key: cord-0004875-3stcx0dd
authors: Mushegian, A. R.; Koonin, E. V.
title: Cell-to-cell movement of plant viruses: Insights from amino acid sequence comparisons of movement proteins and from analogies with cellular transport systems
date: 1993
journal: Arch Virol
DOI: 10.1007/bf01313766
sha: 2a6c2b454445036e0c621b7282464b5d69e4f922
doc_id: 4875
cord_uid: 3stcx0dd

Cell-to-cell movement is a crucial step in plant virus infection. In many viruses, the movement function is secured by specific virus-encoded proteins. Amino acid sequence comparisons of these proteins revealed a vast superfamily containing a conserved sequence motif that may comprise a hydrophobic interaction domain. This superfamily combines proteins of viruses belonging to all principal groups of positive-strand RNA viruses, as well as single-stranded DNA containing geminiviruses, double-stranded DNA-containing pararetroviruses (caulimoviruses and badnaviruses), and tospoviruses that have negative-strand RNA genomes with two ambisense segments. In several groups of positive-strand RNA viruses, the movement function is provided by the proteins encoded by the so-called triple gene block including two putative small membrane-associated proteins and a putative RNA helicase. A distinct type of movement proteins with very high content of proline is found in tymoviruses. It is concluded that classification of movement proteins based on comparison of their amino acid sequences does not correlate with the type of genome nucleic acid or with grouping of viruses based on phylogenetic analysis of replicative proteins or with the virus host range. Recombination between unrelated or distantly related viruses could have played a major role in the evolution of the movement function. Limited sequence similarities were observed between i) movement proteins of dianthoviruses and the MIP family of cellular integral membrane proteins, and ii) between movement proteins of bromoviruses and cucumoviruses and M1 protein of influenza viruses which is involved in nuclear export of viral ribonucleoproteins. It is hypothesized that all movement proteins of plant viruses may mediate hydrophobic interactions between viral and cellular macromolecules.

It is generally accepted that plant viruses exploit plasmodesmata to move from initially infected cells, which are usually negligible in amount, into neighbouring healthy cells. Without cell-to-cell movement productive virus infection is not established. Also it is known that specific movement proteins encoded by virus genomes are essential for cell-to-cell spread. Frequently, a plant virus protein is referred to as movement protein if i) it is not a capsid protein and ii) disruption of the coding sequence of this protein abolishes infection in whole plants but has no effect on virus replication in protoplasts. Putative movement proteins have been identified in this manner in about 15 different plant virus groups (see [4, 19, 40] for recent reviews). In some cases, additional virus proteins are also required for movement, e.g. coat protein [13, 52, 58] or a specific segment in a replication protein [55] , but these phenomena are not universal. On the other hand, a specific movement protein or at least movement domain is thought to be present in all viruses that are able to spread from cell to cell [4] .

Apart from this genetic evidence for the movement function, the current knowledge of mechanisms which facilitate intercellular spread of plant virus genomes is insufficient. Recently certain properties of some movement proteins, notably of the 30 kDa movement protein of tobacco mosaic virus (TMV), were characterized in vitro and in vivo and models for movement of plant viruses have been prompted by these studies [ 16, 19] . Additionally, movement proteins were subjects of comparative sequence analysis, which allowed to delineate common motifs in movement proteins from otherwise unrelated groups implying existence of common features among many movement proteins [33, 43] .

In this work, we sought to critically analyze currently available biochemical and ultrastructural data on movement proteins. We then investigated whether sequences of movement proteins could provide information on their function based on similarities with other viral or cellular proteins.

Fluorescent probes injected into a plant cell are either transported into neighbouring cells via plasmodesmata or retained in the initially injected cell. The crucial determinant for the exclusion of a particle from plasmodesmatal movement is its hydrodynamic (Stokes) radius [51] . Studies of transgenic tobacco plants constitutively expressing TMV 30 kDa movement protein (MP + plants) revealed that certain molecules, which were too large to be transported in wild type plants, gained such an ability in MP + tobaccos. Specifically, FITC-labelled dextran (Mr 9.4 kDa) moved from injected to neighbouring cells only in plants transformed with and expressing TMV 30 kDa protein but not in control MP-Plant virus movement proteins 241 plants [57] . Interestingly, plasmodesmata in MP + and MP-plants appeared to be structurally indistinguishable [19, 57] .

These data indicate that movement protein of TMV functionally modifies plasmodesmata. It is not understood yet whether these modifications and the changes required for moving virus genome through plasmodesmata are the same or different. Virus nucleic acid would not be expected to exist in naked form in the cell; rather, it seems reasonable that, like all nucleic acids in eukaryotic cells, it is associated with cellular and/or virus proteins throughout its lifetime. While Stokes radius of a 10 kDa dextran molecule is estimated to be about 3.1 nm [20, 57] , ribonucleoproteins (RNPs) are clearly larger, many of them having sizes within the range of 10-50 nm [41] , which would exclude them from movement even in MP + plants described in [57] .

Recently, plasmodesmatal permeability in Nicotiana clevelandii leaves infected with tobacco rattle virus (TRV) has been investigated [20] . Movement protein of TRV, the 29 kDa protein, shares high sequence similarity with TMV movement protein, and both proteins are considered to be functionally very similar. Simultaneous injection of virus and fluorescent tracers into cells of leaf trichomes allowed to observe movement of the labels under the conditions when the virus itself was known to move in the same cells. It has been shown that FITC-labelled dextran (Mr 4.4kDa) moved from cell to cell only in TRVinfected cells and synchronously with the virus. However, Lucifer Yellow-labelled dextran (Mr 10 kDa) was excluded from plasmodesmatal movement even in infected cells. Thus, certain molecules with the dimensions resembling the smallest of the known RNPs did not move under conditions when virus RNPs readily moved. Apparently, either an extremely thin virus-specific RNP should be postulated (see below), or it might be concluded that slight increase in plasmodesmatal permeability induced by movement proteins hints at some important circumstances but is not sufficient to explain celt-to-cell movement of the virus genome.

Binding of movement proteins to single-stranded nucleic acids TMV 30 kD movement protein has been overexpressed in E. coli, and purified protein has been shown to bind single-stranded (ss) DNAs and RNAs nonspecifically and cooperatively [14] . A model has been proposed, in which movement protein binds genomic RNA stoichiometrically to unfold and shape it into a thin complex [ 14, 16, 17] . Electron microscopy of such complexes formed in vitro revealed thin structures at the limit of microscope resolution; it was concluded that, as it was barely seen under EM, the complex was thin enough to fit in plasmodesmata with size exclusion limit of 3 nm [17] . Additionally, ss nucleic acid-binding properties have been reported for movement proteins of cauliflower mosaic caulimovirus (CaMV) [ 15] , red clover necrotic mosaic dianthovirus (RCNMV) ( [47] , D. Cookmeyer, pers. comm.) and alfalfa mosaic virus (AIMV) [-40] . Interestingly, P1, the movement protein of cauliflower mosaic virus, has been shown to preferentially bind RNA rather than DNA 242 A.R. Mushegian and E. V. Koonin [15] ; it was speculated that 35 S transcript, a genome length virus replication intermediate, might be the form which is transported from cell to cell.

These in vitro observations offer new insights into a mechanism of virus cell-to-cell movement. However, is should not be forgotten that virus RNAs in vivo are associated with proteins, as revealed for TMV-specific RNAs [22] ; it is thought that such an association is maintained throughout the whole life of the RNPs in the cell; a mechanism of stripping virus RNA from these proteins has not yet been proposed. Also, as binding of movement proteins to RNA is thought not to be sequence-specific [14, 17] , it could be anticipated that the majority of movement protein will bind to cellular RNAs, which in the case of TMV infection are in excess over virus RNAs at the time when 30 kD protein is transiently expressed [18] . The "ss nucleic acid-binding" model, therefore, has to be modified to deal with these difficulties.

Biochemical fractionation of infected cells revealed that in many virus-plant interactions movement proteins are found in cell wall-enriched fraction. Immunogold labelling has confirmed these observations and demonstrated localization of TMV, A1MV, CaMV, and RCMV movement proteins in plasmodesmatal channels or in cell wall near plasmodesmata [36, 40] . Movement protein of cowpea mosaic comovirus (CPMV), the 58 kD/48 kDa protein, has been found to form tubular structures perpendicular to and intruding into plasma membrane and cell wall. These tubules have been found both in whole leaves and in protoplasts [-56] . Virions have been observed inside the tubules [56] . It has been speculated that the tubular inclusions are the structures required for movement.

Cucumber mosaic virus (CMV) movement requires the 32 kDa movement protein [7] . This protein has not been found to associate with the cell wall or from tubules; instead, it was located in nucleoli of infected cells [-37] .

Observations on movement proteins of TMV and CPMV have given rise to the idea of two different types of virus movement, "tobamo-type" and "comotype". It has been proposed that the tobamo-type is characterized by movement protein localization in the plasmodesmatal area of the cell wall and by the virus infectious entity capable of cell-to-cell movement in the absence of the viral capsid protein. Como-type of movement was thought to require both movement protein and coat protein and to involve formation of tubular structures protruding from cell wall into cytoplasm and formed by the movement protein [19, 56] . Complementation data argued, however, that there is certain compatibility among both types of movement as tobamoviruses were able to complement comovirus movement [4] .

When all available ultrastructural data are considered, the picture becomes not so clear-cut. It should be expected that RCMV, a comovirus closely related to CPMV, moves according to the como-type; instead, movement protein of RCMV was found exclusively in plasmodesmata of the infected plants [-40] which had been proposed to be a tobamo-type feature. CaMV movement protein is found in cell wall fraction and, more specifically, in plasmodesmata [2, 36] ; this seemingly supports the attribution of caulimovirus movement to the tobamo-type. However, tubular structures of unknown composition have been observed in the cytoplasm of CaMV-infected cells [36] and recently it has been shown that P1, the movement protein of CaMV, forms tubular structures on the surface of inoculated protoplasts [49] . These findings seem to be more consistent with the como-type action of the caulimovirus movement protein.

The above discussion shows that it may be premature to draw too sharp a distinction between different types of movement based on ultrastructural observations. It also should be remembered that we see movement proteins at the sites where they are most easily observed and/or are retained for longer periods, but not necessarily at the sites of their essential activity.

With full-length DNA copies of numerous virus genomes available, attempts have been made to genetically dissect movement protein coding sequences. In most cases, lengthy deletions have been introduced, and effect of these deletions on the overall infectivity or individual properties of the movement proteins has been investigated. It has been shown that even small changes near the N-terminus of TMV 30 kD movement protein abolish virus infectivity [6] ; at the C-terminus, up to 33 of the 268 amino acid residues of this protein could be deleted without apparent effect on infectivity or cell wall localization of the movement protein.

When 55 C-terminal amino acids were deleted, virus moved slowly though movement protein was still found in plasmodesmata [-6] . Deletion of 74 Cterminal residues (195 to 268) was lethal, and, at the same time, movement protein was no longer found in plasmodesmata. It has been noted that essential residues from 195 to 214 could comprise a domain with a specific function, or their deletion might simply cause misfolding of the protein I-6].

In another study, C-terminal and internal deletions have been introduced into TMV movement protein to determine location of ss nucleic acid-binding site(s) [-17] . Initially, it was thought that residues 65 to 86 constitute RNAbinding domain [14] . Later, this domain has been shown to rather decrease protein solubility [17] . Up to 84 amino acids from the C-terminus of TMV movement protein could be eliminated without effect on ss nucleic acid binding capacity of the protein [-17]. In the remaining part, deletion of amino acids 111-125 abolished ss nucleic acid binding, but when the gene segment encoding these amino acids was fused to an unrelated sequence, the resulting himeric protein did not bind to ss nucleic acids [17] .

Movement protein of A1MV associates with cell wall upon infection and in transgenic tobacco plants; however, this did not happen when N-proximal amino acids (12-77) have been deleted [-24] .

Strains of TMV have been characterized that are temperature sensitive in 244 A.R. Mushegian and E. V. Koonin: Plant virus movement proteins cell-to-cell movement or break the host resistance which is based on plant genes restricting virus movement. Involvement of mutations in 30 kD movement protein cistron could be implied, and, indeed, in many cases nucleotide changes responsible tbr the altered phenotype were shown to be confined to the movement protein gene [11] . Many of these mutations were scattered around the middle one-third of 30 kD movement protein and resulted in a change in polarity of the encoded amino acid [11] . As no known function could be mapped to the area, these data are difficult to interpret.

With representative sequences of plant virus genomes from many groups available, efforts were made to identify regions of similarity between known movement proteins and to predict movement functions for uncharacterized ORFs. Early observations made by this approach have been reviewed previously (e.g. [4] ) and included strong similarity between tobamovirus and tobravirus movement proteins [29] as well as less pronounced but significant local similarity between segments of tobamovirus and caulimovirus movement proteins [30] . Later, similarities between movement proteins of tricornaviruses and dianthoviruses were observed [59] . In fact, the movement function for the respective proteins of tobraviruses, caulimoviruses and dianthoviruses has been implied by these observations and only subsequently confirmed by site-directed mutagenesis [54, 59, 60] . More detailed analyses performed by Melcher [43] and by ourselves [33] revealed longer (ca. 150 amino acids) segments of sequence similarity between movement proteins from different virus groups. Interestingly, one family of movement proteins included proteins encoded by viruses with positive-strand RNA genomes, single-strand DNA geminiviruses and by pararetroviruses with virion DNA, namely caulimoviruses [33] .

In the time past after the publication of these analyses, further significant increase in the number of sequenced virus genomes was achieved and it has been argued that the current collection of such sequences may represent the majority of the existing virus groups [32, 34] . With this in mind, we undertook a systematic comparison of the sequences of the movement proteins with each other and with the current sequence databases.

Using the BLAST program [3] , statistically highly significant similarities were observed only between proteins of viruses belonging to a single group or two closely related groups. Further comparisons were done using programs MA-CAW [53 a] and OPTAL [27] to generate multiple alignments according to the progressive alignment strategy [21] , i.e. starting with pairs of most closely related sequences and proceeding to incorporate more and more distant ones. It was demonstrated that movement proteins of viruses belonging to 17 groups contain a weak but confidently identified conserved motif consisting of approximately 30 amino acid residues (Fig. 1) . This motif includes a nearly invariant aspartic acid residue (replaced by asparagine in dianthoviruses and ilarviruses) preceded by a region enriched in hydrophobic and nonpolar residues. In some of the movement proteins this region was predicted to form a transmembrane helix using the algorithm of Rao and Argos ( [50] , data not shown; see also discussion below). Remarkably, this conserved motif unites viruses with single-stranded DNA genomes (geminiviruses of group II), pararetroviruses with double-stranded DNA genomes (caulimoviruses and badnaviruses), negative-strand RNA viruses with ambisense genome segments (tospoviruses), and numerous groups of positive-strand RNA viruses. Among the latter, all the three major phylogenetically defined divisions [31, 34] are represented, namely ruN-like viruses (tobamoviruses, tobraviruses, cucumoviruses, bromoviruses, ilarviruses, capilloviruses, furoviruses, raspberry bushy dwarf virus), picornalike viruses (comoviruses, parsnip yellow fleck virus, pea enation mosaic virus), and flavi-like viruses (dianthoviruses and tombusviruses). Several groups of movement proteins showed significant similarity on longer sequence stretches. In Fig. 2 we show the amino acid sequence alignments for four of these groups, delineation of which included identification of new putative movement proteins. Such new identifications are: i) 38 kDa protein of apple stem grooving virus, which is related to the putative movement proteins of the other capilloviruses, namely apple chlorotic leaf spot virus and potato virus T (Fig. 2 a) ; ii) 37 kDa protein of soil-borne wheat mosaic furovirus related to the movement proteins ofdianthoviruses (Fig. 2 b) ; iii) 27 kDa protein of pea enation mosaic virus and 21 kDa protein of tombusviruses related to the movement proteins of bromoviruses and cucumoviruses (Fig. 2 c) ; and iv) NSm protein of Fig. 2 . Amino acid sequence alignments for groups of proteins including newly identified putative movement proteins. For each group of boundaries of the conserved segments were determined using the MACAW program and the alignment itself was constructed hierarchically using the OPTAL program. In the consensus U indicates a bulky aliphatic residue (I, L, V, or M), a indicates an aromatic residue (F, I7, or W) & indicates a bulky hydrophobic residue (aliphatic or aromatic) and x indicates any residue. The conserved motif of the "30 K superfamily" is highlighted by bold typing, a Putative movement proteins of capilloviruses. ACLSV Apple chlorotic leaf spot virus; PVT potato virus T; ASGV apple stem grooving virus; b movement protein of red clover necrotic mosaic dianthovirus (RCNMV), putative movement proteins of raspberry bushy dwarf virus (RBD V) and soil-borne wheat mosaic furovirus (SB WMV); e movement proteins of cucumber mosaic cucumovirus (CMV) and brome mosaic bromovirus (BMV), putative movement proteins of pea enation mosaic virus (PEMV) and cucumber necrosis tombusvirus (CNV); the similarity between 21 kDa proteins of tombusviruses and movement proteins of bromoviruses and cucumoviruses has been independently reported by U. Melcher (pers. comm.); d putative movement proteins of caulimoviruses, badnaviruses, parsnip yellow fleck virus, and tomato spotted wilt tospovirus; asterisks show identical residues and colons show similar residues between the sequences of PYFV and FMV, and CoYMV and TSWV tospoviruses and the N-terminal domain of the parsnip yellow fleck virus polyprotein related to the putative movement proteins of caulimoviruses and badnaviruses (Fig. 2 d; the similarity between caulimovirus movement proteins and N-terminal domains of badnavirus polyproteins was noted by U. Melcher and cited in [8] ). Even among these smaller groups, the latter three brought together proteins of very different viruses. Group iv) is particularly notable in that it compiles proteins of a positive-strand RNA virus, negative-strand (ambisense) RNA viruses, and DNA-containing pararetroviruses. Curiously, the array of domains, namely N-movement domain-capsid protein(s)-replicative domains-C in the polyproteins of such seemingly unrelated viruses as parsnip yellow fleck virus and the badnaviruses is very similar. A relevant observation from our previous work is the relationship between movement proteins of two component geminiviruses (BL1/BC1) with those of tobamoviruses and tobraviruses [33] . All the proteins containing the conserved motifs shown in Fig. 1 should be considered a single vast, and highly diverged superfamily of plant virus movement proteins. We would like to coin the nickname "30 K superfamily", after the most thoroughly studied movement protein of tobamoviruses, and also because most of the proteins of this superfamily have the size around this value (Fig. 1 a) .

Several groups of positive-strand RNA viruses, namely potexviruses, carlaviruses, hordeiviruses, and a group including beet necrotic yellow vein virus, Nicotiana velutina mosaic virus, and peanut clump virus, encompass the triple gene block that consists of two small membrane proteins and a putative RNA helicase [34, 44] . It has been demonstrated that disruption of any of the triple block genes abolished the cell-to-cell movement of hordeiviruses, potexviruses and BNYVV [5 a, 25 a, 49 a]. The putative helicases in the triple block comprise a distinct group within superfamily I of DNA and RNA helicases and contain several unusual amino acid substitutions in the conserved helicase motifs [26, 34] . Unlike the genes coding for the proteins of the "30 K superfamily", the triple block is found only in positive-strand RNA viruses. Carmoviruses and necroviruses encode a pair of small proteins, one of which is predicted to be an integral membrane protein. It has been shown that disruption of either of these genes precluded cell-to-cell movement of turnip crinkle carmovirus [28] . These two genes perhaps may be considered a "truncated triple block" although it was hard to demonstrate this at the level of sequence comparison due to the small size of the hydrophobic proteins.

A third type of movement protein has been revealed in tymoviruses. A 75 kDa proline-rich protein that is encoded in the 5' portion of the genomic RNA and overlaps with the replicative polyprotein gene is essential for the movement of TYMV in infected plants [10] . This protein may have specific secondary and tertiary structure similar to that of other proline-rich filamentous proteins.

Interestingly, upon screening of the amino acid sequence databases with the sequence of the tymovirus movement protein the highest (albeit moderate) similarity was observed with the plant cell wall-associated protein extensin (data not shown).

Recently it has been shown that cell-to-cell movement of geminivirus MSV is dependent on the small protein V1 [9] . This protein and the related proteins of other one component geminiviruses did not show appreciable sequence similarity to movement proteins of other viruses. Interestingly, however, alignment of their amino acid sequences revealed the conservation of a highly hydrophobic central domain terminating at an invariant aspartic acid residue and thus distantly resembling the conserved motif of the 30 K superfamily (Fig. 3) . This domain was flanked by two proline-rich domains showing some similarity to extensins (Fig. 3) . Thus, in a sense the movement proteins of one component geminiviruses combined the 30 K superfamily and the tymovirus structural themes.

For several virus groups, in which at least one member has been completely sequenced, the movement function has not been genetically mapped and identification of a putative movement proteins by amino acid sequence comparison was very uncertain if possible at all. Among these, uncharacterized proteins of nepoviruses and closteroviruses showed a very limited sequence similarities to some members of the 30 K superfamily, in particular to caulimovirus movement proteins [33] .

The observations discussed in the previous sections show that classification of movement proteins based on comparison of their amino acid sequences does not correlate with the type of genome nucleic acid or with grouping of viruses based on phylogenetic analysis of replicative proteins 1-31, 34, 61] . Also, we were unable to notice any correlation between grouping of movement proteins and biological properties of the viruses encoding these proteins (e.g. host ranges). Recombination between unrelated or distantly related viruses could have played a major role in the evolution of the movement function. This is clearly the preferential explanation for situations when related movement proteins are found in viruses with different type of genome nucleic acid (e.g. the striking similarity between the putative movement proteins of caulimo/badnaviruses Fig. 2 d) . In some cases recombinational transfer of movement protein genes appears very likely also between remote groups of positive-strand RNA viruses (e.g. dianthoviruses and soil borne wheat mosaic furovirus; Fig. 2 c) .

Another equally important trend in the evolution of movement protein genes is the apparent high rate of mutational change leading to the very limited level of sequence conservation as featured above. Even within compact groups of viruses that share common genome organization and functional properties (e.g. tobarnoviruses), the divergence between the movement protein sequences is much higher than that between the principal repticative domains although comparable to that between capsid proteins.

It has been proposed that plasmodesmata are plant analogues for certain animal cell-cell contacts, namely gap junctions, based mainly on studies in permeability of both to fluorescent dye-labelled molecules [5, 51] . Details of protein organization of plasmodesmata are poorly understood, although a protein serologically related to a major gap junctions protein, connexin, was reportedly found in plasmodesmata and gene encoding this plant protein was cloned [42] . In animal kingdom, gap junctions are important for electrical coupling of cells in various tissues [51] . Interestingly, permeability of both plasmodesmata and gap junctions for some low molecular weight tracer dyes is down-regulated by activation of protein kinase C [5] . However, movement of macromolecules, in particular nucleoproteins, through gap junctions, has not been described.

A case of movement of a nucleoprotein through a channel in a membrane is represented by nucleocytoplasmic trafficking of RNPs (reviewed in [38, 45] ). Among the features of the nucleocytoplasmic export of cellular RNAs revealed thus far, several might be relevant to the mechanisms of movement of plant virus genomes. In particular, it has been shown that at all stages of nucleocytoplasmic export, mRNA is attached to specific protein frameworks, being transferred from nucleoskeleton to nuclear pore complex to cytoplasm [12] ; specific proteins in mRNP are thought to chaperone association-dissociation of the mRNA to and from these frameworks [1, 53] . Components of mRNP activate enzymes within the nuclear pore complex which are essential for energydependent translocation of RNP through nuclear pore; two of these enzymatic activities are ATPase and RNA helicase [53] . It is tempting to speculate that mechanistic analogies may exist between the nucleocytoplasmic transport of cellular RNPs and intracellular and/or intercellular movement of viral RNPs. More specifically, the parallel between the RNA helicase in the nuclear pore and the putative viral RNA helicase in the triple block that is involved in virus movement (see above) certainly is provocative. It should be noted that plant viruses, which undergo some stages in the nucleus of the infected cells (i.e. pararetroviruses and geminiviruses), should be able to perform both nucleocytoplasmic movement and cell-to-cell movement of their genomes; one may wonder to what extent these two processes are similar and whether movement proteins are involved in both of them.

Comparison of the amino acid sequences of dianthovirus movement proteins with amino acid sequence databases revealed a marginally significant but provocative similarity with membrane proteins of the MIP family [48] . With the BLAST score of 67 and probability of random matching of 0.23, the similarity of the movement protein of RCNMV with tonoplast membrane protein TIP 1 from Arabidopsis was the highest in the entire database except for the similarities with the homologous proteins of the other dianthoviruses. Analysis of the multiple alignment of the dianthovirus movement proteins with a representative set of the MIP proteins (Fig. 4) showed that the region of similarity included the conserved motif of the "30 K superfamily" described above and the characteristic conserved motif of the MIP family [48] . Each of these motifs consisted of a putative transmembrane helix succeeded by a loop containing conserved amino acid residues (Fig. 4) . The drastic difference between the two types of proteins is that MIP proteins contain six transmembrane helices whereas in the virus movement proteins the region shown in Figs. 1 and 4 is the only such predicted helix. Nevertheless, we believe that the observed sequence similarity may be functionally relevant indicating that the conserved motif of the "30 K superfamily" may mediate membrane interaction. For many proteins of this superfamily, membrane interaction is compatible with the observed localization in the cell wall [40] . However, caution is due in the interpretation of these observations as the hydrophobicity of the conserved region in different movement proteins differed significantly and not for all of them a membrane-spanning helix was confidently predicted ( Fig. 1 b and Fig. 4 . A conserved sequence motif in the movement proteins of dianthoviruses and the MIP family of membrane proteins. Asterisks show identical residues and colons show similar residues in the sequences of the movement protein of RCNMV and TIP1. The consensus (for symbols see legend to Fig. 2) includes the conserved residues, with one possible exception in the MIP-related proteins. The hydrophobic regions predicted to form transmembrane helices are underlined. The motif was extracted from an alignment generated using the MACAW program in some of these proteins the conserved motif may be involved in hydrophobic interactions other than membrane spanning. As mentioned above, membrane localization also has been proposed for the triple block proteins and strongly predicted for the movement proteins of monopartite geminiviruses.

The observations on the involvement of influenza virus M 1 protein in nuclear export of virus ribonucleoproteins [-39] prompted us to directly compare the M 1 sequence to those of plant virus movement proteins. Moderate similarity was revealed between M 1 and the movement proteins of bromoviruses and cucumoviruses (Fig. 5) . The region of similarity included the conserved motif of the "30 K superfamily", with the invariant aspartic acid residue and the preceding hydrophobic stretch. M 1 protein chaperones newly formed influenza virus RNPs from the nucleus to the cytoplasm; upon formation of virions, this protein is thought to provide the link between the RNP and the virion membrane (reviewed in [35] ). Both of these functions are apparently based on the ability of M 1 to interact both with RNP and the membrane providing a provocative analogy with the plant virus movement proteins. Very recently, a mutation in M 1 protein gene resulting in impaired nucleocytoplasmic movement was mapped within the region of similarity to bromovirus and cucumovirus movement proteins [23] .

Finally, our previous observations indicated a weak sequence similarity between a domain of cellular 90 kDa heat shock proteins with some of the movement proteins, specifically those of caulimoviruses [33] .

A common denominator for all these observations may be that the function of plant virus movement proteins is determined by domains mediating hydrophobic interactions. Taking into account the limited sequence conservation and the absence of strictly invariant amino acid residues, it is very unlikely that these proteins (with the exception for the putative helicase in the triple block) have any enzymatic activity.

Among the known groups of plant viruses [25] , movement proteins so far have been identified genetically or by sequence comparison for about 75 percent. The variety of these proteins could be reduced to only three types, namely the "30K superfamily", with the distantly similar proteins of monopartite geminiviruses; the triple block; and the proline-rich proteins of tymoviruses. Obviously, future studies may lead to characterization of movement proteins unrelated to any of these groups. Different types of movement proteins may affect different aspects in viruscell interaction leading to virus spread in planta. However, it appears likely that in many cases these proteins mediate hydrophobic interactions between viral and cellular components, in a general analogy with molecular chaperone action [16, 33] .

As discussed in the first part of this review, a large fraction of experimental data on plant virus movement proteins is very hard to be interpreted unequivocally. Conceivably, this reflects the highly complex nature of virus-plant interaction. In this situation, it appears reasonable to focus further experiments on probing the specific functions fo conserved domains revealed by amino acid sequence comparison. Hopefully, this approach will highlight new important aspects of plant virus lifestyle.

Recently, it has been shown that TMV movement protein expressed by transgenic tobacco behaves as an integral membrane protein [Moore PJ, Fenczik CA, Deom CM, Beachy RN (1992) Developmental changes in plasmodesmata in transgenic tobacco expressing the movement protein of tobacco mosaic virus. Protoplasma 170:115-127].

Between nucleus and cytoplasm

Cauliflower mosaic virus gene I product detected in cell wall-enriched fraction

Basic local alignment search tool

Expression of a plant virus-coded transport function by different viral genomes

Dynamic continuity of cytoplasmic and membrane compartments between plant cells

Triple gene block proteins of white clover mosaic potexviurs are required for transport

The TMV movement protein: role of the C-terminal 73 amino acids in subcellular localization and function

Mutational analysis of cis-acting sequences and gene function in RNA 3 of cucumber mosaic virus

An analysis of the complete nucleotide sequence of a sugarcane bacilliform virus genome infectious to banana and rice

Replication of maize streak virus mutants in maize protoplasts: evidence for a movement protein

Expression of ORF-69 of turnip yellow mosaic virus is necessary for virus spread in plants

Nucleotide sequence analysis of the movement genes of resistance breaking strains of tomato mosaic virus

A three-dimensional view of precursor messenger RNA metabolism within the mammalian nucleus

Potato virus X as a vector for gene expression in plants

The P 30 movement protein of tobacco mosaic virus is a single-stranded nucleic acid binding protein

Gene I, a potential cell-to-cell movement locus of cauliflower mosaic virus, encodes an RNA-binding protein

How do plant virus nucleic acids move through intercellular connections

Visualization and characterization of tobacco mosaic virus movement protein binding to single-stranded nucleic acids

Relationship of tobacco mosaic virus gene expression to movement within plant

Plant virus movement proteins

Increase in plasmodesmatal permeability during cell-to-cell spread of tobacco rattle virus from individually inoculated cells

Of URFs and ORFs. A primer on how to analyze derived amino acid sequences

The Plant virus movement proteins 255

informosome-like virus-specific ribonucleoprotein (vRNP) may be involved in the transport of tobacco mosaic virus infection

An influenza virus temperature-sensitive mutant defective in the nuclear-cytoplasmic transport of the negative-sense viral RNAs

An N-proximal sequence of alfalfa mosaic virus movement protein is necessary for association with cell walls in transgenic plants

Classification and nomenclature of viruses. Fifth Report of the International Committee on Taxonomy of Viruses

Efficient cell-tocell movement of beet necrotic yellow vein virus requires 3' proximal genes located on RNA 2

A novel superfamily of nucleoside triphosphate-binding motif-containing proteins which are probably involved in duplex unwinding in DNA and RNA replication and recombination

An ATP-binding motif is the most conserved sequence in a highly diverged group of proteins involved in positive strand RNA viral replication

Turnip crinkle virus genes required for RNA replication and virus movement

The complete nucleotide sequence of tobacco rattle virus RNA-1

The sequence of carnation etched ring virus DNA: comparison with cauliflower mosaic virus and retroviruses

The phylogeny of RNA-dependent RNA polymerases of positivestrand RNA viruses

Virus evolution: time for Sturm und Drang

Diverse groups of plant DNA and RNA viruses share related movement proteins that may possess chaperonelike activity

Evolution and taxonomy of positive-strand RNA viruses: implications of comparative analysis of amino acid sequences

Genes and proteins of the influenza viruses

The subcellular localization of gene I product of cauliflower mosaic virus is consistent with a function associated with virus spread

Ultrastructural location of non-structural protein 3A of cucumber mosaic virus in infected tissue using monoclonal antibodies to a cloned chimeric fusion protein

Nuclear mRNA export

Nuclear transport of influenza virus ribonucleoproteins: the viral matrix protein (M1) promotes export and inhibits import

Virus movement in infected plants

Translocation of a specific premessenger ribonucleoprotein particle through the nuclear pore studied with electron microscope tomography

Gap junction protein homologue in Arabidopsis thaliana: evidence for connexins in plants

Similarities between putative transport proteins of plant viruses

Probable reassortment of genomic elements among elongated RNA-containing plant viruses

Nuclear import-export: in search of signals and mechanisms

Detection of the movement protein of red clover necrotic mosaic virus in a cell wall fraction from infected Nicotiana clevetandii plants

Cooperative binding ofthe red clover necrotic mosaic movement protein to single stranded nucleic acids

Evolution of the MIP family of integral membrane transport proteins

Cauliflower mosaic virus gene I product (P 1) forms tubular structures which extend from the surface of infected protoplasts

Identification of barley stripe mosaic virus genes involved in viral RNA replication and systemic spread

A conformation preference parameter to predict helices in integral membrane proteins

Spray DC (eds) (1990) Parallels in cell-to-cell junctions in plants and animals

Effects of deletions in the N-terminal basic arm of brome mosaic virus coat protein on RNA packaging and systemic infection

Evidence for involvement of a nuclear envelope-associated RNA helicase activity in nucleocytoplasmic RNA transport

A workbench for multiple alignments construction and analysis

A mutation in cauliflower mosaic virus gene I interferes with virus movement but not virus replication

Deletion analysis of brome mosaic virus 2 a protein: effects on RNA replication and virus spread

Tubular structures involved in movement of cowpea mosaic virus are also formed in infected cowpea protoplasts

Plasmodesmatal function is probed using transgenic tobacco plants that express a virus movement protein

Cell-to-cell transport of cowpea mosaic virus requires both the 58/48 K proteins and the capsid proteins

The roles of the red clover necrotic mosaic virus capsid and cell-to-cell movement proteins in systemic infection

Tobacco rattle virus RNA-1 29 K gene product potentiates viral movement and also affects symptom production in tobacco

Evolution of RNA viruses

We appreciate helpful discussions with Dr. V. V. Dolja and Dr. U. Melcher. We would like to thank Drs. D. Cookmeyer, S. A. Demler, R. Kormelink, U. K. Melcher, N. Olszewski and Yu. Shirako for communicating data prior to publication. A. M. is grateful to Dr. R. J. Shepherd for constant support and encouragement.

Received June 9, 1993