RNAi-mediated downregulation of AcCENH3 can induce in vivo haploids in onion (Allium cepa L.)

Generation and molecular confirmation of AcCENH3 knockdown lines

A hairpin RNAi construct targeting AcCENH3 was designed using a 397 bp sense/antisense fragment of AcCENH3 and designated as AcCENH3-RNAi (Fig. 1A, Supplementary Table S1). A total of 3942 embryogenic calli of onion cultivar Bhima Super was transformed with AcCENH3-RNAi construct by using the Agrobacterium-mediated transformation protocol. Five independent AcCENH3 knockdown lines regenerated, with a transformation efficiency of 0.13%, which was lower than the 1% observed in our previous study using a GUS construct¹⁹. Putative transgenic lines were validated by PCR amplification with T-DNA-specific primers for hptII, sense and antisense strands of AcCENH3-RNAi construct (Supplementary Table S1 and Supplementary Fig. 1). Additional thermal asymmetric interlaced PCR (TAIL-PCR) analyses of T₀ plants confirmed single copy insertion of transgene and identified the T-DNA-plant genome junctions (Supplementary Fig. 2). Further analysis of the junction sites revealed that none of these insertions disrupted functional genes, confirming the independence of all 5 events (designated as E1–E5). Based on these insertion sites, event-specific markers were designed (Supplementary Table S1). These markers were also used to screen for the zygosity of plants in the T₁ generation. All five independent events were acclimatized under greenhouse conditions, where they successfully formed bulbs. These bulbs were then planted in the subsequent season to complete their biennial life cycle. Unfortunately, all plants from E4 failed to survive. Among the surviving events, E3 did not produce seeds (Supplementary Fig. 3). Thus, for subsequent analyses, T₁ plants from E1, E2, and E5, along with bulb-derived T₀ plants from E3, were used.

Biased segregation distortion in AcCENH3 knockdown lines

When the selfed progenies of events E1, E2, and E5 were screened using T-DNA specific primers, the population was found to be not in confirmation with the segregation ratio of 3:1 (Supplementary Table S2). We observed 24.24%, 23.53%, and 37.5% positive plants out of totals of 99, 85, and 24 plants in events E1, E2, and E5, respectively, which is a significant deviation from the expected frequency of 75% (Supplementary Table S2). Further, PCR analysis using event-specific T-DNA and plant junction markers in the T₁ population of AcCENH3 knockdown lines E1, E2 and E5, revealed the presence of heterozygous and azygous plants. We observed a single homozygous plant in E2, which died early in its development (Supplementary Fig. 4A,B and Supplementary Table S3). This indicated that the survival of homozygous knockdown lines was severely affected probably due to higher suppression of AcCENH3 than in heterozygous lines. F₁ generation obtained from the reciprocal crosses of AcCENH3 knockdown lines (E1 and E2) with WT did not conform to the 1:1 segregation of the transgene. The E5 test cross (with WT) population conformed to 1:1 segregation but segregation distortion was observed in the selfed population. Interestingly, the transgene segregation distortion is more pronounced when the two events, E1 and E2, were used as the male parent. In maize, the inheritance of the knockout allele from the female parent (25%) was higher than that in the male parent (12.1%)¹². We also observed that AcCENH3-RNAi transgene inheritance was higher from the female parents than males. In the events E1 and E2, the inheritance of the transgene was much higher from the female parent with a 5% lower inheritance observed from the male parent (Supplementary Table S4). CENH3 is essential for male gamete maturation²⁰, and reduction or loss of CENH3 can significantly impact transgene inheritance from the male gamete^14,21,22. In summary, our findings demonstrate a significant deviation from the expected segregation ratios only in lines E1 and E2. As we were not able to recover homozygous knockdown lines, we used heterozygous AcCENH3 knockdown lines in all further analyses.

AcCENH3-RNAi transgenic lines show decreased AcCENH3 transcript and protein levels

To assess the impact of the dsRNA hairpin RNAi cassette on native AcCENH3 transcripts and protein levels, we analysed the heterozygous knockdown lines using quantitative RT-PCR, and ELISA (T₀ plants were used for E3). qRT-PCR showed that the transcript levels of native AcCENH3 were significantly reduced in all knockdown lines compared to the WT. The relative downregulation of AcCENH3 transcripts in the RNAi transgenic events E1, E2, E3, and E5 was 0.74, 0.79, 0.59 and 0.89-fold, respectively, compared with WT (Table 1 and Supplementary Fig. 5A). ELISA was carried out with protein extracted from leaf tissue to quantify the AcCENH3 protein levels from knockdown lines E1, E2, E3 and E5. ELISA results showed that native AcCENH3 protein levels in all transgenic events decreased compared to the WT. The relative protein levels were 77.84%, 81.50%, 71.80% and 87.83% of WT AcCENH3 in knockdown lines of E1, E2,E3 and E5, respectively (Table 1; Supplementary Fig. 5B). Our findings from qRT-PCR, and ELISA analyses consistently indicated that decrease in AcCENH3 transcript levels are commensurate with reduction in AcCENH3 protein levels. A substantial reduction of CENH3 transcript was achieved in A. thaliana (27–43%)²¹, cotton (over 80%)²³ and maize (21–65%)¹⁸. It is interesting to note that plants with CENH3 knockout alleles in heterozygous state in A. thaliana and maize also survive^6,12,14. Thus, it indicates these species i.e., Maize and A. thaliana can tolerate substantial reduction in CENH3 transcript. Segregation distortion and nonrecovery of homozygous lines for AcCENH3-RNAi locus in onion shows that slight reduction in CENH3 could affect its survival. This could be possibly due to insufficiency of AcCENH3 chromatin to support normal segregation of large chromosomes in onion knockdown lines.

AcCENH3 knockdown lines show poor seed set in self-fertilized and outcross progeny

In the second season, all T₁ plants from all events produced seeds upon self-fertilization except for E3 (Table 1 and Supplementary Figs. 3, 6). The T₂ seed yield per umbel from the flowering transgenic events was lower than that from the WT. The seed set efficiency in transgenic events E1, E2, and E5 was 27.10%, 29.86%, and 66.50%, respectively, relative to the WT (Table 1 and Supplementary Fig. 6). Furthermore, a correlation analysis between AcCENH3 quantity (OD values from ELISA) and relative seed set efficiency revealed a correlation coefficient of 0.962. This suggested that a decrease in the AcCENH3 protein impacted the seed set (Supplementary Fig. 7). The seed set efficiency upon outcrossing was significantly lower in the transgenic lines with WT compared to the WT × WT crosses. When AcCENH3 knockdown lines served as the female parent, relative seed set efficiency was 49.32%, 67.34%, and 81.57% in events E1, E2, and E5, respectively (Table 1 and Supplementary Fig. 8). Conversely, when AcCENH3 knockdown lines were used as the male parent, the relative seed set efficiency was 37.89%, 61.82%, and 83.76% for events E1, E2, and E5, respectively (Table 1 and Supplementary Fig. 8).These findings further suggested that the knockdown of AcCENH3 affected the seed set. The differential seed set could be one of the reasons for the biased segregation distortion of the transgene observed in the reciprocal crosses. Our observations are similar to the previous studies, emphasizing that a decline in CENH3 levels is associated with a decrease in seed set efficiency both during selfing²¹ and when crossed with WT^6,14,18.

AcCENH3 knockdown lines show in vivo haploid induction

In our study, the F₁ population derived from reciprocal crosses of AcCENH3 knockdown lines (heterozygous) with WT was analysed using FCM for estimating the ploidy level (Fig. 1C). Progenies from events E1 and E2 showed haploid induction (Table 2). In these events, the HI efficiency (HIE) ranged from 2.04 to 4.63%. The highest HIE was observed in E1 (4.63%). While aneuploids were observed in progenies of all events, there was no HI in the progenies of E5 and WT × WT crosses. Haploids identified by FCM were further confirmed for the ploidy by cytological staining of root tips (Fig. 1E). The haploids showed poor growth when compared to diploid progenies (Fig. 1B). Microscopic analysis of leaf surface revealed a 41.25% decrease in the size of stomata in haploid plants, indicating cell size reduction (Fig. 1D and Supplementary Fig. 9). Our result was in accordance with Foschi and co-workers²⁴. Interestingly, the HIE was noticeably higher when the AcCENH3 knockdown parent served as the male, but statistical analysis analysis of male vs female based HI using Fisher’s exact test showed no significant difference (Table 2). A comparative analysis of HIE across various plant species, revealed that A. thaliana recorded the highest HIE at approximately 40%⁶. In crop plants, HIE was comparatively lower than A. thaliana. Cotton exhibited the highest HIE (8%)²³, followed by wheat (7%)¹³ and maize (5.2%)¹². In other species, rice demonstrated an HIE of 1%, similar to cucumber and melon, while tomato had a slightly higher rate of 2.3%, and these rates pertain to EMS mutants²⁵. In this study, we achieved an HIE of 4.63% in onion, which is close to the efficiencies observed in maize.

Since the breakthrough study of Ravi and Chan, several methods have been devised to develop CENH3-based HI lines⁶. Among them, a simple approach has been to maintain the knockout allele under a heterozygous state for inducing haploids upon crossing with WT lines¹². The heterozygous knockout allele for CENH3 in maize might have led to a reduction in the centromere size by dilution of CENH3 in the cellular environment and could have resulted in genome elimination in a competitive environment when crossed with WT^12,26. Similar observations were reported in A. thaliana i.e., heterozygous knockout when crossed with WT resulted in 0.83% of haploids in the progeny¹⁴. These results indicate that Arabidopsis can tolerate a significant reduction in CENH3. On the other hand, in onion, even a slight reduction in CENH3 levels leads to seed sterility and HI (Tables 1 and 2). These observations in Arabidopsis, maize and onion suggest the existence of a species-specific critical threshold for CENH3 chromatin. Our approach to develop knockdown-based HI lines builds upon the work by Wang and co-workers in maize¹², offering an alternative when the generation of knockouts is not feasible due to a higher CENH3 threshold.

In this report, we demonstrate successful generation of HI lines in onions. The induction rates are comparable to those achieved through in vitro gynogenesis²⁷. However, our in vivo HI method offers breeders a cost-effective approach to produce haploids from both male and female gametes. This method also eliminates the dependency on genotype-specific regeneration processes. Furthermore, our approach facilitates the transfer of traits such as male sterility and genome editing across different genotypes in a single generation, bypassing the labour-intensive backcross breeding. As a result, this methodology holds significant promise for transforming onion breeding practices.