Accumulation of interspersed and sex-specific repeats in the non-recombining region of papaya sex chromosomes

Composition of interspersed repeats in the sex determining region of papaya

Long terminal repeats (LTRs) accounted for more than 97% of all identifiable retroelements in all three sources of sequences and the Ty3-gypsy element was the most abundant LTR in the HSY and the corresponding X, whereas Ty1-copia element was more abundant in V. monoica (Table 1). The number and sequences of Ty3-gypsy elements increased notably along the increase of sequence length in the HSY and the corresponding X (Figure 1A and B). To examine the portion of papaya-specific repeats accounting for interspersed repeats, the HSY and the corresponding X sequences were also masked by only known repeats, consisting of Repbase and TIGR repeats excluding papaya repeats. The known repeat content was approximately 19.5% in the HSY; 2% higher than 17.5% in the corresponding X (Table 1). As a result, papaya-specific repeats were at least 57.8% and 46.3% in the HSY and in the corresponding X, respectively.

Although the HSY and the corresponding X were highly occupied by interspersed repeats, there were potential gene rich regions with significantly low repeat accumulation. Two large sequence blocks with scarce or no repeats were detected from 1.8 to 2.2 Mb and from 3.2 to 3.7 Mb in the corresponding X [21], whereas only one large block with low repeat content was found from 4.6 to 5.3 Mb region in the HSY [21].

Identification of sex-specific repeats in the papaya sex determining region

To examine the localization of the sex-specific repeats in the sex determining region, all positions aligned with the sex-specific repeats were plotted to their corresponding locations in the HSY (Figure 2A) or in the corresponding X (Figure 2B). Most HSY-specific repeats were located in two regions in the HSY, from 2.0 to 4.0 and 5.0 to 7.5 Mb (Figure 2A), but rarely found in the corresponding X except for X-R55 (Figure 2B), an X-specific repeat. Remarkably, the two regions in the HSY with high HSY-specific repeats were matched to two HSY expansion regions very well [21].

Among the 21 potential sex-specific repeats, two HSY-specific repeats, HSY-R29 and HSY-R162, and one X-specific repeat, X-R55, were selected for further analyses. Hermaphrodite specificity of both HSY-specific repeats was confirmed by PCR using genomic DNA samples as templates. Both repeats were confirmed to be present only in SunUp hermaphrodite plants having both the hermaphrodite Y^h and X chromosomes, but not in SunUp females having two X chromosomes (Figure 3A). Since all HSY-specific repeats were present in more than 10 copies (Additional file 3: Table S2), it was of interest to examine the phylogenetic relationship among copies of each HSY repeat. Twenty-two aligned sequences to HSY-R29 and thirteen aligned sequences to HSY-R162 were retrieved from the HSY sequences for phylogenetic analysis. Phylogenetic analysis revealed that there was no correlation between distance and sequence identity among copies (Figure 3B and C), which was further confirmed by the Mantel test (Additional file 4: Figure S1). The correlation coefficient of the Mantel test (Rxy) and the one-tailed p-value (rxy-rand ≥ rxy-data) were 0.014 and 0.511 for the HSY-R29 and -0.033 and 0.378 for the HSY-R162, respectively.

The potential X-specific repeat, X-R55, appeared 22 times (> 100 bp) only in the corresponding X, but not in the HSY and papaya genome sequences (excluding X chromosome). The copies of the X-R55 repeat were present within a small range with ~50 kb (Figure 4A) in the corresponding X, and the repeat size ranged from 109 to 306 bp. Except for the two copies at both ends, the rest of the 20 copies were oriented in the same direction and tandemly repeated in three separate repeat blocks, apart from each other by about 8 to 15 kb (Figure 4A). The first repeat block near a zinc finger protein consisted of 4 tandem repeats, the second block of 11, and the third block of 5 (Figure 4A). Among the 22 copies of X-R55 repeat, the 17 copies longer than 200 bp were used for phylogenetic analysis. Similar to the result of phylogenetic analysis of two HSY-specific repeats, the distance among individual copies did not show correlation to the sequence similarity (Figure 4B, Additional file 4: Figure S1). An interesting feature of the X-R55 repeat was that it showed very high sequence identity (91%) with the third exon of a potential Carica papaya (Cp) zinc finger protein nearby (Figure 4A), whose expression was confirmed by an expressed sequence tag (GB: EX272522.1). If the X-R55 repeats originated from the third exon of the Cp zinc finger protein, the second repeat block could be the most recently duplicated, on the basis of phylogenetic analysis (Figure 4B). The presence of the X-R55 repeat was confirmed by PCR (Figure 4C). The expression of the Cp zinc finger protein was examined by RT-PCR and detected in all sex types of flowers and leaf tissues from ‘SunUp’ and ‘AU9’ papayas, and also in seed and half ripened fruit of ‘SunUp’ (Figure 4D). Phylogenetic analysis revealed that the Cp zinc finger protein was closely related to Arabidopsis zinc finger gene (NP_565037) (Figure 4E).

Accumulation of SSRs in the sex determining region

Accumulation of repetitive sequences is one of the key elements for the degeneration of sex chromosomes. Accordingly, high repetitive sequence accumulation was observed in papaya HSY and in the corresponding X compared to that of papaya genome [2, 21]. However, not only interspersed repeats but also tandem DNA repeats were accumulated in sex chromosomes [22]. Simple sequence repeats (SSRs) of the short tandem DNA repeats normally originate from slippage during DNA replication. Therefore, SSRs were examined in the HSY and the corresponding X (Figure 5 and Table 3). The SSR densities were much lower in the HSY (one per 8.1 kb) than in the corresponding X (one per 5.4 kb) and the papaya genome (one per 3.2 kb). On the other hand, SSR densities of V. monoica BAC sequences were even higher (one per 2.3 kb) than papaya genome (Table 3). SSRs have been categorized into two classes, class I and class II. Class I includes hypervariable SSRs ≥ 20 bp, whereas class II consists of less variable SSRs ≥ 12 bp and < 20 bp [23]. Class I and class II SSR densities were lower either in the HSY or in the corresponding X compared to those in the papaya genome (Table 3). Class I and Class II SSR densities in V. monoica BAC sequence were comparable to those in V. monoica shotgun genome (Table 3). Class I SSR densities in the corresponding X, V. monoica BACs and shotgun sequence, and papaya genome were approximately two-fold less than that of class II SSR, but much less in the HSY compared to the rest. SSR density of di-nucleotide SSR units were similar between papaya genome and V. monoica shotgun or BAC sequences, but SSR density of tri-nucleotide SSR units in V. monoica genome was significantly higher than that of the papaya genome (Table 4). Therefore, it was evident that SSR frequency in the sex determining region was lower than that in papaya genome and V. monoica genome.

Distribution of repetitive sequences

Y chromosomes are featured by degeneration, duplication, and abundance of repetitive sequences due to a non-recombining property. The HSY sequences on the papaya Y^h chromosome were occupied by higher repetitive sequences, compared to its X counterpart [2, 21]. The average repeat content of the HSY was approximately 77%, 13% higher than the 64% of the corresponding X (Table 1). These numbers were different from what we reported previously [2, 21], which was caused by the analysis of all 5.3 Mb X sequences including 1.8 Mb Knob 1 sequences that were not included in the previous report. The high accumulation of repetitive DNA sequences was shown in ancient Y chromosomes in human [24] and Drosophila melanogaster[25], also in the nascent Y chromosome in Drosophila Miranda[26] and Silene latifolia[22]. Repeat contents of the HSY and the corresponding X increased dramatically when masked by a repeat library containing both papaya and public repeat sequences, compared to the repeat contents of 19.5% of the HSY and 17.5% of the corresponding X when both sequences were masked by only publicly available known repeat sequences (Table 1), indicating that the majority of repeats accumulated in the sex determining regions are most likely papaya-specific [20]. It is worth noting that the conserved repetitive sequences in the HSY and its X counterpart were more than the genome-wide average of 14% [2] and also higher than the repeat content in V. monoica, which has no sex chromosomes, reinforcing the notion that increased repetitive sequences are a feature of the sex determining region.

Ty3-gypsy elements were highly accumulated in the sex determining region and accounted for 46.3% of the HSY and 37.7% of the corresponding X (Table 1 and Figure 1). The Ty3-gypsy content of the HSY was ~8% lower than that the previous study estimated from sequences of seven HSY BACs where it was 54% [2], which might be due to uneven distribution of Ty3-gypsy elements throughout the HSY. On the other hand, the Ty1-copia elements were less abundant compared to Ty3-gypsy in both the HSY and the corresponding X (Table 1 and Figure 1). Ty1-copia content in the corresponding X was 1.3% higher than that in the HSY (Table 1), suggesting that Ty1-copia elements were not a major contributor to repeat accumulation in both the HSY and the corresponding X. This result is different from the retroelement accumulation in S. latifolia Y chromosome where Ty1-copia elements are more abundant than Ty3-gypsy elements [27]. It could be due to the incomplete sequences of S. latifolia Y chromosome or the feature of the very early evolutionary stage of homomorphic Y^h chromosome in C. papaya compared to heteromorphic Y chromosome such as in S. latifolia, Rumen acetosella, and Marchantia polymorpha.

Decreased SSR frequency in the sex determining region

SSR density was significantly lower in the HSY and in the corresponding X compared to that in papaya or V. monoica genomes (Table 3) due to the increase of the overall repetitive sequence and the decrease of gene content [21]. These results suggest that the HSY is less vulnerable to mutation caused by replication slippage compared to other chromosome regions to maintain its unique sequence feature. Class I and class II densities in V. monoica BAC sequences were comparable to those in V. monoica shotgun genome, whereas those densities in the sex determining region were much lower than those in the papaya genome (Table 3), indicating that low SSR density in the HSY and the corresponding X was most likely caused by the process of the evolution of papaya sex chromosomes accompanied by the insertion of repetitive sequences. The density of Class I SSR (longer than class II SSR) in the HSY was lower than the rest, suggesting that the longer SSRs might be more susceptible to degeneration in the HSY.

Sex-specific repeats

Papaya sex types are determined by a small non-recombining region of recently evolved sex chromosomes [2]. The suppression of recombination in the HSY accompanies the accumulation of repetitive sequences and chromosomal rearrangements. These changes might ultimately result in the evolution of sex-specific repeats and the differentiation of sex chromosomes from their ancestral autosomes. Phylogenetic analysis and Mantel test of three sex-specific repeats, HSY-R29, HSY-R162, and X-R55, revealed that the distance and sequence similarity among copies of each repeat had no correlation (Figure 3B and C, Additional file 4: Figure S1), indicating that the insertion of repeats occurred either in a random manner regardless of the physical distance between original and new target sites or rearrangements occurred after tandem duplications. The HSY-R29 and HSY-R162 did not show any similar sequence match from the NCBI nucleotide database and TIGR plant repeat database (http://plantrepeats.plantbiology.msu.edu/). However, many HSY-R29 flanking sequences (~500 bp) showed similarity to chloroplast DNA of papaya and other plant species (data not shown), suggesting that the possible origin of HSY-R29 might be associated with chloroplast DNA insertions. DNA fragments transferring from organelles are not rare. For example, there is over 100 kb chloroplast DNA in rice chromosome 10 [28]. The papaya genome also contains nearly 1 Mb chloroplast DNA [2]. The papaya HSY accumulated a staggering amount of chloroplast DNA due to its lack of recombination with the corresponding X chromosome. The chloroplast DNA insertion could be another means of sex chromosome evolution.

Several sex-specific repeats were identified in other plant species, such as the RAYSI - III family in the plant Rumex acetosa[17, 29], MADC1 in Cannabis sativa[30], and the tandem Y-specific DNA repeats in Marchantia polymorpha[31]. The RAYSI-III family is satellite DNAs and MADC1 is homologous to LINE-like retrotransposons with a site-specific accumulation of the long arm of the Y chromosome [30]. Like the Y-specific repeats in M. polymorpha[31], the HSY-R29 and HSY-R162 were identified as sex-specific repeats and exhibited no similarity to any known repetitive sequences such as retroelements or satellite DNAs, indicating that these repeats are specific to the sex determining region of the papaya genome.

The Y-specific repeats of M. polymorpha are not only tandemly duplicated, but also contain male-specific genes [31]. In humans, it was also reported that the active gene could be multiplied as a result of tandem duplications and large sequence inversions, such as the AZFc region of the Y chromosome [32, 33] and the ZNF91 gene family in chromosomes 19 and 7 [34, 35]. In this study, we identified a potential X-specific repeat X-R55, which contained the third exon of a papaya zinc finger protein (Figure 4A). The tandem duplication of X-R55 was quite similar to the ZNF91 subfamily of primate-specific zinc finger genes, consisting of large gene clusters with some dysfunctional copies [34]. Another interesting feature of the ZNF91 gene family was that the large gene clusters are located near the centromere of chromosomes 19 and 7 [34]. In papaya, gene duplication was reported [20], and some of those genes may be clustered as similar to the ZNF 91 gene family. Nevertheless, this finding raises questions about whether the X-R55 repeats are located near the centromere of papaya X chromosome and whether the duplication of the X-R55 passed through a similar process as ZNF91 after duplication, such as loss of function and alternative splicing. These questions remain to be further investigated.

DNA sequences

The sequences of 13.4 Mb consisting of 8.1 Mb of HSY and 5.4 Mb of the corresponding X chromosome [21, 36] were used to examine repetitive genomic features and SSR distribution. In addition, a 245 Mb of V. monoica genomic shotgun sequences and a 1.1 Mb of 11 V. monoica BAC sequences [19] were used to compare accumulation and distribution of repetitive sequences and SSRs with those in the HSY, the corresponding X, and papaya genome.

Tandem repeats

A perl program, MIcroSAtellite identification tool (MISA; http://pgrc.ipk-gatersleben.de/misa/download/misa.pl), was used to mine SSRs in the given sequences. SSRs with a length greater than 12 nucleotides, motif lengths of 2 to 6 bp, and a minimum of 5 repeats were detected and analyzed.

Interspersed repeats analysis against known repeat databases

The repeat library was generated by combining Repbase [37], TIGR plant repeats (ftp://ftp.tigr.org/pub/data/ TIGR_Plant_Repeats), and papaya repeats [20]. For analyzing the repeat composition in the HSY and the corresponding X, RepeatMasker (http://www.repeatmasker.org) was used to analyze the repeat composition in the HSY and the corresponding X using the repeat library with default settings.

Identification of new repeats in the sex determining region

To identify new repeats, the sequences of the HSY and the corresponding X were first run on RepeatScout [38] to generate putative repeat sequences. Then, the resulting repeats were run on RepeatMasker (http://repeatmasker.org) to mask the HSY or the corresponding X sequences and to screen the repeats with the occurrence of more than 10 times and aligned length longer than 100 bp. Next, the non-redundant repeats passing above criteria were determined as new repeats by comparing them to previously identified papaya repeats from female papaya genome sequence [20] using CD-HIT software [39] with a cutoff of 70% similarity. Finally, the new repeats were blasted against the HSY and the corresponding X sequences using Standalone BLAST software (NCBI) and screened based on the following more stringent criteria: 1) at least a 50% alignment over a consensus sequence, 2) occurrence of at least 10 times, and 3) an aligned region with at least 100 bp and > 75% identity. The repeats that met these criteria were re-screened with property of less than 10 hits in the papaya genome in order to obtain potential sex-specific repeats. Clustalw [40] and MEGA [41] software were used for phylogenetic analysis of the repeats.

PCR for sex-specific repeats

Samples from SunUp female, SunUp hermaphrodite, AU9 female, and AU9 male were used to isolate genomic DNAs as described by K Edwards, C Johnstone and C Thompson [42] with slight modifications. PCR was carried out with 5 ng of DNA as a template with the following primer sets: HSY-R162 (Forward: 5′-TTTGTTCTCCTCTCAGCTTGC-3′; Reverse: 5′-GCCATACACGTAATGGGAAAA3′), HSY-R29 (Forward: 5′-GAAACCCATGCGAAGGAATA-3′; Reverse: 5′-TGGGATTCTTTTTGGGTCAG), and X-R55 (Forward: 5′-CCTTAGGAAGTTGCATTATGCTG; Reverse: 5′-ATTTATGAATTGAAAAGTTCAAGCAA). One of the papaya BAC end sequences was used to amplify a positive control for PCR of sex-specific repeats using the following primers: Foward 5′-TGACTCCATTGCCTGAATTTT-3′, and Reverse5′-TCCTCTCCATACCTTCTCGTG-3′.

RT-PCR analysis

Total RNAs were extracted from samples (SunUp female, hermaphrodite plant, seeds and half ripen fruit, and AU9 female and male plant leaf) using the hot phenol extraction method (Sambrook et al., 1989). The cDNA was synthesized using SuperScript II™ reverse transcriptase according to the manufacturer’s instructions (Invitrogen). The expression of the papaya zinc finger protein was examined by RT-PCR using the following primers: (F: 5′-CACTGGTTTTGCGGAAATTG; R: 5′-TGCACTTAGCATCATTGCAATG). As an internal control for RT-PCR analysis, papaya Actin gene was used [43].

Mantel test

To examine relationships between physical distances and sequence identities among sex-specific repeats, all pairwise sequence identities were obtained from the clastalW2 online tool (http://www.ebi.ac.uk/Tools/msa/clustalw2) and all pairwise physical distances were calculated manually. The sequence identities and the physical distances were used for Mantel test implemented in Genetic Analysis in Excel (GenAlEx 6.5) program [44]. Briefly, the sequence identities were manually arranged to Y matrix and the physical distances to X matrix as described in GenAlEx Tutorial 3. Then, Mantel test was performed with default set except for permutations of 9,999.