Development and Fertility Restoration of CMS Eggplant Lines Carrying the Cytoplasm of Solanum violaceum

A functional cytoplasmic male sterile (CMS) eggplant line carrying the cytoplasm of Solanum violaceum was developed in the past, but the fertility restoring genes (Rf-genes) were not identified. This work aimed to produce the CMS lines of three Hellenic eggplant cultivars (viz., ‘Langada’, ‘Emi’ and ‘Tsakoniki’) using the cytoplasm of S. violaceum and study the inheritance of the Rf-genes. The respective CMS eggplant lines were developed by the backcross method and examined for their fertility parameters. The results demonstrated that female fertility was not affected by the cytoplasm of S. violaceum. In contrast, the occurrence of three male fertility phenotypes (male sterile, male fertile and potentially male fertile) indicated that male fertility was affected by nuclear/cytoplasmic interactions. Male sterile plants were characterized by indehiscent anthers, low pollen viability and abnormal anther morphology. Male fertile plants formed dehiscent anthers with high pollen viability and normal morphology. Potentially male fertile plants initially formed dehiscent anthers, but in later stages formed exclusively indehiscent anthers. Male fertile plants were obtained in the advanced backcross populations of CMS ‘Tsakoniki’, but not in CMS ‘Langada’ and CMS ‘Emi’. The genetic analysis of fertility restoration indicated that male fertility in the genetic background of cv. ‘Tsakoniki’ is controlled by one essential genetic locus, affected by a secondary modifying locus. Molecular analysis of cp-DNA and mt-DNA in the CMS lines indicated maternal inheritance of the cytoplasm organelles. Our findings demonstrate that the genotype of the eggplant parent can affect the expression of CMS as well as fertility restoration.


Introduction
Eggplant (Solanum melongena L., Solanaceae) is an Old World vegetable crop of significant economic importance and its fruits are worldwide marketed for fresh consumption or processed.Today, commercial F 1 hybrids are widely used in eggplant cultivation because they exhibit heterosis in several agronomical traits including earliness and yield (Kakizaki, 1931;Rodríguez, Prohens, & Nuez, 2008;Sambandam, 1962).In the 2015 European catalogue of registered varieties, more than 75.0% of the eggplant entries were F 1 hybrids (European Commission, 2015).Since the eggplant flower is androgynous, emasculation and hand pollination are required in hybrid seed production, resulting in a higher seed cost.A strategy to reduce that cost is to use eggplant male sterile lines as female parents in hybrid seed production.In these lines flower emasculation is not necessary and pollination can be achieved manually or with insect pollinators.
In eggplant, a number of genic (Chauhan, 1984;Nuttall, 1963;Phatak, Liu, Jaworski, & Sultanbawa, 1991), cytoplasmic (Fang, Mao, & Xie, 1985;Hasnunnahar, Khan, & Isshiki, 2012;Isshiki & Kawajiri, 2002;Khan & Isshiki, 2011) and genetically engineered (Cao, Huang, Chen, & Lei, 2010;Toppino & Kooiker, 2011) male sterility systems were developed.Genic male sterility has a limited practical application because of its complex mode of maintenance and utilization (Budar & Pelletier, 2001).Besides, there is some evidence that expression of genic male sterility in eggplant is affected by the environmental conditions (Hazra, Roy, & Choudhury, 2008).In addition, the cultivation of genetic engineered crops for human consumption is not widely accepted in many countries and has received mixed public acceptance, especially in the EU (Costa-Font, Gil, & Traill, 2008).On the other hand, the maintenance of CMS and their utilization in hybrid seed production is preferable because of the maternal inheritance of the male sterile character and the mode of fertility restoration (Budar & Pelletier, 2001).Therefore, the development of reliable CMS systems accompanied by genes that restore fertility (Rf-genes) is an efficient way to reduce the cost of eggplant hybrid seed.
The identification and incorporation of the appropriate Rf-genes in the male parent is essential for the effective utilization of an eggplant CMS line in a commercial hybridization system.To the best of our knowledge, the Rf-genes are available only in the pollen-non formation eggplant CMS lines carrying the cytoplasms of S. grandifolium, S. integrifolium and S. anguivi (Hasnunnahar et al., 2012;Khan & Isshiki, 2010, 2011;Saito et al., 2009).Taking into account that long term utilization of limited cytoplasm sources in eggplant cultivation has the potential risk of genetic vulnerability against new pathogens and pests (Khan & Isshiki, 2008), the identification of the respective Rf-genes in the rest of CMS eggplant lines could be beneficial for both practical and research purposes.
The aim of the present study was: firstly, to develop and study the CMS lines of the Hellenic eggplant cultivars 'Langada', 'Emi' and 'Tsakoniki' by utilizing the cytoplasm of S. violaceum, and secondly, to study the inheritance of the Rf-genes for this type of CMS.

Plant Material
Three Hellenic eggplant cultivars viz., 'Langada' (L), 'Emi' (E) and 'Tsakoniki' (T), were used as recurrent parents (nucleus donors) and the wild species Solanum violaceum (V) was used as maternal parent (cytoplasm donor) for the development of the respective eggplant CMS lines (Figure 1).The aforementioned cultivars were selected because they are of significant economic importance and are highly appreciated in the local market.Briefly, interspecific crosses were made using the wild species as the female parent and the eggplant cultivars as the male parents.The resulting interspecific hybrids viz., F 1 (VxL), F 1 (VxE) and F 1 (VxT) were continuously backcrossed to their respective eggplant parent and five backcross generations were produced for each hybrid.Progenies obtained after self-pollination of individual male fertile plants were also cultivated and used in the genetic analysis of fertility restoration.The experiments were carried out following standard cultural practices in the experimental fields of Hellenic Agricultural Organization 'Demeter' (formerly, National Agricultural Research Foundation) located in Thermi, Thessaloniki.

Female Fertility
Female fertility of the backcross progenies was assessed by investigating the seed germination rate, percentage of successful backcrosses and number of seeds per fruit.Backcross seed germination rate was examined in eight plants of each backcross population, with the exception of BC 1 where one sample of 200 seeds was used per population.One hundred seeds per plant were sown in plastic pots in a heated glasshouse and the germination rates were recorded four weeks later.For investigating the percentage of successful backcrosses, three to five flowers per plant were emasculated and hand pollinated with fresh eggplant pollen and the percentage of successful backcrosses was calculated in each population.In addition, each year 12 flowers of each eggplant cultivar were emasculated and manually self-pollinated using pollen of the same cultivar.Average number of seeds per fruit was examined in eight randomly selected plants of each backcross population by counting the number of seeds per fruit.

Male Fertility
Male fertility was evaluated by pollen release ability, anther morphology and pollen viability examination in the first 10 flowers per plant.Pollen release ability was investigated by observing the terminal pore end of the anther in freshly opened flowers under a stereomicroscope.The plant material was classified as male fertile (MF), potentially male fertile (PMF) or male sterile (MS) according to the ability of the anther to release its pollen at the stage of anthesis.The observed frequencies of each class were recorded and chi-square (χ 2 ) goodness-of-fit test was used for the statistical analysis.Since, the presence of three fertility phenotypes, including a potentially male fertile one, indicated the action of a major genetic locus plus at least one modifying locus, the observed frequencies were compared to the appropriate expected frequencies.
For the examination of anther morphology, anthers from 3 freshly opened flowers per plant were visually observed and classified as normal (having typical size and appearance), petaloid (transformed to petal like structures) or atrophic (having reduced size and dark brown color) and the observed frequency of each class was recorded.Pollen viability was estimated by the acetocarmine staining method.Concisely, anthers of the plant material were squashed in a 2% acetocarmine solution and placed under a light microscope (Qureshi, Khan, Arshad, Rashid, & Ahmad, 2009).More than 300 pollen grains per plant were observed with three replications and the percentage of viable (stained) pollen grains was calculated.Abnormal anthers were excluded from pollen release ability and pollen viability examination because it was preliminary confirmed that they did not contain pollen grains.

Cytoplasm Inheritance
The inheritance of the cytoplasm in the three CMS eggplant lines was confirmed by chloroplast (cp) and mitochondrial (mt) DNA analysis.Total DNA was isolated from fresh leaves by using the CTAB method (Doyle & Doyle, 1987) from 3 individuals of the following plant material: S. violaceum, cv.'Langada', cv.'Emi', cv.'Tsakoniki', MS plants from populations BC 4 (V×L) 6 , BC 4 (V×E) 14 and BC 4 (V×T) 24a and MS and MF plants from population BC 3 (V×T)S 1 26 (Figure 1).High resolution melt (HRM) analysis and sequencing of the chloroplast trnL intron were employed in order to study the inheritance of cp-DNA.The normalized HRM curves of trnL amplicons were generated from the examined plants by using the Ph-trnL oligonucleotide set following the procedure described by Madesis, Ganopoulos, Argiriou, and Tsaftaris (2012) and used to discriminate the different genotypes.In addition, the chloroplast trnL intron was sequenced by using the set of primers F:CGAAATCGGTAGACGCTACG and R:GGGGATAGAGGGACTTGAAC in both directions and the obtained sequences were compared.Mitochondrial DNA was analyzed by RFLP analysis according to Khan & Isshiki (2008).Briefly, PCR amplified fragments of nad7/3-4 region and V7 region of mitochondrial small ribosomal subunit RNA (srRNA) gene were digested with AluI and ScrFI, respectively.The resulting DNA restriction fragments were separated on a 1.5% agarose gel containing ethidium bromide and photographs were taken under UV illumination.

Develo
The flowc consisted o occurred in (data not s subsequen

Female
The Note. z Superscripted numbers indicate the cross from which the examined population was obtained and correspond to the cross numbers given in Figure 1; letters (a and b) indicate origin from different maternal plants of the former population; y N = sample size; x Means±standard errors.For the eggplant cultivars means of six years are presented; w Not investigated.

Pollen Release Ability
Pollen release ability was normal in the eggplant cultivars, S. violaceum and their corresponding interspecific hybrids.Observations under the stereomicroscope showed that the terminal pore of the anther dehisced at anthesis and released pollen (Figure 2a).On the contrary, in the BC 1 populations three distinct classes of plants were identified with respect to the pollen release ability: (I) Pollen releasing class: plants that formed exclusively dehiscent anthers and set several fruits following self-pollination (Figure 2c); (II) Pollen non-releasing class: plants that formed exclusively indehiscent anthers and did not set fruits following self-pollination (Figure 2b); (III) Potentially pollen releasing class: plants that originally were of the pollen releasing type, but later (usually one month after flowering) formed anthers of the pollen non-releasing type.Such plants set a few seeded fruits after self-pollination of the first flowers, while the rest of the flowers were aborted.In addition, formation of seedless parthenocarpic fruit occasionally occurred in the pollen non-releasing and potentially pollen releasing progenies derived from the hybrid F 1 (V×E) (data not shown).Since anther non dehiscence was the main reason of male sterility, the plant material was classified based on pollen release ability as male fertile (MF), male sterile (MS) and potentially male fertile (PMF).

Anther Morphology
Normal anther morphology was observed in S. violaceum, the eggplant cultivars, their respective interspecific hybrids and the MF plants (Table 3).On the contrary, MS plants formed flowers with either normal or abnormal anthers and the frequency of abnormalities tended to increase in the succeeding backcross generations (Table 3, Figures 2d-2f).PMF plants initially developed some flowers with normal anthers, but the flowers formed in later stages had anthers with either normal or abnormal morphology.It was noticed that the type of abnormalities in anther morphology varied with respect to the eggplant recurrent parent.For example, both the petaloid and atrophic anther types were present in the BC populations with the genetic background of cv.'Langada' and cv.'Emi', whereas, only the atrophic type was found in the genetic background of cv.'Tsakoniki' (Table 3, Figures 2e and 2f).It was also observed that petaloidy was the prevalent morphological abnormality of the anther in the advanced backcross progenies of F 1 (V×L), but diminished in the respective progenies of F 1 (V×E) (Table 3).Table 3. Number of flowers (in percent) with normal (Nor), petaloid (Pet) or atrophic (Atr) anthers in Solanum violaceum, three eggplant cultivars, their interspecific hybrids and the respective backcross and selfed progenies in three fertility phenotypes   Note.z Superscripted numbers indicate the cross from which the examined population was obtained and correspond to the cross numbers given in Figure 1; letters (a, b and c) indicate origin from different maternal plants of the former population; y N 1 = total number of examined plants per fertility phenotype.N 2 = total number of examined flowers per fertility phenotype; x For S. violaceum and the eggplant cultivars data for six years are presented.

Pollen Viability
In the examined plant material the viable pollen grains were round with typical size and were stained after acetocarmine treatment, while non-viable grains appeared shriveled and empty (Figure 2g-2i).It was established that pollen viability was high in the eggplant cultivars and S. violaceum and moderate in their corresponding interspecific hybrids, whereas, pollen viability of the MS plants was generally low and tended to decrease further in the subsequent backcross generations (Table 4).On the other hand, it was observed that pollen viability in the PMF and MF plants was moderate and moderate to high, respectively, and remained high and rather stable in the advanced BC populations (Table 4).

Cytoplasm Inheritance
HRM analysis of the cp-DNA showed that the examined plants in BC 4 and BC 3 S 1 populations grouped together with S. violaceum regardless of their fertility phenotype; while the eggplant parents were clearly separated (Figure 3A).Similarly, the trnL intron sequencing data indicated that the examined BC 4 and BC 3 S 1 plants shared the same sequence with the wild species, but these sequences were different in the eggplant cultivars (Figure 3B).In addition, the RFLP analysis of the mt-DNA demonstrated that the restriction patterns of the examined MS and MF plants were identical to those of S. violaceum, but different from those of the eggplant parent (Figure 4).
Table 4. Mean pollen acetocarmine stainability (in percent) in S. violaceum, three eggplant cultivars, their interspecific hybrids and their backcrossed and selfed progenies in three fertility phenotypes Note.z Superscripted numbers indicate the cross from which the examined population was obtained and correspond to the cross numbers given in a , nts ment n the were successfully produced via the backcross method using S. violaceum as the female parent (Figure 1).The stability of the male sterile phenotype confirms that S. violaceum is a suitable cytoplasm donor for the induction of CMS in eggplant, as it was previously reported by Isshiki & Kawajiri (2002).The presence of some dwarf and chlorotic plantlets was restricted in the BC 1 and BC 2 (V×T)S 1 22 populations and indicated adverse effects of interspecific hybridization on plant vigor.Reduced vigor was observed in the reciprocal interspecific hybrids S. melongena × S. macrocarpon (Bletsos, Roupakias, Tsaktsira, & Scaltsoyjannes, 2004).However, in the present study these abnormalities were eliminated from the subsequent generations where only healthy plants were recovered.
A common feature of CMS is that the female gametophyte is not affected from the presence of the alien cytoplasm (Nair, 1993).In an eggplant CMS line, high fruit set and seed number are essential because hybrid seed is the final product.Our results demonstrate that the CMS induced by the S. violaceum cytoplasm had no negative effect on the female fertility of the examined plant material.That was apparent from the observed gradual increase in the percentage of successful backcrosses, the number of seeds per fruit and the germination percentage in the BC populations (Table 1).The seed number per fruit was lower in the BC 5 populations when compared to the respective eggplant parents (Table 1), perhaps due to incomplete replacement of the eggplant alleles for increased seed number.However, selection of individuals with increased number of seeds is expected to resolve that problem (Isshiki & Yoshida, 2002).From a practical point of view, CMS 'Tsakoniki' is a more suitable seed parent because of the higher backcrossing percentage and seed number per fruit when compared to the other CMS lines (Table 1).The overall female fertility results reported here are in accordance with those previously reported in several other CMS eggplant systems (Isshiki & Kawajiri, 2002;Khan & Isshiki, 2008, 2009, 2010, 2011).
On the other hand, CMS can adversely affect the appearance and functionality of the male organs of the flower, as well as male gametogenesis (Hanson & Bentolila, 2004).The CMS phenotype in alloplasmic eggplant was associated with the transcription patterns of novel open reading frames (ORFs) in the 5' flanking region of atp1 mitochondrial gene (Yoshimi et al., 2013).In the present study the negative effect of the S. violaceum cytoplasm on male fertility of the MS plants was expressed with reduced pollen release ability, abnormalities in anther morphology and low pollen viability (Figure 2b, Tables 2-4).Since the terminal pore of the anthers in our CMS lines did not dehisce at anthesis (Figure 2b), this type of CMS can be classified as a functional CMS.Anther non dehiscence was a common feature in our three CMS lines and the one developed by Isshiki and Kawajiri (2002), which also carried the cytoplasm of S. violaceum.Therefore, it can be concluded that the specific cytoplasm induces the functional CMS type in eggplant, regardless the genetic background of the eggplant nucleus donor.Anther non dehiscence was attributed to the malfunction of the tapetum tissue in GMS eggplant and S. insanum, a species closely related to eggplant (Chauhan, 1984;Karihaloo & Malik, 1995).
The replacement of the anthers with petaloid structures in the lineages of CMS 'Langada' and CMS 'Emi' (Figure 2e), indicated that cytoplasmic/nuclear interactions induced homeotic changes in flower morphology.Similar CMS induced alterations were reported in eggplant and several other plant species including carrot, leaf mustard, rapeseed and wheat (Carlsson, Leino, Sohlberg, Sundström, & Glimelius, 2008;Fang et al., 1985;Isshiki & Yoshida, 2002;Linke, Nothnagel, & Börner, 2003;Yang et al., 2005;Zhu et al., 2008).However, the petaloid anther type was absent from the lineage of CMS 'Tsakoniki' (Table 3), suggesting that the type of anther abnormalities depends on the genotype of the eggplant parent.This is further supported by the fact that petaloidy was the prevalent anther abnormality in CMS 'Langada', but the less frequent in CMS 'Emi' (Table 3).On the other hand, the atrophic anther type was common in the three CMS eggplant lines (Table 3, Figure 2f).Perhaps this type of abnormality is induced by the cytoplasm of S. violaceum through a different mechanism.Similar abnormalities including underdeveloped, abortive and brown anthers were attributed to pre-meiotic malfunction of the tapetum in GMS eggplant, CMS carrot and CMS Rosmarinus officinalis L. (Chauhan, 1984;Hidalgo, Hesse, Ubera, & Frosch-Radivo, 1999;Kozik, Nowak, Nowakowska, & Dyki, 2012).
Pollen viability of the MS plants was very low and did not recover in the advanced BC populations (Table 4).These findings combined with the increased female fertility of the advanced BC populations (Table 1), suggest that low pollen viability is induced by a kind of disharmony between the S. violaceum cytoplasm and the eggplant nucleus, rather than meiotic abnormalities.Although meiotic associations were not studied in the present work, available cytogenetic evidence regarding the same species combination strongly supports this hypothesis.For example, near perfect meiosis was reported in the interspecific hybrids between S. violaceum × S. melongena (Isshiki & Kawajiri, 2002;Rajasekaran, 1970) and in several MS plants in the respective BC 4 progenies (Isshiki & Kawajiri, 2002).Problems during pollen maturation including incomplete starch accumulation and hydrolysis may account for the low pollen fertility observed in our functional CMS lines, as it was recently demonstrated in the eggplant functional CMS lines carrying the cytoplasm of S. violaceum and other wild species (Khan, Hasnunnahar, Iwayoshi, Ogura-Tsujita, & Isshiki, 2015).
In the CMS eggplant systems studied so far, the Rf-genes are available only in the pollen non-formation type caused by the cytoplasms of S. integrifolium, S. anguivi and S. grandifolium (Hasnunnahar et al., 2012;Khan & Isshiki, 2010, 2011;Saito et al., 2009).In the abovementioned CMS systems, the fertility restored plants exhibited marked differences in the procession of the co-transcripts of a novel orf and atp1 in relation to the CMS counterparts (Yoshimi et al., 2013).However, the presence of the Rf-genes was not confirmed in the pollen non-releasing (functional) CMS eggplant lines carrying the cytoplasms of S. violaceum, S. virgianum and S. kurzii (Isshiki & Kawajiri, 2002;Khan & Isshiki, 2009, 2010).In the present study, some plants appeared to possess the appropriate Rf-genes that suppress or compensate for the adverse effects of the S. violaceum cytoplasm on male fertility and were characterized by normal anther functionality and morphology, high pollen viability (Figure 2c, Tables 3 and 4).Therefore, our study is the first to report the inheritance of the Rf-genes in the advanced BC progenies in eggplant functional CMS lines.
In the pollen non-formation eggplant lines with the cytoplasms of S. integrifolium, S. anguivi and S. grandifolium the fertility restoration was reported to be under the control of two independent duplicated loci (Hasnunnahar et al., 2012;Khan & Isshiki, 2010, 2011); whereas, Saito et al. (2009) reported that the CMS caused by the cytoplasm of S. grandifolium was restored by a single dominant allele.However, the segregation ratios observed in our study clearly deviated from those previously reported in the other eggplant CMS/Rf-genes systems.These differences may be attributed to the different cytoplasm and nucleus donor combinations used for the development of the previous eggplant CMS/Rf-genes systems.The presence of PMF plants in our segregating populations suggested a modifying gene action on the fertility restoration reported for the first time in eggplant functional CMS.Modifying gene action was also reported in some GMS and CMS/Rf-genes systems studied in the past.For example, the expression of anther dehiscence in GMS eggplant was suggested to be affected by environmentally sensitive modifiers (Hazra et al., 2008), showing an interesting analogy with our results.In addition, in the fertility restored CMS maize some contrasting segregation ratios observed in diverse environments were also attributed to modifier gene action (Duvick, 1956).
In our interpretation, the observed segregation ratios of the fertility phenotypes in the examined populations with the genetic backgrounds of cultivars 'Emi' and 'Tsakoniki' can be better explained by the action of one major genetic locus plus a duplicated modifier locus (Table 2; crosses 8 to30).In both genetic backgrounds, the presence of a dominant allele in the major locus (designated Rf 1 ) is essential for fertility restoration; thus, genotypes homozygous for the recessive allele (rf 1 rf 1 ) are male sterile, regardless the allelic constitution of the modifying loci.The presence of a dominant allele in both modifying loci (Rf m1 and Rf m2 ) is also needed for stable expression of male fertility.On the other hand, recessive homozygosity in one of these modifying loci seems to inhibit the expression of Rf 1 under certain conditions, resulting in the PMF phenotype (Table 2).According to the proposed model, S. violaceum and cv.'Tsakoniki' have the same allelic constitution in the second modifying locus (Rf m2 Rf m2 , Table 2; crosses 16 to 30).
The presence of the Rf-gene was confirmed in the BC 1 (V×E) 9 and BC 1 (V×T) 17 populations, but not in BC 1 (V×L) 2 .In addition, MF plants were obtained via backcrossing only in the advanced BC populations of F 1 (V×T) 16 and their selfing progenies.Besides, our results indicated a more complex genetic control of fertility restoration in BC 1 (V×E) 9 than in BC 1 (V×T) 17 , probably due to different allelic constitution in the second modifier locus (Table 2).While these findings seem rather inconsistent, it is important to note that MF plants could not be obtained in the CMS line of eggplant cv.'Uttara' with the cytoplasm of S. violaceum (Isshiki & Kawajiri, 2002).Perhaps, when diverse eggplant nuclear genomes are combined with the cytoplasm of S. violaceum, a different number of genetic loci and/or gene interactions are involved in fertility restoration, resulting in different segregation ratios.This is supported from studies in CMS rape-seed, rice and pepper where the gene interactions between the fertility restoring loci varied according to the restorer line used in the crosses (Ma et al., 2013;Pahwa, S. K. Banga, Gogna, & S. S. Banga, 2004;Sarkar, Zaman, & Singh, 2002).In addition, Hazra et al. (2008) reported a strong genotypic effect on the expressivity of the genes involved in eggplant GMS and fertility restoration.
The inconsistency observed in the inheritance of the fertility restoration between the three BC 1 populations could also be attributed to the special characteristics of each recurrent eggplant parent.For example, cv.'Langada' is considered to be a medium late cultivar and its anthesis occurs about 2 weeks later than cvs.'Emi' and 'Tsakoniki'.It is possible that in BC 1 (V×L) the action of the modifier(s) on the Rf-gene took place before the anthesis so all plants except one appeared to be MS.This may also apply to the backcross progenies of plant BC 2 (V×E) 11b where only MS plants were obtained, contrary to the expected 1PMF:1MS ratio (Table 2; cross 13).
In the case of BC 1 (V×E) the formation of parthenocarpic fruits may have masked the effect of low pollen viability and led to the miss-selection of PMF plants instead of the MF ones.For example, plant BC 1 (V×E) 9b was initially characterized as MF due to pollen release ability throughout the observation period, but the respective backcrossing segregation ratio suggested that it was probably a PMF one (Table 2; cross 11).In addition, it cannot be ruled out that some alleles involved in the fertility restoration were lost during the backcrossing process due to the minor effect on the fertility phenotype, relative small population size, environmental effects on phenotypic expression of the fertility (Duvick, 1956;Yoshimi et al., 2013) or linkage with the dwarf or chlorotic phenotypes mentioned in the results section.Nevertheless, further experiments should be pursued in order to incorporate the Rf-genes in the genetic backgrounds of cvs.'Langada' and 'Emi'.This process may be facilitated by the development of molecular markers linked to the Rf-genes.That was demonstrated to be applicable for the Rf-genes of the pollen non-formation eggplant CMS lines (Khan, Hasnunnahar, Iwayoshi, & Isshiki, 2014).
From the genealogy of the plant material and the direction of the crosses (Figure 1, Table 2), it can be derived that the dominant alleles responsible for fertility restoration were inherited from S. violaceum.Similarly, the recessive alleles responsible for the CMS were inherited from the eggplant parent.The fact that backcrossing of MS plants to the eggplant parent constantly produced MS progenies further supports this hypothesis suggesting the fixation of the eggplant alleles (Figure 1, Table 2).Moreover, the occurrence of MS plants in the selfing progenies of MF plants indicates that the CMS induced by the cytoplasm of S. violaceum is of the sporophytic type.Probably, these MF plants were heterozygous for the fertility restoring loci (Rf 1 rf 1 ) and the recessive allele was transmitted through the functional pollen grains (Table 2; crosses 26 to 28).The 3MF:1PMF segregation ratio observed in the population BC 2 (V×T)S 2 29c suggests that the essential Rf 1 allele was fixed (Table 2; cross 29).This, along with the high pollen viability of the MF plants (Table 4), denotes the possibility of obtaining lines homozygous for the Rf-genes through selfing, which is necessary for the development of self-maintained restorer lines.
The inheritance of cytoplasm organelles in angiosperms is mainly maternal with only some exceptions (Mogensen, 1996).In our study, analyses of cpDNA and mtDNA confirmed the maternal inheritance of the chloroplasts and mitochondria in the BC 4 and BC 3 S 1 progenies from S. violaceum (Figures 3 and 4).It seems that maternal inheritance of cytoplasm organelles is a common situation in the interspecific hybrids between eggplant and its wild relatives (Khan & Isshiki, 2009, 2010, 2011), although one case of biparental transmission of the chloroplast was documented (Khan & Isshiki, 2008).Moreover, the results obtained by HRM analysis, a fast molecular approach with a wide applications range (Ganopoulos, Madesis, Darzentas, Argiriou, & Tsaftaris, 2012;Madesis et al., 2012;Madesis, Ganopoulos, Bosmali, & Tsaftaris, 2013;Reed, Kent, & Wittwer, 2007) showed that this method could be used as an additional tool in similar experiments in the future.Overall cytoplasmic analysis data indicated that the eggplant cytoplasm was successfully substituted with the cytoplasm of S. violaceum by the backcrossing method.
The comparative study of three CMS eggplant lines developed in the present work may have some important implications for the development of CMS lines of the pollen non-release type.First, since the seed number per fruit of the CMS lines corresponded to that of the eggplant parents, it may serve as a criterion for the selection of eggplant cultivars prior to the development of CMS lines.Second, since the expression of male sterility/fertility is affected by the genotype of the nucleus donor (eggplant), it would be desirable to employ as many eggplant cultivars as possible in the interspecific crosses with the cytoplasm donor.Third, the populations segregating for the Rf-genes should be studied, if possible, in different climatic conditions or cultivating seasons in order to detect potential effects of modifier loci.Fourth, it would be interesting to ascertain if the Rf-genes discovered herein can restore the fertility in the rest of the pollen non-releasing CMS eggplant lines which lack the Rf-genes, in a similar way reported in the pollen-non formation CMS lines (Khan et al., 2014).In such a case, these pollen non-releasing CMS lines will be made available for eggplant hybrid seed production broadening the cytoplasmic sources of eggplant CMS, without the burden of re-developing the respective Rf-gene carrying lines.

Conclusions
The present study confirmed the suitability of S. violaceum as a cytoplasm donor for the development of functional CMS eggplant lines.The female fertility of CMS eggplant was not affected by the alien cytoplasm; however, the male fertility was negatively affected.The flowers of CMS plants were characterized by abnormal anther morphology, low pollen viability and prevention of anther dehiscence.Most importantly, the inheritance of the appropriate Rf-gene for this type of CMS was confirmed for the first time in eggplant functional CMS.Consistent inheritance of the Rf-gene was observed only in the advanced backcross progenies of CMS 'Tsakoniki'.The Rf-gene was dominantly inherited by the wild species and compensated for the negative effects of the alien cytoplasm, thus restoring the male fertility.In addition, a modifying gene action on the Rf-gene was observed for the first time in eggplant functional CMS.The development of molecular markers linked to the Rf-genes would facilitate the incorporation of these genes in various eggplant genetic backgrounds.It was established that the nuclear genotype of the eggplant parent affected the phenotypic expression of CMS in the cytoplasmic background of S. violaceum as well as the mode of fertility restoration.In conclusion, our findings contribute towards the development and understanding of a novel CMS/Rf-gene system in eggplant.
z cross numbers correspond to cross numbers given in Figure 1; y Superscripted numbers indicate cross number from which the maternal parent was obtained; letters (a, b and c) indicate different maternal parent of the same population; x MF = male fertile, PMF = potentially male fertile, MS = male sterile.Symbol ⊗ indicates self-pollination of individual MF plants; w Superscripted numbers represent cross number from which the examined population was obtained; letters (a, b and c) indicate origin from different maternal parent of the former population.

Table 1 .
Seed germination (in percent), fruit setting after backcrossing and number of seeds per fruit in three eggplant cultivars, their interspecific hybrids with Solanum violaceum (V) and the respective backcross populations