(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



A catalogue of recombination coldspots in interspecific tomato hybrids [1]

['Roven Rommel Fuentes', 'Bioinformatics Group', 'Wageningen University', 'Research', 'Wageningen', 'The Netherlands', 'Chromosome Biology', 'Max Planck Institute For Plant Breeding Research', 'Cologne', 'Ronald Nieuwenhuis']

Date: 2024-07 

Increasing natural resistance and resilience in plants is key for ensuring food security within a changing climate. Breeders improve these traits by crossing cultivars with their wild relatives and introgressing specific alleles through meiotic recombination. However, some genomic regions are devoid of recombination especially in crosses between divergent genomes, limiting the combinations of desirable alleles. Here, we used pooled-pollen sequencing to build a map of recombinant and non-recombinant regions between tomato and five wild relatives commonly used for introgressive tomato breeding. We detected hybrid-specific recombination coldspots that underscore the role of structural variations in modifying recombination patterns and maintaining genetic linkage in interspecific crosses. Crossover regions and coldspots show strong association with specific TE superfamilies exhibiting differentially accessible chromatin between somatic and meiotic cells. About two-thirds of the genome are conserved coldspots, located mostly in the pericentromeres and enriched with retrotransposons. The coldspots also harbor genes associated with agronomic traits and stress resistance, revealing undesired consequences of linkage drag and possible barriers to breeding. We presented examples of linkage drag that can potentially be resolved by pairing tomato with other wild species. Overall, this catalogue will help breeders better understand crossover localization and make informed decisions on generating new tomato varieties.

Ensuring a stable food supply in a changing climate hinges on enhancing plants’ natural defenses and resilience. Breeders achieve this by crossing cultivated plants with their wild counterparts, mixing specific gene variants through genetic recombination. Yet, in some cases, certain genomic areas lack this recombination, particularly when crossing diverse plant types, limiting the mix of beneficial genes. In this study we used a method called pooled-pollen sequencing to map regions where genetic recombination does and doesn’t occur when breeding tomatoes with five wild relatives. We discovered specific regions where genetic exchange is rare, mainly due to structural differences in DNA. These regions tend to be rich in repetitive DNA sequences and are less likely to swap genes during reproduction. Importantly, many of these "recombination coldspots" contain genes related to plant health and stress resistance, inadvertently limiting the effectiveness of breeding efforts. By identifying these regions, breeders can now make more informed choices when developing new tomato varieties, potentially by incorporating genes from other wild species to overcome these limitations and improve crop resilience.

Considered as one of the most cultivated vegetables crop, tomato has become a model system for genetic, developmental and physiological studies of fleshy fruits and is among the most well-studied crops for meiosis [ 18 – 21 ]. The availability of at least 12 wild relative species of tomato [ 22 ] makes it ideal for the study of recombination patterning in relation to genetic features such as SVs. Although previous studies addressed the role of SVs as recombination barriers [ 6 , 23 ], a genome-wide analysis of decreased or absent COs related to SVs and other genome features in tomato and multiple hybrid crosses is currently lacking, due to the absence of cost-effective and high-resolution crossover detection methods and accurate SV prediction. In particular, it is currently unclear whether coldspots are conserved (i.e. occur in the same genomic region in various genotypes) or more specific. To better understand the occurrence of recombination coldspots, we profiled the recombination landscape in multiple crosses of tomato and wild relatives by sequencing pools of pollen gametes. We identified coldspots in each hybrid cross and related their occurrence to genomic features. Our results suggest a major role for SVs and transposable elements in shaping the recombination landscape in hybrids, specifically in suppressing COs in a group of linked genes that relate to adaptation, speciation, and domestication. Finally, we show examples on how this catalogue can help determine bottlenecks in tomato introgressive hybridization breeding.

Genomic rearrangements may exist between related species and between different genotypes of the same species and can lead to recombination coldspots, some of which are associated with resistance genes or adaptive traits [ 9 , 10 ]. Due to absent or diminished crossovers in SV regions, clusters of tightly linked alleles known as supergenes are inherited together as a single locus, contributing to local adaptation and reproductive isolation [ 11 – 13 ]. Suppression or absence of recombination has been found essential in speciation and domestication by allowing the fixation of alleles [ 14 , 15 ]. In the backcross descendants of a Solanum habrochaites introgression into cultivated tomato (S. lycopersicum), an inversion containing the Ty-2 resistance genes and at least 35 more genes causes linkage drag, rendering selection of desirable agronomic trait combinations in the offspring impossible [ 16 , 17 ]. The instances of CO suppression presented here underscore the importance of investigating the impact of genomic rearrangements on recombination patterns and overall crop adaptation. Addressing these challenges is crucial for enhancing the effectiveness of introgressive breeding strategies and ensuring the successful development of crops with desirable traits.

Crop breeding relies on the availability of genetic diversity to generate novel allele combinations that are agronomically valuable. However, long term selection by inbreeding often causes loss of essential genetic variation. To reintroduce lost alleles, breeders introgress new genetic material by crossing crops with wild relatives, followed by repeated backcrossing and selection. Among the most desirable traits to be incorporated into the breeding material are abiotic stress tolerance and disease resistance, yield, and fruit quality [ 1 ]. The success of introgression breeding largely depends on the process of meiotic recombination to introduce genetic material from the donor into the recipient crop [ 2 ]. Meiotic recombination, commonly referred to as crossover (CO), facilitates the exchange of chromosomal segment between parental chromosomes, shuffling alleles to make new combinations. Lack or even complete absence of recombination in a genomic region leads to linkage drag, i.e. the introgression of deleterious alleles along with the beneficial one. This can severely limit the ability to develop novel desired allele combinations. Chromosome regions where recombination is suppressed are found in pericentromeres (PER), including retrotransposons and other DNA-methylated regions [ 3 , 4 ]. Moreover, genomic rearrangements, in particular structural variants (SVs), affect recombination patterns, especially in hybrids [ 5 – 8 ].

Results

Unique recombination patterns between hybrids All hybrids show similar recombination landscapes, with COs mostly in distal, gene-rich chromosome regions. Yet there are distinct local patterns of COs, as illustrated in Fig 1B. Comparisons of recombination profiles among different hybrids are crucial to understanding variability and genomic factors contributing to CO patterns. To assess similarities between hybrids, we initially identified overlapping COs and found a significantly higher fraction than expected by chance for every pair of hybrids (Fig 1C; Fisher’s exact test; all P < 2.56 x 10−9). The highest overlap of COs is observed between hybrids with wild parents that are evolutionarily closely related to each other (NE and CH, or HB and PN). In contrast, CO sites in PM have more overlap with PN than with other closely related species, which does not align with their evolutionary distance (Fig 1D). A low but significant overlap is also observed when comparing recombination hotspots in natural populations of wild and domesticated rice, cocoa and tomato [15,30,31]. COs per hybrid cover around 2% of the genome, whereas they cover 10% (77.6 Mbp) when combined. This apparently extensive non-overlapping coverage suggests divergent CO regions between the hybrids, or the need to generate more CO data per hybrid to exhaust all possible sites. Given the low rate of CO region overlap between hybrids, we investigated whether the overall recombination landscapes across the genome are significantly correlated. Fig 1E shows that NE and CH have the most similar landscape. The low CO overlap (4%; Fig 1C) between CH and HB does not translate to a low landscape correlation (ρ = 0.64). Similarly, despite the high overlap between PM and PN COs (7%), the correlation between their landscapes is one of the lowest (ρ = 0.52), consistent with their evolutionary distance. Hybrids with wild parents that are closer to each other show higher correlation (e.g CH and NE, HB and PN) while PM, which is distant to the other wild parents, is least correlated with the other landscapes. Although the number of overlapping COs is higher than expected by chance, it is far less than the number of non-overlapping COs, which contribute more to shaping the overall recombination landscape. These results suggest that variation in CO landscapes is related to the evolutionary distance between parental genomes. The patterns of genomic regions without recombination in the hybrids differ as well. To analyze these patterns, we identified CO coldspots of more than 1Mb, covering 72–79% of the genomes, with the highest coverage in HB and PN (S3 Table). All coldspots overlap SNP markers, confirming that the absence of COs is not due to the lack of markers in these regions. Grouping by genomic position and size, we assigned coldspots into 325 unique and 101 shared clusters (Fig 1F). Approximately 63.6% of the genome (6.4Mb euchromatic; 485Mb heterochromatic) lacks CO in all five hybrids, which we refer to here as conserved coldspots. PM has significantly shorter coldspots than the other hybrids (pairwise Wilcoxon rank-sum test; P < 1.4 x 10−2) and a large number of unique coldspot regions. Our comparison of CO landscapes indicates that although the chromosome-wide distribution is similar, a closer inspection of CO locations between hybrids reveals distinct differences: patches of CO-suppressed regions. These divergent patterns of CO regions and coldspots imply that hybridization of tomato with different wild parents results in variable recombination along the genome, revealing potential complexity in introgressive breeding.

Absence of COs in structural variants With the results above indicating clear variation in the occurrence of COs in the different hybrids, we speculated that large genomic rearrangements between species may underlie the varying patterns of recombination. To investigate this, we detected SVs between the parental species S. lycopersicum and the wild relatives. Given that heterozygous SVs may exist in the wild species genomes, allowing the F1 hybrid to inherit an allele that is similar to the reference genome, we genotyped SVs in the F1 hybrid pollen sequences and retained only the heterozygous ones (Fig 2A). Combining all parental wild species genomes, we detected 59,265 SVs larger than 50bp. Among the wild species, S. habrochaites and S. pennellii have the highest number of SVs, which are also significantly longer than in the other parental genomes (Figs 2B and S3). In order to validate the accuracy of the filtered set of structural variants (SVs), we randomly selected SVs from S. pennellii and visually compared the assemblies of S. lycopersicum and S. pennellii using dot plots (S4 Fig). These confirmed the presence of SVs. For instance, we found that 88% of 50 randomly selected deletions were supported, with an additional 10% belonging to more complex translocation events, and only 2% (1 case) identified as false positives. Similarly, for inversions, we observed a 76.7% true positive rate. Overall, condordant with previous studies [32–34], we find a varied landscape of SVs that are either unique to one, or shared by few, species. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Lack of crossover in structural variations. A) Selection of parental SVs causing heterozygosity in the F1 pollen genomes. B) Frequency of SVs per wild relative. Inversions only include events > 30 kb. C) Distance of COs to the nearest SV compared to the 10,000 permutation sets represented by gray lines. The vertical lines marks the boundaries of COs. D) Genome coverage of COs (blue) and SVs (orange) in the PER (left y-axis). The gray squares show the number of PM COs that overlap with SV regions in the wild genome (right y-axis). E) Crossover density of selected PN chromosomes (gray peaks) plotted together with Marey map (green dots) of EXPEN2012. The blue dots are genetic markers within coldspot regions (blue box). The yellow distribution line indicates the recombination rate obtained by taking the derivative of the Marey map. The gray horizontal segment in the middle of the chromosome marks the PER. F) Rate of synteny in coldspot (C) and non-coldspot (NC) regions of PN (Wilcoxon rank-sum test; P < 2x10-16). G) Heatmap of linked reads within the Ty-2 region in chromosome 11 long arm. H) Recombination coldspot overlapping the Ty-2 inversion. https://doi.org/10.1371/journal.pgen.1011336.g002 To examine the relationship between SVs and recombination, we identified rearrangements and syntenic regions between S. lycopersicum and S. pennellii assemblies and compared them against PN COs. We found that 94% of PN COs are in syntenic segments in distal chromosomal regions (Fisher’s exact test; P <0.001; S5 Fig), corresponding to the essential role of synteny in synapsis and crossing-over of homeologous chromosomes during meiosis [10,35]. 62–74% of SVs in the genomes of wild relatives overlap with coldspots. Using a permutation test, we found strong reduction of recombination in SVs across all hybrids, specifically for SVs larger than 1kb (S6 Fig); further analyses will only consider SVs larger than 1kb. Most SVs are located a few to tens of kilobases away from COs (Fig 2C) and SV size is not correlated to distance from the CO site (S7 Fig). Comparing the distal euchromatin (DEU) and PER compartments of the chromosomes, we found more SVs in DEU than in PER regions, with an average ratio of 1.55 to 1. This agrees with previous observations that wild and domesticated tomato accessions have higher SV density in DEU than in PER [34,36]. In addition, SVs in PER are on average longer than those in DEU (Wilcoxon rank-sum test; P < 5.8 x 10−16; S8 Fig). A higher genome coverage by SVs regions in PER coincides with fewer CO events (Fig 2D). PM has the largest total number of CO events in PER, while PN has the lowest number. These PM COs overlap with SVs in the other wild genomes, suggesting that the higher SV content in other wild genomes leaves less room for recombination. Overall, our results implicate SVs as one of the major modifiers of CO landscapes in hybrids, especially in CO-rich distal regions. We identified prominent spots in the PN DEU without CO. To validate whether these represent real coldspots, we compared them against the recombination coldspots in the EXPEN2012 linkage map [37], which is derived from a cross between S. lycopersicum and S. pennellii. Large coldspots are observed in the genetic map (Fig 2E), spanning 0.14 to 7.64 Mb, matching the coldspots we found in PN. Closer inspection of these large PN coldspots revealed significantly lower levels of synteny compared to non-coldspots (Fig 2F). These coldspots, however, may be specific to PN or may not fully overlap coldspots in other hybrids, as we found 518 COs in the same region in other hybrids. Among the PN coldspots, we found that at least two, specifically in the short arm of chromosomes 6 and 7, contain large inversions relative to the reference genome as previously validated using BAC-FISH [9]. We were able to confirm a large inversion in chromosome 7 by comparing genome assemblies and inspecting linked reads (S9 Fig). In addition to the inversion, this 2.4 Mbp coldspot region apparently also contains other rearrangements, like translocations, that could inhibit proper synapsis and recombination. Interestingly, this suppression in the short arm is not present in all hybrids, suggesting the absence of linkage drag when tomato is crossed with specific wild relatives. Another known SV we examined was the Ty-2 inversion [17] in the chromosome 11 euchromatic long arm. We confirmed that a CO coldspot is located in the inversion but only present in three wild parents (Fig 2G and 2H). This presents the possibility of using alternative parental genomes without SVs in target regions to overcome CO suppression.

Widespread coldspots in TE regions Aside from SVs, studies on other species have also linked the presence of transposable elements (TEs) with CO incidence, specifically retrotransposons with COs suppression [4]. Our data shows that most retrotransposons (Class I), except SINEs and RTE-BovBs, indeed exhibit suppression of COs (Fig 3A). However, Stowaway and Tip100 (Class II) TEs, as well as simple repeats and low complexity regions, are enriched with COs. TEs associated with CO suppression (Gypsy, Copia) are densely distributed in the PER, whereas Stowaway and Tip100 are located mostly in the DEU (Fig 3B), consistent with the CO distribution along the chromosomes. This association with TE superfamilies was also reported for historical recombination hotspots of wild and domesticated populations of tomato [15]. As shown in Fig 3C, the density of retrotransposons such as Gypsy, Copia and L1 in a genomic region correlates with CO suppression. In contrast, Stowaway and Tip100 show positive correlation with CO incidence (S10 Fig). About 98.6% of the conserved coldspots are in PER, where retrotransposon presence is dense. Furthermore, the retrotransposon superfamilies that are linked with CO suppression cover 450Mb (~52%) of the tomato genome, implying a wide span of suppression due to retrotransposons. This underscores the importance of transposable elements in shaping recombination patterns, both in hybrids and inbreeding materials and predominantly in regions with high retrotransposon density. Based on this varying association of TE superfamilies with COs, we propose that TE dynamics during tomato evolution may have a more complex impact on recombination landscapes than just suppressing COs in pericentromeric regions. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. TE-associated crossovers. A) TE superfamilies and repeats showing enrichment of COs. Elements are clustered into DNA transposons (yellow), retrotransposons (brown) and other repeats (gray). B) Recombination landscape of acrocentric chromosome 2 from multiple hybrids (colored peaks) with layers of density heatmaps representing different features, including class I (red) and II (blue) TEs, and meiotic ACRs (gray). The horizontal grey line represents the PER. C) Spearman’s rank correlation of crossover count and retrotransposons (Gypsy, Copia, L1) coverage in a sliding genome window. Each dot indicates a window. The red line is the local regression fitting. D) Total coverage of ACR per region. E) Normalized enrichment of ATAC-seq read coverage over repetitive elements of meiotic and somatic cells. F) Total ACR coverage per genome feature. Upstream and downstream covers 1 kb from the transcription start and termination sites, respectively. https://doi.org/10.1371/journal.pgen.1011336.g003 The occurrence of COs correlates with lower nucleosome occupancy and reduced DNA methylation [38]. To investigate the chromatin state of TE elements with and without COs, we performed an ATAC-seq analysis of S. lycopersicum meiotic and somatic cells and found 52,802 and 25,101 accessible chromatin regions (ACRs), respectively (S4 Table). These ACRs, with an average size of 733bp, represent accessible chromatin in the S. lycopersicum parent. Read distributions over the genome were highly correlated between biological replicates (S11 Fig). We found significant overlap between COs and meiotic ACRs (permutation test, z-score = 87.2), confirming that COs occur in regions accessible to recombination machinery. Fig 3D shows that crossover regions are more accessible than random genomic regions. Upon comparing meiocyte ACRs with TEs, we found that TE superfamilies enriched with COs are found in accessible chromatin segments, whereas retrotransposons like Gypsy, Copia and L1 are not associated with accessible chromatin (Fig 3E). This is similar to reports in A. thaliana, where DNA transposons show nucleosome depletion and high SPO11-1-oligo levels, and retroelements like Gypsy, Copia and L1 have very few SPO11-1-oligos with high DNA methylation and nucleosome occupancy [38]. Furthermore, shown in Fig 3E, the chromatin accessibility of TE superfamilies flips between somatic and meiotic cells, hinting at a preference to keep specific superfamilies inaccessible during meiossis. Our results emphasize the major role of chromatin structure in the suppression or enrichment of COs in TEs and the need to particularly analyze meiocytes, to account for tissue-specific ACRs. Similar to the association of COs with proximal promoter regions [24], it was previously reported that ACRs are strongly associated with transcription start sites (TSSs) [39]. To evaluate this, we examined the average ATAC-seq signal in genes and their flanking regions, finding the highest coverage at the TSS in both meiotic and somatic cells (S12 Fig). The majority of ACRs are located near or within genes (Fig 3F; Fisher’s exact test; P < 0.05), similar to COs (Fig 1A). Normalized by the total genome coverage of the feature, promoter regions and UTRs (untranslated regions) have the highest ACR density, which may explain the excess of CO in these regions and the need for accessible chromatin to initiate recombination.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011336

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/