Based on interbreeding, the biological species concept stated, “Species are groups of interbreeding natural populations that are reproductively isolated from other such groups.” (Mayr, 1996; Coyne and Orr, 2004). It literally means that the boundaries between the population of one species and that of another are defined by reproductive barriers, and the mechanism itself is referred to as “reproductive isolation”, in which genetically based intrinsic barriers prevent gene flow between populations of different species (Nosil, 2013). Reproductive isolation maintains species identity, and evolutionary biologists are highly interested in this area of study. Reproductive isolation is considered to be incidentally acquired as a by-product of other divergences between species' populations (Nosil, 2013). The genetic variations gradually increase over time, and the accumulation of those changes and divergences is probably either neutral or wholesome to its own genetic background while it works as a deleterious factor in other alternative genetic backgrounds, exclusively in a heterozygous state (Turelli, 1998). It is possible for hybrid dysfunction to appear due to a divergence in multiple genomic regions across the whole genome in a wide variety of species populations. The Bateson-Dobzhansky-Muller (BDM) model is a model in which negatively epistatic interaction between divergent alleles contributes deleterious effects within hybrid populations (Bateson, 1909; Dobzhansky, 1937; Muller, 1942; Kubo et al., 2008; Nosil, 2013; Xie et al., 2019a) and is a widely accepted model of the emergence of reproductive barriers. However, are genes for reproductive isolation always neutral or wholesome when they spread in a population? By summarizing genes for hybrid sterility, a form of reproductive barrier, we will discuss how selfish genes act to build species barriers in rice.

In plants, reproductive barriers can be basically divided into two categories, namely prezygotic reproductive isolation and postzygotic reproductive isolation, based on the developmental stage in which they give rise (Koide et al., 2008b; Ouyang and Zhang, 2013; Zin Mar et al., 2021). While prezygotic isolation occurs during the formation of the zygote, the latter restricts the introgressive gene flow in crossed populations, inducing hybrid arrest after fertilization at different developmental stages and/or other advanced generations (Chen et al., 2016). Hybrid sterility, the gametic disorder at their reproductive stage with the failure to produce fertile male and/or female gametes in normally grown hybrid plants, is one form of postzygotic reproductive isolation. The genus Oryza, which consists of two cultivated rice species and 22 wild rice relatives (Ammiraju et al., 2010), is a valuable pool for improvement of agricultural traits. In rice genetics, interspecific and intraspecific hybrid sterility, which is one of the most pronounced forms of postzygotic reproductive isolation, has been the most extensively investigated subject across a wide variety of genomic regions in a substantial number of different rice populations (Ouyang and Zhang, 2013; Li et al., 2020).

Based upon either sporophytic or gametophytic function and the number of loci involved, altogether four separate genic models underly rice hybrid sterility: namely (1) One-locus sporo-gametophytic interaction, (2) Duplicate gametic lethals, (3) One-locus sporophytic interaction, and (4) Complementary sporophytic interaction (Oka, 1957; Sano et al., 1979; Koide et al., 2008b). Among them, one-locus sporo-gametophytic interaction fits with genetic mechanisms for the majority of sterility loci in rice (Oka, 1957; Xie et al., 2019a). Therefore, we will focus on single or tightly linked sterility loci in this review (see Table 1). Hybrid sterility loci, such as S1, S2, and S5, can perfectly explain how one-locus sporo-gametophytic model works in rice sterility (Sano et al., 1979; Chen et al., 2009; Yang et al., 2012, 2016; Xie et al., 2017; Koide et al., 2018; Zin Mar et al., 2021). As an example, for S2 locus, although S2^g (allele derived from oryza glaberrima) and S2^s (allele derived from O. sativa) are neutral in their respective backgrounds, the incompatible interaction between these two alleles occurred in heterozygous hybrids. As a result, neither male nor female gametes which carry S2^g allele survive, causing preferential transmission of selfish allele S2^s in later generations (Sano et al., 1979; Zin Mar et al., 2021). Such a preferential transmission of one of the two alleles is referred to as transmission ratio distortion (TRD). Selfish genes causing TRD take advantage over their alternative alleles, and preferentially promote their own distribution in populations, with the help of incompatibilities that occur in heterozygous hybrids.

Although several underlying mechanisms behind the TRD phenomenon, including non-random segregation of chromosomes during meiosis (Pardo-Manuel de Villena and Sapienza, 2001; Birchler et al., 2003; Fishman and Willis, 2005; Koide et al., 2008a), unequal gametic success in fertilization (Price, 1997; Diaz and Macnair, 1999; Seymour et al., 2019), and embryo lethality (Lyttle, 1991; Silver, 1993; Price, 1997; Diaz and Macnair, 1999; Ubeda and Haig, 2005; Moyle, 2006) has been reported in plants, selective gametic abnormality (Lyttle, 1991; Silver, 1993; Ubeda and Haig, 2005; Moyle, 2006; Koide et al., 2008a) caused by hybrid sterility locus is the most frequently observed in Oryza species. The selective abnormality occurs in either male/female-gametes or sex-independently. In this report, the former and latter are termed mTRD (male specific transmission ratio distortion)/f TRD (female specific transmission ratio distortion) and siTRD (sex-independent transmission ratio distortion), respectively, (Maguire, 1963; Sano, 1983; Koide et al., 2008c; Ouyang and Zhang, 2013). Therefore, in general, hybrid sterility genes in rice can be divided into three subcategories: pollen killer (PK: which induces hybrid male sterility and mTRD), egg killer (EK: which results in hybrid female sterility and f TRD), and gamete eliminator (GE: which eliminates both pollen and embryo sac causing siTRD) (Figure 1).

Several loci causing TRD have been cloned and their underlying mechanisms have been studied. The S5 locus is the most renowned hybrid sterility locus in the study of rice reproductive barriers. This locus is an example of “Killer-Protector system”, and its effect is shown in the failure of embryo sac fertility (We note that “Killer” in the Killer-Protector system and “Killer” in PK or EK have different actions. The former “Killer” induces the abortion of gametes with both alleles. The “Protector” selectively protects gametes with one of two alleles from the “Killer” function. As a result, the preferential dysfunction of gametes with one of two alleles occurs. The factor causing preferential function is referred to as PK, EK, or GE. Thus, PK and EK is a concept of a factor with the combined function of “Killer” and “Protector”.). Three different tightly linked genes (ORF3, ORF4, and ORF5) regulate the tripartite complex in indica-japonica crossed species. The indica-derived S5 allele (ORF3+, ORF4–, and ORF5+) contained both functionally active Protector and Killer genes, ORF3+ and ORF5+, while its allelic japonica-derived S5 region (ORF3–, ORF4+, and ORF5–) carries inactive Protector and Killer genes, ORF3– and ORF5–. However, the indica-derived Killer factor ORF5+ enacts its killing function with the assistance of its japonica-derived partner ORF4+ in the heterozygous combination. Meanwhile, ORF3+, which derives from the indica, selectively protects the embryo sac in which it resides. In the absence of stress responsive gene ORF3+, the unsolved endoplasmic reticulum (ER) stress induced by ORF5+ in partner with ORF4+, triggers programmed cell death (PCD), which results in embryo sac abortion (Yang et al., 2012).

The S1 locus is one of the major reproductive barriers between Asian and African rice species and a remarkable siTRD locus. The O. glaberrima allele (S1^g) takes advantage over O. sativa allele (S1^s) within populations, thus preferentially transmitting S1^g and eliminating both male and female gametes when two alleles meet in a heterozygous state (S1^s/S1^g) (Koide et al., 2008c, 2018). According to Xie et al. (2017), CRISPR/Cas9-generated knockout mutants of OgTPR1 on S1^g region (which encodes a protein with two trypsin-like peptidase domains and one ribosome biogenesis regulatory protein domain) produce normally fertile pollen and embryo sacs in crossing to O. sativa parent. The tripartite gamete Killer-Protector complex involving S1A4, S1TPR (OgTPR1), and S1A6 (SSP) of the S1^g region, generates a sterility signal in sporophytic cells and protect itself with S1TPR in gametophytic cells, whereas its allelic S1^s region lacks both functional killer and protector (Xie et al., 2019b).

The evolution of Killer-Protector system should be explained by models in which the deleterious effect of the Killer did not occur in a lineage (e.g., BDM model or parallel-sequential divergence model, Ouyang and Zhang, 2013), because without Protector, Killer induces abortion of gametes with both alleles highly reducing the chance of its fixation in a population. In the study of S5 sterility locus, the combination of non-functional Protector, and functional Killer and its partner (ORF3–, ORF4+, and ORF5+) was not found in the surveyed population (Yang et al., 2012). It may suggest that the combination that has Killer alone, without having functional Protector, cannot survive long in a population and also suggest that Killer can exist with functional Protector. However, after the emergence of Killer-Protector system, a single hybrid sterility locus with two tightly linked genes produces incompatibility when it meets the divergent allele in a different population. Then, the selfish nature of the system (i.e., TRD) caused by Killer-Protector system may facilitate its spread in a population.

In Oryza genus, reproductive isolation is excessively influenced by PK rather than EK and GE, which results in preferential occurrence of mTRD in contrast to the other two types, f TRD and siTRD. Compared to single sex-specificity, factors controlling siTRD are less frequently observed, mostly in interspecific hybridizations. Unveiling the underlying cause(s) behind this disproportionate pattern of TRD systems will shed light on the evolutionary process of reproductive barriers between rice relatives. Since our understanding on TRD systems remains very limited with confined experimental factors, further efforts are required to extend our investigation on many other selfish genes that exist and their distribution in Oryza genus.

Zin Mar Myint and YK contributed to conception, analyzed, and wrote the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

YK was funded by JSPS KAKENHI Grant Number 21H02160.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.