Cassava mosaic geminiviruses: actual knowledge and perspectives

SUMMARY

Cassava mosaic disease (CMD) caused by cassava mosaic geminiviruses (CMGs) is one of the most devastating crop diseases and a major constraint for cassava cultivation. CMD has been reported only from the African continent and Indian subcontinent despite the large‐scale cultivation of cassava in Latin America and several South‐East Asian countries. Seven CMG species have been reported from Africa and two from the Indian subcontinent and, in addition, several strains have been recognized. Recombination and pseudo‐recombination between CMGs give rise not only to different strains, but also to members of novel virus species with increased virulence and a new source of biodiversity, causing severe disease epidemics. CMGs are known to trigger gene silencing in plants and, in order to counteract this natural host defence, geminiviruses have evolved suppressor proteins. Temperature and other environmental factors can affect silencing and suppression, and thus modulate the symptoms. In the case of mixed infections of two or more CMGs, there is a possibility for a synergistic interaction as a result of the presence of differential and combinatorial suppressor proteins. In this article, we provide the status of recent research findings with regard to the CMD complex, present the molecular biology knowledge of CMGs with reference to other geminiviruses, and highlight the mechanisms by which CMGs have exploited nature to their advantage.

CASSAVA AND CASSAVA MOSAIC DISEASE

Cassava (Manihot esculenta Crantz, Family Euphorbiaceae), which originates from Latin America, is a major source of food for more than 700 million people in tropical developing countries and is cultivated in a total global area of 18.6 million hectares with a total production of 238 million tonnes, half of which is contributed by the African continent (FAOSTAT, 2008). The status of cassava cultivation today is changing from subsistence farming to an industrialized system designed to process cassava into a diverse spectrum of products, including starch, sago grains, flour, chips, animal feed and, potentially, biofuel, all derived from a crop that has the ability to grow in poor soils (Thresh, 2006). Unfortunately, cassava is affected by several pests and diseases and is vulnerable to at least 20 different viral diseases as a result of its vegetative propagation. Cassava mosaic disease (CMD), caused by cassava mosaic geminiviruses (CMGs) (Family Geminiviridae: Genus Begomovirus), transmitted by the whitefly Bemisia tabaci, is one of the most important viral diseases and a major constraint for cassava production in Africa and the Indian subcontinent (Legg and Fauquet, 2004; Legg et al., 2006).

The earliest report of CMD was from Tanzania by Warburg (1894) and it was first proposed to be a viral disease by Zimmermann (1906). This virus was initially called cassava latent virus (Bock et al., 1981) until Stanley and Gay (1983) first published its sequence and Koch's postulate was satisfied (Bock and Woods, 1983), when it was renamed the African cassava mosaic virus (ACMV). Recent reports suggest that the CMD pandemic has affected at least nine countries in East and Central Africa, covering an area of 2.6 million square kilometres and causing an estimated annual economic loss of US$1.9–2.7 billion, and it has been termed the most damaging plant virus disease in the world, causing famine and the death of thousands of people (Legg et al., 2006). Although most of the molecular studies concerning CMGs have focused mainly on ACMV, it would be difficult and illogical to restrict this review only to this virus species, because it is now known that members of several different geminivirus species cause CMD and are replacing ACMV in nature. This article summarizes the current understanding, highlights recent research findings and discusses the new concepts concerning CMGs from both Africa and the Indian subcontinent.

TAXONOMY AND NOMENCLATURE

Recently, taxonomic guidelines have been developed to provide a framework on which to base the definitions of species and strains. Based on this new approach, in which the sequence identity demarcation between members of different species has been set at 89% for the DNA‐A component of begomoviruses, seven African and two Indian CMG species have been recognized (Fauquet et al., 2008; Table S1, see Supporting Information). The corresponding virus names are African cassava mosaic virus (ACMV), East African cassava mosaic virus (EACMV), East African cassava mosaic Cameroon virus (EACMCV), East African cassava mosaic Kenya virus (EACMKV), East African cassava mosaic Malawi virus (EACMMV), East African cassava mosaic Zanzibar virus (EACMZV) and South African cassava mosaic virus (SACMV), reported from Africa, and Indian cassava mosaic virus (ICMV) and Sri Lankan cassava mosaic virus (SLCMV), identified from the Indian subcontinent. In addition to these species, several strains have been recognized: East African cassava mosaic virus‐Uganda (EACMV‐UG), ‐Kenya (EACMV‐KE), ‐Tanzania (EACMV‐TZ); South African cassava mosaic virus‐South Africa (SACMV‐ZA), ‐Madagascar (SACMV‐MG); East African cassava mosaic Cameroon virus‐Cameroon (EACMCV‐CM), ‐Tanzania (EACMCV‐TZ); Indian cassava mosaic virus‐India (ICMV‐IN), ‐Kerala (ICMV‐Ker); Sri Lankan cassava mosaic virus‐India (SLCMV‐IN), ‐Sri Lanka (SLCMV‐LK) (Fig. 1; Table S1).

Figure 1.

Geographical distribution of cassava‐infecting geminiviruses. Representatives of all species and strains of cassava mosaic geminiviruses (CMGs) in Africa and the Indian subcontinent. The green colour indicates the area under cassava cultivation and each different coloured dot represents a unique species. The virus name abbreviations are given in the text.

GEOGRAPHICAL DISTRIBUTION AND EXPANSION

Although the cassava geminivirus pandemic continues to expand, CMD has been reported only from the African continent and Indian subcontinent despite the large‐scale cultivation of cassava in Latin America and many South‐East Asian countries (Fig. 1; Fargette et al., 2006). The absence of CMD in South America and several countries in South‐East Asia, despite the occurrence of other geminivirus diseases, has been mainly attributed to the inability of the polyphagous B. tabaci B biotype to colonize cassava effectively in this part of the world (Carabali et al., 2005).

Cassava mosaic geminiviruses in Africa

The earliest reports on the increasing prevalence and geographical expansion of the disease in both East and West African countries appeared in the 1920s and 1930s (Legg and Thresh, 2000). In the early 1990s, a new epidemic of severe CMD occurred in Uganda, which progressed steadily from the north‐east towards the south‐west, and the progress of the epidemic ‘front’ was mapped during regular monitoring surveys in 1992 (Legg and Thresh, 2000). The subsequent discovery of the association of a novel recombinant virus, the EACMV Uganda strain (EACMV‐UG), in severely affected plants from the epidemic area boosted research interest in CMD (Deng et al., 1997; Zhou et al., 1997). This was followed by an extensive series of surveys in different parts of the African continent during which the co‐occurrence of ACMV and EACMV was discovered in western Kenya and north‐western Tanzania (Ariyo et al., 2005). From these surveys, many virus isolates were sequenced and new species related to EACMV (EACMV‐like viruses) were identified from East and West African countries. The sequence analyses of both genomic components of these viruses revealed the presence of recombined fragments, a feature commonly found in begomoviruses (Padidam et al., 1999b).

The initial surveys conducted during the early period of the epidemic in central‐southern Uganda showed that ACMV occurred more frequently than EACMV‐UG, although there were some incidences of mixed infections (Harrison et al., 1997). However, subsequent surveys revealed more frequent occurrences of EACMV‐UG than ACMV and a significantly decreased proportion of mixed infections (Legg et al., 2006). Such a drastic change has been attributed to the selection of planting material by the farmers, in which they avoided severely diseased plants that were usually dually infected. It has also been hypothesized, but not proven, that EACMV‐UG DNA‐A, which contains a fragment of the ACMV coat protein (CP), might be more efficiently transmitted by whitefly vectors, thus leading to its widespread distribution in Africa (Pita et al., 2001b).

From the work of Ndunguru et al. (2005) it is evident that EACMVs may be indigenous to Tanzania, thus making East Africa the cradle of origin for many cassava viruses, and, in particular, the Eastern Arc and Coastal Forests of Tanzania/Kenya as the main hot spots for CMD begomovirus biodiversity in Africa. Thus, it is possible that the geminiviruses infecting the local hosts were spread throughout Africa for millions of years and later colonized cassava after its introduction in the 16th century and onwards. In addition to the transmission of CMD by whiteflies, human intervention can be another important cause for rapid geographical movement and exchange of geminiviruses. Recombination between members of different virus species has been a driving force for biodiversity, as exemplified by the EACMV/SACMV gradient between East Africa and South Africa, favoured by a natural corridor along the Eastern Rift Valley (Fig. S1, see Supporting Information; Ndunguru et al., 2005). Very recently, Bull et al. (2006) conducted an exhaustive survey of the genetic diversity of CMGs across the major cassava‐growing areas of Kenya, mainly representing EACMV and EACMZV, as well as a novel begomovirus named East African cassava mosaic Kenya virus (EACMKV). The DNA‐B components of these EACMV‐like viruses were much less diverse than their corresponding DNA‐A components, but clustered as western and eastern (coastal) groups.

Comparison of the CMG sequences from Africa indicates that all isolates of ACMV, irrespective of their geographical origin, are clustered together with little variation in their genomic sequence. In contrast, the genomes of EACMV‐like viruses are more genetically diverse as a result of the frequent recombination between members of different species (Fig. 3). In addition, there is a large range of phenotypic symptom variation for each of these virus isolates, irrespective of their location and their belonging to a particular species (Bull et al., 2007; Pita et al., 2001a). Thus, the explosion of the cassava mosaic pandemic in Africa is currently explained by a conjunction of several mechanisms: (i) the key synergistic interaction between two viruses (discussed in detail later in this article), leading to a huge increase in viral titre in the top of the plants, thus enhancing the whitefly transmission capacity; (ii) the CP recombinant fragment that may provide an advantage to EACMV‐UG, as it has been shown that the whitefly transmission epitopes are coded by this fragment and that there has been co‐adaptation, as far as the rate of transmission is concerned, between the local whiteflies and viruses (Maruthi et al., 2002); and (iii) the tremendous increase in whitefly populations adapted to cassava in Uganda, where the epidemic started.

Figure 3.

Recombination linearized map of representatives of the nine species of cassava‐infecting geminivirus and their strains. Each horizontal line represents the genotype of one virus isolate, and the different patterns and shades represent the tentative origins of the putative recombinant fragments. The length of the genome is indicated at the top of each diagram and the genome organization is depicted at the bottom; the abbreviated names of the virus isolates are listed on the left. The shades and patterns for the different original parental genome regions are provided below the genome organization. The numbers at the top of the diagram correspond to the length of the genome, starting at the nicking site of the conserved nona‐nucleotide TAATATT/AC. The virus name abbreviations are given in the text.

Cassava mosaic geminiviruses on the Indian subcontinent

In India, cassava is mainly grown in three southern states: Kerala, Tamil Nadu and Andhra Pradesh. In addition, it is sparsely cultivated in several other parts of India with a total production of 7.7 million tonnes (FAOSTAT, 2008). The first report of CMD in the Indian subcontinent was documented by Abraham (1956) and later elaborated by Alagianagalingam and Ramakrishnan (1966). A similar disease in Sri Lanka caused by a different begomovirus, Sri Lankan cassava mosaic virus (SLCMV), was reported by Austin (1986). The first clone of Indian cassava mosaic virus (ICMV‐Ker) was obtained by Hong et al. (1993) and, subsequently, Saunders et al. (2002) cloned SLCMV. The DNA‐A and DNA‐B components of ICMV share 65% and 30% nucleotide sequence identity with ACMV DNA‐A and DNA‐B, respectively. Although SLCMV has similar iteron sequences to those of ACMV, it shares a greater identity with ICMV.

Patil and Dasgupta (2005) and Patil et al. (2005) reported the presence of SLCMV in southern India through differential polymerase chain reaction (PCR) studies, and developed a distribution map for ICMV and SLCMV. Subsequently, several ICMV and SLCMV isolates from southern India were cloned and their biodiversity and infectivity were studied (Dutt et al., 2005; Patil et al., 2007; Rothenstein et al., 2006). Most of the CMGs cloned from India were isolates of SLCMV with lower diversity compared with the variability of CMGs from Africa (Rothenstein et al., 2006). Based on DNA‐A sequence data, all of these isolates were classified as isolates of ICMV or SLCMV, and their phylogenetic analysis revealed recombination events between ancestors of ICMV and SLCMV (Rothenstein et al., 2006).

THE WHITEFLY VECTOR AND CASSAVA MOSAIC DISEASE TRANSMISSION

The abundance of the whitefly vector (B. tabaci; Hemiptera: Aleyrodidae) is the second major distinguishing biological feature of the CMD pandemic in Africa, particularly at the epidemic ‘front’ between severely affected and relatively unaffected areas (Legg and Fauquet, 2004). CMD was first shown to be transmitted by Bemisia sp. in Africa and it is now considered to be a whitefly complex, with different populations occurring in different regions (Brown et al., 1995). The CP of a geminivirus is specifically adapted for transmission by the local whitefly population, which explains the antigenic similarity of the CPs of begomoviruses from the same area (Harrison and Robinson, 1999) and co‐adaptation between CMGs and their local whitefly populations (Maruthi et al., 2002). Mutation studies made in ACMV to disrupt the CP showed that ACMV was infectious in plants, either by mechanical or agro‐inoculation, but was not transmitted by whiteflies (Klinkenberg et al., 1989). In addition, it has been shown that efficient whitefly transmission of ACMV requires products of both genomic components; the results of Liu et al. (1997) indicated that virus accumulation and transmission to new plants were determined by DNA‐B (mostly BC1 coding for the movement protein) and virus acquisition was determined by CP; it was clear that the loss of whitefly transmissibility could be caused by the prolonged vegetative propagation of a virus isolate.

The whitefly B‐biotype is more fecund and has an extremely broad host range (Colvin et al., 2004), which might have contributed to the transmission of new viruses from weed hosts to cultivated crop plants, thereby leading to the emergence of a number of geminivirus diseases throughout the world. Since the onset of the CMD epidemic in Uganda, an increased whitefly population has been reported, mainly comprising two distinct clusters of B. tabaci (Ug1 and Ug2) (Legg et al., 2002; Sseruwagi et al., 2005). In India, two distinct biotypes of whitefly, cassava biotype and the sweet potato biotype, have been reported through sequencing of the mitochondrial cytochrome oxidase I gene (mtCOI) (Lisha et al., 2003). The inability of the B‐biotype to colonize cassava in the Americas has been postulated as a major reason for the absence of CMD in that region (Carabali et al., 2005). Although the problem of CMD has been partially resolved using resistant cassava cultivars, the abundant whitefly populations still continue to fuel the virus pandemic across Africa. It is not known why these whiteflies have adapted so well to cassava, but it is important that we improve our understanding of whitefly behaviour and adaptation, as they are essential elements in the emergence of new diseases.

PLANT HOST RANGE OF CASSAVA MOSAIC GEMINIVIRUSES

The host range of viruses is also considered to be an important criterion to differentiate between geminiviruses and to better understand their epidemiology. Each CMG has different preferences for different hosts in addition to their natural host, cassava. Previous studies have shown that the experimental host range of ACMV is largely restricted to the Solanaceae family, within which it is more readily transmitted to members of the genera Nicotiana and Datura (Bock and Woods, 1983). Both ICMV and SLCMV also infect other species of Nicotiana (for example, N. tabacum and N. glutinosa), and SLCMV, in particular, is highly virulent with a very broad host range extending to Arabidopsis, as reported recently (Mittal et al., 2008), and even to Ageratum conyzoides in association with ageratum yellow vein betasatellite (Saunders et al., 2002). The host range of EACMV has not been studied extensively, but it is not known to infect Nicotiana species, except for the commonly used experimental host N. benthamiana, and this is also the case with SACMV (Berrie et al., 2001). Thus, CMGs can be classified into two major groups based on their host range: ACMV, ICMV and SLCMV forming one class infecting all species of Nicotiana in addition to cassava, although with different levels of virulence, and EACMV‐like viruses and SACMV in another class, which infects only N. benthamiana and cassava. Recently, there has been a report of an association of ACMV and EACMV‐Ug with Manihot glaziovii, a wild species of cassava, which has been used as a source of resistance in CMD breeding programmes (Sserubombwe et al., 2008).

GENOME ORGANIZATION AND GENE REGULATION

Geminiviruses are a large and diverse group of plant viruses characterized by geminate shaped particles of 30 × 20 nm that replicate their circular single‐stranded DNA (ssDNA) genome (with a size of 2.7–2.8 kb) via double‐stranded DNA (dsDNA) intermediates, either by rolling circle replication or a recombination‐dependent replication mechanism in the nuclei of infected cells (Hanley‐Bowdoin et al., 1999; Jeske et al., 2001). The Family Geminiviridae is divided into four genera on the basis of genome organization, host range and insect transmission (Hull, 2002). CMGs belong to the genus Begomovirus which contains whitefly‐transmitted, dicotyledonous plant‐infecting geminiviruses (Fauquet et al., 2008). The DNA‐A encodes four genes in the complementary sense (AC1–AC4) and two in the virion sense (AV1 and AV2), responsible for replication, transcription enhancement and encapsidation functions, whereas DNA‐B encodes one gene in the virion and complementary senses (BV1 and BC1), respectively, required for intra‐ and intercellular movement (Fig. 2; Hull, 2002; Jeske, 2009).

Figure 2.

Genome organization of cassava mosaic geminiviruses (genus: Begomovirus). The genome is split into two components, termed DNA‐A (left) and DNA‐B (right). DNA‐A comprises six open reading frames (ORFs), and each ORF encodes a specific protein. AC1, replication‐associated protein (Rep); AC2, transcriptional activator protein (TrAP); AC3, replication enhancer protein (REn); AC4, RNA‐silencing suppressor; AV1, coat protein (CP); AV2, precoat protein. DNA‐B has two ORFs: BV1 encodes nuclear‐shuttle protein (NSP), and BC1 encodes movement protein (MP). C, Complementary‐sense ORFs; V, virion‐sense ORFs. The non‐coding intergenic region, also referred to as the common region (CR), with the stem‐loop structure is also depicted. The fragmented circles outside of the virus genome represent the DI‐DNAs of Indian cassava mosaic virus (ICMV) and Sri Lankan cassava mosaic virus (SLCMV) with their respective positions; dark and light grey lines represent DNA‐A and DNA‐B sequences, respectively. (Adapted from Patil et al., 2007.)

The geminate particles which encapsidate the ACMV genomic components are known to release their genomic DNA through breaks occurring at the top and shoulders of the virus particles (Bottcher et al., 2004; Kittelmann and Jeske, 2008). The begomovirus genomic components DNA‐A and DNA‐B share a common region of approximately 200 nucleotides having a high nucleotide sequence identity of more than 80% (Harrison and Robinson, 1999). The common region contains several regulatory elements, including two TATA motifs and also multiple copies of cis‐elements known as iterons, which are the binding sites for the replication‐associated protein (Rep) (Hanley‐Bowdoin et al., 1999). Different types of iteron sequences have been identified in CMGs and they can be classified into three groups: the ACMV type with isolates of ACMV, EACMZV and SLCMV; the EACMV type encompassing all the other EACMV‐like viruses and SACMV; and the ICMV type with ICMV isolates alone (Fig. 4).

Figure 4.

Sequence comparison of intergenic regions of cassava‐infecting begomoviruses depicting the iteron sequences and their relative positions. Comparison of DNA‐A components of representatives of all nine species and their strains, showing the location of iterons (highlighted with different shades for the different types of iteron) with reference to the TATA box and the stem‐loop with the nona‐nucleotide sequence embedded in it (boxed), and their orientation indicated by arrows, all located upstream of the stem‐loop structure. Spaces (‐) have been introduced to align the motifs, and each shade represents a unique iteron sequence. The virus name abbreviations are given in the text.

Transcriptional regulation in begomoviruses has been studied extensively in both transgenic plants and protoplast systems (Frey et al., 2001; Shivaprasad et al., 2005). The dsDNAs formed during rolling circle replication serve as transcriptional templates and the geminivirus genomes are transcribed in a bidirectional manner by the bidirectional promoters located in the intergenic regions (Hanley‐Bowdoin et al., 1999). The begomovirus open reading frame (ORF) AC2 is a viral transcription factor that trans‐activates the late viral genes AV1 and BV1 (Sunter and Bisaro, 1992), and recently it has been shown that AC2 of ACMV trans‐activates several plant genes (Lacatus and Sunter, 2008; Trinks et al., 2005). In ACMV and other Old World begomoviruses, the AV1 gene (CP) is overlapped at the 5′ end by another ORF (AV2). Characterization of AV2 of the CMGs indicated it to possess the redundant function of AV2 in monopartite begomoviruses functioning as a movement protein; presumably, it is an evolutionary relic of a monopartite geminivirus that may still increase virus fitness, but is no longer indispensable in a bipartite genome (Bull et al., 2007; Rothenstein et al., 2007). In the case of the ACMV genome, five putative promoter regions have been identified and tested for their transcriptional activity in both protoplasts and transgenic plants, and, in contrast with DNA‐A, the complementary‐sense DNA‐B promoter shows weak activity and the viral‐sense promoter appears to be regulated by host factors (Frey et al., 2001; Haley et al., 1992; Zhan et al., 1991). However, despite all of these studies, there are still major gaps in our understanding of the structural organization of the geminivirus promoter, gene regulation and transcription.

RECOMBINATION, PSEUDO‐RECOMBINATION AND TRANS‐REPLICATION: THE SOURCE OF MOLECULAR DIVERSITY

Recombination between geminiviruses was first revealed with the molecular analysis of EACMV‐UG, the virus responsible for the pandemic on cassava in Africa (Deng et al., 1997; Zhou et al., 1997). Soon after, it was established that geminiviruses and, in particular, begomoviruses are subject to very intense recombination rates, leading to a rapidly changing and large molecular diversity (Lefeuvre et al., 2009; Padidam et al., 1999b) to respond to changes in the environment and to invade new ecological niches. CMGs are no exception, and Fig. 3 depicts the recombination maps of DNA‐A and DNA‐B of CMGs. However, it is striking that ACMV seems to be an exception to this rule, as all the available ACMV sequences originating from several places in Africa do not show any evidence of recombination. This could result from the lack of ancestral parents of this virus, or could indicate that not all begomoviruses are subject to recombination, although ACMV donated 550 nucleotides to an EACMV isolate to create EACMV‐UG, associated with the CMD pandemic in Africa. By contrast, EACMV‐like viruses seem to be highly prone to recombination, as shown by the diverse number of strains encompassing stretches of DNA from various known origins (Fig. 4).

The complementary‐sense gene transcription initiation and termination sites and virion strand origins of replication on the circular DNA molecules at which the recombination repairs of double‐stranded genome breakages are resolved have been identified as the recombination hot spots in geminiviruses (Cromie et al., 2001; 2007, 2009; Varsani et al., 2008). The intergenic regions showed a greater tendency to be moved between the genomes than did the actual genes and, among the coding regions, the complementary‐sense genes showed relatively higher frequency of recombination relative to the virion‐sense genes. It has also been shown that there are very few recombination events occurring within the stretches of genome coding for the structural proteins, and most of the recombinations occur on the periphery of the ORFs (Lefeuvre et al., 2009).

Viruses can acquire additional DNA components through a mechanism called regulon grafting, whereby the transfer of intergenic regions harbouring replicational and transcriptional control elements to further DNAs occurs, and this has been reported in the case of ACMV (Jovel et al., 2004; Roberts and Stanley, 1994). It has been hypothesized that the DNA‐B component of SLCMV is actually a captured ICMV‐B, following component acquisition by a hypothetical monopartite strain of SLCMV (Saunders et al., 2002). This hypothesis illustrates how the compatibility of DNA‐A and DNA‐B can be obtained, enhanced and maintained, and indicates a way in which a begomovirus might acquire a new DNA‐B during mixed infection of two begomoviruses, even if they are too distantly related to produce a viable pseudo‐recombinant. Thus, this mechanism provides an explanation for the evolution of bipartite begomoviruses from monopartite ancestors (Rojas et al., 2005).

Pseudo‐recombination is a term used to describe the situation in which the DNA‐A and DNA‐B of the two components of a geminivirus originate from two different geminiviruses. In general, pseudo‐recombination occurs when there is re‐assortment of the genomic DNA‐A and DNA‐B of isolates of the same begomovirus species, but not those of distinct begomovirus species (Chakraborty et al., 2008). Pseudo‐recombination in which the DNA‐A of one virus coexists with and trans‐replicates the DNA‐B of another virus has been reported for CMGs occurring in Uganda. The most frequent naturally occurring CMG infections in Uganda involve the co‐occurrence of EACMV‐UG2 DNA‐A and EACMV‐UG3 DNA‐B (Pita et al., 2001a). EACMZV seems to be a recombination product of EACMV with an unknown begomovirus from which the 5′ terminal Rep coding sequence and part of the intergenic region have been captured, whereas EACMV and SACMV recombined to form members of a new species, EACMKV (Bull et al., 2006) (Fig. 4). EACMZV has ‘GGAGA’ as the iteron sequence and is incompatible with other EACMVs, but surprisingly ACMV and EACMZV did not form a viable pseudo‐recombinant despite having identical iteron sequences, indicating that there could be additional factors required for trans‐replication (Fig. 4). In contrast, both EACMV and EACMKV, sharing the same iteron sequences (GGGGG), can form viable pseudo‐recombinants (Bull et al., 2007).

The diversification of geminiviruses mainly occurs because of recombination and the exchange of genomic components in mixed infections (Padidam et al., 1999b). CMGs are known to undergo recombination in mixed infections (Fig. 4) and the survival of such recombined molecules depends on several factors, including their ability to replicate (Pita et al., 2001b). Trans‐replication is an important mechanism, wherein the DNA‐A component of a given geminivirus is capable of replicating a DNA‐B component of a heterologous virus, thereby offering an additional possibility for new interaction and providing a new source of CMG biodiversity. Trans‐replication and trans‐complementation have been experimentally demonstrated among CMGs that exhibit distinguishing characteristics; moreover, ACMV DNA‐B genes can functionally interact with TGMV genomic components (Saunders and Stanley, 1995; Stanley et al., 1985).

Recently, natural trans‐replications within different EACMV strains (EACMV‐UG2 DNA‐A and EACMV‐UG3 DNA‐B) or between ACMV DNA‐A and EACMV‐UG3 DNA‐B have been reported from Uganda with epidemiological implications (2001a, 2001b). Evidence for trans‐replication between EACMCV‐CM DNA‐A and ACMV‐CM DNA‐B and between EACMCV‐CM DNA‐A and EACMV‐UG3 DNA‐B has also been obtained in tobacco protoplasts (Pita et al., 2001a). It appears that a DNA‐A molecule can recruit a DNA‐B molecule provided that there is compatibility between their Rep and iterons, although such an incompatibility can be overcome by the exchange of intergenic regions (Roberts and Stanley, 1994; 2001, 2002). Both infectivity studies and leaf disc assays in N. benthamiana have demonstrated the following: ACMV DNA‐A trans‐replicated DNA‐B of SLCMV, but not that of ICMV; SLCMV DNA‐A trans‐replicated ACMV DNA‐B but not that of ICMV; and ICMV DNA‐A was unable to trans‐replicate either ACMV or SLCMV DNA‐B (Saunders et al., 2002). This highlights the importance of iteron sequences (Fig. 4) in forming a successful pseudo‐recombination complex, and thus in playing an important role in the evolution of geminiviruses through recombination and pseudo‐recombination events.

DEFECTIVE AND SATELLITE DNAs ASSOCIATED WITH CASSAVA MOSAIC DISEASE

In addition to genomic components, smaller sized DNAs, referred to as defective DNA (def‐DNA), often occur naturally in several geminivirus‐infected plants. Some def‐DNAs have been shown to interfere with virus proliferation, as they are associated with a delay in and an attenuation of symptoms; consequently, they are named ‘defective interfering DNAs (DI‐DNAs)’ (reviewed by Patil and Dasgupta, 2006). def‐DNAs are usually half the size of the full‐length genomic component and may be formed by sequence deletion, duplication, inversion or rearrangement of viral DNA and, in some cases, insertion of foreign sequences (Fig. 2; Stanley et al., 1997). def‐DNAs always contain the intergenic region, harbouring all cis‐acting sequences required for replication, and, for bipartite viruses, def‐DNAs are predominantly derived from the DNA‐B component (Fig. 2). In most cases, the BV1 ORF and the C‐terminus of BC1 are deleted (Stanley and Townsend, 1986). def‐DNAs are mostly formed at higher frequency in laboratory host plants, such as N. benthamiana, and rarely in their natural hosts, indicating that they are not well adapted to the experimental hosts.

Stanley and Townsend (1986) first characterized the def‐DNA forms associated with ACMV DNA‐B as being usually encapsidated in isometric particles of 18 × 20 nm in size, whereas genomic DNA was encapsidated in geminate particles (Frischmuth et al., 2001). Recently, DI‐DNA derived from the DNA‐A component of EACMV and SLCMV has been reported, albeit at a lower frequency (Ndunguru et al., 2006; Patil et al., 2007). Patil et al. (2007) also reported defective molecules generated from intercomponent recombination between DNA‐A and DNA‐B components of CMGs from India, in which the sites of deletions or recombinations were mapped to the potential recognition sites of replication‐associated enzymes topoisomerase I and II. Symptom amelioration by DI‐DNA is one of its most fascinating aspects, which could be mainly a result of the slowing down of the replication of the cognate helper virus caused by the sequestration of limited amounts of Rep proteins by the DI‐DNA (Patil and Dasgupta, 2006).

In addition to the presence of def‐DNA molecules as a subgenomic component, there are satellite DNA molecules, termed alphasatellites and betasatellites, associated with the begomovirus disease complex, which mostly depend on the proteins encoded by the helper virus for their replication, movement and encapsidation (Briddon et al., 2008). Hitherto, most of the ssDNA satellites reported have been associated with monopartite begomoviruses, but the recent discovery of resistance‐breaking satellite molecules, termed ‘satDNA’, associated with CMD in Tanzania has posed serious threats to cassava cultivation (J. Ndunguru and C. M. Fauquet, personal communication). Experiments on the trans‐replication of begomovirus betasatellites and alphasatellites by cassava‐infecting geminiviruses and their further effect on symptom modulation have demonstrated that each species of CMG shows a differential interaction with the DNA satellites (B. L. Patil and C. M. Fauquet, personal observation).

VIRULENCE OF CASSAVA MOSAIC GEMINIVIRUSES

The aetiology of CMD has increased in complexity during the past decade, particularly on the African continent, reflected by the increasing number of novel species of CMGs from different locations. The presence of EACMV‐based recombinants, such as EACMV‐UG, has been correlated with the severe CMD epidemic in Uganda (Deng et al., 1997; Zhou et al., 1997), although, as of today, there is no proof that this recombination was responsible for the CMD pandemic.

The response of the host to virus infection is characterized by an initial onset of systemic symptoms from which the plant may or may not recover. The detailed analysis of representatives of four distinct CMG species by Chellappan et al. (2004b) has allowed us to distinguish two distinct infection types in cassava: a rapid onset from which plants recover (ACMV and SLCMV), and a slow onset from which plants do not recover (EACMV and EACMCV). In addition to gene silencing, the DNA‐B component can have a profound effect on symptom modulation (Bull et al., 2007). With the recent identification of two novel CMGs, i.e. EACMKV and EACMZV, and their isolates, it has been shown that symptom severity or symptom recovery could be a manifestation of the type of DNA‐B component involved in the infection, which has been revealed to modulate the symptom phenotypes of these viruses (von Arnim and Stanley, 1992; Bull et al., 2007). Thus, such symptom variability in infected cassava plants, which ranges from mild to extremely severe, has been attributed to variation in virus strains or their genes, virulence, host susceptibility or modification of vector activity (Fig. 5). Host susceptibility is variable, with some germplasm, such as the cassava landrace TME3, considered to be extremely resistant to CMGs, controlled by a major dominant resistance gene locus CMD‐2 (Akano et al., 2002). Although this landrace is highly resistant, it is not immune to infection from several CMGs, including some isolates of EACMV and EACMZV, which initially give significant symptoms in TME3; later, the cassava plants recover completely from the infection, possibly indicating that the CMD resistance in TME3 could be through the gene silencing mechanism (B. L. Patil and C. M. Fauquet, personal observation).

Figure 5.

Differential virulence of cassava mosaic geminivirus (CMG) isolates. Symptoms of isolates of African cassava mosaic virus (ACMV) (top panels) and East African cassava mosaic virus—Uganda (EACMV‐UG) (bottom panels) reported from Uganda in 1998. Mild isolates and severe isolates are shown on the left and right, respectively.

GENE SILENCING, SUPPRESSION AND SYNERGISM IN CASSAVA MOSAIC GEMINIVIRUSES

One of the most important biological functions for RNA silencing, first established in plants, is an adaptive defence system against viruses (Ding and Voinnet, 2007; Waterhouse et al., 2001). Recently, it has been shown that all the CMGs investigated trigger post‐transcriptional gene silencing (PTGS) in infected plants with the production of virus‐specific short interfering RNAs (siRNAs) (Chellappan et al., 2004b). This was attributed to the formation of dsRNA caused by the overlap at the 3′ ends of transcripts produced from opposite polarity through bidirectional transcription, and also possibly by dsRNA formation in the secondary structure of the virus transcripts. The symptom recovery phenotype exhibited by ACMV and SLCMV in cassava has been correlated with the production of virus‐derived siRNAs through PTGS (Chellappan et al., 2004b). Detailed analysis of the composition of these siRNAs has revealed that most of the siRNAs in ACMV‐infected plants are 21 nucleotides in size; the majority are derived from the DNA‐A component, particularly from a region that corresponds to the C‐terminus of AC1 overlapping with the N‐terminus of the AC2 gene, and those derived from the DNA‐B component are mostly from the C‐terminus of BC1 (Chellappan et al., 2004a). Virus‐derived siRNAs implicating virus‐induced PTGS have also been demonstrated for EACMCV, EACMV and ICMV, although the infected plants do not recover. These observations indicate that, although the plant responds to virus infection through the induction of gene silencing, the plant might not always recover from virus infection. The interaction of host gene silencing machinery with other viral factors or suppressors might counteract this effect, and thus might not lead to the recovery of plants from the infection.

In addition, virus–host interactions are strongly modified by environmental factors, particularly temperature and light conditions (Chellappan et al., 2005b). In CMGs, irrespective of the recovery‐ and non‐recovery‐type viruses, there was a general trend for symptom recovery with an increase in temperature, suggesting increased PTGS or transcriptional gene silencing (TGS) at high temperatures (Chellappan et al., 2004b). Recently, Akbergenov et al. (2006) have characterized siRNAs of three different size classes (21, 22 and 24 nucleotides), which were derived from both the coding and intergenic regions of ACMV.

To circumvent the constraints imposed by gene silencing triggered in the host plant, geminiviruses have evolved suppressor proteins that counteract this host defence response. Voinnet et al. (1999) first showed that AC2 of ACMV, when transiently expressed, could suppress green fluorescent protein (GFP) silencing. Transient expression of ACMV AC2 also induced the expression of about 30 genes in Arabidopsis, such as Werner exonuclease‐like 1, which was capable of suppressing RNA silencing, thus suggesting that AC2 suppresses silencing indirectly by activating the expression of cellular proteins (Trinks et al., 2005). The AC2 protein could suppress silencing by two independent mechanisms, one of which targets cytoplasmic RNA silencing and the other siRNA‐directed methylation (Bisaro, 2006). In this regard, it is interesting to note that geminivirus AC2 also interacts with adenosine kinase (ADK), an enzyme involved in sustaining methylation through methyltransferase activity, although it has not yet been investigated whether ADK is related to PTGS (Wang et al., 2003). In addition to its interaction with ADK, AC2 also suppresses sucrose non‐fermenting 1 (SNF1) kinase, which plays a central role in the regulation of metabolism, indicating that SNF1‐mediated responses constitute a novel defence pathway in plants (Hao et al., 2003; Wang et al., 2005). Thus, further research might unravel the intricate network between these diverse defence pathways.

The AC4 gene is entirely embedded within AC1 and, despite the high level of conservation of AC1, AC4 is one of the least conserved genes of all the geminiviruses. Functional analysis of AC4 has proved to be elusive, as the mutagenesis and/or transgenic expression of some AC4 genes results in no phenotype, whereas others produce phenotypes consistent with a movement protein or a symptom determinant (Krake et al., 1998; Vanitharani et al., 2004). AC4 has been found to be capable of suppressing gene silencing for some CMGs, enhancing disease symptoms. Transient studies on the suppression activity of AC4 from four different CMGs in N. benthamiana 16c plants showed that two AC4 proteins, from viruses giving recovery phenotypes in cassava (ACMV and SLCMV), showed suppressor activity with inhibition of GFP‐specific siRNAs and increased accumulation of GFP mRNA, whereas two other AC4 proteins from non‐recovery‐type viruses (EACMCV and ICMV) did not show similar silencing suppression activity. This may reflect considerable functional diversity within AC4 sequences of distinct begomoviruses. Transgenic expression of AC4 of ACMV, but not EACMCV, leads to developmental defects in Arabidopsis that resemble virus disease symptoms, which has been explained by its interference in the microRNA pathway (Chellappan et al., 2005a). Conversely, in the case of non‐recovery‐type viruses, the AC2 proteins were effective silencing suppressors, whereas those from recovery‐type viruses were less effective (Vanitharani et al., 2004).

Mixed infections of ACMV and EACMV have been shown to be an important feature of the severe CMD first reported from Uganda and subsequently in neighbouring countries (Harrison et al., 1997; Pita et al., 2001a). A synergistic effect in a mixed infection occurs when there is a significant difference in AC2 and AC4 functions of these individual viruses. The difference in the phenotypes exhibited by these viruses further suggests that AC2 and AC4 may act at different steps in the silencing pathway, and that the effect of AC4 could be more transient, allowing some hosts to overcome it. These experiments revealed that the synergism between ACMV and EACMCV was mediated by the complementation of two different PTGS suppressors, AC4 and AC2 present in ACMV and EACMCV, respectively, acting in a temporal and spatial manner (Fig. 6; Fondong et al., 2000; Vanitharani et al., 2004). Therefore, having more than one type of PTGS suppressor provides an advantage to viruses, synergistically interacting in mixed infections, and thus helps in the establishment of a severe/successful disease.

Figure 6.

Symptom severity of cassava plants. Dual infection leads to synergistic severe disease and corresponding post‐transcriptional gene silencing (PTGS) capacity of each virus. Second panel (from left to right): non‐infected healthy control plant; plant infected with African cassava mosaic virus (ACMV); plant infected with East African cassava mosaic Cameroon virus (EACMCV); dual infection with both viruses, which leads to synergistic severe disease (candle‐stick symptoms). The third panel indicates the function of AC2‐ and the bottom panel indicates the function of AC4‐type PTGS suppressor proteins. EACMCV‐AC2 (EAC2; third panel, second red image from right) and ACMV‐AC4 (AAC4; bottom panel, third red image from right) are positive for PTGS suppression. Synergism is explained by dual suppression of PTGS (red panels on the right). Green control panels (left) represent frame‐shift mutants of ACMV‐AC4 and EACMCV‐AC2 that have lost the capacity to suppress PTGS. ACMV‐AC2 is considered to be negative, as PTGS suppression is weak, and EACMCV‐AC4 is not a PTGS suppressor. (Adapted from Vanitharani et al., 2005.)

STRATEGIES FOR CONTROLLING CASSAVA MOSAIC DISEASE

Several approaches have been used to control CMD, including improved cultural practices, breeding of natural resistant varieties and improved biotechnological strategies, which have been reviewed extensively by Thresh and Cooter (2005), Thresh (2006), Sudarshana et al. (2007) and Vanderschuren et al. (2007b), and thus they are described only briefly here. Of all the methods used for the control of CMD resistance, genetic engineering is a rapid method of transferring resistance genes to traditional cultivars, bypassing the possibility of the appearance of undesired traits. AC1 is an excellent target for interference by the expression of mutant proteins because of its crucial functions in the replication cycle and its multiple interactions with both the viral and host proteins, and targeting it may also provide broader resistance against different geminiviruses (Brunetti et al., 2001).

To engineer a virus‐inducible hypersensitive reaction, attempts were made to express apoptotic non‐viral proteins controlled by the ACMV promoter trans‐activated by the AC2 protein on infection (Zhang et al., 2003); however, the use of such proteins for practical purposes has biosafety concerns. Symptom amelioration by DI‐DNA was tested as a possible strategy to control ACMV in both the model host N. benthamiana and in cassava, but did not lead to significant levels of resistance in cassava (Frischmuth and Stanley, 1993; Taylor et al., 2003).

An alternative approach is antisense RNA technology, also known as RNAi/gene silencing technology, in which resistance to ACMV infection of cassava has been achieved with high efficacy by expressing anti‐sense RNAs, or hairpin RNAs, against viral transcripts encoding Rep, TrAP and REn (Chellappan et al., 2004a; Vanderschuren et al., 2007a; Zhang et al., 2005). However, the major concern in using gene silencing strategies to control viruses is the fact that many viruses encode one or more PTGS suppressors that could suppress the enhanced resistance mechanism (Vanitharani et al., 2005) and, above all, that point mutations or recombinations in the target viruses could ‘break’ the engineered resistance; therefore, different strategies, such as the G5 strategy, are now being developed to counter this issue.

The non‐specific DNA‐binding activities of some proteins, such as the G5 protein from isolates of Enterobacteriophage M13 which binds to ssDNA, competing with CP and affecting geminivirus movement, are being used to engineer broad‐spectrum resistance to cassava‐infecting geminiviruses and their satellites (Fig. S2, see Supporting Information; Padidam et al., 1999a; J.S. Yadav, N. Taylor and C.M. Fauquet, personal communication). More recently, peptide aptamers, a class of recombinant proteins that bind to and inactivate the target protein, and artificial zinc finger proteins, which block the Rep protein, are being tested to develop geminivirus‐resistant transgenics (Takenaka et al., 2007; L.A. Lopez‐Ochoa and L. Hanley‐Bowdoin, personal communication).

Although further improvements in the group of transgene‐derived CMD resistance strategies are required before field deployment, these approaches offer great promise. Most importantly, they can provide a valuable complement to conventional breeding approaches to control CMD, and could effectively allow farmers to continue to grow locally preferred cultivars. In view of the difficulties that have been experienced in some pandemic‐affected areas in addressing farmer quality preferences, this represents a very important development. Currently, two major research institutes, namely the Donald Danforth Plant Science Center (St. Louis, MO, USA) and the Swiss Federal Institute of Technology (Zurich, Switzerland), in collaboration with several African countries, India and other laboratories across the globe, are using different biotechnological approaches to find a remedy for CMD, together with improving other quality traits of cassava. The two Consultative Group on International Agricultural Research (CGIAR) institutes, CIAT (International Centre for Tropical Agriculture, Cali, Colombia) and IITA (International Institute of Tropical Agriculture, Ibadan, Nigeria), are using molecular markers to tag genes for CMD resistance from multiple sources of resistance as a means of pyramiding different genes for durable resistance (M. Fregene, personal communication).

CONCLUSION

In the last two decades, significant progress has been made in our knowledge of CMGs in terms of their structure, diversity, movement and interaction with their host. This has been supplemented by a wealth of information generated by genomic studies and the description of hundreds of virus sequences. Despite this progress, it has not been possible to understand the presence of this vast diversity in the case of EACMV‐like viruses and ICMV in comparison with ACMV and SLCMV. The diversification and rapid spread of EACMV‐like viruses, their high rate of recombination combined with synergistic interaction with other CMGs and the population explosion of their whitefly vector are major threats to cassava cultivation in Africa and the Indian subcontinent. In addition, there is a huge threat of CMD introduction in South America and South‐East Asia. Thus, strict quarantine regulations need to be exercised to prevent the import of CMGs and cassava‐adapted whitefly biotypes.

More recently, dual pandemics have been reported of two major viral diseases infecting cassava, ‘cassava mosaic disease’ and ‘cassava brown streak disease’, caused by an RNA virus (putative member of the genus Ipomovirus of the Family Potyviridae), in the coastal regions of eastern Africa. The rapid spread of these diseases into eastern and central regions of Africa is alarming the cassava community (Alicai et al., 2007).

Undoubtedly, the discovery of the gene silencing mechanism will have a profound impact on the understanding of plant–virus interactions and also on the control of plant viruses. Further exploitation of this wealth of knowledge is essential to improve the efficiency and durability of all existing strategies, each approach with its own limitations, as well as for developing potential new strategies to cope with the emergence of resistance‐breaking, cassava‐infecting geminiviruses and satellites. These novel approaches for the management of CMD appear to have great potential in the near future, a prospect that certainly justifies further investment in this technology to herald another green revolution.

Supporting information

Fig. S1 Linear recombination pattern along the Rift Valley in East Africa, from Uganda (left) to South Africa (right), for the cassava mosaic geminivirus (CMG) DNA‐A components, showing a gradient between East African cassava mosaic virus (EACMV)‐type (purple colour) and South African cassava mosaic virus (SACMV)‐type (green colour) sequences. The virus name abbreviations are given in the text.

Fig. S2 Example of cassava mosaic geminivirus (CMG)‐resistant plants produced by genetic engineering. (A) Control 60444 cassava, not infected. (B) Control 60444 cassava, infected by East African cassava mosaic virus—Uganda (EACMV‐UG). (C) Control 60444 cassava, infected by African cassava mosaic virus (ACMV). (D) Transgenic cassava plant resistant to ACMV via the gene silencing strategy. (E) Transgenic cassava plant resistant to EACMV‐UG via the gene silencing strategy. (F) Transgenic cassava plant resistant to ACMV via the G5 strategy.

Table S1 List of full‐length clones of cassava mosaic geminiviruses (CMGs) isolated from Africa and the Indian subcontinent. The species name, isolate name and their abbreviations, and the accession numbers for DNA‐A and DNA‐B, are provided.

Please note: Wiley‐Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Supporting info item

Supporting info item

Supporting info item