Plant cell growth and development depend on continuous cell proliferation which is restricted to small regions of the plant called meristems. Infection by geminiviruses, small DNA viruses whose replicative cycle relies on host cell factors, is excluded from those proliferating areas. Since most of the replicative factors are present, almost exclusively, in proliferating cells, geminivirus infection is believed to induce a cellular state permissive for viral DNA replication, e.g. S‐phase or, at least, some specific S‐phase functions. The molecular basis for this effect seems to be the interference that certain geminivirus proteins exert on the retinoblastoma‐related (RBR) pathway, which analogously to that of animal cells, regulates plant cell cycle activation and G1–S transition. In some cases, geminiviruses induce cell proliferation and abnormal growth. Mechanisms other than sequestering plant RBR probably contribute to the multiple effects of geminivirus proteins on cellular gene expression, cell growth control and cellular DNA replication. Current efforts to understand the coupling of geminivirus DNA replication to cell cycle and growth control as well as the directions in which future research is aiming are reviewed.
The evolutionary trends of DNA replication and cell cycle and growth control are mechanistically well conserved among eukaryotes, from yeast to humans, from flies to plants (Stillman, 1996; Gutierrez, 1998; Huntley and Murray, 1999; Mironov et al., 1999; Sherr and Roberts, 1999). Some of the key regulatory components, their activators and inhibitory cofactors, their upstream effectors and downstream targets are conserved. Furthermore, many of the building blocks are so similar that, in some cases, they can substitute functionally between different species. It is noteworthy, however, that the kind of stimuli perceived by different organisms, their response to them and the pleiotropic roles that cellular regulatory proteins play, other than those as components of the cell cycle machinery, are quite distinct in multicellular organisms at the physiological and developmental levels. An illustrative example is provided by studies of the retinoblastoma (RB) tumour suppressor pathway in different animal systems (RB and interacting proteins; Brook et al., 1996; Neufeld et al., 1998; Du and Dyson, 1999; Hsieh et al., 1999; Macleod, 1999).
Plants have, in addition to specific metabolic pathways, unique growth characteristics, developmental patterns and body architecture. These are consequences of several plant‐specific features such as the plasticity of the plant cell, which largely contributes to its ability to dedifferentiate and regenerate, the continuous post‐embryonic body remodelling, which requires a continuous proliferative potential, the frequent occurrence of endoreduplication cycles and the lack of cell migration, among others. Since the original isolation of plant cdc2 homologues (Feiler and Jacobs, 1990; John et al., 1990), the past decade has witnessed significant progress in the identification of plant cell cycle regulators (see, for example, John, 1996; Verma and Gu, 1996; Gutierrez, 1998; Hesse et al., 1998; Huntley and Murray, 1999; Mironov et al., 1999; Wei and Deng, 1999). Comparatively, study of plant DNA replication enzymes and other regulatory proteins is scanty and still very far behind studies in other eukaryotes. Steps towards identifying cell cycle and DNA replication genes in plants are necessary in order to begin to understand their roles in plant growth and development (Clark and Schiefelbein, 1997; Cebolla et al., 1999; Riou‐Khamlichi et al., 1999) and their differences from and similarities to animal systems.
The study of geminivirus biology, in particular their DNA replication mechanism and its connection to the host cell physiology, is a research line complementary to others based on homology or genetic screenings, functional complementation analysis and genomics. This has been exploited in recent years as (i) the geminivirus replicative cycle relies extensively on host cell factors and (ii) infection appears to have a significant impact on several cellular processes. Below, I will discuss these aspects and how these studies are providing new insights into understand the coupling of DNA replication, cell cycle and growth control in plants. Comprehensive reviews covering the molecular biology of geminiviruses have appeared recently (Bisaro, 1996; Palmer and Rybicki, 1998; Gutierrez, 1999; Hanley‐Bowdoin et al., 1999; Lazarowitz, 1999).
Geminivirus DNA replication follows the model of ‘simple genomes’ with complex virus–host interactions
A large variety of plant species are susceptible to infection by geminiviruses, a family of viruses named after their unique geminate (twinned) virion morphology. The viral genome consists of one or two small circular single‐stranded DNA (ssDNA) molecules (2.6–3.0 kb). Transcription of geminivirus genes occurs bidirectionally on a double‐stranded DNA (dsDNA) template from divergent promoters located in a DNA region containing cis‐acting signals required for viral transcription and DNA replication (Figure 1).
Geminivirus DNA replication follows a rolling circle strategy (Figure 2; Saunders et al., 1991; Stenger et al., 1991) which resembles that of prokaryotic ssDNA replicons (Baas and Janzs, 1988; Novick, 1998). The initial stage encompasses the conversion of the ssDNA genome into a dsDNA intermediate product (Figure 2; Kammann et al., 1991; Saunders et al., 1992). This step, poorly understood in molecular terms, must be carried out entirely by cellular enzymes. DNA primase, α‐like and δ‐like DNA polymerase activities have been identified in plants (Richard et al., 1991; Bryant et al., 1992; Laquel et al., 1993; Coello and Vazquez‐Ramos, 1995a; Garcia et al., 1997), but a direct demonstration of their participation in geminivirus DNA replication still awaits further work (for a detailed discussion, see Gutierrez, 1999). This line of research is of special importance as it should help in the identification of plant genes encoding DNA polymerase and accessory factors.
Initiation of DNA replication during the second stage, the rolling circle phase, requires the concerted action of the viral Rep protein (and perhaps other viral proteins) with cellular factors, and leads to the production of new dsDNA and ssDNA viral forms (Figure 2; Stenger et al., 1991; Heyraud et al., 1993; Stanley, 1995). In this respect, it conforms to the so‐called ‘simple replicon model’ (DePamphilis, 1996). The DNA replication origin has a modular architecture (Fontes et al., 1994a; Sanz‐Burgos and Gutierrez, 1998) and falls into two major categories: the mastrevirus‐type origin, e.g. wheat dwarf virus (WDV; Kammann et al., 1991; Hofer et al., 1992; Schneider et al., 1992; Sanz‐Burgos and Gutierrez, 1998), consists of a large cis‐acting region where the initiator Rep protein forms multiple complexes (Castellano et al., 1999), and the begomovirus‐type origin, e.g. tobacco golden mosaic virus (TGMV; Fontes et al., 1994a), containing one binding site for Rep (Fontes et al., 1992, 1994b; Lazarowitz et al., 1992; Orozco and Hanley‐Bowdoin, 1998). Further details on initiation of geminivirus DNA replication have been reviewed recently (Gutierrez, 1999; Hanley‐Bowdoin et al., 1999).
After initiation, recruitment of cellular replication factors is a necessary step to complete viral DNA replication. Future efforts to identify these components should concentrate on at least two complementary approaches: the biochemical characterization of replication factors and the identification of potential interactions with one or more viral proteins. In this context, preliminary results have revealed interactions between the initiator Rep protein and cellular DNA replication proteins (A.P.Sanz‐Burgos and C.Gutierrez, unpublished results). The absolute requirement for host cell DNA replication factors has forced geminiviruses to develop a number of complex interactions with the host cell which are discussed in the following paragraphs.
Geminivirus proteins and the retinoblastoma pathway: one strategy, two mechanisms?
The prokaryotic‐like DNA replication strategy used by geminiviruses has adapted to a eukaryotic cellular environment and life style. A characteristic feature of multicellular organisms is that most of their cells are non‐proliferating, differentiated cells in which DNA replication factors are absent or functionally inactive. Plant architecture depends on a post‐embryonic morphogenetic process whereby new cells are produced continuously in the meristematic regions where cell proliferation activity is confined. Also in plants, changes in proliferating activity are associated with changes in the levels of DNA replication enzymes (Coello and Vazquez‐Ramos, 1995b; Benedetto et al., 1996).
It is somewhat surprising that geminivirus DNA replication seems to be excluded from the actively dividing regions of the plant, called meristems (Rushing et al., 1987; Horns and Jeske, 1991; Lucy et al., 1996). As a consequence, geminivirus DNA replication has been suspected for a long time to be associated somehow with a virus‐induced permissive state where replication factors would be available. Consistent with this view, geminivirus dsDNA replication intermediates are up to ∼20‐fold more abundant in S‐phase nuclei of cultured cells (Accotto et al., 1993). Further support came from the observation that proliferating cell nuclear antigen (PCNA), an accessory protein of DNA polymerase δ, is undetectable in non‐proliferating cells but accumulates in TGMV Rep‐expressing, terminally differentiated cells (Nagar et al., 1995). Expression of other S‐phase gene markers, e.g. histone H2B, however, does not correlate with maize streak virus (MSV) distribution in infected tissues (Lucy et al., 1996).
In animal cells, the human tumour suppressor RB protein, together with other members of the so‐called ‘pocket family’ (p107 and p130), regulate the passage of cells through the G1 phase and the G1–S transit of the cell cycle by modulating the activity of the E2F–DP family of transcription factors (Weinberg, 1995; Helin, 1998; Brehm and Kouzarides, 1999). A sophisticated mechanism allows the sequential phosphorylation of RB by CDK–cyclin complexes leading to the release of RB‐bound E2F–DP factors needed to activate transcription of genes required for the G1–S transition and S‐phase progression (Mittnacht, 1998; Sherr and Roberts, 1999). Animal oncoviruses are able to bypass the normal RB control pathway in G1 by the action of one oncovirus‐encoded protein whose LxCxE amino acid motif mediates binding to RB (Ludlow, 1993; Moran 1993; Vousden, 1993).
The initial finding that a geminivirus protein, e.g. WDV RepA protein, contains an LxCxE motif and interacts with human pocket proteins both in yeast and in vitro (Xie et al., 1995) was a first clue as to which mechanism could be used by geminiviruses to induce a permissive cellular state. This observation together with the identification of plant D‐type cyclins (Dahl et al., 1995; Soni et al., 1995), which also contain an LxCxE motif, provided strong support for the notion that a plant RB‐related (RBR) pathway might exist in plants. This was confirmed by the cloning of cDNAs encoding a plant RBR protein which, as expected, is able to interact with WDV RepA (Grafi et al., 1996; Xie et al., 1996). Currently, plant RBR proteins have been identified in several plant species (Ach et al., 1997a; Fountain et al., 1999; Nakagami et al., 1999; D.Dudits, personal communication; W.Gruissem, personal communication), and their study will serve to define the role of the RBR pathway in plant cell cycle transitions, cell growth and development. The LxCxE motif is conserved in most mastrevirus RepA proteins (Xie et al., 1995), where it also mediates RepA–RBR interaction (Horvath et al., 1998; Liu et al., 1999b). Point mutations within the LxCxE motif of RepA greatly reduce or abolish binding to pocket proteins (Xie et al., 1996; Liu et al., 1999b).
The mastrevirus RepA and Rep proteins have identical primary sequence in their ∼200 N‐terminal residues (see Figure 1). The LxCxE motif is located in the region common to both proteins. In spite of this, mastrevirus Rep protein is unable to interact with RBR (Horvath et al., 1998; Liu et al., 1999b). It is likely that the unique C‐terminal moiety of Rep somehow hinders the LxCxE motif, as C‐terminal deletions of MSV Rep (Horvath et al., 1998) and WDV Rep (E.Ramirez‐Parra and C.Gutierrez, unpublished) are able to interact with RBR. Quite interestingly, however, in begomoviruses, e.g. TGMV, Rep, which lacks an LxCxE motif but is otherwise highly homologous in all geminiviruses, binds to RBR (Ach et al., 1997a). Residues required for TGMV Rep–RBR interaction have been mapped to a region in close proximity to the Rep oligomerization domain (L.Hanley‐Bowdoin, personal communication). It should be mentioned that other plant RBR‐binding proteins lacking an LxCxE motif, such as the MSI‐like proteins (Ach et al., 1997b) or the E2F transcription factor (Ramirez‐Parra et al., 1999; Sekine et al., 1999), have been cloned recently. The identification of amino acids crucial for RBR binding should contribute to our understanding of the mechanisms of geminivirus RepA–RBR and Rep–RBR interactions.
Therefore, it is conceivable that interaction with a plant RBR cell cycle regulatory pathway is a strategy common to all geminiviruses, although two distinct mechanisms have evolved: one, the mastrevirus‐type, which depends on the RepA–RBR interaction through an LxCxE motif and another, the begomovirus‐type, relying on an LxCxE‐independent Rep–RBR interaction (Figure 3). Consistent with this, studies in cultured cells have shown that: (i) WDV vectors carrying point mutations in the LxCxE motif exhibit a significant reduction in their ability to replicate (Xie et al., 1995); and (ii) expression of human or plant pocket proteins strongly impairs WDV DNA replication (Xie et al., 1996). Furthermore, in typical dicot‐infecting geminiviruses, e.g. TGMV, point mutations that reduce or impair binding to RBR alter symptom production and tissue specificity (L.Hanley‐Bowdoin, personal communication). However, recent findings indicate that the situation regarding the coupling of geminivirus replication to cell cycle control is likely to be much more complex than previously anticipated. Thus, in dicot‐infecting mastreviruses, e.g. bean yellow dwarf virus (BeYDV), point mutations in the LxCxE motif of the RepA protein reduce binding to plant RBR but the virus can replicate in plants and produce wild‐type symptoms (Liu et al., 1999b). One possibility which still has to be examined is that mastreviruses adapted to replicate in dicots have lost the functionality of the LxCxE motif, present in their RepA protein, and have gained a begomovirus‐type of interaction between Rep and RBR that is independent of an LxCxE motif.
The complexity of geminivirus–host cell interactions in relation to the RBR pathway could be comparable with that of animal oncoviruses for which a bypass of the RB pathway is well established (Ludlow, 1993; Moran, 1993; Vousden, 1993). In papovaviruses, for example, an intricate circuitry of interactions occurs between viral proteins and cell factors during the viral replication cycle, which makes some results difficult to reconcile. Thus, polyomavirus T‐ag mutants with reduced RB binding have a decreased capacity to replicate viral DNA in cultured cells, but this effect is only clearly visible after infection of growth‐arrested cells (Söderbärg et al., 1993) or under conditions of limited growth factor availability (Tevethia et al., 1997). Furthermore, although polyomavirus large T‐antigen with an intact RB‐binding motif is required for immortalization, an apparent lack of correlation exists between immortalizing ability and the absolute levels of binding to RB (Pilon et al., 1996). Clearly, further studies are still needed before we can fully understand in molecular terms the correlation between interference with the plant RBR pathway and geminivirus infection, not to mention the many differences that may derive from the as yet unknown plant‐specific regulatory functions of RBR.
Effects of geminivirus proteins on other cellular pathways
Several lines of evidence strongly support the notion that impinging on the RBR cell cycle pathway might be just one of several mechanisms that geminiviruses may use to modify the cellular physiology (Figure 4). It should be kept in mind that interference with several cell growth‐related regulatory pathways, other than the RBR pathway, has well‐characterized precedents among animal oncoviruses (Eick and Hermeking, 1996; Jansen‐Durr, 1996).
Recently, a group of proteins termed GRAB (for geminivirus RepA binding) has been identified by their ability to interact with WDV RepA protein (Xie et al., 1999). Three aspects are relevant to the present discussion. First, GRAB proteins share their ∼170 N‐terminal residues with an apparently plant‐specific family of proteins that contain the so‐called NAC domain (Aida et al., 1997). It is noteworthy that these proteins play roles as diverse as in pattern formation in embryos (Souer et al., 1996), flower development (Sablowski et al., 1998) and leaf senescence (John et al., 1997). Secondly, residues required for interaction with GRAB proteins are located within the C‐terminal domain of RepA, a region which has a significant degree of conservation in all mastrevirus RepA proteins (Xie et al., 1999). Thirdly, GRAB expression inhibits WDV DNA replication in cultured cells (Xie et al., 1999). The functional significance of RepA–GRAB interaction, and perhaps interaction with other NAC domain‐containing proteins, still needs to be addressed in whole plants, but it reveals a potentially novel pathway for geminivirus proteins to interfere with growth‐ and development‐related proteins (Figure 4).
While some geminiviruses are phloem restricted (Abouzid et al., 1988; Sanderfoot and Lazarowitz, 1996), where they could take advantage of the meristematic activity of pre‐cambial cells, others have been detected outside the vascular system, in differentiated cells (Rushing et al., 1987; Nagar et al., 1995; Lucy et al., 1996). In this context, one attractive question is whether, as a consequence of interference with cell cycle arrest controls, geminivirus proteins actually induce dedifferentiation and re‐entry in the cell cycle with a concomitant passage through S‐phase and subsequent cell division. TGMV‐infected cells do not seem to show a commitment to divide (Nagar et al., 1995). However, infection with curtoviruses (beat curly top virus; BCTV) frequently is associated with a hyperplastic response which leads to vein swelling and leaf curling (Latham et al., 1997), an effect dependent on the viral‐encoded C4 protein (Stanley and Latham, 1992; Latham et al., 1997). Cotton leaf curl disease also results in tissue reorganization, cell proliferation, dedifferentiation and abnormal growth in infected leaves. Recently, a novel circular ssDNA, associated with the disease and resembling a nanovirus‐like DNA component, has been isolated (Mansoor et al., 1999), although the significance of this component is still unknown. These observations again reveal significant differences among geminiviruses. Therefore, the identification of the cellular targets of viral proteins encoded by abnormal growth‐inducing viruses would be very helpful in understanding cell cycle regulation and its relationship to plant cell growth.
One possibility is that geminiviruses do not necessarily induce cell proliferation but, rather, a cellular state in which S‐phase functions are up‐regulated. This can be mediated through interference with the RBR pathway by allowing E2F‐dependent activation of S‐phase‐specific genes. However, this might not be the only mechanism, as transcription of some S‐phase‐specific genes is mediated by cis‐acting signals other than the E2F consensus sequence (Taoka et al., 1999). Another way of affecting cellular gene expression could be achieved directly by geminivirus proteins, e.g. Rep, TrAP or RepA, for which a transcriptional activity has been well documented (reviewed in Lazarowitz, 1992; Bisaro, 1996; Palmer and Rybicki, 1998; Hanley‐Bowdoin et al., 1999).
Interestingly, changes in the nuclear architecture seem to be an early consequence of TGMV infection, and both viral and chromosomal DNA incorporate bromodeoxyuridine (cited in Hanley‐Bowdoin et al., 1999). If a geminivirus‐dependent stimulation of semi‐conservative cellular DNA replication is demonstrated conclusively, it will be crucial to determine its consequences on the infected cell and, eventually, on the whole plant. Questions such as (i) whether an S‐phase‐like state is sensed as abnormal by the infected cell and is able to trigger a cell death‐like process, (ii) whether a cellular defence mechanism is activated, e.g. a mechanism mediated by post‐translational gene silencing (PTGS; Baulcombe, 1996; Brigneti et al., 1998) and (iii) whether any viral‐dependent countermeasure exists, are virtually unexplored areas. In this context, although far from being fully understood, are two recent observations. First, consistent with previous results that suggested a probable function of the African cassava mosaic virus (ACMV) TrAP protein as a suppressor of an RNA‐mediated defence mechanism (Hong et al., 1997), this protein is an efficient suppressor of PTGS in Nicotiana benthamiana (Voinnet et al., 1999). Secondly, TGMV TrAP transgenic plants exhibit an enhanced susceptibility to infection by DNA and RNA viruses (D.M.Bisaro, personal communication).
Studies on geminiviruses are important to the understanding of the molecular and cellular biology of geminivirus infection as a basis for a rational design of strategies for virus control. In addition, they offer us extraordinarily powerful tools with which to approach the study of basic processes in plants and the genes controlling them. Although certainly some mechanistic aspects are shared with other eukaryotes, these studies should help to delineate some of the unique properties of plant cell growth, differentiation and body architecture. Therefore, by taking advantage of the combined approaches that are under way, we should look forward to exciting developments in the near future.
Current developments in our laboratory would not have been possible without the contribution of past and present members, whose work is greatly acknowledged. I also thank David Bisaro, Margaret Boulton, Denes Dudits, Willi Gruissem, Linda Hanley‐Bowdoin, Ed Rybicki and John Stanley for communicating results prior to publication, and E.Martinez‐Salas for comments on the manuscript. I apologize to the colleagues whose publications have not been included in this article due to space limitations. Our research is partially supported by grants PB96‐0919 (Dirección General de Enseñanza Superior), 07B/0020/98 (Comunidad de Madrid) and ERBFMBI‐CT98‐3394 (European Union), and by an institutional grant from Fundación Ramón Areces.
- Copyright © 2000 European Molecular Biology Organization