The Role of Viral Infection in Inducing Variability in Virus-Free Progeny in Tomato

Experiment 1: cytogenetic analysis

Following studies of chromosome behavior at meiotic diakinesis, the number of chiasmata per pollen mother cell (PMC) has been found to increase substantially in 3 F₁ hybrid genotypes and in one of two cultivars included in the cytogenetic analysis. This was due to an increase in the number of bivalents with two or more chiasmata per bivalent (Table 1).

At diakinesis, the chiasma frequency was 1.38–1.46 per bivalent in the non-infected plants of the F₁ hybrids and 1.53–1.65 in infected plants, and 1.36–1.45 chiasmata in non-infected cultivars and 1.52–1.63 for infected one.

Typical meiotic bivalent types were attested in healthy and infected plants (Figure 1). F₁ hybrids and cultivars have been found to significantly differ in the bivalents with one chiasma (rod bivalents) (Table 1). In all attested cases the number of rod bivalents have decreased in virus infected plants. On the contrary, the ring bivalents have increased significantly or remained unchanged (cv. Prizior). At the same time the numbers of other types of bivalents (univalent as well as bivalents with three chiasmata) have increased too, in hybrids and cultivars genotypes.

Modified frequencies of types of bivalents classified according to the chiasma number were observable in both hybrid genotypes and cultivars when comparing infected plants with non-infected ones. The number of bivalents with one terminal chiasma (T) decreased from 7.50 down to 4.14 (P < 0.001) upon infection of plants of the Fakel cultivar, and from 6.22 down to 5.34 (P < 0.001) following infection of plants of the Prizior cultivar (Table 2). Concurrently, the number of bivalents with two (IT, P < 0.01) and three (ITT, P < 0.001 and P < 0.05 for Fakel and Prizior, respectively) chiasmata per bivalent increased significantly. In other words, infected plants exhibited a significant increase in the number of bivalents with two and three interstitial chiasmata, whereas the number of bivalents with a single chiasma (T, I) decreased or remained unchanged.

Analysis of the correlation between various bivalent types and the total number of chiasmata (S) per PMC revealed differences for infected plants. In conformity with the diagrams presented (Figure 2A–F), the increase in the total number of chiasmata (S) per PMC is dependent on the increase in the number of TT and ITT bivalents, followed by II and IT bivalents, and the degree of correlation varies according to the variant (infected or non-infected) and genotype.

Recombination frequency

The frequency of recombinants in marked segments of the virus-infected progeny of F₁ hybrid Mo × K differed from that of the non-infected progeny.

The exhibition of the traits determined by genes d, aw, c, and m-2 in homozygous genotypes remained unaffected in both infected plants and their progeny. The ratio of dominant to recessive phenotype classes was within the expected range of 3:1 in both the control and treatment (Table 3).

An increase in the frequency of recombinants from 10.39 ± 1.11 to 13.75 ± 0.86 (P < 0.001) (%) was noted for the d-aw segment (chromosome 2), as the result of the appearance of different rates of the phenotypes determined by the genotypes D-/awaw and dd/Aw- in the treatment variant of F₂ Mo × K compared with control one (Table 4). At the same time, no difference in rf was observed in the c-m-2 segment (chromosome 6) (rf = 26.58 ± 1.12, %) between the control and the treatment variant (rf = 27.49 ± 0.94, %) of the F₂ hybrids Mo × K and La × V (data not shown).

Quantitative traits

Quantitative traits were examined in virus-free populations (F₂–F₄, G₁–G₂) descending from infected F₁ plants and their parental forms. The results suggested that although viral infection was not present in the progeny, the difference (from the control) was still observable. In F₂–F₄ populations derived from infected F₁ hybrids, a significant difference in mean values were observed for 40.91–90.48% of the traits analyzed (Table 5). Some traits tended to evolve in a similar way, exhibiting mean values higher or lower than those of the control over three generations analyzed (data notshown).

The substantial differences in fruit weight between forms involved in crossing resulted in a wide range of distribution for this trait in F₂ segregating populations (Table 6). In each of the four categories of F₂ plants examined for the fruit weight, plant frequency in the population derived from infected F₁ hybrids differed from the control. The largest deviations were noted in F₂ populations of the La × V and N × K crosses. Thus, the frequency of plants with a fruit weight of 20–50 g was 23.81% in the treatment variant of the F₂ hybrid N × K, compared with 7.94% in the control. At the same time, the frequency of plants with a fruit weight of more than 100 g was 2.38% in the treatment variant, compared with 13.49% in the control.

In each hybrid combination, one of the parental forms – that with a smaller fruit weight – was characterized by an earliness of ripening while the other parental form – the one with a larger fruit weight – showed a later maturity (Figure 3). It can be assumed, therefore, that the F₂–F₄ plants with an earlier maturity will also tend to have a smaller fruit weight since, according to the evidence in the literature, these traits are inherited as linked characters.

Earlier floral initiation and fruit ripening were observed in the treatment variant as compared with the control; at the same time, fruit weight also showed larger values or no significant difference (Figure 3). According to our results, fruit reached maturity later in the C group of plants than in the B group in treatment variants of the F₃–F₄ populations of the Nw × C cross and the F₃ population of the N × K cross, or the differences were negligible (the A group in population F₃ of the La × V cross and the C group in population F₄ of the N × K cross) (Figure 3A–C).

At the same time, mass ripening of fruit occurred 1–12 d earlier in the treatment variant than in the control. Thus, in the F₃ population of the Nw × C cross, the control exhibited delayed maturing (118–120 d), whereas the treatment variant plants in the B and C groups showed medium (112–114 d) and early (105–107 d) maturity, respectively, in line with Brezhnev's (1982) classification of maturity dates.

Concurrently with an increase in mean values of fruit weight and earliness in treatment variants, the proportion of recombinants with higher mean values of these two traits increased. Therefore, it can be stated that in the treatment variants, early fruit ripening occurred more frequently in those plant forms that had larger fruit weights. Approximately 94.4% and 82.0% of the F₃ plants of the N × K cross in the treatment variant showed early maturity (105–107 d), whereas in the control they constituted 69.6% and 29.2% for the B and C groups, respectively (Figure 4A).

Earliness of ripening was often associated with a smaller fruit weight in the control (and in the parental form K) in the F₃ of the N × K cross, while in the treatment variant of this cross earliness of ripening was observed in plant forms with a larger fruit weight (a combination of traits inherited from both parental forms). The above traits differed dramatically compared with the control in the B group in the F₃ of the Nw × C cross and less so in the C group (Figure 4A,B). As for the population that had not been subdivided into fruit weight categories, the frequency of plants combining earliness of ripening and large fruit weight increased from 2.22 up to 8.24 in the treatment variant. In the treatment variant of the F₃ population of the La × V cross, along with earliness of ripening noted for this population, a proportion of plants exhibited very late maturity (>120 d) – something that was not observable in the control.

Generally, a shorter “seeds germination–fruit maturing” period results in lower fruit biochemical indices since most of these indices are negatively correlated with early maturity (Raijadhav et al. 1996). In view of the above, of interest are individuals that combine such traits as large fruit weight, early maturity and increased soluble solids (SS) content. anova revealed a significant contribution (P < 0.001) of the factors “fruit weight” and “virus” to the variation in SS in populations descending from three virus-infected F₁ hybrids (Table 7).

According to some evidence (Eshed and Zamir 1996; Tanksley et al. 1996; Bernacchi et al. 1998; Causse et al. 2002), the SS content is negatively correlated with fruit weight. At the same time, in the treatment variant of F₂ population the correlation between fruit weight and SS content was not always similar to that in the control. Sometimes, this correlation was positive as well, as in Grandillo and Tanksley (1996). Thus, in the progeny comprising the treatment variants, fruit with large weight often had a higher SS content as compared with the large-weighted fruit in the control or with the smaller weight fruit in the same treatment variant. Similarly, the values of coefficients of correlation between SS and pH or fruit weight and pH changed in the treatment variant populations with those of the control (data not shown).

Besides the facts mentioned, we have to add that in recombinant forms the favorable combinations of traits were frequently associated with less desired ones, such as fasciated or cracking fruits.

Basic experimental materials

The material used in the present study consisted of four tomato (Solanum lycopersicum L.) intraspecific hybrids F₁ Nota × Krasnoyarskii rannii (N × K), Novichok × Colocolchik (Nw × C), La 0651 × Victorina (La × V), Mo 500 × Krasnoyarskii rannii (Mo × K), parental forms (N, Nw, C, K, V, La, and Mo) and two additional cultivars – Fakel and Prizior. The genotypes used in producing hybrids differed in the traits under study. Tomato F₁ hybrids were obtained by artificial cross pollination.

Two of the F₁ hybrids – Mo × K and La × V – heterozygous for genes D, Aw, C, M-2 and C, M-2 respectively, were used to determine the recombination frequency (rf) in the marked segments. The genotypes La 0651 (c-m-2/ c-m-2) and Mo 500 (d-aw/ d-aw; c-m-2/ c-m-2) served as maternal parents. The genes d-aw (d– dwarf and aw– without anthocyan) and c-m-2 (c– potato leaf and m-2– mottled-2) are located on chromosome 2 and chromosome 6, respectively. Locations and designations of these genes are according to Atherton and Rudich (1986).

Inoculation of plants

In our previous research (Chiriac et al. 2006) we have attested the effect of TMV or PVX alone on chiasmata frequency in virus-infected plants as well as inducing variability in virus-free progenies. Synergetic effect was attested in some cases of mixed infection. A double infection inoculum – TMV + PVX, was used in our case.

Tomato plants were inoculated by rubbing the TMV + PVX inoculum into two leaves at the stage of the first truss differentiation. Under similar conditions, plants of the control variants were inoculated with a tissue extract from healthy plants. Viruses were isolated from tomato and potato plants infected with TMV and PVX, respectively. The TMV + PVX combination causes the inoculated plants to develop visual symptoms manifested as diffuse mosaic. The progeny lacked the symptoms. The presence of virions in the infected plants and their absence in the progeny of these plants as well as in the control variants was ascertained through the immunosorbent electron microscopy (ISEM) test (Derrick and Brlansky 1976).

Field experiments

Experiment 1

Tomato hybrid plants, parental forms and additional cultivars were inoculated with TMV + PVX. At meiotic prophase I, flower buds were excised from both infected and non-infected plants and analyzed for chiasma frequency. The infected plants were examined concomitantly with non-infected ones but on different experimental plots to avoid infection of healthy plants. F₁ hybrid plants were grown in the field, next to parental plants. Some 12–15 plants were infected for hybrid combinations and 10–20 plants for parental forms.

Experiment 2

The seeds collected from fruits of the first cluster from infected plants were used to produce the first generation of progeny from infected plants (F₂ for hybrid combinations and G₁ for parental forms herein referred to as the treatment plants and those collected from non-infected plants comprised the control).

Progenies of F₂ and F₃ plants, both the control and treatment, were used to produce the F₃ and F₄ generations, respectively. Under similar conditions, descendants of healthy and infected plants of parental forms were used to produce the first (G₁) and second (G₂) generations for the control and treatment, respectively.

With a view to inactivating viral particles, the seeds representing the control and the treatment and coming from healthy and virus-infected plants, respectively, were subjected to thermal treatment at 70 °C for 72 h (Wolf and Schmelzer 1972). Heat treatment had no significant influence on seed germination rate.

The seeds derived from selfing of four F₁ hybrids (N × K, Nw × C, La × V, and Mo × K) and of parental forms of the control and treatment were sterilized with 1% KMnO₄ for 10 min, rinsed in running water for 30 min, germinated in Petri dishes and then sown in the soil in a greenhouse.

Thirty five days after germination, some 250–424 F₂ seedlings, depending of genotype, were transplanted in the field. Plants of both variants – the control and treatment – were planted in the field in three replications with a minimum between-variant disturbance.

The F ₂ generation of plants derived from infected and healthy F₁ hybrids were examined for fruit weight and classed into a few categories: <20, 20–50, 50–100, >100 g. Seeds from fruits of each plant were selected separately and then pooled into groups according to the weight of fruit from which they had been extracted. These seeds were used to produce F₃–F₄ plants, for separate groups, according to their derivation in F₂ from plants with a particular fruit weight: the 20–50 and 50–100 groups (mentioned as B and C groups, respectively) for the N × K and Nw × C hybrids. The exception was La × V for which the <20 and 20–50 groups (mentioned as A and B groups, respectively) were included in the study for only F₃ plants. Included in the study were groups that were more representative numerically in F₂. Each plant group was analyzed separately.

The F₂ hybrids Mo × K and La × V were evaluated for the expression of qualitative traits c, m-2, d, and aw at the stage of two true leaves.

Quantitative traits

Trait values were assessed for each plant with subsequent estimation of the mean value per group. Approximately 20 traits were examined for each plant (of hybrid populations and cultivars): internode length, number of internodes to the first truss, number of trusses per main stalk, leaf length, plant height, number of fruit per plant per growing season, number of flowers and fruit in the first, second and third cluster, mass flowering and fruit ripening; fruit weight, diameter and height, and number of locules according to Brezhnev (1982). Besides, soluble solids content and pH were measured (Ermakov et al. 1987).

At the fruit ripening stage, five fruits were harvested from each plant. These were used to determine the mean weight and other fruit parameters. In order to determine biochemical characteristics of fruit (soluble solids content and pH), five fruits were homogenized and filtered through a cotton cloth. Soluble solids (SS) content was measured with the aid of a refractometer and pH level – with a pH-meter (Ermakov et al. 1987), each parameter being estimated with three replications.

Flowering: The number of plants with at least one flower in the first truss was counted every 4–6 d for a period of 5 weeks, and mass flowering was represented by the number of day at which 60% of the plants were in flower.

Ripening: The number of d from germination to the appearance of ripe fruit in the first cluster in 60% of the plants.

Cytological analysis

Genotypes of three tomato hybrids (N × K, Nw × C, and La × V) and of two cultivars (Fakel and Prizior) were used to ascertain the correlation between the presence of infection in the plants and chiasma frequency. The hybrid genotypes were the same as those for which quantitative traits had been examined. The cultivars were chosen from a similar and parallel study based on the estimation of chiasma frequency in non-hybrid genotypes.

Chiasma frequency was measured in 43–53 PMCs from buds of five infected plants and five non-infected plants for each genotype. Flower buds corresponding to stage II (4–5 mm) of the meiotic prophase I (Sawhney and Bhadula 1988) were collected from the first truss of infected and non-infected plants. The flower buds were fixed according to Clark's procedure (Pausheva 1988).

Chiasma numbers were determined in temporary mounts produced by staining with acetocarmine and pretreated with iron alums. The mounts were visualized under a light microscope BIOLOR, at resolutions of ×100 (objective) and ×20 (ocular).

The bivalents visualized (12 pairs) were classed according to the number and position of chiasmata at diakinesis (Imai and Moriwaki 1982): 1 interstitial (I), 2 interstitial (II), 1 terminal (T), 1 interstitial + 1 terminal (IT), 1 terminal + 2 interstitial (TII), 2 terminal (TT), 2 terminal + 1 interstitial (TTI), and univalents (0).

Analysis of recombination frequency in marked segments

F₂ populations of the Mo × K and La × V hybrids were produced by selfing virus-infected and healthy F₁ plants.

Heterozygous F₁ plants exhibited uniformity in dominant traits. Recombination frequency in marked segments was measured in segregating populations and it is expressed as a percentage. Included in the analysis were 4 795 plants for the treatment variant and 2 581 for the control of the F₂ hybrid Mo × K, and 1 552 and 1 235 plants for the treatment and the control, respectively, of the F₂ La × V genotype.

Mathematical data processing was carried out using the RISS software package of the Institute of Genetics, Academy of Sciences of Moldova. The segregation was tested by χ² analysis for their goodness-of-fit statistics to the expected Mendelian segregation ratio (3:1). Two methods were used to calculate the recombination fraction: the maximum likelihood method and the product ratio method.

Statistical analysis

Mean values of traits and error or standard deviation were estimated for F₁–F₄ hybrid plants and parental forms. The data were processed using the software package Statgraphics Plus for Windows (version 2.1; Microsoft Corp., Redmond, WA, USA) and Microsoft Excel. Significant differences in mean values of traits were established using the t-test. anova of SS was carried out by means of a two-way anova (Clewer and Scarisbrick 2001).

Cluster analysis was carried out by the Nearest Neighbor method based on comparing distances between the parameters analyzed and placing these in clusters by the closest distance between them. The distance is expressed in Squared Euclidean units.

(Handling editor: Daoxin Xie)