Massive crossover suppression by CRISPR–Cas-mediated plant chromosome engineering

Recent studies have demonstrated that not only genes but also entire chromosomes can be engineered using clustered regularly interspaced short palindromic repeats (CRISPR)–CRISPER-associated protein 9 (Cas9)1–5. A major objective of applying chromosome restructuring in plant breeding is the manipulation of genetic exchange6. Here we show that meiotic recombination can be suppressed in nearly the entire chromosome using chromosome restructuring. We were able to induce a heritable inversion of a >17 Mb-long chromosome fragment that contained the centromere and covered most of chromosome 2 of the Arabidopsis ecotype Col-0. Only the 2 and 0.5 Mb-long telomeric ends remained in their original orientation. In single-nucleotide polymorphism marker analysis of the offspring of crosses with the ecotype Ler-1, we detected a massive reduction of crossovers within the inverted chromosome region, coupled with a shift of crossovers to the telomeric ends. The few genetic exchanges detected within the inversion all originated from double crossovers. This not only indicates that heritable genetic exchange can occur by interstitial chromosome pairing, but also that it is restricted to the production of viable progeny. Manipulation of genetic exchange is an important objective of plant breeders. Using chromosome engineering to invert a 17.1 Mb fragment on chromosome 2 in Arabidopsis thaliana, meiotic recombination could be suppressed in nearly the entire chromosome.

derived inversion on chromosome 4 in the Arabidopsis thaliana ecotype Col-0, we could show that CO can be induced in a previously recombination-cold region 1 .
In this study, we aimed to investigate whether we could exclude a substantial part of the Arabidopsis genome from genetic exchange by CRISPR-Cas-mediated chromosome engineering. To exclude as much genetic information as possible from meiotic recombination, we decided to invert nearly an entire chromosome and investigated the effect of the inversion on CO frequency and distribution. To achieve this goal, we used a well-established protocol for obtaining rare chromosome restructuring events in Arabidopsis 26 . By combining the Staphylococcus aureus orthologue of Cas9 27 with an egg cell-specific promoter, two double strand breaks (DSBs) were induced near the telomeric ends of chromosome 2 to invert the interjacent fragment with a size of 17.1 Mb. This left only a 2-Mb-and a 0.5-Mb-long telomeric end in their original orientation (Fig. 1a). The two selected guide sequences had both been previously used for the successful induction of inversions and translocations 28,29 . The selection and detection of the inversion were performed as previously described 1 . Out of the 40 screened T2 pools, one T2 pool was identified as positive for the inversion. Subsequently, the 40 descendants of the positive line were screened individually and one plant of the respective line was found to carry the inversion. The inversion junctions were amplified by PCR and sequenced to analyse the molecular nature of the junctions (Fig. 1b). The sequencing data revealed that the DNA ends were ligated precisely without any sequence loss or gain. To determine the genotype, the plant was also screened for the wildtype (WT) junctions and was found to carry the inversion hemizygously. The plant was grown until seed set and propagated. For subsequent experiments, T3 seeds were harvested and grown for 2 weeks on germination medium without selection and their genotype was determined by PCR screenings for the WT and inversion junctions. The phenotype of homozygous, hemizygous and WT plants did not differ from each other (Extended Data Fig. 1). The plants were also tested for segregation on the basis of Mendel's laws of segregation by using a chi-squared test with the critical value χ² (1; 0.95). Mendelian segregation was confirmed (Supplementary Table 1).
At the cytological level, chromosome inversion was confirmed in the hemizygous and homozygous mutant by fluorescence in situ hybridization (FISH). To obtain purely hemizygous seeds for the analysis, the homozygous inversion line was back-crossed with Col-0. Metaphase I meiocytes of the resulting F1 plants were prepared and BAC clones (Supplementary Table 2), which spanned Letters Nature PlaNts the two CRISPR-Cas9 cleavage sites, were chosen for red and green fluorescent labelling. In case of an inversion, both probes should be detected near each other. In WT, they should be located apart. Indeed, two separate signals were detected in WT (Fig. 2a). In the hemizygous sample, the WT signal pattern, as well as the inversion signal pattern were detected (Fig. 2a). Both newly formed junctions can be identified by adjacent red and green fluorescence signals. The FISH analysis of the hemizygous sample also shows that the WT chromosome and the chromosome carrying the inversion are indeed able to pair during meiosis. The pairing occurs outside of the inverted region at the telomeric end of chromosome 2 that carries a red label (Fig. 2a). Additionally, the inversion was confirmed in the homozygous mutant. In the homozygous sample, only the adjacent red and green fluorescence signals of the newly formed junctions were detected (Fig. 2a).
To assess the effect of the chromosomal rearrangement on fertility, a fertility assay was performed on homozygous, hemizygous and WT plants by counting the seed number of 10 siliques from 12 plants per genotype. We found that the seed number per silique was decreased by 1/3 in the hemizygous plants compared with homozygous and WT plants (Fig. 2b). Additionally, we observed empty spaces in place of seeds in several siliques of the hemizygous plants (Extended Data Fig. 2). The total seed number did not differ significantly between homozygous and WT plants.
Next, we investigated the effect of inverting approximately nine-tenths of the chromosome on the recombination frequency on chromosome 2 and, for purposes of control and comparison, on chromosome 3. The homozygous inversion line was crossed with Ler-1. F1 seeds were subsequently harvested. Then, they were propagated to obtain F2 seeds. The recombination frequency was determined by analysing 400 plants that were grown from F2 seeds. As a control, 400 plants of the progeny of crosses between WT Col-0 and Ler-1 were analysed. Using a TaqMan single-nucleotide polymorphism (SNP) genotyping assay, we selected 23 SNP markers on chromosome 2 and 9 SNP markers on chromosome 3 to detect marker changes between Col-0 and Ler-1 alleles (Supplementary Table 3). On chromosome 2, 8 SNP markers were selected within the inverted area. Outside of the inversion borders, 10 SNP markers were selected at the 5′ telomeric end and 5 markers at the 3′ telomeric end. The distance between the markers outside of the inversion was shorter to achieve a higher resolution of CO events at the telomere ends. The analysed plants were diploid; therefore, the marker analysis sums up the genetic state of two individual chromosomes that both underwent two meioses independently. Thus, we first analysed the presence or absence of marker changes in each sample.
In the offspring of the 'Inversion x Ler-1' (Inv x Ler-1) line, we detected a massive reduction of marker changes within the inverted area compared with the Col-0 x Ler-1 control. We detected 32 marker changes within the inverted region, compared with 386 in the Col-0 x Ler-1 control line in the same area. This corresponds to a roughly 92% reduction of recombination events in the inverted area of the Inv x Ler-1 line (Fig. 3a,b). Comparing the total number of marker changes detected outside of the inversion area in the Inv x Ler-1 line and the control, we found that the total number of marker changes was increased about 1.5-fold at the 5′ telomeric end and 2-fold at the 3′ telomeric end in the inversion line compared with the control line ( Fig. 3a,b), pointing towards a shift of COs to the telomere ends as a consequence of the inversion. In total, we detected marker changes in 206 out of 400 samples on chromosome 2. As COs only need to occur in one of the two sister chromatids of a bivalent chromosome to ensure correct segregation, one would expect that about a quarter of the diploid progeny plants would show a lack of CO. We assume that the difference between the expected and detected number of non-recombinations is, at least in part, due to cases in which the CO required for chromosome segregation occurred at one of the subtelomeric ends outside of our tested marker intervals: about 68 kb on the 5′ telomeric end and about 21 kb on the 3′ telomeric end were not covered by our marker intervals. By comparison, in the offspring of the Col-0 x Ler-1 control, we detected marker changes in 331 out of 400 samples. No recombination was detected in 69 samples, which was in line with our expectations. In contrast to chromosome 2, no notable difference in CO distribution between the two crosses was detected on chromosome 3 in the same plants (Extended Data Fig. 3).
Subsequently, by analysing the marker change patterns in more detail (Fig. 3c-f and Extended Data Figs. 4-7), we deduced the genetic state of the individual chromosomes from the diploid genomes. In the Inv x Ler-1 line, of the 206 samples in which marker changes were detected, 142 samples had marker changes originating from a single CO on one chromosome (Fig. 3c,d). All of these single COs were located outside of the inversion in the unchanged subtelomeric ends: 128 samples had single COs detected at the 5′ telomeric end (Fig. 3c) and 14 samples at the 3′ telomeric end (Fig. 3d). Six samples showed a single CO at both telomeric ends (Fig. 3e). Forty-two samples showed marker changes originating from CO events that could not be clearly assigned to either of the two chromosomes (Extended Data Fig. 4a,b). Within the inversion, no single COs were detected. Among the 16 samples with detected marker changes within the inverted area, 9 samples showed marker changes originating from a double CO in one of the chromosomes (Fig. 3f). In 7 samples, we detected a double CO within the inversion, combined with a single CO at one telomeric end outside of the inversion (Extended Data Fig. 4c).
Out of the 331 samples with detected marker changes in the Col-0 x Ler-1 control, 204 samples had marker changes originating from single CO events (Extended Data Fig. 5). In 127 samples, we detected CO events originating either from a double CO in combination with a single CO, or from a combination of single COs stemming from two independent meioses (Extended Data Figs. 6 and 7).

Nature PlaNts
As our marker analysis sums up the genetic state of two individual chromosomes that both underwent two meioses independently, it was not possible to clearly distinguish between double CO or multiple single CO events in the control line. CO suppression in one area can lead to the enhancement of meiotic recombination on other chromosomes, which is referred to as interchromosomal effect and has been observed in various organisms 30 . In our experiments, we did not detect an increase of CO events on chromosome 3 in the Inv x Ler-1 line compared with the control. However, we did detect 1.5-and 2-fold increases in the number of marker changes on chromosome 2 outside of the inversion borders compared with the control. Therefore, the lack of COs within the inversion might have been compensated by an increased number of COs in the telomeric regions of chromosome 2 and the obligate CO required for proper segregation of the chromatids might have been shifted to both telomeric ends of the chromosome. The observed increase in recombination could also be due to the elimination of non-viable gametes with a single CO in the inversion, which would subsequently lead to the detection of an increased number of chromatids with CO outside of the  White arrows indicate meiocytes in which the location of both fluorescently labelled BAC clones is visible. Two separate signals were detected in WT. In the hemizygous sample, the WT signal pattern, as well as the inversion signal pattern were detected. Both newly formed junctions can be identified by adjacent red and green fluorescence signals. In the homozygous sample, only the adjacent red and green fluorescence signals of the newly formed junctions were detected. b, Fertility analysis of homozygous, hemizygous and WT plants. Twelve homozygous, hemizygous and WT plants were randomly selected for fertility analysis (n = 12). Ten siliques per plant were randomly selected and the number of seeds per silique counted. The seed number of the hemizygous plants was significantly reduced, approximately by 1/3, compared with WT and homozygous plants. The seed number of homozygous and WT plants did not differ significantly. The boxplot shows the 12 independent biological replicates for each genotype. The average of the seed number of 10 siliques was calculated and is presented in the boxplots. The middle line represents the median, box edges represent the first and third quartiles, and whiskers extend to the minimum and maximum, respectively. Individual data points are represented as open circles. P values were calculated using a two-tailed t-test (***P < 0.001; NS, not significant).

Nature PlaNts
Border Border   I1  I2  I3  I4  I5  I6  I7  I8  I9  I10  I11  I12  I13  I14  I15  I16  I17  I18  I19  I20  I21  I22 Inv Letters Nature PlaNts inversion. In line with this observation, we were able to detect pairing of a WT chromosome and a chromosome carrying the inversion at the telomeric end of the smaller arm of chromosome 2 in our cytological analysis. Applying our CRISPR-Cas-based strategy for inverting chromosomal regions 29 , we were able to invert almost nine-tenths of chromosome 2 in Arabidopsis, which led to a near-complete exclusion of more than one-tenth of the Arabidopsis genome from genetic exchange. Indeed, within the inversion borders, the number of marker changes was drastically reduced (by 92% compared with the control), although recombination was not completely abolished. In 4% of the samples, double COs were detected within the inverted region, indicating that in rare cases, meiotic pairing and recombination can occur within the inverted region and can lead to viable gametes. However, as known from Drosophila, a single CO between a chromosome carrying a pericentric inversion and its WT homologue leads to the formation of chromosomes carrying duplications or deletions, which ultimately causes embryonic lethality 18 . In the case of the induced 17.1 Mb inversion, a single CO within the inverted area led to the formation of two Arabidopsis chromosomes carrying two identical telomeric ends as opposed to differing 5′ and 3′ telomeric ends (Fig. 4). This results in a massive loss of genetic information and, therefore, non-viable progeny. In contrast, as shown in Fig. 4, only COs in the telomeric ends or double COs within the inversion can lead to the formation of viable gametes.
In its hemizygous state, the inversion led to a notable reduction (by about one third) of viable seeds, whereas the homozygous mutant was fully fertile. The simplest explanation is that in about every third meiosis, the presence of the inversion hindered the formation of viable gametes due to the occurrence of single CO events within the inversion. However, the majority of seeds were still viable, making the exclusion of large parts of the genome from meiotic recombination by chromosome engineering a nonetheless useful tool for breeders to maintain favourable genetic linkages. This is also supported by the fact that natural inversions do not always lead to measurable fertility defects 31,32 .
Our study shows that it is possible to suppress recombination at the chromosomal level in a multicellular eukaryote by CRISPR-Cas-mediated chromosome engineering. We were able to develop a new instrument for the ever-growing toolbox of genome engineering, allowing the suppression of recombination anywhere in the genome in a predesigned fashion. This approach should be widely feasible among eukaryotes. Moreover, we showed that it is possible to almost completely abolish recombination in nearly an entire chromosome representing a substantial part of the Arabidopsis genome.
Accordingly, our results have important consequences for plant breeding, as they demonstrate that chromosome engineering can be a valuable tool for breeders in excluding favourable allelic combinations of any size from recombination, thereby stabilizing genetic linkages.  Table 3) outside and inside of the inverted area in the Inv x Ler-1 and Col-0 x Ler-1 lines. We detected a massive reduction of COs within the inversion borders, and 1.5-and 2-fold increases in marker changes at the 5′ and 3′ telomeric ends, respectively, compared with the control. Panel colour: grey background, intervals that lie outside the inverted area; light grey background, intervals covering the inversion sites; no background colour, intervals that lie inside the inverted area. c, Overview of marker changes that originated from a single CO on the 5′ telomeric end on one chromosome (126; rows [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] or from an identical single CO on both chromosomes (2; row 19). The Col-0 allele is represented by blue bars and the Ler-1 allele by red bars (same colour coding in d-f). d, Overview of marker changes that originated from a single CO on the 3′ telomeric end on one chromosome (14). e, Overview of marker changes that originated from a single CO on both telomeric ends of one chromosome (6). f, Overview of marker changes detected inside the inverted area, which can be clearly assigned to one of the two chromosomes. In 9 samples, marker changes inside the inversion that all originated from a double CO on one chromosome were detected. In another 7 samples, marker changes inside the inversion that originated from a combination of a double CO on one chromosome and a single CO on one or both chromosomes were detected. These samples are listed in extended Data Fig. 4c.

Interstitial single crossover
Interstitial double crossover Subtelomeric crossover(s)

Cloning of T-DNA constructs.
Cloning of the transfer DNA (T-DNA) constructs was based on the Gateway-compatible plasmids pEn-Sa-Chimera and pDe-Sa-Cas9 with S. aureus Cas9 under the control of an egg cell-specific promotor (pDe-Sa-Cas9 EC) 1,27 . The spacer sequences (Supplementary Table 4) were cloned as annealed oligonucleotides into individual pEn-Sa-Chimera vectors via BbsI restriction digest. Via MluI restriction digest, the first guide RNA (gRNA) cassette was integrated into pDe-Sa-Cas9 EC. The second gRNA cassette was transferred via a Gateway LR reaction.

Plant cultivation and transformation.
A. thaliana seeds were stratified overnight at 4 °C and transferred to soil consisting of a 1:1 mixture of Floraton 3 (Floragard) and vermiculite (2-3 mm, Deutsche Vermiculite Dämmstoff). For propagation, the plants were cultivated in the greenhouse under 16 h light/8 h dark conditions at 22 °C for 6-7 weeks until seed set. For transformation, 4-5-week-old Col-0 plants were transformed with the CRISPR-Cas expression construct via Agrobacterium-mediated floral dip transformation. The transformed plants were cultivated for 4-5 weeks until seed set. For sterile plant culture, seeds were surface-sterilized with 4% sodium hypochlorite and stratified overnight at 4 °C. For transgenic plant selection, the stratified seeds were sowed on germination medium (4.9 g l −1 Murashige and Skoog medium, 10 g l −1 saccharose, pH 5.7 and 7.6 g l −1 plant agar) with phosphinotricin and cefotaxime. T1 plants were propagated in the greenhouse and the seeds harvested. T2 seeds were stratified and sowed on the germination medium without further additives. The plates were placed in a growth chamber at 22 °C under 16 h light/8 h dark conditions for 2 weeks.
DNA extraction. For individual plant screenings, one leaf per plant was cut off and separately placed into a 1.5 ml reaction tube. For bulk screenings, one leaf each from 40 plants was cut off and leaves combined in the same 1.5 ml reaction tube. The plant material was ground with a pestle and 500 µl extraction buffer (200 mM Tris-HCl (pH 9.0), 400 mM LiCl, 25 mM EDTA, 1% SDS, pH 9.0) was added to the tubes. The mixture was centrifuged for 5 min at 17,000 g at room temperature. The supernatant (400 µl) was transferred to a new 1.5 ml tube containing 400 µl 2-propanol. The sample was thoroughly inverted and the DNA was pelleted for 10 min at 19,500 g at room temperature. The supernatant was discarded and the pellet was dried either for 1.5 h at 37 °C or overnight at room temperature. Afterwards, the pellets were resuspended in 50 µl (individual DNA extraction) or 100 µl (bulk DNA extraction) TE buffer (10 mM Tris-HCl (pH 9.0), 1.0 mM EDTA, pH 8.0).

Establishment of the homozygous inversion line.
T2 seeds were cultivated on germination medium for 2 weeks in a growth chamber. Subsequently, bulk DNA extractions were performed with 40 plants per T2 line. These T2 pools were screened for the presence of the inversion via PCR using junction-specific primers (Supplementary Table 4). Sequencing of the junctions was performed by Eurofins Genomics, and the software ApE (v2.0.55) was used for alignment and analysis of Sanger sequencing data. One T2 line tested positive for the inversion. The individual plants of this line were subjected to individual DNA extraction to identify the individual plants harbouring the rearrangement. The individual DNA samples were again screened by PCR. The identified positive plant was transferred to the greenhouse and grown for 6-7 weeks until seed set. In the T3 generation, the seeds were genotyped by PCR using primers specific for the WT and inversion junctions (Supplementary Table 4). The line was also tested for Mendelian segregation on the basis of the genotyping results using a chi-squared test with the critical value χ² (1; 0.95). Homozygous plants and, as a control, Col-0 plants were crossed with the Ler-1 ecotype for subsequent determination of CO frequencies. F1 seeds were propagated in the greenhouse and ripe F2 seeds were harvested after 6-7 weeks.
Phenotypic analysis and fertility assays. Homozygous, hemizygous and WT plants, as determined by PCR-based genotyping in the T3 generation, were grown in the greenhouse for 5-6 weeks under 16 h light/8 h dark conditions at 22 °C. After representative pictures were taken for phenotype analysis, 10 mature siliques of 12 plants per line were collected and incubated overnight in 70% EtOH to determine the effects of the inversion on fertility. The number of seeds per silique was counted and pictures of the siliques were taken using a binocular microscope. P values were calculated using a two-tailed t-test. Graphs were created using R Studio (Version 1.1.423) and CorelDRAW 2020 (Version 22.1.1.523).
Determination of CO frequency. CO frequency was determined by the detection of marker changes via SNP-based genotyping using cultivar-specific probes. For chromosome 2, 23 TaqMan probes and specific primers were designed. For chromosome 3, 9 TaqMan SNP genotyping assays were designed. All probes and primers are listed in Supplementary Tables 3 and 4. The analysis was performed using a Lightcycler 480 II (Roche) and the PerfeCTa qPCR ToughMix (Quantabio) according to the manufacturer's protocol but with a total reaction volume of 10 µl. The data were analysed with LightCycler 480 SW 1.5. Graphs were created using Microsoft Excel 2016 and CorelDRAW 2020 (Version 22.1.1.523).
FISH. FISH analysis was performed using differently labelled DNA probes specific for the inversion breakpoint regions of chromosome 2. Pools of contiguous BAC clones of chromosome 2 were used to paint the neighbouring regions of the CRISPR-Cas9 cleavage sites (Supplementary Table 2) 33 . An epifluorescence microscope (BX-61, Olympus) equipped with a UPlan(F) ×100/1.30 objective lens (Olympus) and a cooled black/white CCD camera (ORCA-R2, Hamamatsu), under the control of the microscope software CellSens Dimension (Olympus) was used to record the micrographs. The following fluorescence filters were used: FITC-2024B-000, Txred-4040C-000n and Sp. Aqua HC (Sembrock). Black and white pictures were pseudo-coloured with the software Photoshop 2020 (Adobe).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings are available within the article or in Supplementary Table 3). The recombination frequency is presented in cM/Mb. No notable difference in CO distribution between the two crosses on chromosome 3 was detected. Source data are provided as Source Data file.

Nature PlaNts
Extended Data Fig. 4 | Overview of the detected CO events in the Inversion x Ler-1 line on chromosome 2 which cannot be clearly assigned to one of the two chromosomes. The Col-0 allele is represented by blue bars and the Ler-1 allele by red bars. The detection of both alleles is represented by yellow bars. a, Overview of marker changes which originated from differing single CO on both chromosomes (24). b, Overview of marker changes which originated either from differing single CO on one or both chromosomes or from double CO outside of the inversion or a combination of double CO outside of the inversion with differing single CO (18). c, In 7 samples, marker changes inside the inversion were detected which originated from a combination of a double CO on one chromosome and single CO on one or both chromosomes.

Nature PlaNts
Extended Data Fig. 5 | Overview of the detected single CO events in the Col-0 x Ler-1 line on chromosome 2. The Col-0 allele is represented by blue bars and the Ler-1 allele by red bars. The detection of both alleles is represented by yellow bars. a, Overview of marker changes which originated from single CO on the short arm on one chromosome (29). b, Overview of marker changes which originated from a differing single CO on the short arm on both chromosomes (3; rows 1-3) or two identical single CO on both chromosomes (1; row 4). c, Overview of marker changes which originated from single CO on the long arm on one chromosome (130). d, Overview of marker changes which originated from differing single CO on the long arm of both chromosomes (22). e, Overview of marker changes which originated from differing single CO, one on the long and one of the short arm of either chromosome (19).

Nature PlaNts
Extended Data Fig. 6 | Overview of the detected double and/or single CO events in the Col-0 x Ler-1 line on chromosome 2. The Col-0 allele is represented by blue bars and the Ler-1 allele by red bars. The detection of both alleles is represented by yellow bars. a, Overview of marker changes which originated either from a double CO on one chromosome or from two single CO from two different meioses on one chromosome (24; rows 1-22) and from a combination of either a double CO on one chromosome and a single CO on the other chromosome or two single CO from different meioses on one chromosome and a single CO on the other chromosome (4; rows [23][24][25][26]. b, Overview of marker changes which originated either from a double CO on one chromosome or from two single CO from two different meioses on one chromosome (60).