Background

Several tropical fruit species are underexploited but have promising economic potential for both industrial processing and in natura consumption. In Brazil, passion fruit is widely cultivated and the most recent survey on agricultural production showed that approximately 58,089 hectares were planted and 838,244 tons of fruit harvested [1]. Commercial plantations are almost exclusively based on a single species, the yellow passion fruit (Passiflora edulis), which is cultivated in over 90% of orchards. It is grown for its edible and aromatic fruits used in juice concentrate blends that are consumed worldwide. A second species, the sweet passion fruit (P. alata) is native to the Brazilian plateau and the eastern Amazonian region, but cultivated fairly widely only in the South and Southeast of Brazil, as reported in Cerqueira-Silva et al. [2]. It is fairly popular for its typical aroma and flavor characteristics, and therefore can fetch up to three times the price of the yellow passion fruit at local fresh fruit markets. Moreover, the vines of P. alata contain bioactive compounds [3] which have been widely used in a traditional medicine, and include a natural sedative, passiflorin.

Passiflora edulis and P. alata belong to the family Passifloraceae, order Malpighiales. Both are self-incompatible [4, 5], diploid species (2n = 18) [6, 7] with perfect, insect-pollinated flowers. Despite the importance of P. edulis as a fruit crop, long-term breeding programs are still in the early stages. On the other hand, the sweet passion fruit is gaining importance for its attractive flowers and exotic flavored fruits, but no commercial varieties are available to farmers. Furthermore, the incidence of diseases can inflict losses in plantations of both species. For instance, bacterial spot caused by Xanthomonas axonopodis pv. passiflorae (Xap) harms both leaves and fruits and has no effective chemical control [8]. It is therefore very difficult to manage.

Currently, plant research tends to incorporate molecular marker-based approaches into conventional breeding programs, including programs for understudied tropical species. In particular, molecular markers derived from DNA sequences characterized as microsatellites or simple sequence repeats (SSRs), and single nucleotide polymorphism (SNPs) located in plant genes have been developed [9, 10] and used to reach the objectives [11, 12]. With this in mind, we used the transcripts from a subtractive library enriched for P. edulis genes expressed in response to Xap infection [13] to develop putative functional SSR and SNP markers for P. alata.

Methods

Origin of the DNA sequences

The aim of the study was to develop a set of functional markers for P. alata from P. edulis transcript sequences induced and repressed in response to Xap, based on two SSH (suppression subtractive hybridization) libraries previously constructed in our laboratory. The forward (F) library was enriched for cDNA fragments of genes whose expression level increased following bacterial inoculation. Conversely, in order to isolate cDNA fragments of genes whose expression level decreased following inoculation, a reverse subtraction (R) was carried out with cDNA from mock-inoculated plants used as ‘testers’ and cDNA from inoculated plants used as ‘drivers’ [14]. We were able to isolate 683 transcripts from the forward (F) library, 274 from the reverse (R) library, and 41 transcripts referred to contigs of sequences derived from both libraries [13], giving a total of 998 sequences that were used for developing putative functional markers as described below.

Plant material and DNA isolation

Two accessions of P. alata and two of P. edulis were used to develop the SSR markers. These accessions were the respective parents of full-sib progenies used for mapping purposes in our laboratory [15, 16]. P. alata accession ‘2(12)’ belongs to a progeny of a wild ancestor that was collected in a locality between the Brazilian Amazon and Cerrado ecosystems [17]. P. alata accession ‘SV3’ is an indoor selection cultivated in the Southeast of Brazil. Both P. edulis accessions belong to the germplasm collection of The Agronomic Institute of Parana State, Brazil. ‘IAPAR-123’ is a selection from the Brazilian commercial population used for industrialized juice production and ‘IAPAR-06’ was introduced from Morocco.

All plant genomic DNA was extracted from young leaf tissue using the cetyltrimethylammonium bromide method adapted from Murray and Thompson [18]. For DNA quantification, samples were electrophoresed in 1.2% agarose gels stained with SYBR SAFE® (Invitrogen, Carlsbad, CA, USA), with 1× TBE as the running buffer, and compared to known concentrations of λ phage DNA. The gels were visualized under UV light and photo documented by a UVP MultiDoc-It digital imaging system. The original samples were diluted to 10 ng µ1−1.

After testing the ability of P. edulis SSR primer pairs to generate reproducible amplicons in P. alata, the loci that revealed polymorphisms between the ‘2(12)’ and ‘SV3’ accessions were used for genotyping 30 individuals of the segregating progeny (F1). For developing SNP markers, the same parental accessions of P. alata and two F1 genotypes (‘F1-67’ and ‘F1-100’) were used for prospecting sequence polymorphisms as explained below.

Microsatellite prospection and development of SSR markers

MISA software [19] was used to search for microsatellites in both the F and R libraries cited above. The following parameters were used to identify microsatellites: mono-, di-, tri-, tetra-, penta- and hexanucleotide repeats with a minimum of 10, 5, 3, 3, 3 and 3 subunits, respectively. SSRs were grouped according to motif, and each group included all the complementary and permuted sequences. SSR markers were developed from a set of genes selected from the F library [13].

Passiflora edulis selected transcripts containing SSRs were analyzed for primer designing. The PRIMER3 [20] and Gene Runner [21] computational programs were used. Next, amplification reactions were performed as follows: 30 ng genomic DNA, 1× PCR buffer, 0.25 mM dNTP, 0.3 µM of each primer, 2 mM MgCl2, and 1.2 U Taq DNA polymerase (Promega, Madison, WI, USA). Ultrapure water was added to make up the volume to 20 µl. Cycling consisted of an initial denaturing step of 5 min at 95 °C, 30 cycles of 40 s at 95 °C, 40 s at 55 °C, 60 s at 72 °C and a final extension of 8 min at 72 °C. PCR products were electrophoresed in 1.2% agarose gels stained with SYBR SAFE® (Invitrogen), with 1× TBE as the running buffer, and compared to the 100 bp DNA Ladder (Invitrogen). PCR conditions were optimized where necessary and adjusted depending on the amplification patterns. Note that tests at different primer concentrations combined with specific annealing temperatures were carried out.

For polymorphism evaluation, DNA samples were first denatured for 5 min at 95 °C and rapidly ice cooled. For each sample, 2.5 µl was loaded on a polyacrylamide gel (0.4 mm) and the products separated by denaturing electrophoresis. The electrophoresis was run in a vertical system (Sequi-Gen® GT Sequencing Cell, Bio-Rad, 38 × 50 cm gels). The gels contained 5% 19:1 acrylamide/bisacrylamide, 7 M urea and 1× TBE, and were run at 80 W for 2 h 30 min at 45° C. The glass plate with the gel firmly adhered to it was rinsed thoroughly in order to stain the DNA bands with silver nitrate (0.2%). Amplicons were visualized after adding sodium carbonate according to Creste et al. [22]. The gels were then analyzed under white light and digitized (Epson Expression 10,000 XL). The genetic configuration at each SSR locus was analyzed as proposed by Wu et al. [23].

Development of SNP putative markers

A total of 122 transcript sequences from P. edulis were selected for designing the primers (Additional file 1: Table S2). Primer sequence designing, amplification reactions and cycling programs, as well as amplicon electrophoresis were performed as described for SSR marker development. Amplifications using P. alata DNA that resulted in specific bands were purified using Illustra GFX 96 PCR Purification Kit® (GE Healthcare, Little Chalfont, BUX, UK). Concentrations of the purified products were estimated by spectrophotometry using a NanoDrop 2000® (Thermo Scientific, Waltham, MA, USA), and purification was checked by electrophoresis in 1.2% agarose gels stained with SYBR SAFE® (Invitrogen), with 1× TBE as the running buffer, and compared to phage λ DNA of known concentrations.

For sequencing reactions, the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Carlsbad, CA, USA) was used. Each reaction contained 50 ng of purified PCR product, 2 µl of 5× buffer, 0.6 µl of BigDye Terminator v3.1, 1.0 µl of each primer (forward or reverse), and ultrapure water to make up to a total volume of 10 µl. The following cycling program was used: 1 min at 96 °C, followed by 25 cycles of 10 s at 96 °C, 5 s at 50 °C and 4 min at 60 °C. Capillary electrophoresis was run on an ABI 3730 DNA Analyzer® (Applied Biosytems).

We routinely sequenced the positive strand (forward) of each genotype DNA but occasionally the negative strand (reverse) was also sequenced. Sequence data were processed using the CodonCode Aligner® 3.7.1 (CodonCode Corporation, Centerville, MA, USA), a base-calling software which assigns a quality phred-based score to each base [24]. Low quality bases (phred < 20) were trimmed. Sequences were then aligned with each other using the assemble function to identify polymorphisms. A single position in DNA sequences with a nucleotide variation was considered polymorphic. The parental genotypes and two F1 genotypes (F1-67 and F1-100) of P. alata were compared.

Results and discussion

We identified 210 SSRs in 998 P. edulis transcript sequences (21.04%), including 22.4% mono-, 23.3% di-, and 31.4% trinucleotide repeats. The remaining 11.9, 4.8 and 6.2% were tetra-, penta- and hexanucleotides. The nucleotide composition of the repeat motifs showed that A/T (98%), AG/CT (67.3%) and AAG/CTT (25.7%) were the most abundant among the mono-, di- and trinucleotides (Fig. 1). In terms of microsatellites (SSRs), 95% were perfect, 2% interrupted, 3% imperfect and 1.5% compound.

Fig. 1
figure 1

Percentage of mono-, di-, tri-, tetra-, penta- and hexanucleotides in the microsatellites found in transcribed sequences of Passiflora edulis (Passifloraceae); the percentage of the most common motif is shown for each case

Full size image

These results are in line with other studies on Malpighiales species, such as Salix spp. and Ricinus communis, in which trinucleotide repeats were the most abundant class in expressed sequences [25, 26]. Some other studies have shown different results, such as those on Manihot esculenta [27]. The variation found in SSR frequency is plausible based on the variation in sample size, search criteria, database size and tools used for the SSR development, as well as intrinsic differences between the species evaluated [28]. Trinucleotides and perfect microsatellites were expected to be predominant, since the SSRs examined were within coding sequences. In terms of nucleotide composition, our results are in line with those reported for Malpighiales, such as Populus euphratica [29] and Jatropha curcas [30].

In all, 41 out of 42 primer pairs designed from selected sequences of P. edulis resulted in amplification. Compared to P. alata, after optimizing PCR conditions, we were able to obtain good amplification patterns for most of the primer pairs (Fig. 2), revealing high transferability rates between P. edulis and P. alata of approximately 80.9% (34/42).

Fig. 2
figure 2

Amplification patterns of 7 SSR loci in Passiflora edulis (first two lanes, accessions ‘IAPAR-06’, ‘IAPAR-123’) and P. alata (third and fourth lanes, accessions ‘2(12)’, ‘SV3’) following agarose gel (1.2%) electrophoresis at 3 V cm−1 for 2 h. Lane L, 100 bp ladder

Full size image

Silva et al. [31] investigated the transferability of 24 primer pairs developed for P. edulis and 7 for P. alata to wild species of Passiflora (P. cacao, P. cincinnata, P. glandulosa, P. gibertii, and P. mucronata). The interspecies cross-amplification rate varied, and the lowest was 14.4% between P. alata and P. cincinnata. Transferability from P. alata and P. edulis to P. cacao was the highest, at 28.5 and 62.5% respectively. In general, the transferability of SSR loci was successful.

Our study shows that genic SSR locus transferability between P. edulis and P. alata is high. In fact, P. edulis and P. alata belong to the same subgenus (Passiflora), though to different botanical supersections (Passiflora and Laurifoliae respectively), and series (Passiflora and Quadrangulares, respectively). Interspecific hybrids of the two species are only 5% fertile [32].

As mentioned above, P. edulis and P. alata are self-incompatible species [4, 5] with an obligatory allogamous mode of reproduction, posing difficulties in the production of inbred lines, which are routinely used to generate hybrids that do not segregate. In the case of outcrossing species, which is enforced by self-incompatibility in P. edulis and P. alata, the F1 populations (or full-sib progenies) do segregate [15, 16, 23]. Therefore, after testing the ability of P. edulis SSR primer pairs to generate reproducible amplicons in P. alata, the loci that revealed polymorphisms between the parental accessions were used for genotyping individuals of the segregating progeny (F1) as reported below.

Herein, the PCR fragments obtained using 34 primer pairs were separated in denaturing polyacrylamide gels and 10 fragments revealed 15 polymorphic loci in P. alata (Table 1). Loci Pa_F-P5B01, Pa_F + R-Contig114, Pa_F-Contig13 and Pa_F-P8E05 showed four alleles. According to Wu et al. [23], these are the most informative loci for generating a high-resolution map, which would require four alleles segregating in a 1:1:1:1 ratio in F1 (when both parents are heterozygous for different alleles), and this configuration is possible only using SSR markers. For the remaining 6 loci, two alleles were identified (Table 1; Fig. 3).

Table 1 Genetic configuration of 10 SSR putative functional markers developed for Passiflora alata
Full size table
Fig. 3
figure 3

SSR locus (Pa_F-P5B01) segregating in an F1 population of Passiflora alata (N = 30). Parental accessions [‘2(12)’ and ‘SV3’] are shown in the first and second lanes. The locus configuration allowed the identification of three alleles (arrowed), and four genotypes. Polyacrylamide gel (1.2%) electrophoresis run at 80 W for 3 h

Full size image

It is worth noting that all SSRs developed herein are putative functional markers that can be deployed in future genetic studies. They are within gene sequences that encode for distinct proteins (Table 1) (Additional file 1: Tables S1, S3). This is the first report on the development of genic SSR markers for P. alata.

In terms of SNP prospection, 74 primer pairs complementary to sequences of P. edulis resulted in unique amplicons in P. alata after performing optimized PCR assays. As expected, a moderate transferability rate between the species was observed (~60%, 74/122).

We selected 37 out of the 74 transcripts to obtain the Sanger sequences of genotypes ‘2(12)’, ‘SV3’, ‘F1-67’ and ‘F1-100’. Comparing the nucleotides that were aligned with each other, it was possible to detect at least one SNP in 15 gene sequences. SNPs were detected on the positive strand of 14 gene sequences in all genotypes, and only one SNP was detected on the negative strand of the Pa_F-P5A12 gene sequence (Table 2). The absence of SNPs was observed in 13 gene sequences, but the remaining 9 gene sequences were of relatively low quality, ruling out any analysis.

Table 2 SNPs found in Passiflora alata putative genes, their position in the contig, and the SNP class classified as intronic (I) and exonic (E)
Full size table

Sequence lengths ranged from 332 to 872 bp, with an average of 625 bp. There were differences between the observed and expected lengths for all 15 gene sequences. Thirteen were longer than predicted and three shorter. We designed primer pairs based on P. edulis transcripts and used them to amplify P. alata genomic DNA.

A total of 34 SNPs were identified in the 15 putative gene sequences of P. alata. We then used BLASTN to compare our results with the cDNA sequences of P. edulis and identify the position of every SNP in the genes, and find out whether they were located within introns or exons (Table 2). The alignment results of the Pa_F-P8F01 gene sequence are shown in Fig. 4.

Fig. 4
figure 4

Alignment results for the Pa_F-P8F01 gene sequence showing an SNP at position 163

Full size image

SNPs were evenly distributed over coding and noncoding regions (i.e. 50%). Of the 17 SNPs found in exons, 13 (76.4%, 13/17) led to a change in the amino acid encoded, resulting in a 3.25:1 ratio (13/4) of non-silent to silent mutations. Five of the non-silent mutations were non-conservative, i.e. the nucleotide change resulted in the substitution of one amino acid for another with different properties that could impair protein function.

All nucleotide substitutions were biallelic. Transversions were slightly more common (53%, 18/34) than transitions (47%, 16/34). Transition mutations are expected to occur at higher frequencies than transversions. Furthermore, transitions are less likely to result in amino acid substitutions in protein sequences [33]. However, numbers can vary depending on sample size and selective pressure on the genes under analysis. For example, in Manihot esculenta, 52.6% (26,030/49,429) of SNPs were found to occur in coding sequences, 51.4% (40,561/78,854) were transition mutations and 48.6% (39,293/78,854) transversions [34], while in Hevea brasiliensis, 60% of SNPs (242,732/404,114) were transitions and 40% (161,382/404,114) transversions [35].

In P. alata, C/T and A/G transitions were predominant, with respective figures of 29.4% (10/34) and 17.6% (6/34). G/C transversions were predominant, accounting for 17.6% (6/34) of total transversions. Similarly, other studies have reported C/T or A/G as the most frequent transition mutations [35, 36].

It was possible to assign a putative function to six gene sequences. The remaining sequences (9) were found to be similar to hypothetical proteins (Additional file 1: Table S3). It is worth noting that the Pa_F-P9A06 locus sequence matches that of a lipoxygenase family protein (GenBank: ACS28586.1). Our group has recently reported that lipoxygenase 2 is highly implicated in P. edulis defense against X. axonopodis infection [13]. Similarly, the expression of the GhLOX2 gene was associated with the hypersensitivity response in cotton following inoculation with X. campestris pv. malvacearum [37]. It is therefore of interest to allocate the Pa_F-P9A06 locus to the linkage map of P. alata [16] as a beneficial complement to QTL (quantitative trait loci) mapping studies already performed in the same population of P. alata [38].

Overall, the 16 gene fragments generated 10,003 bp, so SNP frequency was estimated as approximately one SNP for every 294 bp. This frequency is lower than that found in a previous study (one SNP for every 177 bp) analyzing 7 genes in ‘2(12)’ and ‘SV3’ P. alata accessions [16]. These differences can be ascribed to the distinct set of genes analyzed. However, the SNP frequencies found herein are low in comparison to other outcrossing species [39,40,41]. Interestingly, the frequency of SNPs in P. alata putative genes is closer to that found in coding regions of self-pollinated species like Arabidopsis thaliana (1 SNP/336 bp) and rice (1 SNP/333 bp) [42, 43].

Finally, the polymorphism rates revealed by SSR and SNP loci were 29.4% (10/34) and 53.6% (15/28), respectively. The accessions did not contribute equally to polymorphisms. The contribution of the wild ‘2(12)’ accession was greater than that of the selected ‘SV3’ accession. In general, the polymorphism rates revealed by SSR markers were similar to those found in our previous studies using the same accessions of P. alata, with rates of 18% (308 AFLPs, amplified fragment length polymorphisms, and 20 SSR markers [44]), 20% (combination of 60 SSRs, 104 SNPs, 567 AFLPs, 12 TRAPs (target region amplification polymorphisms) and 23 RGAs (resistance gene analogs) [16]) and 35% (72 SSR markers [45]). Studying two distinct accessions of P. edulis, polymorphism rates were 26.4, 13.0 and 24.7% based on 113 RAPD (random amplified polymorphic DNA) [46], 174 AFLP [47] and 21 SSR loci [15], respectively. All these rates are considered low in view of the high levels of molecular polymorphism found in outcrossing species, like chicory (69,3%, based on 730 SSR markers [48]), cocoa (59%, based on 227 SSR markers [49]) and apple (50.5%, based on 338 SSR markers [50]).

Conclusions

We were able to use gene sequences of Passiflora edulis to develop a set of genetic markers for P. alata, which can be used for genetic mapping purposes and also in diversity studies. A high to moderate level of transferability occurred between species (80.9% for SSR and 60.6% for SNP markers), providing tools for detecting polymorphic loci. We have provided additional information on the abundance of SSRs and SNPs in gene sequences of P. alata, and also confirmed the occurrence of low levels of molecular polymorphism in this outcrossing species.