INTRODUCTION

Potato (Solanum tuberosum L.) is an important crop grown over ~23,500 ha, and with an estimated production of 385,000 tons in Tunisia (FAOSTAT Database Results, 2014). It is grown as three distinct, though partly overlapping, season crops, where the main and late seasons are the most important with regard to acreage and production but differ in that the late season seeds are mostly self-produced, thus with substantial levels of virus (Khamassy, 1999). Spunta is the prevalent cultivar (75% of the planted acreage) compared to the others cultivated in Tunisia, such as Nicola, Liseta, Atlas, Safrane, and Pamina (Khamassy, 1999). Viral diseases are the major cause of poor seed quality and low yield. Potato virus Y (PVY) is reported to be one of the most detrimental potato viruses responsible for the major yield losses worldwide (Gray et al., 2010) and in Tunisia (Boukhris-Bouhachem et al., 2010; Djilani-Khouadja et al., 2010). The type-species is of the genus Potyvirus in the family Potyviridae, and this virus is a single-strand, positive-sense genomic RNA of ~9.7 kb. Within a suitable host, the PVY RNA genome is translated into a large precursor polyprotein, which is cleaved into 10 mature proteins (Berger et al., 2005). PVY has a large diversity of strain groups and genetic variants within strains, linked to its biological, serological and molecular properties (Singh et al., 2008; Karasev et al., 2011). According to the reaction of potato cultivars carrying different resistance genes and Nicotiana tabacum, potato isolates of PVY are classified into five strain groups: PVYO, PVYC, PVYN, PVYZ, and PVYE (Galvino-Costa et al., 2012). According to their sequences, isolates of the PVYN strain group fall into two genetic subgroups, which are the European (PVYN) and North American (NA-PVYN) (Chikh Ali et al., 2010). Among these strain groups and subgroups, PVYO and PVYN are the most frequent strains in potato, whereas PVYC is not common in potato fields and has less economic importance (Kerlan et al., 2011). Genomic recombination plays a significant role in PVY evolution and has led to the emergence of new genotypes/strains including the recombinant PVYNTN, PVYNW and PVYNTN-NW (Chikh Ali et al., 2010). PVYNTN variant has become common in potato fields in most potato production areas (Glais et al., 2005; Kerlan et al., 2011) causing potato tuber necrotic ringspot disease (PTNRD) and was reported to be able to overcome the field resistance of potato cultivars (Van den Heuvel et al., 1994). The destructive effects of this threatening disease are a particular problem in consumption and seed potato production, as it makes tubers of susceptible cultivars unmarketable as fresh potato and since infected seed lots submitted for certification are rejected (Beczner et al., 1984). This disease is now well-known in Tunisia (Boukhris-Bouhachem et al., 2010) and in most countries cultivating potatoes (Kerlan et al., 2011). Several DNA-based molecular tools have been developed as an alternative to serological assays to distinguish between PVYN-W and PVYO isolates (serotype O) and between PVYNTN and PVYN isolates (serotype N) (Singh et al., 2008). Immunocapture RT-PCR was used in this study for specific detection of PVYNTN isolates on the basis of the nucleotide sequence polymorphism observed in the 5'NTR/P1 region (Glais et al., 2005). Moreover, studies of the genetic structure of populations of plant viruses are important for understanding the evolution of virus/host interactions and the geographic distribution of viruses (Gibbs and Ohshima, 2010).

In this context, surveys of potato plants were undertaken in this study, and PVY strain and variant diversity was identified for different growing areas in Tunisia. Consequently, the 5'NTR-P1 nucleotide sequences of Tunisian PVY isolates (PVY-TN) were obtained and genetic structure of PVY isolates was determined. The pairwise mismatch distribution was combined to the network shape, and the statistical neutrality test was performed to look for a possible population expansion. The phylogenetic relationships of PVY-TN isolates with other worldwide PVY isolates from different countries were discussed to determine whether they represent a novel type or share common origins with strain variants reported elsewhere. Lastly, the correlation between these partial nucleotide sequences and the collection site of PVY isolates was considered.

MATERIAL AND METHODS Surveys

Surveys were carried out in late season crops (autumn 2007) in several fields of nine regions in the Northern and Central Tunisia belonging to different bioclimatic zones (Emberger, 1966): Bizerte (humid), Cap Bon (subhumid), Manouba (subhumid), Jendouba (subhumid), Kairouan (very arid), Monastir (subsemiarid), Mahdia (subsemiarid), Sidi Bouzid (arid), and Gafsa (arid) (Figure 1). Regions were 60 to 230 km apart and the distance between sites within regions varied from 1 to 5 km (Figure 1). A total of 385 leaf samples were collected from six potato cultivars (Spunta, Nicola, Atlas, Pamina, Gredine, and Belle de Fontenay) showing symptoms associated with virus infection (mosaic, dwarfing, mottle, yellowing, necrosis, and/or rugosity). The presence of aphids and weeds was especially noted in Bizerte and Monastir.

Figure 1

Geographic origin of the 385 collected leaf potato samples. Designation in brackets indicates numbers of collected samples in each region.

Standard viral isolates

For biological, serological and molecular characterization, 5 standard isolates maintained on N. tabacum cv. Xanthi and representing the main strains and variants of PVY, were used in each experiment: PVYN-605 (accession No. X97895), PVYNTNH (accession No. M95491), PVYNTNNZ (accession No. AM268435), PVYN-W-iP (accession No. AJ889868), and PVYO-139 (accession No. UO9509).

For phylogenetic analysis, nine nucleotide PVY sequences retrieved from the GenBank database (11627-12, NTNTK1, HN1, L26, Ditta, Satina, GBVC, NIB-NTN, and NTNH) were selected and included in this study (Table 1).

Table 1Origin and characteristics of PVY sequence isolates. Full name of isolate Origin Country Strain or varianta Accession No. Corresponding haplotype 11627-12 S. tub United Kingdom NTN KC634007 Hap 18 NTNTK1 S. tub Japan NTN AB711146 Hap 18 HN1 S. tub China NTN HQ631374 Hap 19 L26 S. tub USA NTN FJ204165 Hap 20 Ditta S. tub Poland NTN AJ890344 Hap 21 Satina S. tub Germany NTN AJ890347 Hap 22 GBVC S. tub Belgium NTN JQ969035 Hap 23 NIB-NTN S. tub Slovenia NTN AJ585342 Hap 24 NTNH S. tub Hungary NTN M95491 Hap 6 Tunisian isolates Bioclimate BS2 S. tub (Pamina) Tunisia (Jendouba) Sub-humid NTN KJ713153 Hap 1 BS7 S. tub (Spunta) Tunisia (Jendouba) Sub-humid NTN KJ713154 Hap 2 BS10 S. tub (Spunta) Tunisia (Jendouba) Sub-humid NTN KJ713155 Hap3 BS13 S. tub (Spunta) Tunisia (Jendouba) Sub-humid NTN KJ713156 Hap 3 BS14 S. tub (Nicola) Tunisia (Jendouba) Sub-humid NTN KJ713157 Hap 4 BS17 S. tub (Nicola) Tunisia (Jendouba) Sub-humid NTN KJ713158 Hap 5 KA1 S. tub (Nicola) Tunisia (Kairouan) Upper arid NTN KJ713159 Hap 17 KA5 S. tub (Nicola) Tunisia (Kairouan) Upper arid NTN KJ713160 Hap 16 KA8 S. tub (Atlas) Tunisia (Kairouan) Upper arid NTN KJ713161 Hap 15 KA10 S. tub (Spunta) Tunisia (Kairouan) Upper arid NTN KJ713162 Hap 14 KA11 S. tub (Spunta) Tunisia (Kairouan) Upper arid NTN KJ713163 Hap 13 CH6 S. tub (Spunta) Tunisia (Mahdia) Under semi-arid NTN KJ713164 Hap 12 MJ2 S. tub (Spunta) Tunisia (Manouba) Sub-humid NTN KJ713165 Hap 11 MJ11 S. tub (Spunta) Tunisia (Manouba) Sub-humid NTN KJ713166 Hap 10 MJ13 S. tub (Spunta) Tunisia (Manouba) Sub-humid NTN KJ713167 Hap 9 MJ15 S. tub (Nicola) Tunisia (Manouba) Sub-humid NTN KJ713168 Hap 8 RJ4 S. tub (Spunta) Bizerte (Ras Jebel) Humid NTN KJ713169 Hap 7

aGrouped according to the molecular structure, NTN: PVYNTN (European or recombinant type); designation in brackets in column 2 indicates the potato cultivar. Designation in brackets in column 3 indicates the city S. tub: Solanum tuberosum.

PVY diagnosis

Detection of PVY infection in potato samples was conducted by double antibody sandwich-ELISA (DAS-ELISA) (Clark and Adams, 1977) using anti-PVY polyclonal antisera (Bioreba, Basel, Switzerland) with microtiter plates (Nunc-immuno plate MaxiSorp surface, Polylabo) Absorbance of alkaline phosphatase reaction end-product was read at 405 nm using a microplate reader (Molecular Devices, Palo Alto, CA, USA).

Biotyping and serotyping analyses

All PVY isolates were further analyzed for strain infection by DAS-ELISA using anti-PVYN and anti-PVYO-C (Bioreba, Basel, Switzerland) specific monoclonal antibodies.

PVY infection of some potato samples was confirmed by mechanical inoculation onto N. tabacum cv. Xanthi plantlets in an insect-proof greenhouse with controlled conditions (20°-23°C). Symptoms in tobacco plantlets were observed daily and recorded at 18-20 days post-inoculation. PVYN strain is known to induce veinal necrosis with smaller distorted leaves and dwarfing of the whole plant in tobacco cv. Xanthi; whereas PVYO and PVYC induce only vein banding or interveinal clearing without any necrosis, crinkling or dwarfing (Kerlan et al., 2011). Tobacco samples were analyzed by DAS-ELISA using the above specific monoclonal antibodies.

Molecular analyses Immunocapture-RT-PCR amplification and sequencing

PVY amplification was performed by immunocapture RT-PCR as modified by Nolasco et al. (1993) and as previously described by Glais et al. (2005) and Djilani-Khouadja et al. (2010). The specific detection of PVYNTN variants was performed using the FR2000/F2-d primer pair based on the nucleotide sequence polymorphism observed in the 5'NTR/P1 region (Glais et al., 2005). The expected amplification fragment from this primer combination is 815 bp in length. Amplified products were separated by electrophoresis on 1.5% agarose gels, stained with ethidium bromide and photographed under UV light.

Sequencing reactions were carried out directly with purified PCR products (Qiaquick PCR purification kit, Qiagen France) with the same FR2000/F2-d primer pairs and analyzed in both strands on an ABI Prism 310 Automated DNA-377 sequencer (Applied Biosystems, Paris, France). Sequencing was performed for seventeen Tunisian isolates belonging to five geographical regions: Ras Jebel, Manouba (extreme north), Jendouba (north), Mahdia (Sahel), and Kairouan (central), and collected from four potato cultivars: Pamina, Spunta, Nicola, and Atlas (Table 1).

Sequence analysis

The whole dataset was first aligned using MEGA5 version (Tamura et al., 2011) and then corrected manually. The identity of PVY sequences were verified by BLAST. Sequences of all the Tunisian PVY isolates were submitted to NCBI GenBank and the relative accession Nos. were obtained (Table 1). The sequences were aligned using the ClustalW package in the MEGA5 program (Tamura et al., 2011). Aligned sequences were analyzed with the DnaSP software version 5.10.01 (Librado and Rozas, 2009) to estimate polymorphism indices and past demographic history. Indices of haplotype diversity (Hd) (Nei and Tajima, 1983), pairwise estimates of nucleotide diversity (Pi) (Jukes and Cantor, 1996), site types (variable, conserved, singleton, and parsimony informative), and number of indels were used to determine genetic diversity between Tunisian PVY sequences. Indels were treated as missing data in each analysis and transition/transversion ratio (ts:tv) was obtained. For each sequence, length and proportion of GC and AT contents were estimated. The averages of pairwise nucleotide differences (K) and minimum number of recombination events (Rm) were also estimated. Thereafter, a codon-by-codon XY-plot graph was drawn for the 5'NTR/P1 nucleotide sequence on the basis of synonymous (dS) and non-synonymous (dN) substitution using the SNAP analysis program available from the HIV database (http://www.hiv.lanl.gov/content/sequence/SNAP/SNAP.html).

Selective neutrality was tested by both Tajima’s D (Tajima, 1989) and Fu and Li’s D* and F* methods (Fu and Li, 1993). Demographic parameters were assessed using the distribution of pairwise sequence differences (mismatch distribution) of Rogers and Harpending (1992) using the program DnaSP software version 5.10.01 (Librado and Rozas, 2009).

Phylogenetic analysis

Phylogeny reconstruction was performed using neighbor joining (NJ) methods based on the maximum likelihood (ML) approach (Saitou and Nei, 1987) applying the MEGA5 software (Tamura et al., 2011). The Tamura-Nei model of substitution was used for nucleotide sequences and both the ts:tv and the shape parameter (a) of the distribution of substitution rate variation between sites (discrete gamma distribution with eight categories) were estimated during the tree reconstruction. The unrooted NJ tree was constructed with 1000 bootstrap replicates using MEGA version 5 (Tamura et al., 2011). The genetic relationship of the detected haplotypes was graphically displayed by the network program NETWORK version 4.6.1.0 (Bandelt et al., 1999).

Correlation analysis

Pairwise correlations between genetic and geographical distances were assessed by the Mantel test (Mantel, 1967) using the TFPGA 1.3 program (Miller, 1997). This method makes it possible to compare two distance matrices which are obtained independently by calculating an association coefficient (z) using the following statistic: z ¼ ∑i∑j → ix ij yij for i ≠ j, where i and j are row and column indices, and where significance of z is assessed by a permutation test (10000 random permutations within one of the matrices). The significance of correlation was estimated at P < 0.05.

DISCUSSION

This paper presents a series of experiments that started with the occurrence and distribution of PVY in Tunisia. Consequently, biological, serological and molecular typing of PVY-TN was determined to assess their strain variant type, genetic structure and phylogenetic relationship with worldwide PVY isolates retrieved from the GenBank database.

Here, preliminary results indicating the high prevalence of PVY (up to 80%) are in close agreement with the situation reported in numerous worldwide potato growing areas (Gray et al., 2010), especially in Northern Africa (Boukhris-Bouhachem et al., 2010; Djilani-Khouadja et al., 2010). All positive samples were largely geographically distributed in Tunisia except for the central part of the country (Sidi Bouzid), which showed the lowest rate. This could be due to climate and seems to be in accordance with other works on seed potatoes in Tunisia (Boukhris-Bouhachem et al., 2010). Several factors may have contributed to the widespread distribution of PVY in Tunisia, which is one of the largest producers of ware potatoes in North Africa. Indeed, it is currently dependent on seed potato importation necessary for subsequent multiplication and propagation in certified seed potato fields. Thus, large volumes of imported seed potatoes may certainly carry a risk of introduction of potato pathogens, including the new recombinant PVY strains. The main sources of seed potatoes imported by Tunisian producers are Holland and France (Khamassy N, personal communication), where a wide range of PVY types have been reported since the 1980s (Kerlan et al., 2011). Furthermore, the absence of local well-established systems for multiplication and distribution of virus-free seed potatoes of high quality, the presence of alternate PVY hosts (weeds or crops) and the large range of aphid species present in Tunisia and able to transmit this virus (Boukhris-Bouhachem et al., 2010) will also contribute to PVY distribution. Furthermore, our data clearly showed that the PVY isolates were mainly necrotic PVYN strains (~73.2%). This is in agreement with other studies in many countries (Kerlan et al., 2011).

Consequently, the analysis by immunocapture RT-PCR of the 5' terminal region of the genome made it possible to separate NTN from the rest of PVYN. This provides host-independent criteria to classify PVY isolates, which can be applied in variability and epidemiological studies (Glais et al., 2005). Such an approach pointed to the spread and predominance of the necrotic PVYNTN variants (~ 67%) in Tunisian potato crops compared to ordinary PVYO-C strains, similar to what was reported in Europe, North Africa (Djilani-Khouadja et al., 2010) and South America (Ávila et al., 2009). On the contrary, PVYO was recently shown to be the most prominent strain infecting potato crops in North America followed by the PVYNTN strain (Gray et al., 2010). This high frequency of the PVYNTN variant in Tunisia, causal agent of PTNRD (Beczner et al., 1984), is not correlated with a significant presence of this disease in Tunisia. This could be due to the fact that the cv. Spunta, widely cultivated in Tunisia, is tolerant to this disease, although susceptible to PVY, i.e., it does not express typical tuber necrosis symptoms under natural conditions of infection (Kerlan C, unpublished results). However, this PVYNTN predominance should be considered a significant risk for growers planting cultivars susceptible to PTNRD, such as Nicola. Recent outbreaks of this disease in the last twelve years have occurred worldwide (Kerlan et al., 2011).

The increased incidence of PVYNTN in potato production across the country noted here is of great concern. It is of considerable importance to highlight mechanisms and factors contributing to virus epidemiology and to establish effective strategies for viral control. Thus, partial genomic sequences of amplified 5'NTR/P1 region were directly determined for seventeen Tunisian isolates belonging to different geographical origins, extreme north (Bizerte), north (Manouba and Jendouba), Sahel (Mahdia) and center (Kairouan), and collected from four potato cultivars: Pamina, Spunta, Nicola, and Atlas (Table 1). The information they provide about the changes that have occurred during worldwide evolution, the migration in PVY populations and the phylogenetic relationship with worldwide isolates was discussed.

Studies of the genetic structure of populations of plant viruses are important for understanding the evolution of virus/host interactions and the geographical distribution of viruses (Gibbs and Ohshima, 2010). There are many reports on the genetic structure of potyvirus populations, notably those on PVY (Ogawa et al., 2008; Karasev et al., 2011). These reports showed that virus populations were shaped by selection, founder effects and genetic recombination. Similarly, we attempted here to understand the genetic structure of the Tunisian PVY isolates. Nucleotide composition and genetic diversity of the 5'NTR/P1 region showed a high level of polymorphism consisting on nucleotide substations (transition) overall (Table 2). Furthermore, the rate of the identical sites was correlated with the pairwise identities, showing high conservation sites. This result confirms the efficiency of the 5'NTR-P1 region as a barcode. It also supports the high level of polymorphism characterizing Tunisian crops. All the sequences analyzed showed a low Rm value (Rm = 3). This makes such an event an implausible mechanism for the generation of the observed Hd (Table 2). Sequence polymorphism devoted to the 5'NTR-P1 region may be associated with the evolution that has allowed mutations rather than recombination to accumulate within this viral segment. PVY isolates propagate as a mixture of closely related mutants termed viral quasi species (García-Arenal et al., 2001).

To understand the selective determinants of sequence variation patterns, dN/dS rates were analyzed. Results showed the lower rates of dN/dS (dN/dS < 1), suggesting selective constraint against non-synonymous mutations acting at this region (García-Arenal et al., 2001). A dS or silent mutation does not increase the likelihood of a change in the structure and may not affect protein function and hence the determinants of pathogenicity of the virus (Kimchi-Sarfaty et al., 2007). Mutations that can be a threat to the functionality of the viruses are eliminated in the virus replication cycles creating a process of “purification” leading to the conservation of involved sequences (Dake et al., 2010). Thus, this region would accumulate dS mutations, described as neutral, along the coding sequences (Lopez et al., 2002). It has long been accepted that dS mutations are neutral until the last two decades, when it was realized that in some populations, especially in highly expressed genes, these mutations do not affect synonymous codons in a random manner. It would appear that this type of mutation would be able to affect the abundance of the protein, disrupting splicing and transport of mRNA while altering its stability (Goymer, 2007). The high genetic stability observed when a virus is in its natural host (host for which it is well adapted) (Koenig and Lennefors, 2000) suggests that it is the action of the stabilizer on the selection of the viral genome that maintains this balanced state. The best example of genetic stability of viral populations is the analysis of European and Australian isolates of turnip yellow mosaic virus, which showed only 3-5% of nucleotide differences (mostly synonymous mutations) between these two populations, which nevertheless diverged about 14,000 years (Keese et al., 1989). Structural and functional constraints on viral populations therefore tend toward temporal and geographical stability of viral populations and may then impose limits to the adaptation of viral populations.

To understand the demographic history of the viral PVY populations, phylogenetic analysis of the sequenced region was performed. The relationship between haplotype network, Tajima’s D neutrality test and mismatch distribution of pairwise differences can provide insights into population genetic structure. The relationship between Hd and Pi can help to determine the population phylogenetic pattern. In this study, Hd was high (0.993) in contrast to Pi value (0.00862). This could be explained by the occurring mutations. This large number of closely related haplotypes may indicate a signature of a rapid population expansion (Lévy-Hartmann et al., 2012). The Tajima’s D neutrality test provides consistent information related to population expansion, genetic hitchhiking and purifying selection (Holsinger et al., 2010). Here, the negative Tajima’s D test predicts an excess of rare alleles (Schmidt and Pool, 2002) and indicates that PVY evolves into stable population’s size, strongly suggesting a recent population expansion of PVY isolates. This may be explained by the good adaptation of Tunisian isolates to their endemic areas. Pairwise mismatch distribution was estimated and a bimodal pattern was revealed for Tunisian PVY isolates. This bimodal distribution with a distinct separation between the two observed modes corresponds to two PVY populations separated by a geographical barrier (Mora et al., 2007). This does not seem to confirm recent demographic expansion of PVY sequences, but an early evolution seems to characterize these sequences. These findings were correlated to the haplotype network, showing the clustering of PVY isolates into two distinct subgroups, regardless of their geographic and potato cultivar origin. In the same prospect, the seventeen Tunisian partial-genome sequences were subjected to a phylogenetic analysis against nine PVY sequences from GenBank. Sequence analysis revealed a strong identity between Tunisian isolates. In contrast, previous analyses at the 5' end have revealed high variability between PVYN field isolates (Blanco-Urgoiti et al., 1998). It is worth noting that Tunisian isolates were distributed independently of geographical origin. However, a correlation between potato cultivar origin and the branching pattern seems to exist. Furthermore, the Belgian isolate (JQ969035) and the Tunisian isolate RJ4 clustered together and showed the same branch length. This could be due to the fact that Tunisia imports a small amount of potato seed from Belgium. Additionally, Nie and Singh (2002) showed that sequence comparison and phylogenetic analysis of the 5'-UTR and P1 region indicated that PVYN isolates from the European Union and North America formed their own separate groups. Intra-group sequence identity for all except one was over 98%, as opposed to the inter-group identity of 90%. On the other side, the PVYNTN isolates from the European Union and North America clustered with their respective PVYN isolates. This indicates a possible evolution of PVYNTN isolates from the PVYN isolates of a geographical region (Nie and Singh, 2002). Thus, sequencing of the P1 gene and use of the competitive RT-PCR approach could be applicable for determining the possible origin of new occurrences of PVYNTN from other geographical regions.