recq4ab figl1 enhance recombination in full hybrid and substitution lines
Previous work combining knockouts for different anti-CO pathways showed that the genotype that gives the strongest increase so far, is obtained by mutating FIGL1 (AT3G27120) together with the two RECQ4 paralogs RECQ4A (AT1G10930) and RECQ4B (At1G60930)23. Here, we combined the same mutations (Figure S1–S2) into F1-hybrid populations with different chromosomal configurations : (i) The hybrid between the Columbia (Col) and Landsberg (Ler) strains (Fig. 1a) and (ii) a set of chromosome substitution lines (CSLs) in which a single chromosome is segregating (Col/Ler), while the rest of the genome is fixed (Fig. 2). Hereby, we will identify these lines as: The full hybrid population (Fig. 1), CSL_chr2_L, and CSL_chr5_L for the Chromosome Substitution Line with chromosome 2 or chromosome 5 segregating, respectively (and the rest of chromosomes homozygous Ler) (Fig. 2a, b), and CSL_chr4_C and CSL_chr5_C for the lines with chromosome 4 or chromosome 5 segregating, respectively (with the rest of chromosomes being Col) (Fig. 2c, d). For each of these hybrids, populations of F2 plantlets were produced by self-fertilization and whole genome sequenced with Illumina short reads. COs were detected using a sliding window approach as described in Lian et al.28. Sequence data were also used to identify potential aneuploids using relative coverage between chromosomes29,30.
a Schematic representation of the full hybrid populations used in this study. b Distribution of crossover number per F2 plant. Wild type and recq4ab figl1 are shown in dark blue and red, respectively. The mean crossover per F2 +/− standard deviation and the number of F2 plants analyzed (n) are indicated. c Distribution of crossovers along all five chromosomes in the hybrid population (recq4ab figl1 in red vs. wild-type hybrids in dark blue), using non-overlapping 500 kb windows. Source data are provided as Source Data files and Supplementary Data 2.
a–d Schematic representation of the chromosome configuration of the parental plant of the populations analyzed in each plot. The analyzed chromosomes are marked by a black arrow. See also Figure S2. e Analysis of crossover frequency and distribution in CSL_chr2_L, in which chromosome 2 segregates and the other chromosomes are fixed Ler, compared to crossover frequency and distribution on chromosome 2 in full hybrid. f Analysis of crossover frequency and distribution in CSL_chr5_L, compared to crossover frequency and distribution on chromosome 5 in full hybrid. g Analysis of crossover frequency and distribution in CSL_chr4_C, in which chromosome 4 is segregating and the other chromosomes are fixed Col, compared to crossover frequency and distribution on chromosome 4 in full hybrid. h Analysis of crossover frequency and distribution in CSL_chr5_C, compared to crossover frequency and distribution on chromosome 5 in full hybrid. All analysis using non-overlapping 500 kb windows. Full hybrid data are from Fig. 1.
In Col/Ler full hybrids, the average number of COs per F2 plant was 8.4 + /− 2.3 for the wild type, ranging from 2 to 16, corresponding to 1.7 CO per chromosome and similar to previous reports28,31. In recq4ab figl1 full hybrids the number of CO per F2 was 50.5 +/− 11.8, ranging from 20 to 87, corresponding to a massive increase of six-fold and an average of ten CO per chromosome (Fig. 1b). This confirms the synergistic effect of RECQ4 and FIGL1 in limiting CO formation23.
Looking at the distribution of COs along the genome in the full hybrid (Fig. 1c), frequencies are increased all along the arms in recq4ab figl1 compared to the wild type. However, the centromeric regions which are off for recombination in wild type, remain off in recq4ab figl1. Further, in regions adjacent to the centromeres, where recombination is high in wild type, CO frequency is not or only slightly increased by the recq4ab figl1 mutations, suggesting that additional factors limit COs in these regions. A lower increase in CO is also observed around position 23 Mb on chromosome 1, which corresponds to a region of high polymorphism associated with a cluster of NBS-LRR disease-resistance genes.
Next, we observed the effect of mutation recq4ab figl1 in the hybrid chromosome of the CSL populations. In CSL_Chr2_L and CSL_Chr5_L, chromosome 2 and 5 segregates, respectively, while the rest of the genome is fixed Ler (Fig. 2a, b). In CSL_Chr2_L, the number and distribution of COs on chromosome 2 are indistinguishable from the same chromosome in the full hybrid context, in both wild type (1.5 COs, Mann-Whitney test, p = 0.92) and recq4ab figl1 (8 COs, p = 0.83) (Fig. 2e). Concerning chromosome 5, it receives 1.9 COs in the wild-type full hybrid context and 2.2 in the wild-type CSL_Chr5_L (Mann-Whitney test p = 0.017), suggesting that a trans factor slightly affects CO number on chromosome 5, with the Col allele favoring COs. As in full hybrid, COs are strongly boosted by recq4 figl1 in CSL-Chr5_L on chromosome 5, to reach similar levels (12 and 11.4, respectively. p = 0.14) (Fig. 2f). Thus, CO frequency can be enhanced in substitution lines, exactly like in full hybrids ( ~ 6 folds).
Strikingly, a different observation was made in CSL_Chr4_C and CSL_Chr5_C, in which chromosome 4 and 5 segregate, respectively, while the rest of the genome is homozygous Col. In wild types, the crossover frequencies are similar in CSL_Chr4_C compared to full hybrids (Fig. 2g, h, 1.6 vs. 1.4; p = 0.16), and slightly increased in CSL_Chr5_C compared to full hybrids (2.2 vs. 1.9, p = 0.002). This suggests that a trans factor slightly affects the CO number on chromosome 5, with the Ler allele favoring COs in the wild type context. In contrast, chromosome 4 receives an average of 3.7 COs in recq4ab figl1 CSL_Chr4_C, compared to 7.4 in recq4ab figl1 full hybrid (p < 10−6; Fig. 2g). Similarly, chromosome 5 receives on average 6 COs in recq4ab figl1 CSL_Chr5_C, compared to 12 in recq4ab figl1 full hybrid (p < 10−6; Fig. 2h).
No aneuploids were detected in any of the wild-type F2 populations (total n = 1183). In recq4ab figl1 F2 populations, no aneuploids were found for full hybrids (n = 329), CSL_chr2_L (n = 175) and CSL_chr5_L (n = 183). However one trisomic was found in the CSL_chr4_C population (n = 148) and one in the CSL_chr5_C (n = 77), both with an extra chromosome 2 (samples 3783_AM and 3786_U). This low frequency of aneuploidy observed in the recq4ab figl1 progenies (2/912 = 0.2%), suggests that chromosome segregation at meiosis is not, or very marginally affected in this mutant.
Hyper-recombination does not affect phenotypic variation
Taking advantage of the Phenoscope robots to carefully phenotype large cohorts of plants, we profiled the most interesting of these populations (full hybrid, CSL_chr2_L, and CSL_chr5_L) for diverse traits related to rosette growth and architecture, as well as phenology (time to bolting). Note that the scoring of time to bolting was done by hand and is, therefore, less resolutive -in time- and less accurate than the other image-based automated phenotypes. Figure 3a–d shows the distribution of relative growth rate, compactness, leaf color -Hue- and bolting time, comparing five repeats (F2 populations derived from five individual F1 plants) of wild-type and recq4ab figl1 full hybrids. For each phenotype, the mean and variation of phenotypic values are essentially indistinguishable between wild-type and recq4ab figl1 F2 populations, showing that neither recq4ab figl1 mutations nor elevated meiotic recombination affected the segregation of macroscopic traits.
For both wild-type and recq4ab figl1, the F2 progeny of five individual F1 plants (line) were scored for a range of phenotypes. The phenotypic value of each F2 population, and the merge, are shown. The red bars indicate the mean. a Relative expansion rate integrated between 18 and 31 Days After Sowing (DAS). b Compactness at 31 DAS. c Hue leaf color parameter at 31 DAS. d Bolting time.
The resolution and power of QTL analysis are improved in hyper-recombinant segregating populations
Next we tested the effect of elevated COs on QTL mapping for the most heritable traits. Figure 4 presents the QTL maps for 3 traits (a. rosette compactness at day 31 after sowing, b. leaf color -hue- at day 31 after sowing, and c. time to bolting) as a comparison between wild-type and hyper-recombinant populations.
Each graph presents LOD-score profiles computed by interval mapping for the indicated trait: (a) Compactness 31 days after sowing, (b) Hue 31 days after sowing and (c) bolting time). For each trait, the top row shows the QTL maps in full hybrids, comparing wild type with figl1 recq4; the bottom row shows the QTL maps in chromosome substitution lines (CSL_L), comparing wild type with figl1 recq4. “n” = number of plants analyzed in each population. The x-axes are in % of the physical position along the chromosome (Chr.). The y-axes represent the LOD-Score with additive allelic effect direction sign (+/−): a positive (negative) LOD Score indicates that the Ler allele at this position increases (decreases) trait value with respect to the Col allele. Conservative LOD-score thresholds of +/−2.3LOD are represented as dashed lines.
Overall, we detect very highly significant QTLs as well as mildly significant QTLs in all contexts. The analyzed traits appear to be essentially controlled by 2 QTLs with major and pleiotropic effects, one on chromosome 2 around position 60 and one on chromosome 5 around position 80. No clear effect of the recombination density on significance emerges, with typically comparable LOD Score values between populations for peaks at similar locations. In contrast, a striking effect of boosting recombination is observed on the resolution of the QTLs, resulting in much finer peaks with significance dropping much more rapidly on each side of the peak. For instance, the most significant QTL (on chromosome 5 for Hue, full hybrid population; Fig. 4) reaches similar maximal significance in both recombination contexts but the extent of -for instance- a 3-LOD drop interval from the LOD maximum spans an interval 3 times bigger in a wild-type recombination context than in the hyper-recombinant population. Similarly, the peak beyond the significance threshold covers more than half of the chromosome in a classical recombinant population, while it is significant over ~15% of the chromosome in a hyper-recombinant background. A colocalizing QTL is mapped in CSL_chr5_L, essentially showing the same pattern, with a much finer peak in the hyper-recombinant lines. The significance for the likely same QTL is lower in the CSL_chr5_L compared to full hybrid, but it should be noted that the number of F2 plants analyzed is larger in the full hybrid sets compared to the CSL sets. Thus, in both full hybrid and CSLs, enhanced recombination resulted in a much finer definition of all the observed large-effect QTLs.
Another general trend is that the refined position of the LOD Score peaks in the hyper-recombinant context allows distinguishing smaller-effect QTLs in the vicinity of large-effect QTLs. For instance, at the bottom of chromosome 5, just south of a major peak, we detect another secondary QTL in the same allelic direction for Compactness and Hue traits (visible in the CSL_chr5_L population), but this one is masked by the shoulders of the main peak in the wild-type background. Similarly, on chromosome 2, a secondary QTL is detectable at position 40 for Compactness in the full hybrid population only in the hyper-recombinant background. For bolting time, the main peak on chromosome 2 seems to hide a QTL around position 90 in the opposite direction (typically a difficult situation to decompose), which is significant only in the hyper-recombinant full hybrid and CSL sets. An extreme and complex case arises when studying bolting time in the CSL_chr5_L set: chromosome 5 appears entirely above the significance threshold in the wild-type background, and this seems to be due to 3 independent QTLs in the same direction according to the hyper-recombinant analysis. The position of at least one of these QTLs is independently confirmed in the full hybrid population.
There are also some examples of significant peaks only detected in one of the backgrounds with no immediate explanation (e.g. the isolated peak at the bottom of chromosome 1 for compactness and hue is only significant in the hyper-recombinant population; the very wide peak on chromosome 1 for bolting time conversely only appears in the wild-type population, while the hyper-recombinant population does not clearly map anything, although there are multiple suggestive QTLs).
QTL fine mapping resolution is enhanced in hyper recombinant lines vs. wild type and fine-mapping gives access to gene-scale resolution
As visible in Fig. 4, the confidence intervals of the QTL are dramatically reduced in the hyper-recombinant background. Because we encountered major-effect QTLs whose phenotypic effect could be discerned in a large proportion of the individual plants, we were able to proceed to fine-scale mapping of two of the QTLs aiming to identify the causal polymorphisms by zooming in and studying single recombination events together with the phenotype of the individual plants (Figs. 5 and 6).
a Each line represents the genotype of a plant with a recombination point (crossover) in the region of interest (green = Ler, red = heterozygous Col/ler). The plant ID is shown on the left, with a color indicating its phenotypic class (in green the plants show a Landsberg-like phenotype and in red the plants show a Col/heterozygous-like phenotype – the Col allele is dominant). The direction in which the causal polymorphism is predicted to be is depicted with a black arrow. The most informative plants are shown for wild-type and recq4 figl1 populations. The interval defined by the closest recombination points are indicated for wild-type and mutant populations. A zoom is made on the closest recombinants. b The positions of the crossovers are shown in the table, start and end defining the borders of the interval in which the crossover is detected.
a Each line represents the genotype of a plant with a recombination point (crossover) in the region of interest (green = Ler, red = heterozygous Col/ler). The plant ID is shown on the left, with a color indicating its phenotypic class (in green the plants show a Landsberg-like phenotype, and in red the plants show a Col/heterozygous-like phenotype – the Col allele is dominant). The direction of where the causal polymorphism is predicted to be is depicted with a black arrow. The most informative plants are shown for wild-type and recq4 figl1 populations. A zoom is made on the closest recombinants. The interval defined by the closest recombination points are indicated for wild-type and mutant populations. b The positions of the crossovers are shown in the table, start and end defining the borders of the interval in which the crossover is detected.
Overlapping major QTL peaks are detected on chromosome 2 for compactness and flowering time. We used compactness for the fine mapping as the phenotypic scoring is more robust in our settings. Combining the wild-type full hybrid and CSL_Chr2_L populations, we could narrow the region of the QTL to an interval of 190 kb (chr2:11045-11235 kb) with one recombinant plant defining each of the left and right borders (Fig. 5). This region contained a few hundred protein-coding genes. When doing the same with recq4ab figl1 populations, and because of the much higher number of COs, we could reduce the interval to 16 kb (Chr2:11195-11212). The defined interval contains only four protein-coding genes (AT2G26300, AT2G26310, AT2G26320, and AT2G26330) (Fig. 5). Remarkably, one of them stands out as the possible causal gene, the receptor-like kinase ERECTA (ER) gene (AT2G26330), which is well known for being mutated in Ler and consequently segregating in our populations. The ER gene is involved in diverse processes regarding somatic development and immune response, but mainly, and as it concerns in this study, giving the specific leaf and rosette shape of the Landsberg erecta accession32,33,34. It is very likely that the erecta mutation (Chr2:11.209.133) is the causal variant under the chromosome 2 QTLs for compactness and flowering time.
Another major QTL was detected on chromosome 5, for compactness, Hue, and flowering time. We used compactness and Hue to fine-map the QTL, as they were both robust and coherent. Using the wild-type populations, the mapping defined an interval of 360 kb (Fig. 6), which contains hundreds of genes as possible candidates. In contrast, using the hyper recombinant figl1 recq4 populations, we could narrow the interval down to 22 kb (Chr5:22608 to 22631). The defined interval contains five protein-coding genes (AT5G55860, AT5G55870, AT5G55880, AT5G55890, AT5G55893) and two transposons (AT5G55875 and ATG555896) (Fig. 6). One very tempting candidate is AT5G55860/TREPH1, whose mutation leads to modification of flowering time and rosette radius in response to mechanical stimulus35. This TREPH1 gene contains two non-synonymous polymorphisms affecting the protein sequence in Ler with respect to Col (Tair10 5:22611342 A > G, Thr376>A; 5:22611934 C > T, Ala>Val) that may affect the protein function.
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.nature.com/articles/s42003-024-06530-w