Introduction

Momordica charantia L. (2x = 2n = 22), commonly known as bitter gourd or bitter melon, is an economically important cucurbitaceous vegetable cultivated in India, China, Malaysia, Africa, and South America [1]. The place of its origin is probably India with secondary center of diversity in China [2]. Compared with other cucurbits, it is highly valuable for its nutritive content of proteins, carbohydrates, vitamins, and minerals mainly ascorbic acid and iron [3, 1], as well as varied medicinal properties [4]. The fruit and juice have been used as traditional medicine for treating diabetes [5], as a remedy for curing anemia, jaundice, malaria, and cholera [6] and for inhibiting breast cancer cell proliferation [7].

Indian bitter gourd provides immense phenotypic variation based on various characters such as growth habit, maturity, fruit shape, size, color, and surface texture [8] and sex expression [9]. Compared with morphological markers, DNA-based molecular markers are more useful for genetic characterization and diversity assessments. Various multi-locus dominant DNA markers such as RAPD [10, 11], ISSR [12], and AFLP [13] have been reported for genetic analyses of bitter gourd. Microsatellite or simple sequence repeat (SSR) markers have gained considerable importance in plant genetics and breeding owing to many desirable attributes like their multi-allelic nature, co-dominant transmission, extensive genome coverage, small amount of starting DNA, and ease of detection by polymerase chain reaction (PCR) [14]. However, the number of microsatellite markers available in bitter gourd is few. Of the 70 SSR markers reported, 16 have been developed using FIASCO technique [15, 16], 11 through genomic library enrichment [17], and 43 through cross-species transferability from other cucurbits [18–20]. SSRs are known to have high hetrozygosity values and are more informative than dominant DNA markers [14]. Although the initial cost of SSR marker development is high, once developed, they are highly repeatable and, consequently, easily transferred across laboratories [21]. Notwithstanding their suitability in various genetic analyses, microsatellite markers provide accurate results with a minimum number of loci/alleles employed in the study. For instance, 350 to 400 alleles were proposed to be needed to detect genetic relationships among common wheat varieties [22]. It is established that greater number of markers are necessary for the development of genetic map and marker-assisted selection [23]. Considering the decisive advantages of having greater number of SSR markers in bitter gourd, we report in this study the development and characterization of 160 new microsatellite markers in M. charantia through a cost-effective strategy of genomic library enrichment. The utility of these markers in analysis of intra-specific genetic diversity, as well as cross-species transferability to six species of Momordica is also demonstrated.

Materials and Methods

Construction of SSR-Enriched Genomic Libraries

Genomic DNA was extracted from fresh leaves of the variety Pusa Vishesh using a modified CTAB method [24]. A small insert genomic library enriched for ten different repeat motifs was constructed with minor modifications [25]. The 10 μg of genomic DNA was nebulized and size-fractionated on 1 % agarose gel to elute fragments in the range of 0.5–1 kb. These were end-polished and ligated to adaptors OG-1 (5′-CTCTTGCTTAGATCTGGACTA-3′) and OG-2 (5′-TAGTCCAGATCTAAGCAAGAGCACA- 3′) using T4 DNA ligase. Adaptor-ligated DNA was PCR-amplified at 94 °C for 5 min, 35 cycles of 94 °C for 45 s, 55 °C for 45 s, and then 72 °C for 1 min with a final extension of 72 °C for 5 min. The amplified DNA was denatured and hybridized to 5’-biotinylated repeat oligonucleotides: (GA)20, (CA)20, (AT)20, (CAT)15, (AGA)15, (GAT)15, (AAT)15, (AAC)15, (AAAT)15, and (AAAG)15. Microsatellite-containing fragments captured by streptavidin-coated magnetic beads (New England Biolabs, Inc., USA) were washed twice with 6× SSC and 0.1 % SDS at 65 °C and room temperature. The various fragments enriched for different repeats were pooled and ligated into pGEM-T Easy vector (Promega, USA) and transformed into DH5α Escherichia coli electro-competent cells. Subsequent to selection on IXA (IPTG, X-gal, and ampicillin) plates, 3,072 white colonies were picked and inoculated in 1 ml LB broth with ampicillin and incubated overnight at 200 rpm and 37 °C. A 50-μl aliquot of each clone was stored as a glycerated culture at −80 °C. Plasmid DNA isolation was followed as per manufacturer’s protocol (Wizard miniprep, Promega, USA).

PCR Screening and Sequencing

Identification of clones containing microsatellite inserts was performed as described previously with minor modifications [26, 27]. One primer complementary to adaptor OG-1 and one non-biotinylated repeat primer used for library enrichment were used for PCR amplification. PCR reaction was carried out in a total volume of 10 μl with 1× Taq buffer, 2 mM MgCl2, 0.25 mM dNTPs, 0.25 μM of each primer, and 0.5 U Taq DNA polymerase (MBI, Fermentas), cycled at 94 °C for 5 min, 35 cycles of 94 °C for 1 min, 55 °C for 1 min, 72 °C for 1 min, and final extension of 72 °C for 5 min. Clones containing two or more amplified fragments were identified by agarose gel (1.5 %) electrophoresis and were sequenced employing M13 universal primers.

Microsatellite Mining, Primer Design, and Validation

The sequence data were mined for microsatellites using the software Simple Sequence Repeat Identification tool [28]. The parameters for detection were a minimum total array size of eight nucleotides [29]. Thus, a minimum number of four and three repeats were considered for di- and trinucleotide motifs, respectively, and that of two repeats for penta-, hexa-, hepta-, and octanucleotide repeats. The identified microsatellites were characterized as perfect, compound, and interrupted (both simple interrupted and compound interrupted). Primers were designed flanking the microsatellite regions using Primer3 software [30] based on the following parameters: G + C content of ∼50 %, primer length (18–24 bases), annealing temperature of the primer in 50–65 °C range, and product size in the range of 150–300 bp. The primer pairs were validated for amplification of the expected size of allele using the total genomic DNA of Pusa Vishesh by optimizing for the annealing temperatures (Table 1).

Table 1 Primer sequences and associated information for the novel 160 SSR loci developed in the study
Full size table

Genetic Diversity Analysis and Cross-Species Amplification

To demonstrate the efficacy of the newly developed primer pairs, 51 of these were employed for genetic analysis of 54 bitter gourd genotypes (Table 2). Genomic DNA (10 ng) was amplified in 20-μl reaction mixture consisting 1× Taq buffer, 2 mM MgCl2, 0.25 mM dNTPs, 0.25 μM of each primer, and 0.5 U Taq DNA polymerase (MBI, Fermentas) in a thermocycler (Biometra). The PCR conditions were as follows: initial denaturation at 94 °C for 4 min, 30 cycles of denaturation at 94 °C for 40 s, annealing at experimentally determined annealing temperature of each primer pair (Table 1) for 40 s, and then extension at 72 °C for 40 s followed by final extension at 72 °C for 5 min. The amplified fragments were separated on 4 % metaphor gel and visualized by ethidium bromide staining. These primer pairs were also tested for their cross-species transferability potential on a representative panel including one accession each of Momordica cymbalaria, Momordica subangulata subsp. renigera, Momordica balsamina, Momordica dioca, Momordica cochinchinesis, Momordica sahyadrica, and M. charantia as control. PCR and electrophoresis conditions were the same as described above. The data for diversity analysis of 54 bitter gourd genotypes at each locus were scored as present (1) or absent as (0). The Free Tree software version 0.9.1.50 [31] was used to calculate the Jaccard’s similarity coefficients [32], and the resulting matrix was subjected to UPGMA cluster analysis. Similar analysis was done for a subset of 15 released Indian varieties of bitter gourd. Parameters of genetic diversity including expected heterozygosity (H e) and the observed heterozygosity (H o) were calculated using POPGENE version 1.3 [33]. The polymorphic information content (PIC) was calculated for each locus using the formula, PIC = 1 − ∑p i 2 [14], where p i is the frequency of the ith allele.

Table 2 List of 54 bitter gourd genotypes used for genetic diversity analysis
Full size table

Sequencing for Validation of Alleles

Amplicons obtained in seven bitter gourd genotypes (including Pusa Vishesh) and seven Momordica species showing allelic variation for the locus McSSR_54 (ATCAT)2 were gel-eluted, purified, and sequenced. The sequences were compared using ClustalW multiple sequence alignment tool [34]. Sequence analysis was also done for the locus McSSR_20 (TCT)9 that showed allelic variation in another set of nine bitter gourd genotypes.

Results

Microsatellite Isolation and Characterization

A small insert genomic library of bitter gourd enriched for ten different repeat motifs was constructed. The repeats employed in the study were three di-repeats: (GA)20, (CA)20, (AT)20; five tri-repeats: (CAT)15, (AGA)15, (GAT)15, (AAT)15, (AAC)15, and two tetra-repeats: (AAAT)15, and (AAAG)15. The library comprised of 3,072 clones, and upon PCR-based screening, 582 clones putatively containing microsatellite repeats were identified. The presence of a microsatellite repeat was indicated by the presence of two bands when amplified using one primer as the repeat used for library enrichment and the other as OG-1. High-quality sequence data were obtained for 466 clones (80 % sequencing efficiency). The sequence information is available for these in NCBI Genbank (accession numbers KC905911 to KC906159 and KC918559 to KC918775).

Sequence analysis with the software SSRIT showed that 437 of the clones sequenced contained in all 1,939 microsatellite repeat stretches of a minimum of eight nucleotides (dinucleotide core repeated four times, trinucleotide core repeated three times, and tetra-, penta-, hexa-, hepta-, and octanucleotide cores repeated twice). These simple sequence repeat stretches comprised 1,814 (93.55 %) simple, three (0.15 %) simple interrupted, 38 (1.95 %) compound, and 84 (4.33 %) compound interrupted repeats. Tetranucleotide repeats were the most abundant class of microsatellites accounting for 1,087 (56 %) of the repeats. These were followed by 335 pentanucleotides (17.2 %), 182 hexanucleotides (9.38 %), 164 trinucleotides (8.45 %), 113 dinucleotides (5.82 %), 47 heptanucleotides (2.4 %), and 11 octanucleotide (0.56 %) repeat containing microsatellites, respectively.

Six dinucleotide repeat motifs, GA/CT, AG/TC, AT/TA GT/CA, AC/TG, and CG/GC were found 40, 35, 24, 6, 4, and 4 times, respectively. Twenty-seven trinucleotide repeat motifs, AGA/TCT, CTT/GAA, AAC/TTG, AAG/TTC, and CAA/GTT were found 26, 19, 13, 12, and 10 times, respectively. As many as 118 tetranucleotide motifs were identified of which AAAT/TTTA, ATTT/TAAA, AAAG/TTTC, AATA/TTAT, CTTT/GAAA, and AGAA/TCTT were found 40, 38, 37, 37, 31, and 30 times, respectively. An assorted penta-, hexa-, hepta-, and octanucleotide repeat motifs were identified and found 175, 123, 34, and 11times, respectively.

Primer Designing and Validation

Based on the availability of sufficient length of sequences flanking the repeat region in the clone and compatible melting temperatures, 160 primer pairs (Table 1) were designed from 437 of the microsatellite containing clones (primer design efficiency was 36.6 %) in the enriched genomic library of bitter gourd. The designed primers were validated for PCR amplification using the genomic DNA of the variety Pusa Vishesh, which was used for library construction. The PCR amplification efficiency of the designed primers was 94.4 %, as sharp discrete fragments of expected size were obtained for 151 primer pairs.

Genetic Diversity Analysis with Newly Developed Microsatellite Markers

Genetic diversity among 54 bitter gourd accessions was evaluated by amplifying markers using 51 of the above developed primer pairs. A representative amplification profile is given in Fig. 1. A total of 68 alleles were amplified, ranging in size from 127 to 300 bp of which 28 (41.17 %) that were generated by ten primer pairs were polymorphic. The number of alleles at these loci ranged from two to five with an average of 2.80 alleles per locus. The indices of utility of these loci are given in Table 3. The observed heterozygosity (H o) ranged from 0.042 to 0.587 (average = 0.106), and the expected heterozygosity (H e) ranged from 0.140 to 0.783 (average = 0.373). PIC values ranged from 0.139 to 0.775 (average = 0.369).

Fig. 1
figure 1

Amplification of alleles of 214 and 205 base pairs at the locus McSSR_20 in 54 bitter gourd genotypes. M is the 50 bp DNA ladder, lanes 1–54 bitter gourd genotypes as appearing in Table 1

Full size image
Table 3 Characteristics of ten polymorphic SSR loci
Full size table

Jaccard’s similarity coefficient-based UPGMA cluster analysis of the SSR marker data revealed three major clusters and three small groups of two to three accessions each (Fig. 2). First major cluster of 26 accessions comprised eleven landraces and nine released varieties. The second major cluster comprised mostly of the exotic Asian germplasm, i.e., from China (two accessions), Taiwan (five accessions), Philippines (three accessions), Pakistan, and Indonesia (one accession each) and two released varieties, Meghana-2 and NDBT-9. The third cluster contained five varieties, three landraces, and three exotic accessions. The genotypes EC620354, Selection-5, EC620330, and EC620332 were out-grouped from the rest of the accessions. The clustering pattern was not in consonance with the geographical distribution of the genotypes used in the study. The maximum similarity value (0.9818) was observed between indigenous genotypes, DBG-8 and DBG-34, and minimum similarity was seen between DBGy-201 and EC620330. A UPGMA dendrogram based on Jaccard’s similarity coefficient was constructed for a subset of 15 released varieties (Fig. 3). All the 15 varieties were discriminated with the maximum similarity (0.96) between the varieties Hirakani and Phule Ujjwala, and, the varieties HABG-22 and Kalyanpur Baramasi were the most distant varieties (0.75).

Fig. 2
figure 2

UPGMA-based dendrogram of 54 bitter gourd genotypes based on 51 STMS loci

Full size image
Fig. 3
figure 3

UPGMA-based dendrogram of 15 bitter gourd varieties based on 28 polymorphic STMS markers

Full size image

Marker Transferability to Other Momordica Species

Cross-species amplification was tested in 51 microsatellite markers in six different Momordica species: M. cymbalaria, M. subangulata subsp. renigera, M. balsamina, M. dioca, M. cochinchinesis, and M. sahyadrica. One accession of each species (including that of M. charantia as control) was used for detecting cross-species transferability potential of these markers (Fig. 4). Forty SSR primer pairs (78.43 %) designed from M. charantia sequences were transferable to other species while 11 primer pairs failed to amplify in any of the six Momordica species. Whereas 12 of these 40 primer pairs gave amplification products in all the six Momordica species, six primer pairs amplified in five species, four primer pairs in four species, five primers pairs in three species, four primer pairs in two species, and nine primer pairs in only one of the six species included in the study (Table 4). Transferability of the 40 SSR markers was the most to M. balsamina (30), followed by M. sahyadrica (29), M. cymbalaria, and M. dioca (27), M. subangulata subsp. renigera (24), and least in M. cochinchinesis (13). UPGMA analyses based on Nei’s genetic distance values (Fig. 5) generated a dendrogram with two major clusters with M. cymbalaria as an outlier. Maximum similarity (0.5483) was observed between M. subangulata subsp. renigera and M. dioca and minimum (0.2321) between M. cymbalaria and M. charantia.

Fig. 4
figure 4

Amplification of alleles of 173 and 165 bp at the locus McSSR_54 in seven Momordica species. M is the 100-bp DNA ladder, lanes 1 M. cymbalaria, 2 M. subangulata subsp. renigera, 3 M. balsamina, 4 M. dioca, 5 M. cochinchinesis, 6 M. charantia, 7 M. sahyadrica

Full size image
Table 4 Cross-species transferability of SSR markers derived from M. charantia
Full size table
Fig. 5
figure 5

UPGMA dendrogram based on Nei’s genetic distance values for Momordica species

Full size image

Validation of Allelic Variation by Sequencing

Amplicons of sizes 173 and 169 bp were amplified with the primer pair McSSR_54. Sequence comparison of the 173-bp fragment eluted by amplification of the genotypes Pusa Vishesh, Arka Harit, HABG-22, Phule Green Gold, and Kalyanpur Baramasi, and the169-bp fragment eluted from the genotypes Priya and Kashi Urvashi revealed that the polymorphism was due to a 4 bp (CATA) indel falling in the repeat motif (ATCAT)2 at nucleotide position 126 (Fig. 6a). Sequence comparison of the fragments amplified at the same locus (McSSR_54) across Momordica species showed an 8 bp (CATATCAT) indel at position 121 in M. charantia that is a part of the (ATCAT)2 repeat motif that was absent in all the other six species (Fig. 6b). Sequence analysis of the 214 bp fragment from the genotypes Pusa Vishesh, DBG-7, DBGy-201, EC620341, and EC620356, and 205-bp fragment from the genotypes DBG-3, DBG-9, Preeti, and EC620326 at the locus McSSR_20 revealed a 9 bp (TTTCTCCAT) indel in DBG-3, DBG-9, Preeti, and EC620326 from nucleotide positions 87–95 in the flanking region of the (TCT)9 microsatellite repeat (Fig. 6c).

Fig. 6
figure 6figure 6

Sequence comparison between alleles amplified at microsatellite loci McSSR54 and McSSR_20. The repeat motifs are represented as bold letters and underlined. Asterisk indicates similar sequences, and dash indicates alignment gaps. Shaded box indicates base insertions/deletions. Left and right arrows indicate primer sequences. a Sequence alignment of alleles amplified using McSSR_54 from bitter gourd varieties: Pusa Vishesh, Arka Harit, Phule Green Gold, Kalyanpur Baramasi, HABG-22, Priya, and Kashi Urvashi. b Sequence alignment of alleles amplified using McSSR_54 from Momordica species: M. charantia var. Pusa Vishesh, M. dioca, M. cochinchinesis, M. sahyadrica, M. subangulata subsp. renigera, M. cymbalaria, and M. balsamina. c Sequence alignment of alleles amplified using McSSR_20 from bitter gourd genotypes: Pusa Vishesh, DBG-7, DBGy-201, EC620341, EC620356, DBG-3, DBG-9, Preeti, and EC620326

Full size image

Discussion

The inherent properties of microsatellite markers make them the most preferred type of markers for various biotechnological applications [35]. Efforts to isolate microsatellite markers in bitter gourd have been few. SSR markers have been developed in bitter gourd using different isolation strategies such as Fast Isolation of by AFLP of sequence-containing repeats (FIASCO) technique [15, 16] and genomic library enriched for a single trinucleotide (GAA)10 probe [17]. Small insert genomic library enriched for different microsatellite repeats is an effective method for targeting many repeat-containing loci [25]. The large number of clones to be sequenced while generating microsatellite markers through construction of enriched genomic libraries is what makes this method more expensive than mining them from databases or transferring them from related species. PCR-based methods to screen for microsatellite containing clones from enriched genomic libraries [27] are a cost-effective means as the number of clones to be sequenced is greatly reduced. The clones that produce two or more bands when screened with the adaptor primer and a repeat-specific primer are considered likely to contain microsatellites [26]. Based on this strategy, the number of clones to be sequenced from the enriched genomic library of the variety Pusa Vishesh was reduced by 81 %, i.e., from 3072 to 582. Quality sequences were obtained for 466 clones of which 437 (93.7 %) contained microsatellite repeats. The detection of a higher percentage of microsatellite-positive colonies following PCR screening makes this a quick and cost-effective method for microsatellite identification.

This is the first report in the genus Momordica where an attempt has been made to develop microsatellite markers from a genomic library enriched for ten different repeat motifs. Most simple repeats and their complementary counterparts can be represented by several basic unit patterns [36]. The dinucleotide motifs used for enrichment in the present study were (CA)20, (GA)20, and (AT)20 which represent all the possible combination of two bases, not considering mononucleotide sequences. The 96 % of the 113 detected dinucleotide motifs showed specific complementarity to these oligoprobes [66 % to (GA) probe, 21 % to (AT) probe, and 9 % to (CA) probe]. Four microsatellites had the motif CG that was not used for enrichment. The relative abundance of (GA/CT)n repeats in bitter gourd followed by the (AT/TA)n and (CA/GT)n repeats is in accordance with that reported for many plant species [37], including related cucurbits [38, 20]. Amongst the 164 trinucleotide repeats detected, 122 (74.4 %) showed sequence complementarity to the five trinucleotide repeat motifs used for library enrichment with those complementary to the (AGA/TCT)n being predominant as in other plants [39, 40], including Curbita pepo and Cucumis sativus [41–43]. As in cucumber, tetranucleotide repeat motifs represent the most abundant (56 %) microsatellites found in the present study. The 53 % of the tetranucleotide repeats were AT-rich followed by 31 % having AT = GC and 16 % being GC-rich. The 24.3 % of the 1,087 tetranucleotide motifs detected showed sequence complementarity to the AAAT and AAAG repeats used for enrichment. These were also the most abundant tetranucleotide motifs, as also in cucumber genomic SSRs [43]. Even though three di-, five tri-, and two tetranucleotide repeat motifs were used for library enrichment, surprisingly the second most abundant repeat motifs found in bitter gourd were the pentanucleotide repeat motifs (335), followed by hexa- (182), tri- (164), di-(113), hepta- (47), and octanucleotide (11) repeats. Comparison of microsatellite repeat motifs across eight plant species shows that the tetranucleotide or trinucleotide repeats are most prevalent followed by the dinucleotide repeat motifs [43]. The few numbers of dinucleotide motifs detected in the present study may be attributed to the high-stringency washes during the enrichment with the dinucleotide motifs or PCR bias during screening. The 71 % of the pentanucleotide repeats detected in the present study were AT-rich, and 29 % were GC-rich. Of the 182 hexanucleotides identified in the present study, 56.11 % were AT-rich followed by 27.22 % with AT = GC, and 16.66 % were GC-rich. Prevalence of hexanucleotides repeats in the exons has been reported in different taxa [44]. In cucumber, a higher percentage of these repeats is present in EST database as compared with genomic sequence [43]. Identification of hexanucleotide repeat motifs, therefore, could be of great significance as they may be present in the protein coding region and be responsible for differential expression or regulation of genes which govern important traits in this crop. The heptanucleotides and octanucleotide repeat motifs that were identified amongst the sequenced clones were mostly complementary to the tetranucleotide probes used for enrichment or were a part of compound interrupted repeats identified.

On the basis of taking eight nucleotides as a threshold for screening microsatellite repeats [29], a majority (93.6 %) of the 1,939 microsatellites identified in the present study were simple repeats. This is similar to what has been reported in sugarcane, pumpkin, wheat, and grass pea [41, 45–47] and far greater than 34 % and 41 % reported in Cucumis melo [38, 48]. Perfect repeats have higher mutation rates as compared with imperfect repeats and, hence, increase variability and evolutionary rates [49]. These loci are therefore more useful for primer development for diversity analysis studies.

Microsatellites can be easily converted into PCR-based assays promoting their widespread use as molecular markers for genetic diversity analysis [50]. Primer pairs were designed for 160 new STMS loci after considering the factors such as little or no flanking regions surrounding the microsatellite regions and melting temperature (Tm) compatibility (±1 °C). The primer design efficiency of 32.5 % was higher than 21 % and 27 % of the microsatellite containing clones in wheat and sugarcane [51, 52]. Amplification conditions were optimized at 151 loci, resulting in an amplification efficiency of 94.4 %. The 51 primer pairs were used for genetic characterization of 54 bitter gourd genotypes and 20 % (10) primer pairs detected polymorphism with number of alleles ranging between 2 and 5. A total of 28 polymorphic markers were generated resulting in 41.17 % average percentage polymorphism which was higher than 36.5 % obtained by RAPD markers [10] and lower to that obtained by ISSR (74.7 %) and AFLP (78.5 %) markers in bitter gourd [12, 13]. The average PIC value of STMS markers was 0.369 which is higher than 0.22 obtained by AFLP markers [13] pointing towards the efficiency of these markers in genetic analysis of the bitter gourd germplasm. The PIC values of the polymorphic primers designed in the study ranged from 0.139 to 0.775, respectively. These are similar to previously reported in bitter gourd and melon [17, 19]. These newly developed markers are thus efficient in evaluating genetic diversity in bitter gourd germplasm and highly useful for constructing SSR-based genetic map in bitter gourd.

The informativeness of microsatellites tends to increase with increase in number of repeats [53]. This was observed for the locus McSSR_20 with the core repeat (TCT)9 which had a high PIC value of 0.775. However, the markers at the loci McSSR_47, McSSR_55, and McSSR_56 with repeat number n = 4 and at McSSR_54 with repeat number n = 2 were equally capable of detecting polymorphism in bitter gourd accessions, revealing that SSR length is not necessarily a benchmark for detecting polymorphism [54].

The UPGMA-derived dendrogram based on the Jaccard’s similarity clustering pattern generated three major clusters and three small clusters with two to three genotypes each. Although the clusters could not be defined on basis of the morphological characters (data not shown), most exotic and indigenous accessions fell in two distinct groups (Fig. 2). The PIC (0.139 to 0.775) and expected heterozygosity (0.140 to 0.783) values of these newly developed co-dominant SSR markers indicate their usefulness in solving issues related to genotype identity, variety protection, and hybrid seed purity testing. An analysis of a subset of 15 Indian-released varieties based on these SSR markers indeed proves the efficiency of the 28 polymorphic markers amplified at ten loci in discriminating these varieties from each other (Fig. 3). These newly developed SSR markers would benefit in discriminating the varieties during plant variety protection processes.

Sequence comparison of the alleles amplified in different varieties at the locus McSSR_54 shows that the allelic variation of 4 bp leading to a 169 bp fragment in the varieties Priya and Kashi Urvashi when compared with Pusa Vishesh was due to a indel of the bases CATA in the repeat motif (ATCAT)n (Fig. 6a) whereas allelic variation at the same locus across the species showed an indel of 8 bp in all the six Momordica species when compared with M. charantia var Pusa Vishesh (Fig. 6b). The allelic variation at this locus may be due to indel slippage which is length independent of the repeat array and is thought to operate at repeats with few repetitions at a constant rate [55]. Comparison of the sequences amplified at locus McSSR_20 in nine genotypes (Fig. 6c) shows that the variation between the alleles is due to 9-bp indel in the flanking region of the (TCT)9 repeat, leading to a 205-bp fragment in genotype Preeti, DBG-3, DBG-9, and EC620326. Indels in the flanking regions have been reported to cause interspecific and intraspecific variations at the microsatellite loci in plants and animals [56–59].

DNA sequences flanking microsatellite motifs are often conserved, and SSR markers are highly transferable among species and genera [60, 54]. Transferability of SSR markers to species within the same genus and to different genera of the same family has been reported in many crops such as Arabidopsis, rice, wheat, sorghum, soybean, sugarcane, and oat [58, 61–66]. SSR markers developed in bitter gourd have been successfully transferred across species and genera [15–17]. SSR markers at 40 of the 51 tested loci were transferable to six other Momordica species, i.e., M. cymbalaria, M. subangulata subsp. renigera, M. balsamina, M. dioca, M. cochinchinesis, and M. sahyadrica. The transferability of the bitter gourd markers to various Momordica species ranged from 32.5 % to 75 % which is much more than the across-genera transferability of markers from bottle gourd and cucumber [18, 20]. Cluster analysis of seven species of Momordica based on these 40 loci show that the dioecious species: M. sahyadrica, M. dioca, M. subangulata subsp. renigera, and M. cochinchinesis formed one major cluster, and while the monoecious species M. charantia and M. balsamina grouped together, M. cymbalaria was distinct from them. These results based on 40 primer pairs are in accordance with the relationship between these Indian species based on morphological [67] as well as molecular (21 RAPD and 12 ISSR primers) markers [68]. The newly developed co-dominant markers, reported here, are thus an important resource not only for intra-specific genetic analyses, but also in discerning species relationship within the genus Momordica.

Conclusions

A novel set of informative microsatellite markers for bitter gourd is provided. One hundred sixty novel microsatellite markers have been developed in M. charantia through sequencing of small insert genomic library enriched for ten different repeat motifs. Evaluation of a small set of these markers for assessment of genetic diversity among 54 bitter gourd genotypes indicates their potential in genetic analysis of bitter gourd for mapping, variety protection, and hybrid seed purity testing. These markers are highly transferable to six other Momordica species and may be used as an efficient tool in phylogenetic and comparative studies among the Momordica species. Therefore, the newly developed M. charantia microsatellite markers are of great significance as they add up to the few available SSR markers in bitter gourd. These are a valuable resource for genetic characterization of bitter gourd genotypes and accomplish the desire for development of large number of SSR markers for constructing SSR-based genetic map in bitter gourd and related species.