloroplasts were also observed; chloroplasts are primary sites for the production of immune signals in plants, and may send out stromules as signal conduits to transmit induced immune signals to the nucleus during effector-triggered immunity and programmed cell death (Caplan et al., 2015; Gu and Dong, 2015). Stromule formation in response to N-mediated resistance to TMV was first reported by Caplan et al. (2008), and may enhance programmed cell death in response to TMV infection (Caplan et al., 2015). In addition, stromule formation has been shown in response to light-sensitive redox signals within the chloroplast (Brunkard et al., 2015), such as might be induced under chloroplast stress and degradation induced by AltMV infection.

An interaction between AltMV TGB3 and a host protein was identified by using TGB3 as bait to screen an Arabidopsis cDNA library by Y2H; of several candidate host genes identified, A. thaliana PsbO1 (AtPsbO1; oxygen-evolving enhancer protein 1-1) was selected as the strongest interactor, and because PsbO1 is a major nuclear-encoded component of the chloroplast oxygen-evolving complex of photosystem II (Tyagi et al., 1987). The N. benthamiana homolog, NbPsbO was also cloned, and used for interaction studies. The interactions initially identified by Y2H were confirmed by BiFC, and by co-localization of GFP:PsbO with DsRed:TGB3.

BiFC interactions of TGB3 and either AtPsbO1 or NbPsbO were observed in both epidermal and mesophyll cell layers. In the mesophyll cells, the interaction was clearly localized around the chloroplasts, with additional fluorescence visible at the cell periphery; in some cases the association was obviously at the chloroplast envelope, but in other instances with a more punctate appearance. In the epidermis, fluorescence was observed primarily at the cell periphery and in cytoplasmic aggregates, probably as a result of overexpression (Jang et al., 2013), whereas expression of GFP-TGB3 or DsRed-TGB3 alone was almost never observed in the epidermal layer, but only in the mesophyll (Lim et al., 2010b).

A 15-residue C-terminal deletion mutant of TGB3 had a weaker BiFC interaction with either AtPsbO1 or NbPsbO, whereas deletion of N-terminal TGB3 residues 2-16 essentially eliminated the interactions, suggesting that the N-terminal domain most important for chloroplast interaction is involved in the TGB3/PsbO interaction. This was further supported by the absence of BiFC interaction between TGB3_VL(17,18)AR with either AtPsbO1 or NbPsbO. As two natural AltMV variants have VL(17,18)AL and are still able to infect N. benthamiana systemically, it is presumably residue L(18) which is critical for both chloroplast and PsbO interactions (Jang et al., 2013).

When expressed separately, GFP:PsbO localized around the chloroplasts in mesophyll cells, while DsRed: TGB3 localized to chloroplasts as punctate aggregates; when the two constructs were co-expressed, almost all of the fluorescent signal was co-localized in mesophyll cells, at what appeared to be contact points between adjacent chloroplasts, thus confirming the results of BiFC localization (Jang et al., 2013). While we have determined that the N-terminal domain of TGB3 appears to be the most important domain for PsbO interaction, we have not determined which domains of PsbO are involved in the interaction, but note that PsbO has an 85-residue SP which is cleaved from the mature PsbO peptide prior to the localization of PsbO to the thylakoid lumen. It may be the PsbO SP which is responsible for the targeting of TGB3 to (and insertion into ?) the chloroplast membrane.

TGB1 interactions

The TGB1 protein has multiple functions, which include roles in viral movement (as a component of the non-virion ribonucleoprotein movement complex, RNC), viral replication (through its helicase activity), and as a suppressor of RNA silencing. The RNA silencing suppression function probably contributes to both movement (by protection against genomic RNA degradation of the RNC), and replication (by protection of the replication complex), although the replication complex is also expected to be sequestered from the general cellular environment within a membrane-bound compartment(s). AltMV TGB1 forms crystalline aggregates in both the cytoplasm and the nucleus of infected cells (Fig. 4A), though it is not clear whether these crystalline structures are active reserves or ‘disposal sites’ where TGB1 accumulates after replication in the cell is completed. Similarly, at present we do not know what differences affect TGB1 localization to cytoplasm versus the nucleus. It is notable that 2-D gel electrophoresis separated two TGB1 isoforms, presumably differing in charge due to different phosphorylation states (Seo et al., 2014). There are a total of seven predicted phosphorylation sites in TGB1 (one each of cAMP-activated protein kinase, casein kinase 2, and tyrosine kinase; and four protein kinase C sites), but we do not know which of these sites are actually phosphorylated, nor how phosphorylation contributes to movement between the cytoplasm and nucleus. However, none of the mutations affecting TGB1 localization involve residues included in any of the predicted phosphorylation sites.

The Y2H method was used to screen an A. thaliana cDNA library with TGB1 as bait, and a number of candidate host genes were identified (Seo et al., 2014). Additional potentially interacting host proteins were identified by two-dimensional (2-D) gel electrophoresis and matrix-assisted laser desorption/ionization-time of flight-mass spectrometry (MALDI-TOF-MS) analysis of proteins co-purifying with a TGB1-containing fraction obtained from an AltMV virus purification (Seo et al., 2014). Y2H resulted in identification of four putative interacting A. thaliana proteins, all of which are nuclear-encoded: AtCpIscA (Iron-sulfur assembly protein, IscA, chloroplast); LHB1B2 (Photosystem II Light-harvesting complex protein B1B2); ATP synthase subunit delta′ (ATPase δ′-subunit; Mitochondrial ATPase subunit delta′); and LHCA4 (Light-harvesting chlorophyll protein complex I, subunit A4). MALDI-TOF-MS led to identification of 20 proteins including two isoforms of TGB1 separated by 2-D gel electrophoresis; two isoforms each of Chloroplast α-ATPase (ATP synthase CF1 alpha subunit), and Chloroplast β-ATPase (H+-transporting two-sector ATPase beta chain). These chloroplast- localized proteins were prioritized for further study based on the known affinity of AltMV TGB3 for the chloroplast, and apparent chloroplast-associated replication; no prior information suggested a chloroplast association for TGB1.

Each of the Y2H-identified proteins, plus the ATP synthase alpha subunit and ATPase beta chain, were examined for TGB1 interactions by BiFC, against TGB1_L88, TGB1_P88, and TGB1_G31R, variants which had previously been identified as differing in silencing suppression efficiency (TGB1_L88, TGB1_P88, and TGB1_G31R), or lacking in self-interaction in Y2H and BiFC (TGB1_G31R; Nam et al., 2013). Positive cytoplasmic interactions were obtained by BiFC between the Chloroplast β-ATPase and TGB1_L88, and between each of AtCpIscA, LHB1B2, ATPase δ′-subunit, and LHCA4 and each of the TGB1 variants. No interaction was detected to the Chloroplast α-ATPase, and we conclude that this subunit was associated not with TGB1 in the virus purification fraction, but probably with the Chloroplast β-ATPase, which itself interacts directly with TGB1 as shown by BiFC (Seo et al., 2014). Interestingly the BiFC interactions of AtCpIscA and LHB1B2 with mutant TGB1_G31R yielded much larger aggregates than with TGB1_L88 and TGB1_P88; TGB1_G31R does not self-associate by either Y2H or BiFC, has negligible silencing suppression activity, and unlike TGB1_L88 and TGB1_P88, accumulates uniformly within the nucleus, without either nucleolar or nuclear periplasm localization (Nam et al., 2013).

Silencing of Chloroplast β-ATPase by TRV-mediated VIGS followed by AltMV challenge yielded a hypersensitive response to AltMV-TGB1_L88 infection, without obvious effects on AltMV replication levels; it thus seems that the normal TGB1/Chloroplast β-ATPase interaction may retard severe symptom expression and protect plants from severe necrosis and early death (Seo et al., 2014). The interaction may normally lead to activation of chloroplast-mediated host immune responses and partial protection.

A number of chloroplast-related genes (either encoded by the nuclear genome, or the chloroplast genome) have been implicated in interactions with various viral proteins. Among reported potexvirus-chloroplast interactions are those of Pepino mosaic virus TGB1 with tomato catalase 1 (CAT1; Mathioudakis et al., 2013); PVX CP with plastocyanin (Qiao et al., 2009); and PVX TGB3 with tobacco protochlorophyllide reductase (Fridborg et al., 2003). Interestingly silencing of PsbO increased accumulation of TMV, but not of either PVX or Alfalfa mosaic virus (Abbink et al., 2002); as noted above, PsbO interacts with AltMV TGB3 at the chloroplast membrane (and may indeed direct TGB3 to the chloroplast), while over-expression of TGB3 causes chloroplast vesiculation and veinal necrosis, suggesting that the TGB3-PsbO interaction may interfere in basal host defenses (Jang et al., 2013). Other plant chloroplast proteins interact with various potyvirus or tobamovirus proteins (see Seo et al., 2014).

Chloroplast ATPase is formed of a large multisubunit complex, CF₁, which is found in the thylakoid stroma attached to integral membrane complex CF₀; together the CF₁-CF₀ ATP synthase catalyzes ATP synthesis and ATP hydrolysis in a balanced reaction to effect proton transfer across the thylakoid membrane (Groth and Strottman, 1999). Recently TMV 183 kDa and 126 kDa replication-associated complexes were shown to be enriched in chloroplast γ- ATPase, which may function to minimize plant damage by limiting TMV replication, as in plants in which γ-ATPase was silenced yielded higher TMV titer and more rapid cell-to-cell and systemic movement (Bhat et al., 2013). Disabling various chloroplast functions may impair host defense signaling (Dardick, 2007) and allow higher levels of viral replication; whether the silencing of chloroplast β-ATPase which led to increased symptom severity and induced a hypersensitive response without increasing AltMV titer reflects another aspect of these complex interactions remains to be determined. However, it is somewhat surprising that there is a cytoplasmic interaction between the chloroplast-encoded α-ATP synthase subunit and TGB1, which is known to accumulate as crystalline aggregates in both the cytoplasm and nucleus (and at the nucleolus, Fig. 4C; Lim et al., 2010c). TGB1 has not been observed within chloroplasts by either transmission electron microscopy or by confocal microscopy (Lim et al., 2010c; Nam et al., 2013; and J. Hammond, unpublished); however, both chloroplast-encoded α-ATPase and β-ATPase co-purified with TGB1 aggregates from AltMV-infected plants, indicating TGB1 as a target for further studies of virus:host interactions.

Silencing suppression

Differences in efficiency of RNA silencing suppression in variants of AltMV was first identified through gene exchanges between a typically severe symptom infectious clone, and a mild symptom variant; both were originally obtained by a population cloning strategy from a strain originally from creeping phlox, which had been maintained by mechanical passage in N. benthamiana for several years prior to generation of infectious clones (Lim et al., 2010a). Gene exchanges of both the POL domain of the RdRp, and of TGB1, affected replication levels and symptom severity, but while there were four amino differences in the POL region, there was only a single amino acid difference in the TGB1 of the mild and severe symptom variants. The severe symptom variant had TGB1_L88, while the mild symptom variant had TGB1_P88, and an agroinfiltration silencing suppression assay of the TGB1 variants revealed that TGB1_L88 had activity comparable to that of the Tomato bushy stunt virus p19; in contrast, AltMV TGB1_P88 had very much lower silencing suppression activity (Lim et al., 2010a). Substitution of TGB1_L88 into the mildest infectious clone 4-1 resulted in an approximately 1000-fold increase in viral replication, and concomitantly increased symptom severity; interestingly, many more cells were infected when plants infected with mild variants were grown at 15℃ rather than 25℃, presumably due to differences in plant RNA silencing activity. At lower temperatures host RNA silencing defenses are less effective, allowing increased viral replication and/or numbers of cells to be infected (Chellappan et al., 2005; Qu et al., 2005; Jovel et al., 2007; Siddiqui et al., 2008).

Y2H assay to compare the homologous and heterologous interactions of TGB1_L88 and TGB1_P88 did not reveal obvious differentiation in their interactions (Lim et al., 2010a). However, significant differences in subcellular localization of GFP:TGB1 fusion proteins was observed, with GFP:TGB1_L88 found to localize to the nuclear membrane and as discrete nucleolar aggregates, whereas GFP:TGB1_P88 accumulated around the nuclear periphery, with no nucleolar accumulation; both variants localize to the cell wall in addition to the nucleus (Lim et al., 2010c). Comparison of AltMV TGB1 to that of PVX, in which seven conserved motifs of the helicase domain have been previously identified (Morozov et al., 1999), revealed a number of perfectly conserved residues, mainly in domain motifs I, II and III (Nam et al., 2013). Helicase motif I has previously been shown to be critical for PVX TGB1 homologous interactions (Leshchiner et al., 2006), so AltMV TGB1 mutations were made in motif I - G(31)R and GK(33,34)RR; motif II - D(81)R; and motif III - Q(101)R and examined for Y2H self-interactions. TGB1_L88, TGB1_P88, and naturally-occurring variant TGB1_L(88)G were also examined; residue 88 lies between motifs II and III. As for PVX, the mutations on motif I disrupted interaction, but each of the other mutants retained self-interactions (Nam et al., 2013). The interactions were also examined by BiFC, with similar results; no interaction was detected with either of the motif I mutants G(31)R and GK(33,34)RR, but all other constructs showed obvious, similar self-interactions yielding a variety of aggregate sizes (Nam et al., 2013). The mutants were also examined for RNA silencing suppression activity (as free TGB1), and for nuclear/nucleolar localization (as GFP:TGB1 fusions). The motif I mutants G(31)R and GK(33,34)RR showed no obvious silencing suppression, while L(88)G had minimal silencing suppression activity comparable to P88, and motif II mutant D(81)R and motif III mutant Q(101)R had obvious activity similar to the WT TGB1_L88. When subcellular localization of GFP:TGB1 fusions was examined, both G(31)R and GK(33,34)RR localized throughout the nucleoplasm, without detectable nucleolar localization, while L(88)G, D(81)R, and Q(101)R, were all observed at the nuclear periphery, with D(81)R and Q(101)R also accumulating at the nucleolus. TGB1 variants which self-associated also localized to the cell wall (Nam et al., 2013). It is thus apparent that self-interaction in motif I is required for both efficient self-interaction, and for localization to the cell wall, the nuclear periphery, and the nucleolus; residue L(88) is also critical for nucleolar localization, as neither TGB1_P88 nor TGB1_L(88)G are able to accumulate at the nucleolus, despite concentration at the nuclear periphery and obvious self-interaction observed by Y2H and/or BiFC. Self-interaction is also necessary for silencing suppression activity, but the residue at position 88 is also a major factor in silencing suppression efficiency (Lim et al., 2010a; Nam et al., 2013). Nucleolar localization is positively correlated with efficiency of silencing suppression, as is cell wall localization in the absence of TGB2 and TGB3, suggesting the possibility that nucleolar localization and cell wall localization may utilize the same signals, or at minimum that both require TGB1 dimerization (Nam et al., 2013). However, AltMV with TGB1_P88, which lacks detectable nucleolar localization, is still able to move systemically, and TGB1_P88 is still detected at the cell wall, though proportionately less so than TGB1_L88.

It has been demonstrated for PVX that cell-to-cell movement is dependent upon suppression of RNA silencing (Bayne et al., 2005). However, Bayne et al. (2005) were unable to identify PVX TGB1 mutants defective for silencing yet functional for movement, but some were functional as silencing suppressor yet not for movement, suggesting that silencing suppression is necessary, but not sufficient for movement. We were able to create a mutant of PVX TGB1 - L(86)P, at a position similar to the L(88)P of AltMV, which significantly reduced silencing suppression efficiency in the same way, while allowing full systemic movement, and creating a more effective PVX VIGS vector (Lim et al., 2010c). A dual L(86)P/L(89)P mutant of PlAMV TGB1 had similar effects on efficiency of silencing suppression, and both the PVX and PlAMV mutants showed the same differences in nuclear (and absence of nucleolar) localization as observed between AltMV TGB1_L88 and TGB1_P88 (Lim et al., 2010c).

Utility as a Viral Vector

As noted above, our original infectious clones of AltMV differed in symptom severity and relative levels of replication, based on differences in the RdRp and TGB1 sequences, with a single residue difference in TGB1 having major effects on the efficiency of RNA silencing suppression activity. The CP subgenomic RNA (sgRNA) promoter was identified by comparison to those of other potexviruses (Skryabin et al., 1988; Lee et al., 2000; Koenig et al., 2006) and to the region upstream of the TGB1 start codon; the TGB1 and CP sgRNA promoters were subsequently also determined for portulaca strain AltMV-MU (Putlaev et al., 2015). We modified the infectious clones by duplication of the CP sgRNA promoter, and insertion of a multiple cloning site (MCS) between the Triple Gene Block and the CP gene, with initial insertion of either eGFP or DsRed as fluorescent marker proteins; these initial infectious clones required in vitro transcription of infectious transcripts using the T7 RNA polymerase (Lim et al., 2010a).

An enhancement to the infectious clones was made by inserting the original full-length clones, including the T7 promoter, between the 35S promoter and a T7 transcription termination region in a binary vector for agroinfiltration. Despite an increased stretch of non-viral sequence between the 35S transcriptional start and the 5′ nucleotide of the AltMV sequence, infection could be obtained by direct agroinfiltration of Agrobacterium tumefaciens carrying the binary vector, with co-infiltration of a second binary plasmid expressing the TBSV p19 to minimize RNA silencing; however, a higher efficiency of infection was obtained, together with a significant reduction in time to symptom appearance, by co-infiltrating the infectious AltMV binary construct with an additional construct expressing the T7 RNA polymerase as well as the standard p19 construct (Lim et al., 2010c).

However, because of the size of the AltMV genome, and restriction sites present in the backbone of the binary plasmid, it was difficult to clone directly into the introduced multiple cloning site of the AltMV vector; a subclone of the 3′-terminal region of the genome including the MCS was useful for introduction of the desired insert, and then the 3′-terminal region including the inserted fragment substituted into the full-length clone. We were able to demonstrate that an agroinfiltrated construct expressing the AltMV RdRp between non-viral 5’ and 3’ UTRs was able to replicate and express eGFP in either orientation between the native AltMV 5′ and 3′ UTRs in trans; and also a defective AltMV genome, with the complete Triple Gene Block, eGFP, and CP gene between the native AltMV 5′ and 3′ UTRs (Lim et al., 2010c). A novel alternative method was therefore developed, allowing insertion directly into a binary plasmid subclone including the T7 promoter, 5′-UTR, an internally-deleted, RdRp fragment fused in-frame, the full Triple Gene Block, a modified MCS, CP, 3′-UTR, and T7 terminator. Presence of the N-terminal fragment of RdRp to initiate translation of the fused RdRp fragment was necessary to obtain effective infectivity and recombination, with the C-terminal region of RdRp essential for recombination to occur (Lim et al., 2010c). The modified MCS contains several unique restriction sites allowing direct insertion into the smaller binary plasmid. The binary plasmid containing the partial genome 3′ fragment is then co-agroinfiltrated with a second binary plasmid containing the 5′UTR, full RdRp, AltMV 3′-UTR, and T7 terminator, and the two plasmids expressing the T7 RNA polymerase and p19 construct; on transcription in planta, the AltMV RdRp and 3′ portion of the genome recombine at the overlap of the RdRp and TGB1 sequences present in both constructs, regenerating a full-length AltMV genome able to infect the plants systemically (Lim et al., 2010c).

The results with expression of eGFP, and replication of the defective genome of 5′UTR-TGB1/2/3-eGFP-CP-3′UTR by co-agroinfiltrated AltMV RdRp between non-viral 5’ and 3’ UTRs suggests that transgenic plants expressing AltMV RdRp with non-viral UTRs would provide an effective form of biocontainment for systemic infection of plants with 5′UTR-TGB1/2/3-eGFP-CP-3′UTR or equivalent constructs containing any gene of interest, as the Triple Gene Block and CP would provide the movement functions, and transgenic RdRp would provide replication functions in trans.

Two different forms of AltMV viral vector were developed by Putlyaev et al. (2015), both of which are of the ‘deconstructed virus’ type (Gleba et al., 2004) and based on a portulaca isolate, AltMV-MU. Vector ‘AltMV-single’ eliminates the entire Triple Gene Block, with the inserted gene expressed from the native TGB1 subgenomic promoter. In this vector the TGB1 start codon within the C-terminus of the RdRp was mutated to ACG so that there would be no initiation within the TGB1/RdRp overlap, and the initiation codon of the inserted gene would serve as the only start of translation. A further construct, ‘AltMV-double’, eliminated all of TGB1 and TGB2, but retained the portion of TGB3 including the subgenomic RNA promoter for the CP gene, and also included the mutation of the TGB1 initiation codon. Importantly, because of the elimination of most or all of the Triple Gene Block, these vectors are unable to move between cells, and can only be delivered to plants by agroinfiltration for local protein expression, providing a high level of biocontainment (Putlyaev et al., 2015). Mutation of the first G residue in the TGB1 sgRNA promoter was shown to abolish transcriptional activity completely (Putlyaev et al., 2015). The AltMV-double vector was found to promote higher levels of protein expression than AltMV-single, and yielded up to 400 µg of human granulocyte colony-stimulating factor per gram of fresh agroinfiltrated N. benthamiana leaf tissue (Putlyaev et al., 2015).

Protein expression

AltMV vectors with effective silencing suppressor TGB1_L88 are appropriate for high-level protein expression, with inserts placed in the MCS between TGB3 and CP, driven by a duplicated CP subgenomic promoter; the equivalent construct with weak silencing suppressor TGB1_P88 expressed significantly lower levels of eGFP (Lim et al, 2010c). Despite duplication of the CP sgRNA promoter, expression of eGFP from AltMV-eGFP was demonstrated to be stable through four serial passages starting from systemically infected leaves of the initially infected plants; in the fifth passage, faint eGFP expression was detected in the inoculated leaves, but not in upper leaves, in which typical AltMV symptoms were observed (Lim et al., 2010c). The AltMV vector thus shows significantly better retention of inserted sequences than many other viral vectors (e.g. Avesani et al., 2007; Shivprasad et al., 1999). AltMV-eGFP was also transmitted into seed of both N. benthamiana and Arabidopsis thaliana, and eGFP expression detected in one germinated seedling of A. thaliana (Lim et al., 2010c).

VIGS utilization

In contrast to the protein expression version of the vector, the weak silencing suppressor TGB1_P88 is preferred for VIGS, as demonstrated by the relative leaf area bleached in plants inoculated by AltMV-TGB1_P88 compared to AltMV-TGB1_L88, each with an insert of the phytoene desaturase gene. The VIGS vector AltMV-TGB1_P88 lacking insert induces negligible symptoms in N. benthamiana, compared to obvious systemic mosaic induced by the empty protein-expression vector AltMV-TGB1_L88 (Lim et al., 2010c); the lack of obvious symptoms means that any phenotype induced in the presence of an insert can be attributed to the effects of silencing the target gene, making AltMV an almost ideal VIGS vector, with a significant host range adding to its potential utility, and including many ornamental plants (see Table 1, Table 2).

The AltMV VIGS vector has also been used to suppress expression of endogenous 4/1 protein of N. benthamiana (known to interact with tospovirus and nepovirus tubule-forming movement proteins; von Bargen et al., 2001; Solovyev et al., 2013), and to affect movement of Potato spindle tuber viroid in 4/1-silenced plants (Solovyev et al., 2013).

High throughput vector system

In order to further increase the utility of the AltMV vector systems, and avoid the necessity to use restriction enzyme digestions to insert target genes for VIGS or protein expression, the full-length AltMV clones were converted into Gateway vectors by insertion of a Gateway cloning cassette into the MCS, creating AltMV-L-att and AltMV-P-att vectors for protein and VIGS applications respectively (Ko et al., 2015). The efficiency of the Gateway cloning system was demonstrated by first testing insertion of five different c.250 bp PCR products in equimolar proportion by a multiplex Gateway LR reaction, resulting in c.11% - 33% of clones for each PCR product, indicating minimal insertion bias (Ko et al., 2015). This was followed by insertion of a fragment of chloroplast β-ATPase and demonstration of a significant reduction of β-ATPase expression in inoculated plants compared to control plants. Utility for high-throughput insertion of inserts was demonstrated by creation of a whitefly (Bemisia tabaci) cDNA library with cDNA fragments predominantly in the range of 250 - 350 bp in a Gateway entry vector, and subsequent recombination into the VIGS vector AltMV-P-att, followed by transformation of the library into E. coli, and randomly-selected clones transformed into Agrobacterium tumefaciens; analysis of 50 randomly selected colonies revealed inserts of between 64 and 771 bp, with almost 60% of colonies having inserts in the range of 200 - 400 bp (Ko et al., 2015).

Conclusions

AltMV has several features that make it attractive as an alternate model system to complement what is known from prior studies with PVX, the type member of the genus Potexvirus, and from some other well-studied members of the genus. Two obvious differences are the targeting of TGB3 to the mesophyll and chloroplasts, and lack of detectable interaction between TGB2 and TGB3, by comparison to PVX. The close association of AltMV replication with the mesophyll cell layer as a function of TGB3 targeting to the chloroplasts presumably functions to bring the virus into association with the vascular system to initiate systemic transport of the virus, as the minor veins are embedded within the mesophyll.

Identification of virus:host interactions involving several of the virus-encoded proteins, and availability of multiple sequence variants should allow gene exchanges and dissection of differences in host range, and of symptom differences between isolates, as already demonstrated with chimeric constructs affecting the RdRp POL domain, TGB1, TGB3, and CP.

The difference in symptom severity and silencing suppression between AltMV variants with either TGB1_L88 (effective RNA silencing suppression) and TGB1_P88 allowed development of AltMV viral vectors suitable for high level protein expression, or VIGs, respectively, making AltMV one of the few viruses well-suited to both purposes; the wide host range of AltMV, infecting species in at least 24 taxonomically diverse plant families including many commercially important ornamentals, also suggests many uses for AltMV viral vectors. The availability of vectors suitable for establishment of infection via either agroinfiltration or inoculation of in vitro transcripts should allow direct infection of many hosts; it is also possible to establish infection in the readily infectible host N. benthamiana, and then to mechanically transfer infection by sap inoculation. Further adaption of the vector systems with the bipartite delivery system, or the Gateway vector system, offer further flexibility.