Single nucleus sequencing reveals spermatid chromosome fragmentation as a possible cause of maize haploid induction

Abstract

Production of maternal haploids using a conspecific haploid inducer is routine and highly efficient in maize. However, the underlying mechanism of haploid induction (HI) is unclear. We develop a method to isolate three nuclei from a pollen grain and four microspores from a tetrad for whole-genome sequencing. A high rate of aneuploidy is observed at the three-nucleus stage (6/22 pollens) rather than at the tetrad stage (1/72 microspores) in one HI line CAU5. Frequent aneuploidy is also observed in another two inducer lines, but not in two regular lines, which implies that HI may be associated with pollen aneuploidy. We further sequence the individual embryos and endosperms of 88 maize kernels crossing between regular and inducer lines. Genome-wide elimination of the CAU5-derived chromosome is identified in eight of 81 embryos. Together, these results suggest that continuous chromosome fragmentation occurring post meiosis in the gametophyte may cause haploidy of the embryo.

Fig. 1

Pollen fertility and viability differs between regular and inducer lines. a Three levels of pollen fertility, (i) typical abortion type, (ii) spherical abortion type, and (iii) wild type, measured by KI/I₂ staining, and counted in the six lines. b Pollen fertility of B73-inducer and B73, stained by KI/I₂. c Five degrees of pollen viability, (i) high, (ii) medium, (iii) low, (iv) no viability, and (v) abortion, measured by the TTC staining and counted in the six lines. d pollen viability of B73-inducer and B73, stained by TTC. Scale bar = 1 mm. The fertility and viability photographs and statistic analysis of all six lines are also provided in Supplementary Fig. 4

Fig. 2

Chromosomal fragment deletions identified in inducer CAU5 pollen. a Bright illumination plus fluorescence shows pollen cut with the tip of a glass micropipette. b One putative trophic nucleus and two putative sperm visualized by fluorescence. In a, b, DNA is visualized by SYTOX Green fluorescence (bar = 100 μm). c–g Pollen grains 1–5 have only one or two defective chromosomes. Some fragments were deleted in one sperm (shown on light green background) or not segregated (deletion in one sperm and diploidy in another sperm, shown on tan color background). h Pollen 6 containing two sperm with a large-scale deletions in almost every chromosome. The darker green background shows the deletion regions that occurred in the two putative sperm. In c–h, blue and red spots represent CNVs in odd and even chromosomes, respectively; the three data sets are results of putative trophic nucleus, putative sperm 1, and putative sperm 2, from top to bottom; the black, blue, and red lines around spots indicate normal haploidy, deletion, and diploidy, respectively. The black triangles point to centromeres

Fig. 3

Chromosomal fragment deletions in different pollen developmental stages. a The number of aneuploid sperm and pollen for B73 and the B73-inducer line. b The degree of chromosome fragmentation (in CAU5) was estimated in meiotic and mitotic stage by single cell and nucleus sequencing and observation under microscope. ^∗ χ ² test, P represents the difference of aneuploid frequency of B73-inducer sperms (14/52) and B73 sperms (3/50). ^#χ² test, P represents the significant difference of aneuploid frequency of B73-inducer pollens (9/26) and B73 pollens (3/25). ^$ χ ² test, P represents the difference of aneuploid frequency of CAU5 pollens (6/22) and CAU5 microspores (1/72)

Fig. 4

Single embryo and endosperm sequencing reveals two classes of aneuploidy. In each panel, the upper dot plot represents CNV by the log₂(CNR) value, CNV is identified if continuous and dispersed obviously Log₂(CNR) values between large segments ( > 10 Mb) are observed. The lower dot plot represents the genetic background SNP-ratio by log₂(RGR) value, which means paternal genome deletion if the value is >0, in contrast, maternal genome deletion if the value is <0. The upper and lower figure of each panel is referring to the same sample. a The normal euploid embryo (embryo 1). A total of 71 embryos and 80 endosperms were identified as euploid. b The complete haploid embryo (embryo 16), in which CAU5-derived chromosomes are eliminated. A total of seven (embryos 16, 19, 21, 32, 38, 45, and 73) were identified as haploid. c One incompletely haploid embryo (embryo 44), in which fragments of chromosomes 1, 7, 8, and 9 remain. d Chromosome 5 of one embryo (embryo 69), in which the CAU5-derived fragments were lost. e Chromosome 1 of one embryo (embryo 78), in which the Zheng58-derived fragments were lost. f Chromosome 4 of one endosperm (endosperm 13), in which the Zheng58-derived fragments were lost. The blue and red spots represent CNVs in odd and even chromosomes, respectively. The black triangles point to centromeres. In the CNV plots, the green, black, and yellow lines around spots indicate haploidy, diploidy, and triploidy, respectively. In the SNP-ratio plots, the blue and red shadings indicate paternal and maternal deletions, respectively

Fig. 5

A model for haploid generation. Two classes of aneuploid pollen grains were identified. In theory, both classes should carry one aneuploid sperm or two aneuploid sperm, and fertilize egg and/or polar nucleus. Class I pollen grains could induce haploid embryos, genome elimination endosperms, and kernels with approximately equal frequency. Class II pollen grains can fertilize egg and/or polar nucleus and may generate normal or abnormal kernels with a few chromosome fragments lost but not haploid kernels. The frequency of class I is higher than class II based on the present data. Eight haploids with paternal genome elimination were detected in 81 embryo samples. However, no paternal genome elimination was detected in 81 endosperm samples. It is possible that Class I endosperm with genome elimination causes abnormal kernels and may have been ignored when sampling