In maize, the Indeterminate gametophyte1 (ig1) mutation was first observed as a spontaneous mutation in the inbred line Wisconsin-23 (W23), resulted in approximately 3% HIR (Kermicle, 1969). The ig1 gene is localized on the long arm of chromosome 3, 90 cM from the most distal locus on the short arm (Kermicle and Demopulos-Rodriguez, 1980; Coe, 1992). It encodes a LATERAL ORGAN BOUNDARIES (LOB)-domain protein (Evans, 2007), which is associated with a large, plant-specific family of transcription factors (Husbands et al., 2007). This single recessive gene causes a range of embryological abnormalities, including atypical fertilization events that lead to ploidy variation in both the embryo and the endosperm (Kermicle, 1969, 1971, 1994; Lin, 1981).

Several embryological abnormalities have been described in ig1 embryo sacs, both in mature (Lin, 1978) and developing stages (Lin, 1981). Lin (1978) observed the phenotype of maize ovules from the Wisconsin dent inbred line W23 (ig1/ig1) when used as the female parent at the time of silk emergence, corresponding to the mature female gametophyte stage. The results revealed that the structure of mature embryo sacs in ig1/ig1 plants differs markedly from that in wild-type (Ig1/Ig1) plants in several respects: (1) Wild-type plants consistently produced a single micropylar cell (egg), whereas ig1 embryo sacs exhibited a range of 0 to 0.5 structurally analogous cells per sac. (2) In ig1 embryo sacs, the primary central cell was frequently accompanied by one or two additional, smaller central cells. These additional cells were occasionally binucleate but more often uninucleate. In contrast, wild-type embryo sacs invariably contained a single central cell. (3) The primary central cell in ig1 sacs exhibited considerable variability, containing between one and six polar nuclei. Wild-type sacs uniformly harbored two polar nuclei. (4) In ig1 embryo sacs, some polar nuclei were dispersed throughout the cell, while others occupied the typical apical position near the egg cell, opposite the micropyle. During double fertilization, only a subset of the micropylar and polar nuclei appeared to be functionally competent. The average number of unfertilized micropylar nuclei was 0.88 per ovule, while unfertilized polar nuclei averaged 0.75 per ovule. These unfertilized nuclei included both those located in accessory central cells and those in the primary central cell, but were displaced from their typical spatial location.

Lin (1981) analyzed 62 mature ig1 female gametophytes from the maize inbred line W23. Of these, only two exhibited the defining characteristics of the wild-type structure: a single micropylar cell —consistent with synergid regression in W23, leaving only the egg cell—and a binucleate central cell with nuclei positioned adjacent to the egg. In the ig1 samples, the number of cells present in the micropylar region ranged from zero to five, averaging 1.96. The number of polar nuclei averaged 3.42 per gametophyte, and approximately 18% of the embryo sacs contained one or more accessory central cells. Notably, among the ten cases examined, fertilization events involving accessory central cells were not observed. Correspondingly, the phenomenon of endosperm doubleness was absent in mature seeds. Further cytological analysis by Lin revealed that in kernels developing on ig1/ig1 plants, the number of maternally derived chromosome sets in the endosperm corresponded to the number of polar nuclei closely appressed to the egg cell, rather than with the total number of polar nuclei present in the primary central cell (Kermicle, 1994).

Huang and Sheridan (1996) conducted a comparative study of nuclear dynamics and microtubule organization in wild-type plants and ig1 mutants. In ig1 mutants, the second and third mitotic divisions within the embryo sacs are not fully synchronized. In addition, supernumerary mitoses may occur, either throughout all nuclei or specifically in the micropylar or central regions. Following cellularization, individual micropylar cells can undergo mitosis. Aberrant microtubule behavior in the mutant leads to irregular nuclear positioning, asynchronous microtubule organization among nuclear pairs, and abnormal phragmoplast formation after the third mitotic event. These results indicate that the ig1 gene plays a primary role in the regulation of nuclear division and a secondary, yet significant, role in modulating microtubule dynamics. The gene’s influence on the cytoskeleton likely contributes to proper polarization and nuclear migration, processes that are critical for normal embryo sac development and cellular differentiation (Huang and Sheridan, 1996).

Evans (2007) positionally cloned ig1 in a genetic cross between ig1-Ow23/+Mo17 females and +/+Mo17 males. Fine mapping using simple sequence repeat (SSR) markers localized ig1 between umc1311 and umc1973 on maize chromosome 3. The gene was subsequently cloned based on the genome sequence of the orthologous region in rice. A second ig1 mutant allele was identified via a non-complementation screen involving active Mutator transposable element lines. Homozygous ig1 mutants exhibit pronounced developmental abnormalities, including altered leaf morphology and defective embryo sac formation. Mutant leaves display disrupted abaxial–adaxial polarity and fail to suppress the expression of meristem-specific knotted-like homeobox (knox) genes in developing leaf primordia. This mis- regulation results in the persistence of proliferative, stem cell-like identity. Notably, despite phenotypic similarities between ig1-O leaves and embryo sacs, ectopic knox gene expression is not detected in ig1-O embryo sacs (Evans, 2007).

Recently, Kyada et al. (2024) identified eight distinct haplotypes of the Ig1 gene across eight wild-type inbred lines (with Ig1 comprising two exons) and two mutant inbreds (ig1, with a single exon). The study reported a haplotype diversity index of 0.956 and a nucleotide diversity (Pi) value of 0.013. The insertion of Hopscotch retrotransposon in ig1 resulted in a truncated protein with 138 amino acids, in contrast to 260 amino acids in wild-type IG1 protein. The study also revealed that the disruption of the leucine zipper domain in the mutant protein accounted for its loss of function. To facilitate genotyping, a functional marker, MGU-IG1-Hopscotch, specific to the Hopscotch insertion, was developed and successfully validated in five F₂ populations. Across these populations, the average HIR was notably higher in mutant homozygotes (ig1/ig1; 2.36%) than in heterozygotes (Ig1/ig1, 1.49%). Furthermore, ig1-positive plants exhibited a twin-embryo seed phenotype at frequencies ranging from 0 to 1.6%.

Since 1959, Coe observed self-pollinated progenies of an inbred line, designated Stock 6, produced 343 haploids out of 10,616 plants, corresponding to an average haploid induction frequency of 2%, with a maximum of 3.23% when used as a male parent. Subsequently, Deimling et al. (1997) identified a major QTL on chromosome 1 (interval 1.03–1.06) from a cross between W23ig and Stock6. Later, Barret et al. (2008) reported the same chromosomal region (bin 1.04) in a study of 471 F₂ plants derived from a cross between DH99, a non-inducer female parent, and PK6, a non-glossy maize line developed from Stock6 with an average HIR of 6%. Genotyping of the 19 highest-inducing F₂ plants using 101 microsatellite markers led to the identification of gynogenesis inducer 1 (ggi1) (Barret et al., 2008).

Prigge et al. (2012) evaluated HIR across 1,061 progenies derived from four segregating populations involving two temperate haploid inducers UH400 (HIR = 8%) and CAUHOI (HIR = 2%), alongside one temperate and two tropical non-inducer inbreds (HIR = 0%). Each population was examined over three successive generations. The QTL analysis identified qhir1 as a major locus, explaining up to 66% of the observed genetic variance in the three populations that involved a non-inducer parent. The allele enhancing HIR at this locus was contributed by UH400.

Dong et al. (2013) conducted fine-mapping of the qhir1 locus and identified markers closely linked to this region. Through fine-mapping efforts involving 14,375 F₂ individuals derived from a cross between 1680 and UH400, polymorphic markers specific to the qhir1 region were developed. This work successfully narrowed the original qhir1 interval from 3.57Mb to a 243-kb region flanked by markers X291 and X263. Subsequently, Hu et al. (2016) employed genome-wide association studies (GWAS) and further dissected the qhir1 locus into two sub-QTLs: qhir11 (0.54 Mb), which contained the 243-kb interval previously identified by Dong et al. (2013), and qhir12 (3.97 Mb), within the broader 50.34 Mb qhir1 region. Then, Nair et al. (2017) characterized the functional contributions of these sub-regions through a cross between CML269 (a non-inducer) and TAIL8 (a tropicalized haploid inducer). Their analysis revealed that only the qhir11 sub-region significantly affected haploid induction ability, and was also associated with traits such as segregation distortion and kernel abortion, hallmark features linked to maternal haploid induction.

In 2017, the ZmPLA1/MTL/NLD characterization was studied and reported by three research groups identified GRMZM2G471240, a gene encoding a patatin-like phospholipase, as a candidate. A 4-bp insertion in its fourth exon induced a frameshift at amino acid 380, resulting in 20 altered residues and a premature stop codon that truncated the protein by 29 amino acids. Multiple functional validations were set through RNA interference (RNAi), transcription-activator-like effector nucleases (TALEN), and genome editing to confirm the role of this gene in haploid induction (Kelliher et al., 2017; Liu et al., 2017; Gilles et al., 2017).

Later, Gilles et al. (2021) observed and characterized NLD localization in maize sperm cells and revealed that the gene is localized on the endo-plasma membrane of sperm cells and is responsible for sperm cells’ PI(4,5)P2 signaling to the embryo sac and helps to deliver the sperm cells. Therefore, they were suggested that the loss of function of NLD would impact sperm cells’ failure to localize in the female egg, leading to the formation of haploid embryos with only the chromosomes from the female parent.

Most recently, Jiang et al. (2022) conducted a comprehensive multi-omics analysis integrating transcriptomic, proteomic, metabolomic, and protein modification datasets to explore zmpla1 mutant anthers. Their findings indicated that differential molecular entities were significantly enriched in pathways associated with the oxidative stress response. Notably, they discovered that inducing a reactive oxygen species (ROS) burst via chemical treatment of pollen could trigger haploid induction, suggesting that ROS-mediated signaling plays a critical regulatory role in the haploid induction process.

Prigge et al. (2012) identified seven QTL across five chromosomes in the CAUHOI × UH400 population, with qhir8 on chromosome 9 accounting for 20% of the HIR variation across three generations. At the time of discovery, qhir8 was classified as a minor QTL. Subsequently, Liu et al. (2015) fine-mapped the qhir8 region and assessed its impact on HIR, segregation distortion (SD), and embryo abortion (EmA). A total of 3,989 F₂ plants derived from the CAUHOI × UH400 cross were screened for recombinants within the qhir8 region. Later, F₂ plants and F₃ progenies from 34 recombinant lines were evaluated for HIR, SD, and EmA, using 31 markers spanning the target region. Their analysis confirmed that qhir8 significantly enhanced HIR and EmA, but had no discernible effect on SD. The qhir8 locus was narrowed down to a 789 kb region flanked by markers 4292232 and umc1867.

Zhong et al. (2019) leveraged a fine-mapping approach using 16 F₃ families derived from a cross between CAU5 and CAUHOI, narrowing the qhir8 region to a 138-kb region flanked by markers ZS4307 and ZS4446. Sequencing within this region utilized two positive bacterial artificial chromosome (BAC) clones from CAU5, yielding detailed sequence information and additional polymorphic markers. Using these markers, the mapping interval was further reduced to a 318-bp region through analysis of 21 newly generated F₃ families. Ultimately, the qhir8 locus was localized within the coding sequence of GRMZM2G465053, which encodes a DUF679 domain-containing membrane protein, subsequently designated ZmDMP. A single-nucleotide polymorphism (SNP) was identified within this region: a thymine (T) in CAUHOI was replaced by a cytosine (C) in CAU5, located 131 bp downstream of the start codon (ATG). This SNP resulted in an amino acid substitution from methionine to threonine, implicating this mutation as the causal variant underlying the enhanced HIR observed in CAU5.

Functional validation was performed by generating ZmDMP knockout lines (zmdmp-ko) using the clustered regularly interspaced short palindromic repeats (CRISPR)–CRISPR-associated protein 9 (Cas9) system. The knockout induced haploid formation with a HIR of 0.1–0.3%, and when combined with mtl/zmpla1/nld mutations, the HIR was elevated by 5–6-fold. The haploid status of the progeny was verified using 10 polymorphic molecular markers and flow cytometry, confirming the presence of only the maternal genotype in all cases.

Consistent with the functional profile of qhir8, ZmDMP disruption also led to an increased frequency of endosperm-aborted kernels (EnAs). Expression analysis using quantitative reverse transcription PCR (qRT–PCR) demonstrated that ZmDMP and zmdmp are highly expressed in mature pollen, with markedly lower expression levels in immature anthers and kernels at various developmental stages. Notably, expression of zmdmp in mature pollen was significantly higher than that of ZmDMP, likely reflecting a feedback regulatory mechanism. These findings suggest that ZmDMP plays a significant role during the late stages of male gametophyte development and establishes it as a key genetic determinant of haploid induction. The results clearly recognized that the zmdmp gene is responsible and plays a critical role in haploid induction in maize (Zhong et al., 2019).

In Li et al., 2021, characterized the role of phospholipase-mediated haploid induction in maize by analyzing RNA-seq data previously reported by Walley et al. (2016), which profiled different tissues of maize inbred line B73. They aimed to identify pollen-specific members of the phospholipase gene family. Among the candidates, only one gene, ZmPLD3, was exclusively expressed in pollen and significantly upregulated in the mtl/zmpla1/nld mutant background. Quantitative reverse transcription PCR (qRT–PCR) further confirmed that ZmPLD3 was highly expressed in mature pollen relative to anthers at various developmental stages.

ZmPLD3 encodes a, putative phospholipase D (PLD), locate on chromosome 6, characterized by its conserved hydrolytic active-site motif (HKD motif, HxKxxxxD) (Wang, 2001). Sequence analysis revealed the presence of two HKD domains within the ZmPLD3 protein. To elucidate its functional role, CRISPR-Cas9-mediated gene editing was employed to generate two independent knockout lines, zmpld3–1 and zmpld3-2, which were subsequently selected for further study. The HIRs of zmpld3–1 and zmpld3–2 were 0.96% and 0.85%, respectively, which did not significantly differ from that of mtl/zmpla1/nld (1.2%).

To assess potential genetic interactions between ZmPLD3 and previously reported haploid induction genes, double mutants (zmpld3-mtl, zmpld3-zmdmp, and mtl-zmdmp) were generated through hybridization of the corresponding single mutants. Statistical analyses revealed that the combination of zmpld3 and mtl/zmpla1/nld elevated the HIR to approximately 4%. In contrast, the HIR of the zmpld3-zmdmp double mutant was not significantly higher than that of the zmpld3 single mutant, although zmpld3 alone exhibited a significantly higher HIR compared to zmdmp.

Subcellular localization of ZmPLD3 was examined using a maize protoplast system co-expressing compartment-specific marker protein. The results indicated that ZmPLD3 localizes to multiple cellular compartments, including the endoplasmic reticulum, plastids, Golgi apparatus, and cytosol, but is likely absent from the plasma membrane, mitochondria, pre-vacuolar compartment, nucleus, and peroxisomes.

Recently, Jiang et al. (2022) discovered that a simple chemical treatment of pollen with ROS reagents could effectively induce haploid formation in maize. Through an integrated multi-omics analysis—including transcriptomics, metabolomics, quantitative proteomics, and protein modification profiling—of zmpla1 mutant anthers, they identified three genes encoding sperm-specific peroxidases: ZmPOD65 (Zm00001d017996), ZmPOD60-1 (Zm00001d027710), and ZmPOD60-2 (Zm00001d048413). Targeted mutagenesis of ZmPOD65 using CRISPR-Cas9 yielded two distinct mutants: one carrying a missense mutation at amino acid (aa) position 73 (methionine to leucine substitution) and another harboring a frameshift mutation between aa positions 50 and 142. Flow cytometric analysis (FACS) of T₁ seedlings from the ZmPOD65^+/Met→Leu^ line and T₂ seedlings from the ZmPOD65^+/frameshift^ line revealed HIRs of 7.7% and approximately 1%, respectively. In contrast, no haploid seedlings were detected among mutants of ZmPOD60–1 or ZmPOD60-2, underscoring a specific role for ZmPOD65 in haploid induction and supporting the critical involvement of ROS dynamics in this process.

Based on these findings, Jiang et al. (2022) proposed a mechanistic model wherein a centrally triggered ROS burst underpins haploid induction during the second mitosis of pollen development: (1) elevated phosphatidylcholine (PC) levels in sperm cells stimulate excessive ROS production; (2) this oxidative stress disrupts the redox homeostasis of the metabolome, leading to sperm DNA damage and initiating multilayered regulatory responses; (3) persistent DNA breakage, extending from centromeric regions to the entire sperm genome, surpasses the capacity of DNA repair mechanisms. Following fertilization, the fragmented male genome is progressively degraded, ultimately resulting in haploid progeny.

Centromere-specific histone H3 (CENH3) is highly conserved of all known kinetochore proteins (Henikoff et al., 2001). Data from several organisms indicate that CENH3 replaces histone H3 on active centromeric DNA (Yoda et al., 2000; Ahmad and Henikoff, 2001; Lo et al., 2001; Blower et al., 2002) and is required to recruit other key kinetochore proteins such as CENP-C (Hooser et al., 2001; Ando et al., 2002). Each of the known CENH3s shares a common histone H3 core sequence, but they diverge in the N-terminal tail and an internal region known as loop 1 (Talbert et al., 2002). The histone fold domain (HFD) of CENH3 comprises six secondary structural elements, including four alpha helices (αN, α1, α2, and α3) and two loops (Feng et al., 2020). Both diverged regions interact with DNA in the nucleosome and show evidence of adaptive evolution, suggesting that CENH3 serves as a linker molecule between the rapidly evolving centromeric DNA and the conserved kinetochore machinery (Malik and Henikoff, 2002).

Because of its close association with DNA in the context of the centromeric nucleosomes, CENH3 has been used as a tool to identify the centromere sequences that interact with the kinetochore. In situ hybridization (Haaf and Ward, 1994; Sart et al., 1997) and chromatin immunoprecipitation (ChIP) have been used to show that maize CENH3 is present in kinetochores throughout the cell cycle. ChIP analysis indicates that maize CENH3 interacts strongly with the centromeric satellite CentC but does not interact with noncentromeric DNA sequences. The ChIP assays also demonstrate that a recently isolated centromeric retrotransposon in maize (CRM) interacts with CENH3 throughout its length. These data provide strong support for the idea that specific sequences confer centromere identity in maize and that a conserved retrotransposable element is part of the functional centromere (Zhong et al., 2002).

Ravi and Chan (2010) studied on genome elimination of the centromere-specific histone CENH3 gene in Arabidopsis thaliana plants by fluorescent protein (GFP)−tailswap. The result found that haploids are efficiently generated at 25–45% from a cross of GFP–tailswap with wild-type. However, several unusual phenotypes in the F₁ progeny were observed, such as approximately 80–95% of fertilized ovules aborted early in development, leading to inviable seeds.

Later, Maheshwari et al. (2015) investigated the function of the CENH3 gene in maize by genetic modification under the control of the endogenous CENH3 promoter. Cenh3-1/CENH3 heterozygotes were created using these transgenes. In the T₁ generation, they discovered Zea mays CENH3 transformants in the homozygous background cenh3–1 are embryo lethality. This finding demonstrates that CENH3 may play a significant role in haploid induction in these plant groups (Rai et al., 2023).

Then Kelliher et al. (2016) engineered two haploid inducer lines, including CENH3−/− and CENH3: RNA interference (RNAi) lines by AcGREEN-tailswap-CENH3. Subsequently, those inducers were test crosses to wild-type plants and evaluated HIRs. Results found that CENH3: RNAi lines did not consistently knock down CENH3 and only occasionally produced haploids. When backcrossed as males, the CENH3−/− showed HIR of gynogenic in many hemizygous individuals reached a maximum of 3.6%, with a lower rate of male sterility. These findings suggest that in vivo haploid induction systems may be engineered into maize plants using CENH3-tailswap transgenes. However, the use of a single RNAi copy line doesn’t prove effective for haploid production, indicating the need for a multi-copy RNAi line (Rai et al., 2023).

After that Wang et al. (2021) presented a method for producing haploid progeny involving mating maize heterozygous lines with wild-type plants in both directions (cenh3 null mutants +/cenh3 (gl1, gl8, and cenh3-mu1015598 transposon insertion lines). When +/cenh3 heterozygotes were crossed as males, 0.5% of the progeny were glossy, but 5.2% of the offspring were glossy when +/cenh3 plants were used as females. genome elimination (GE) in the gametophyte was detected by the genotype for CENH3, proving that CENH3 dilution during postmeiotic cell divisions before gamete production is what causes centromere failure.

Recently Meng et al. (2022) developed maternal haploid inducer lines derived from Stock6 (3.2%) with genetic modifications in CENH3 gene. Subsequently, overexpressed fluorescently protein-tagged maize CENH3 to modify the Stock6-derived inducer lines (CAUHOI and CAU5), resulting in an increased maternal HIR of 6.1%. The HIR increased to 16.3% when replacing the full-length CENH3 in the tagged expression cassette with a tail-altered version in CAU5^{M-tailswap-RFP}, while it increased to 8.8% from 5.3% in CAUHOl^{M-tailswap-RFP}. These findings provide a potentially workable hypothesis for further improving the induction rates of maize CAU5 (mtl/zmpla1/nld + zmdmp) and CAUHOI (mtl/zmpla1/nld) inducer lines (maternal haploid) and indicate the potential of combining two in-vivo haploid induction strategies to improve the success rate of generating haploid populations in maize (Rai et al., 2023).

Most recently, it was found that CENH3 not only has broader applicability as both maternal and paternal parents in haploid induction but also exhibits temperature sensitivity due to reduced HIR when the temperature decreases. As the previous report indicated that CENH3 disruption caused embryo lethality (Maheshwari et al., 2015). Yang et al. (2025) proposed moderate modifications of CENH3 using microRNA-induced gene silencing (MIGS) driven by the RPS5a promoter (Felippes et al., 2012), which is activated only during gametogenesis and early embryogenesis (Weijers et al., 2001) to knock down the gene in Arabidopsis thaliana. The result found that at 22°C, a transgenic line (pRPS5a::MIGS #1) displayed a low HIR of 0.2%, while the HIR increased significantly to 0.2%–3.4% at 25°C and 6.1%–14.2% at 30°C across all lines, and aborted seed rates also rose with temperature. Also found that the combination of null cenh3–1 mutant with pRPS5a::MIGS and GFP-CENH3 in heterozygous form can increase the HIR by 1.31-fold to 1.97-fold compared with that of the null cenh3–1 mutant with GFP-CENH3 alone, and increase to 66.7% and 61.4% in T₂ and T₃ generations in homozygous form at 22°C. And found that all the plants were maintaining healthy, fertility, and pollen viability.