Adaxial-abaxial bipolar leaf genes encode a putative cytokinin receptor and HD-Zip III, and control the formation of ectopic shoot meristems in rice

In plants, pattern formation occurs in a group of cells called meristems, which contribute to the modeling of the architecture of the plant throughout its life. Shoot apical meristems (SAMs), the main meristems that form above-ground plant structures during the vegetative stage, are established during embryogenesis and maintained until the transition to reproductive meristems. For SAM maintenance, cell proliferation and differentiation are regulated by intricate interactions between genes and plant hormones. In Arabidopsis, the gene network for SAM maintenance is centered at WUSCHEL (WUS) (Eshed Williams, 2021). WUS is expressed in a small group of cells just below the stem cell domain in SAM and activates stem cell maintenance. WUS induces the expression of CLAVATA3 (CLV3), which encodes a signaling peptide that is specifically expressed in the stem cell domain (Schoof et al., 2000). WUS also positively regulates the cytokinin (CK) response sensitivity, and the CK signaling pathway activates WUS expression with CKs providing positional cues for WUS patterning (Leibfried et al., 2005). CK biosynthesis is upregulated by the Class Ⅰ Knotted 1-like homeobox (KNOX) genes, which are essential for meristem development and maintenance (Jasinski et al., 2005; Yanai et al., 2005; Zhao et al., 2010).

In rice, WUSCHEL RELATED HOMEOBOX4 (WOX4) is known as a positive regulator of SAM maintenance, orthologous to the Arabidopsis WOX4 (Ohmori et al., 2013). Analyses of flower mutants in rice revealed the functions of FLORAL ORGAN NUMBER 1 (FON1), an ortholog of CLV1, and FON2, an ortholog of CLV3, on flower meristem maintenance (Suzaki et al., 2004, 2008). FON2 regulates stem cell maintenance through FON1 in rice flowers (Suzaki et al., 2006). Although further investigations are needed to validate their molecular mechanisms, FON2-LIKE CLE PROTEINs (FCPs) and WOX4 are likely to function together in rice SAM (Ohmori et al., 2013). Oryza sativa homeobox1 (OSH1), one of the KNOX genes, is known to play an indispensable role in the maintenance of SAM and in activating CK in rice (Tsuda et al., 2011). In addition to OSH1, SHOOTLESS (SHL) genes are also known to function during shoot development in rice (Satoh et al., 2003). SHL genes are involved in trans-acting small interfering RNA production and regulate the expression of Class III homeodomain leucine zipper (HD-ZIP Ⅲ) via microRNA (Nagasaki et al., 2007).

Besides SAM maintenance, the initiation pattern of axillary meristems (AMs) is also an important factor for shaping plant body plans. AMs have been considered as ‘de novo meristems’, which start from differentiated cells, regain a meristematic fate and establish a stem cell population, or as ‘detached meristems’, which derive from the SAM and are maintained before being amplified to generate the new AM (Nicolas and Laufs, 2022). Recently, the mechanism of AM initiation has been studied intensively both in Arabidopsis and rice. Based on the data obtained from both plant species, a two-step regulation model was proposed (Wang and Jiao, 2018). In the model, AM formation is divided into two steps: meristematic cell lineage maintenance and stem cell activation. For the former, low expression of the KNOX gene SHOOT MERISTEMLESS (STM) is required. The minimal level of auxin in leaf axils keeps STM expression low, which is required to reserve the competence of leaf axil cells to form AMs (Wang et al., 2014). During the AM activation stage, gene expression of several transcription factors, including CUP-SHAPED COTYLEDON (CUC), LATERAL SUPPRESSOR (LAS), REGULATOR OF AXILLARY MERISTEMS (RAX), REVOLUTA (REV), upregulate STM expression (Nicolas and Laufs, 2022). STM promotes CK biosynthesis and, in turn, CKs induce de novo WUS expression. It is also proposed that the premeristem stage would exist after WUS expression is initiated during AM initiation (Tanaka et al., 2015; Nicolas and Laufs, 2022). The premeristem stage is a transient state in which the regulatory networks are distinct from the ones acting during the establishment of AMs. In Arabidopsis, the expression patterns of WUS and CLV3 are reorganized between premeristem and established AM (Xin et al., 2017; Nicolas and Laufs, 2022). In rice, the difference between premeristem and established AM is marked by the expression of TILLERS ABSENT 1, an ortholog of WUS, in axillary premeristem whereas WOX4 is expressed in established AM.

Here, we detail our effort to uncover a new genetic pathway involved in the patterning of meristem initiation by isolating mutants that are defective in proper leaf initiation. We identified three dominant mutant alleles carrying point mutations in a gene coding for a putative histidine kinase CK receptor and a recessive mutant carrying a loss-of-function mutation of a homeodomain-Leucine zipper III transcription factor. These mutants, called adaxial-abaxial bipolar leaf, form ectopic shoot meristems, distinct from AMs, at the base of leaf primordia. Our findings reveal the presence of a previously unknown mechanism governing meristem activity and formation of ectopic meristems at a specific developmental stage during rice germination following dormancy.

The abl1-d and abl2 mutants were isolated by screening chemically mutagenized M₂ populations of two Japonica rice cultivars based on their atypical foliage phenotypes observed at the fourth leaf position. We obtained three dominant alleles of abl1-d (abl1-1d, abl1-2d and abl1-3d) and one recessive abl2-1 allele through the mutant screening. Although the mutants showed no significant difference in overall shoot morphology (Fig. 1A-C), both mutants exhibited an abnormality in the formation of the fourth leaf, often comprising two leaves fused at their abaxial side (Fig. 1). In other words, the fourth leaf of the mutants appeared to have an ectopic leaf attached at the midrib in a mirror image, resulting in abaxial–abaxial fusion. Moreover, the original fourth leaf and the ectopic fourth leaf shared a connection at the base of the leaf sheath (Fig. 1D,E). The opposite polarity of the ectopic leaf was further confirmed through examination of leaf cross-sections. Bulliform cells, distinct large bubble-shaped epidermal cells, which are characteristic features of the adaxial surface of the leaf blades, were observed on the outer surface of fused leaves (Fig. S1). In addition, the position of the xylem, which formed on the adaxial side whereas the phloem was localized on the abaxial side within the wild-type (WT) vascular bundles, also indicated that the fusion of two leaves occurred on two abaxial surfaces (Fig. S1). Based on the observed phenotypes, we designated the mutants as adaxial-abaxial bipolar leaf (abl) mutants. Regarding basic leaf structure, ectopic leaves showed no significant differences to WT leaves in terms of interval length of the vascular bundles, cell numbers between vascular bundles, or cell sizes (Table S1).

The morphology of bipolar leaves varies within a range. We categorized bipolar leaves into three types based on the ratio of the width of the ectopic fourth leaf to that of the original fourth leaf. In Type I bipolar leaves, the ectopic leaf width is more than 80% that of the original leaf. In such case, the ligules and auricles (Fig. 2A) of both leaves were formed within a height difference of less than 3 mm, and auricles sometimes fused together (Fig. 2B). Additionally, a well-formed ectopic shoot was usually visible at the adaxial side of the ectopic fourth leaf, i.e. at the abaxial side of the original fourth leaf (Fig. 2H, Table 1). In Type III bipolar leaves, ectopic leaf width is less than 50% of the original fourth leaf. In such cases, the ligules and auricles were positioned significantly lower (often 1 cm or more) than the original fourth leaf (Fig. 2D). Moreover, few ectopic shoots were observed on the adaxial side of the ectopic fourth leaf (Fig. 2I, Table 1). Plants with Type II bipolar leaves showed a phenotype that was intermediate between Type I and Type III, showing 50-80% of the width of the original fourth leaf (Fig. 2C, Table 1). Therefore, the size of the ectopic fourth leaves correlates with the formation of ectopic shoots at the base of the fourth leaf, indicating that well-developed ectopic leaves are associated with well-developed ectopic shoots. The degree of fusions of the original and ectopic fourth leaves varied. In most cases, they were fused from the base of leaf sheath to the blade tip, but a significant number of bipolar leaves were fused at the sheath and bifurcated from the bottom of the leaf blades (Fig. 2E,F, Table 2).

In addition to the bipolar leaf phenotype, the abl1-d and abl2-1 mutants also showed aberrant phyllotaxy (Table 2, Fig. 2J,K). In WT rice plants, leaves are formed in alternate phyllotaxy, which means they occur at offset positions on opposite sides of the stem (Fig. 2G). Typical phyllotaxy defects in abl1-d and abl2 can be classified into two groups. The first group has a spiral phyllotaxy (Fig. 2J) and the second has an alternate decussate phyllotaxy at the fourth and fifth leaves (Fig. 2K). Aberrant phyllotaxy of the main shoot was sometimes observed in association with the bipolar leaf phenotype in a mutant plant (Fig. S2).

In rice, macro hairs are typically found on the adaxial surface of leaf blades and abaxial surface of leaf sheaths. In the abl1-d mutants, we observed that all leaves had more and longer macro hairs (Figs S3 and S4). In addition, leaves of abl1-d mutants also had macro hairs on the abaxial surface of blades where no macro hairs are found in WT. Although the appearance of the bipolar fourth leaf was seen in 8-11% of the abl1-d mutants (Table 2), the hair phenotype showed full penetrance, meaning that 100% of abl1-d plants exhibited the excessive macro hair phenotype. In addition to leaves, more hairs were also seen on the stigma of abl1-d mutants (Fig. S5). The ovules of abl1-d mutants were fertile, as evidenced by the ability to produce seed through self-pollination of heterozygous plants and pollinating with WT pollen. However, pollen of the abl1-d homozygote was abnormal and sterile (Fig. S6). By contrast, abl2-1 homozygous mutant plants were fertile and did not exhibit the excess hair phenotype.

To explore possible genetic interactions between abl1 and abl2 mutants, we analyzed the phenotype of double mutants in various genetic combinations. For this investigation, we crossed abl1-1d heterozygote, abl2-1 heterozygote and homozygote plants together and then genotyped the F₂ progeny using PCR (see Materials and Methods). Key mutant phenotypes, such as ectopic fourth leaf formation and aberrant phyllotaxy, were scored and the frequencies of each genotype were calculated (Table S2). The frequency of aberrant phenotypes in homozygous single mutants was 50% in abl1-1d and 3% in abl2-1 (Table S2). As for the abl1-1d homozygous mutant, its phenotype was enhanced by adding the abl2-1 allele in heterozygosity, resulting in an increase in the proportion of visible mutant plants from 50% to 60% (Table S2). In this case, the proportion of plants with aberrant phyllotaxy was also increased (Fig. 3A). For the abl2-1 homozygous single mutant, the phenotype output was increased from 3% to 76% in abl1-1d heterozygosity (Table S2). In this case, phenotypic variation clearly expanded from solely the presence of the fourth bipolar leaf to include twin shoots and the occurrence of third and fourth bipolar leaves and so on. Double mutants were richest in phenotypic variety (Fig. 3A). In a double mutant, several shoots, bipolar leaves, and aberrant phyllotaxy were observed (Fig. 3B,C). Moreover, the presence of several phenotypic traits were observed in one plant with at least a double heterozygote of both abl1-1d and abl2-1 mutations. These mutant analyses confirmed synergistic interactions between abl1-1d and abl2-1 mutations.

To understand how the ectopic fourth leaf is formed in abl1 and abl2 mutants, we analyzed the pattern of gene expression associated with the shoot meristem activity. The gene selected for analysis was OSH1, a homeobox gene recognized as a molecular marker for meristematic cells in rice (Sato et al., 1996). OSH1 is expressed in meristematic cells but not in leaf primordia of WT SAM. In both abl1-d and abl2 mutants, when the fourth leaf primordium was initiated in the SAM 2 days after germination, ectopic expression of OSH1 was observed at the base of the fourth leaf primordium (Fig. 4A-C). The region where OSH1 was ectopically expressed appeared to swell to form a new meristem. We could observe bipolar leaf P1 primordia and ectopic meristem formation morphologically 3 days after germination (Fig. 4G) as well as its subsequent growth (Fig. 4I). This region of ectopic OSH1 expression in Fig. 4B,C could be identified as a tiller meristem existing at the adaxial side of the second leaf. To investigate the distinction between the ectopic meristem and the tiller meristem, we made cross-sections of the abl1-d and abl2-1 mutant shoots. As shown in Fig. S7, the tiller meristem at the axil of the second leaf was formed independently of the ectopic shoot formation at the base of the fourth leaf primordium (Fig. S7). This indicates that the ectopic meristem is distinct from a tiller meristem that could rise at the axil of the second leaf later in development. Thus, the abl1 and abl2 mutations induced ectopic OSH1 expression resulting in the formation of a new SAM at the base of the fourth leaf primordium.

The phenotype of abl1-d and abl2-1 mutants appears only on the fourth leaf. The peculiarity of the fourth leaf position lies in the developmental timing within rice plant. In rice, the first to third leaves differentiate from SAM during embryogenesis, after which the seed becomes dormant. Thus, the fourth leaf is the first leaf to be initiated from the reactivated SAM after germination. To examine the effect of embryogenesis and dormancy on the formation of bipolar leaf mutant phenotypes, we cultured immature embryos of mutants that had initiated first, second or third leaves 7-12 days after fertilization. Rice embryos between 7 and 12 days after fertilization develop the SAM and the primordia for the first three leaves. However, they have not entered the desiccation phase of seed maturation, which triggers dormancy, a process that occurs around the 20th day after fertilization (Itoh et al., 2005). Those tissue-cultured embryos formed differentiating fourth leaves without undergoing complete embryogenesis and dormancy.

We used the selfed progeny of a plant carrying a heterozygous mutation of abl1-1d and homozygous abl2 plants for our study. Thus, in abl1-1d, 22% of the 158 individuals in the control group and 25% of the 175 individuals in cultured embryo group indicate a WT genotype with no macro hair phenotype (Table 3). Among the remaining individuals with heterozygous or homozygous mutations of abl1-1d, only 1% of the 175 individuals in the cultured embryo group showed the bipolar leaf and aberrant phyllotaxy phenotype, whereas 17% of the 158 individuals in the control group showed abnormal phyllotaxy or bipolar leaves. The frequency of the plants with at least one abl1-1d mutation that showed a normal phenotype was significantly higher in cultured individuals than in controls derived from seeds (Table 3). For the abl2-1 mutation, the tendency was same as for abl1-1d. Only 2.5% of tissue-cultured abl2-1 mutants showed the bipolar leaf and aberrant phyllotaxy phenotype, whereas mutant plants derived from seeds, with completed embryogenesis and dormancy, showed those phenotypes with a frequency of 62%. Again, the frequency of abl2 plants with normal phenotype was significantly higher in cultured individuals than in controls derived from seeds (Table 3). These results show that the abl1-d and abl2 mutations require proper embryogenesis and dormancy to form an ectopic meristem at the base of the fourth leaf primordium.

To understand the molecular function of the ABL1 gene, we performed a positional cloning of the gene. Through rough mapping, the abl1-d mutation was localized between markers at 125.6 cM and 139 cM on chromosome 2 (Table S3). Fine-mapping using 368 F₂ and 204 BC₂ plants allowed us to narrow down the mutation to a 162 kb region between 31,659,000 bp and 31,821,000 bp of chromosome 2. This region encodes 35 genes according to the Rice Genome Project annotation performed at the Michigan State University (see Materials and Methods). By sequencing these 35 genes in the abl1-d mutants, we found that all three mutants carry mutations in the gene Os02g0738400/LOC_Os02g50480. The mutations were identified as a C-to-T substitution (leucine 234 to phenylalanine 234) in abl1-1d, a C-to-T substitution (leucine 320 to phenylalanine 320) in abl1-2d, and a G-to-A substitution from (valine 250 to methionine 250) in abl1-3d. Os02g0738400/LOC_Os02g50480 encodes a histidine kinase known as OsHK6, an ortholog of AHK4/CRE1 in Arabidopsis and ZmHK1 in maize. Further analysis revealed that the mutations of abl1-1d and abl1-3d are localized in the CHASE (Cyclase/Histidine Kinases Associated Sensory Extracellular) domain and the abl1-2d mutation in the transmembrane domain between the CHASE domain and the histidine kinase domain (Fig. 5A, Fig. S8). The CHASE domain is known as the region that interacts with ligands such as ethylene and CK in plants (Nongpiur et al., 2012; Schaller et al., 2011). A number of dominant mutations that result in constitutive activation of histidine kinases without the presence of ligands have been mapped to this domain (Chang et al., 1993; Tirichine et al., 2007; Bartrina et al., 2017; Muszynski et al., 2020). The position of the abl1-2d mutation in the transmembrane domain coincides with the dominant mutation known in the Arabidopsis AHK2 gene (Bartrina et al., 2017).

To investigate whether ABL1 functions as a histidine kinase like AHK4/CRE1, we conducted a complementation analysis using a yeast mutant (ΔSLN1), which lacks its endogenous histidine kinase gene. It has been demonstrated that ectopic expression of the AHK4/CRE1 protein can complement the loss-of-function mutation of the SLN1 gene in yeast (Inoue et al., 2001). We performed a similar analysis using ABL1/OsHK6. As shown in Fig. 5B, ABL1/OsHK6 complements the lethality of ΔSLN1 in the presence of 10 μM of trans-Zeatin (tZ), in a manner similar to Arabidopsis CRE1. This suggests that ABL1/OsHK6 is acting as a putative receptor activated by CK in a similar fashion as AHK4/CRE1. Furthermore, we tested the dominant mutant alleles of ABL1 that we have identified. The yeast SLN1 mutant expressing the mutant proteins escaped from lethality even without addition of CK (Fig. 5B). These results further show that the ABL1 protein carrying the dominant mutations is constitutively activated even in the absence of CK.

We also identified the ABL2 gene by using a map-based cloning approach. We narrowed down the genomic region of the abl2-1 mutation between two markers at 100 cM and 104 cM on chromosome 12 through rough mapping using 50 progeny plants of an F₂ mapping population (Table S3). By analyzing the whole genome sequence of the abl2-1 mutant, we found only one base substitution from G to A at the 5′ splicing site of the first intron of gene Os12g0612700/LOC_Os12g41860 within this genetic interval. Os12g0612700/LOC_Os12g41860 is annotated as a Class III homeodomain leucine zipper gene (HD-ZIP III), OSHB3 (Fig. 6A). The mutation is localized at the donor splice site of the first intron, changing the conserved GT nucleotides to AT. By sequencing the cDNA obtained from the abl2 mutant, we confirmed that the mutant transcripts exhibited aberrant splicing of the first intron. We observed that GT, located 20 nucleotides upstream of the mutation site, is used as a cryptic donor splice site (Fig. 6A). This aberrant splicing causes a frameshift, producing a truncated peptide of 105 amino acid residues (instead of the expected 855 amino acid residues) lacking the leucine zipper, START and MEKHLA domains, which are essential for its function as a transcription factor (Fig. 6B). Owing to the massive truncation of the protein, the abl2 mutation is considered to result in loss of function of the gene. To confirm that the phenotypes of abl2 were truly caused by the disruption of the OSHB3 gene, we created additional mutant alleles of the gene using CRISPR/Cas9 with a guide RNA designed to target the first exon of OSHB3 (Fig. 6A). The abl2-CR1 mutant plants were confirmed to be harboring a 47 bp deletion and 7 bp substitution around the target sequence of OSHB3. This mutation causes a frameshift, producing a 96-amino-acid peptide that lacks the functional domains of the protein (Fig. 6A,B). The abl2-CR2 mutant plants contain a 4 bp deletion around the target sequence that causes a frameshift, generating a truncated protein similar to abl2-CR1 (Fig. 6A,B). In addition, we did not find any mutation in the potential off-target sequences of the other HD-ZIP III genes (OSHB2/Os10g0480200/LOC_Os10g33960), which have sequences that differ from the target site by only three nucleotides. Seedlings of homozygote plants (abl2-CR1 and abl2-CR2) exhibited bipolar leaves, extra tillers and aberrant phyllotaxy, which are similar phenotypes to abl2 but more severe (Fig. S9). Based on these results, we conclude that OSHB3 is responsible for the regulation of proper meristem organization and growth in rice.

According to previous expression studies of ABL1/OsHK6 and ABL2/OSHB3 genes, both genes are highly expressed in inflorescence among plant tissues (Ding et al., 2017; Itoh et al., 2012; see also Fig. S10). To unravel the functions of the ABL1 and ABL2 genes in the shoot apex, we examined their precise expression patterns by performing in situ hybridization experiments. We found that ABL1 is expressed in the SAM and in P₁ leaf primordia (Fig. 7A,B). More precisely, it is strongly expressed at the margin of P₁ leaf primordia (Fig. S11A). ABL1 expression was also observed at the leaf margin of P₂ and P₃ leaf primordia and phloem (Fig. S11A,B). In addition, sporadic expression in those leaves was also found (Fig. S11C). ABL2 expression was detected in the SAM, adaxial cell layer of leaf primordia (Fig. 7C,D) and phloem (Fig. S11E), which is consistent with a previous study (Itoh et al., 2012). Furthermore, ABL2 expression was also observed in axillary shoot meristems (Fig. S11D). There was no clear difference in ABL1 and ABL2 expression between fourth leaf and fifth leaf (Fig. 7A-D).

To investigate the relationship between ABL1 and ABL2, the expression of each gene was analyzed in 2-day-old shoot apices by qRT-PCR. We found that the expression levels of ABL1 and ABL2 in abl2-1 and abl1-1d respectively were not significantly different from those in the WT, whereas the expression level of ABL2 in abl2-1 was significantly lower than in the WT (Fig. S12A-C). We also performed in situ hybridization experiments to determine whether there is a defect in spatial expression patterns of ABL1 and ABL2 in abl2-1 and abl1-1d, respectively. The expression patterns of ABL1 and ABL2 were maintained even in bipolar leaves in the mutants (Fig. 7E-H). That is, ABL1 was expressed in P₁ leaf primordia in the abl2-1 mutant and ABL2 was expressed in the adaxial epidermal cells in leaf primordia in the abl1-1d mutant. We further analyzed the expression levels of B-type response regulators, OsRR21, OsRR22 and OsRR24, which are known to be involved in CK signaling, to detect any interaction between the CK signaling pathway and the molecular genetic pathways of ABL1 and ABL2. However, the expression levels of B-type response regulators were not significantly different in the abl2 mutant compared with the WT (Fig. S12D-F).

A typical phenotype exhibited by the abl1-d and abl2 mutants is characterized by presence of adaxial-abaxial bipolar leaves. By studying the development of the ectopic leaf and its fusion with the fourth leaf in the opposite topology, we demonstrated that the formation of an ectopic meristem at the base of the fourth leaf primordium is associated with the phenotype. The formation of the ectopic meristem was confirmed by the ectopic expression of OSH1, a rice ortholog of the Arabidopsis STM gene. OSH1 expression is specific to the SAM, but not to the L1 layer of meristems and leaf primordia. The onset of ectopic OSH1 expression at the base of the fourth leaf primordia coincides with the formation of the fourth leaf primordia. As two leaf primordia develop simultaneously in close proximity, one from the original SAM and the other from the ectopic meristem, they tend to fuse along their axis as they develop into full leaves.

Both abl1-d and abl2 mutants show a gradient of severity concerning the development of ectopic leaves and the formation of ectopic meristems. For instance, the width of ectopic leaves correlates with the degree of development of ectopic shoots. As for phyllotaxy, some show spiral and others show decussate, affecting the development of leaves produced after the fourth leaf. Based on these observations, we propose a model that accounts for the variation of mutant phenotypes (Fig. 8). In the model, we adopt the concept of zones of competent meristematic cells proposed for wheat embryos (Fischer et al., 1997). This concept is based on the hypothesis that a region of cells in wheat embryos is competent to receive a signal for giving rise to meristems. Adopting this concept helps explain the variable phenotypes observed in the abl1-d and abl2-1 mutants during germination, i.e. there is a zone of competent meristematic cells around the SAM in germinating rice shoot. In abl1-d and abl2-1, the size or competency of the zone increased in size or was upgraded. This is reflected in the size of the SAM and the activity of the ectopic meristems. As shown in Fig. 8, we modeled several scenarios regarding the varying size and activity of the SAM and the ectopic meristem formed at the base of the fourth leaf primordia. As the activity of the original SAM is enhanced, an overall aberrant phyllotaxy is produced (Fig. 8B). Previous reports have shown that the size of the meristem influences phyllotaxy in plants (Jackson and Hake, 1999; Giulini et al., 2004; Itoh et al., 2012). When the activity of the ectopic meristem is relatively low, the formation of the first leaf from the ectopic meristem is limited, resulting in no ectopic leaf or a narrow, small ectopic leaf fused to the fourth leaf produced from the original SAM (Fig. 8C). When the ectopic shoot meristem becomes as active as the original SAM, an ectopic leaf as large as the original fourth leaf is produced (Fig. 8D). When the ectopic shoot meristem surpasses the original SAM for activity, the ectopic meristem affects phyllotaxy to produce leaves in a decussate manner (Fig. 8E). The isolation of additional mutant alleles of ABL1 and ABL2 with different degrees of severity, coupled with the identification of the molecular signals that give rise to meristems, would further help correlate the size of the zone of competent meristematic cells around the SAM with the phenotypes.

To understand the molecular mechanism behind the formation of the ectopic shoot meristem in abl1-d mutants, we identified the genes corresponding to the abl mutations by using map-based cloning approaches. The ABL1 gene was identified as the OsHK6 gene, encoding a histidine kinase. A loss-of-function mutation in OsHK6 affects root growth, leaf width, inflorescence architecture and/or floral development, as well as the tolerance of root development to benzyl adenine and kinetin (Ding et al., 2017). Analysis of the double mutant hk5 hk6 has shown that OsHK5 and OsHK6 redundantly function in root and shoot growth through size maintenance of the SAM and the root apical meristems (Burr et al., 2020). As for the molecular function of OsHK6, our yeast complementation test suggested that OsHK6 is a putative CK receptor. Previous reports also suggested that it is preferentially for isopentenyladenine rather than t-zeatin (Choi et al., 2012).

Gain-of-function mutation analyses have shed light on the function of CK receptors. In Arabidopsis, gain-of-function mutants of AHK2 and AHK3 showed a relatively minor increase in shoot growth, such as shoot length and stem diameter, without causing any aberrant meristem phenotypes (Bartrina et al., 2017). In maize, gain-of-function mutants called Hairy sheath frayed 1 (Hsf1), have been reported. Point mutations have been identified in the CHASE domain of the Histidine Kinase 1 (ZmHK1) gene (Muszynski et al., 2020). ZmHK1 encodes a CK receptor histidine kinase, one of the closest homologs of rice ABL1/OsHK6 based on amino acid sequence comparison (Muszynski et al., 2020). The dominant Hsf1 mutants are characterized, not only by defects in leaf patterning, but also by pleiotropic phenotypes such as increased macro hair size and density on leaves, short and narrow leaves, and a short stature. Although the maize Hsf1 mutants are not known to affect shoot meristem activity to produce ectopic shoots, they show a proliferative shoot phenotype when combined with a mutation of ZmRR3, a negative regulator of CK signaling (Muszynski et al., 2020).

We showed that the dominant abl1-d mutants constitutively activate OsHK6 even in the absence of CK in yeast (Fig. 5B). By contrast, ABL1/OsHK6 expression is localized to the entire P₁ primordia with strong expression in the leaf margin at the base. The expression in the leaf margin was also observed in P₂ and P₃ primordia. The hair phenotype of the abl1-d mutants seems to correspond to the sporadic ABL1 expression pattern in leaves. Specifically, excess CK signal correlates with the presence of more and longer macro hairs on rice leaves. A similar correlation between CK and hairs on leaves has also been reported in maize (Muszynski et al., 2020). This suggests that CK signaling, especially in the leaf margin at the base of P1 primordia in the apexes of abl1-d mutants, is constitutively activated. Interestingly, during germination, the third leaf margin is close to the area where an ectopic meristem is formed at the base of the fourth leaf primordia in abl1-d and abl2.

In Arabidopsis, genes encoding CK receptor histidine kinases, AHK2, AHK3 and CRE1/AHK4, are known to be strongly expressed in the central domain of the SAMs (Gordon et al., 2009; Gruel et al., 2016). It has been discussed that CKs provide positional cues for WUS patterning and are required for maintaining pluripotency of the SAM (for a review, see Eshed Williams, 2021). The difference in expression patterns and phenotype of gain-of-function mutants between ABL1/OsHK6 and AHK4 suggests that there is a unique relationship between CK and the activity of SAM in rice, which has not yet been revealed in Arabidopsis. Investigating the localization of CK signaling would bring us one step closer to understanding how the activation of CK signaling in the ABL1/OsHK6 expression domain leads to the formation of an ectopic meristem in the vicinity.

Our morphological analysis and map-based cloning revealed that ABL2/OSHB3 functions to suppress the formation of ectopic meristems during rice germination. In another study, a RNAi knockdown of OsHox33/OSHB3 resulted in an acceleration of leaf senescence, but no developmental defects were observed (Luan et al., 2013). Our findings reveal a previously unknown function for the HD-ZIP III gene.

The HD-ZIP III gene family is a well-studied gene family, especially in Arabidopsis. There are five Class III HD-ZIP genes known in Arabidopsis: REVOLUTA (REV), PHABULOSA (PHB), PHAVOLUTA (PHV), CORONA (CNA) and ATHB8. ABL2/OSHB3 is in the same clade as PHV, PHB and REV, which is distinct from the other clade of CNA and ATHB8 (Agalou et al., 2008). Furthermore, Arabidopsis PHB and PHV are closest to the rice ABL2/OSHB3 and OSHB4 genes. PHB, PHV, ABL2/OSHB3 and OSHB4 are expressed in a similar pattern in the SAM and the adaxial area of lateral organs, indicating that they likely have the same or similar functions in development of shoot meristems and leaf primordia (McConnell et al., 2001; Itoh et al., 2008). However, loss-of-function mutations of each of these genes in Arabidopsis do not result in any significant phenotypic alteration. Instead, it has been observed that the triple mutant phv phb rev creates seedlings without SAM (Emery et al., 2003; Prigge et al., 2005). By contrast, another triple mutant, phv phb cna, has extra cotyledons, an enlarged SAM, fasciated stems, and flowers with extra organs (Prigge et al., 2005). These results, with seemingly opposing phenotypes, suggest complicated genetic interaction between HD-ZIP III genes. Further analyses have indicated that PHV, PHB, CNA and REV regulate stem cell formation in a WUS-independent pathway (Lee and Clark, 2015).

The HD-ZIP III genes are known to be controlled by miRNAs and their gain-of-function mutations have been studied extensively. PHV and PHB gain-of-function mutants turn abaxial identity of leaves into adaxial identity (McConnell et al., 2001). In rice, analysis of plants ectopically expressing the OSHB3 gene without miRNA regulation sites suggested the involvement of the gene in leaf polarity and meristem initiation (Itoh et al., 2008). LATERAL FLORET1 (LF1), encoding OSHB1, influences the development of the three-floret spikelet and lateral organ polarity in rice (Zhang et al., 2017, 2021). These studies shed light on the relationship between the HD-ZIP III genes and auxin, already known in Arabidopsis (Turchi et al., 2015; Ramachandran et al., 2017; Sessa et al., 2018; Manuela and Xu, 2020). LF1 regulates the expression of OsYUCCA6, one of the indole-3-acetic acid synthetic pathway genes (Zhang et al., 2021). However, OSHB3 is induced by auxin in the SAM (Itoh et al., 2008). Thus, although further studies are required for elucidating the mechanism, one can postulate that auxin plays a role in the ABL2 pathway to suppress the formation of ectopic shoot meristems at the base of fourth leaf primordia during germination in rice.

The phenotypes of the abl1-d and abl2-1 are very similar and are specific to the fourth leaf, which is the first leaf formed after embryogenesis in rice. Complete embryogenesis and dormancy appear to be a prerequisite for the mutant phenotype as cultured immature embryos largely lack the formation of bipolar leaves (Table 3). The mutant phenotype revealed the distinctiveness of the first leaf produced post-embryonically and the unique role of ABL1 and ABL2 in suppressing the formation of ectopic meristem during rice germination. In our study, we demonstrated that two genes with distinct functions exhibit a similar mutant phenotype of an ectopic shoot meristem. The ABL1/OsHK6 gene, encoding a putative CK receptor histidine kinase, transmits the CK signal in yeast (Fig. 5B). The ABL2/OSHB3 gene is an HD-ZIP III transcription factor, homologous to the Arabidopsis genes PHV and PHB. We found that the expression domains of the two genes overlap. Both genes are strongly expressed at the edge of the young leaf primordia (Fig. S11A; Itoh et al., 2008) and showed slightly weak expression in the SAM. As mentioned above, we can hypothesize that CK signaling is activated constitutively in the ABL1/OsHK6 expression domain of abl1-d mutants during germination. At the same time, auxin is involved in the ABL2 pathway. In other words, excess CK signaling caused by the abl1 mutations and the alteration of auxin signal transduction by the abl2 mutation appear to result in a similar outcome of ectopic shoot meristem formation. The excess CK signal caused by abl1-d can be enhanced by a decreased auxin response caused by the abl2 mutation. Such cross-talk between CK and auxin has been demonstrated in other studies (El-Showk et al., 2013). However, in the case of ABL1/OsHK6 and ABL2/OSHB3 genes, we did not observe any reciprocal transcriptional regulation of the genes (Fig. S12A,B). In addition, the difference in expression of B-type response regulators in WT and abl2 mutant was not significant (Fig. S12D-F). This indicates that the interaction between ABL1/CK and ABL2/auxin is not direct, but rather there is an indirect interaction between the pathways.

In a developmental stage-wise and in spatially focused manner, further studies on genes downstream of ABL1 and ABL2 should shed more light on the interactions between ABL1/CK and ABL2/auxin in the regulation of ectopic shoot meristems during germination in rice.

The abl1 and abl2 mutant lines were isolated by screening M₂ populations of Akitakomachi and Taichung 65 (T65) backgrounds treated with N-methyl-N-nitrosourea. The mutants were back-crossed three times with their respective WT lines, Akitakomachi and T65, to clean their backgrounds. Because abl1 homozygous mutants are sterile, we used heterozygous plants for the maintenance of dominant mutations. The macro hair phenotype, which showed 100% penetrance in mutants, was used as the indicator of plants carrying dominant abl1-d mutations to conduct morphological analyses. Thus, the results of abl1-d mutants analysis presented in Table 2 are based on individuals with at least one abl1-d mutations. As for the abl2 mutant plants, fertile homozygous plants were used for morphological analyses. The self-fertilized heterozygous plants carrying abl1-d mutations segregated in a 3:1 fashion. However, in the case of the abl2 mutation, the ratio of mutant plants from the heterozygous plants carrying the abl2 mutation was around 10%.

For next-generation sequencing, recessive homozygous individuals isolated from the F₂ generation obtained by crossing the abl2 mutant with T65 were used.

For germination, seeds were soaked overnight in a 200-fold diluted solution of fungicide (Kumiai Chemical Industry) at room temperature and washed with water. After 2 days in water at 15°C, the seeds were transferred into a 28°C incubator and placed in soil after germination. One month after planting, morphological observations were conducted.

To distinguish homozygous and heterozygous individuals of each mutant, genotyping was performed as follows. DNA was extracted from individual plants by grinding a small amount of plant tissues in liquid nitrogen with a multi-bead shocker (Yasui Instruments), adding 300 µl of TPS buffer (0.1 M Tris-HCl pH 8.0, 1 M KCl, 0.01 M EDTA) and incubating at 70°C for 30 min. For the precipitation, samples were centrifuged at 2900 rpm (1100 g) for 15 min at 4°C, then 200 µl of isopropanol was added to approximately 200 µl of the supernatant and the mixture was centrifuged at 4°C, 2900 rpm (1100 g), for 15 min after mixing. For washing, 300 µl of 70% ethanol solution was added followed by centrifugation at 4°C, 2900 rpm (1100 g), for 15 min. The pellet was dissolved in 50 µl of sterile water.

PCR was performed using dCAPS primers and the PCR products were subjected to restriction enzyme analysis. Primer pairs for each mutant allele and enzymes used for genotyping are listed in Table S4. PCR reactions were performed with TaKaRa Taq (Takara Bio) under the following conditions: 94°C for 3 min; 38 cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 30 s; and 72°C for 7 min. After the PCR reaction, restriction enzymes (New England Biolabs) were to digest the products, and digested samples were electrophoresed to detect polymorphisms.

Samples were fixed in 4% of paraformaldehyde at 4°C overnight for 2 days. Then, they were dehydrated through an ethanol series and 2-butanol. Subsequently, samples were embedded in Paraplast plus (McCormick Scientific), sectioned into 8 µm thickness, and stained with Hematoxylin (Sakura Finetek).

Embryos were removed from immature seeds at 7-12 days after flowering. Sterilized immature embryos were placed on MS medium and incubated under continuous light conditions at 28°C. After 14 days, the phenotype of the fourth leaf was scored. As a control, mature mutant seeds were sterilized, cultured on MS medium, and the phenotype was recorded. We performed χ² goodness-of-fit test using Excel.

The heterozygous abl1-1d mutant was crossed with cultivars Kasalath (Ka), Chokoukoku (CKK) and 93-11. A total of 368 F₂ progenies from the cross between abl1 and Ka and abl1 and CKK that displayed the WT phenotype were used for mapping the abl1-d mutation. For the third population, F₁ plants with abl1 phenotypes derived from the cross with 93-11 were backcrossed into 93-11 and all of the 204 BC₂ progeny with a mutant phenotype were used for mapping. For cloning of the ABL2 gene, homozygous abl2-1 mutant plants were crossed with Ka and a total of 50 F₂ progenies that exhibited mutant phenotypes were used for initial mapping.

With the aid of cleaved amplified polymorphic sequences and sequence-tagged site markers (Heubl, 2010), the abl1-d and abl2-1 mutations were mapped to chromosomes 2 and 12, respectively. For high-resolution mapping, we designed PCR-based markers using on the sequence information at the MSU website (rice.plantbiology.msu.edu/cgi-bin/gbrowse/rice/) (see Table S3).

Total RNA was extracted using TRIzol (Invitrogen), following the manufacturer's protocol, and 1 µg of total RNA was used for cDNA synthesis using the Superscript III Kit (Thermo Fisher Scientific). The coding region of ABL1 was amplified by PCR in five fragments (Fr.1 - Fr.5) using the primers listed in Table S5. PCR was performed using KOD -Plus- DNA polymerase (TOYOBO) under the following conditions: 94°C for 3 min; 6 cycles of 94°C for 30 s, 67°C (−0.5°C per cycle) for 30 s and 68°C for 30 s; 29 cycles of 94°C for 30 s, 62°C for 30 s and 68°C for 30 s; and 68°C for 7 min. PCR products were purified using the QIA quick PCR Purification Kit (QIAGEN). To recover the full-length sequence of ABL1 from the five cDNA fragments, equal amounts of each PCR fragments were mixed and used as template for PCR using the above-mentioned conditions, with the exception of a 3 min elongation time. Primers used were: forward primer (5′-CCGCTCTCTAGAACTAGTAAAATGGGGGAAGCCGGGAGGCGAGGAG-3′) and reverse primer (5′-CGGTATCGATAAGCTTTCATACCGCTGATTCCTTCGTTCCCAC-3′). The amplified fragment was cloned into pBluescript SK (−) [pBSK(−)]. We isolated full-length cDNAs from Akitakomachi, abl1-1d, abl1-2d and abl1-3d mutants. The coding region of the ABL2 gene was amplified using the following PCR primers: forward primer (5′-CGGGCTGCAGGAATTCATAGCTCCTTCTCGACTTGGAC-3′) and reverse primer (5′-CGGTATCGATAAGCTTTGTTCATCCACACAGCAGCAG-3′).

For studying ABL1 expression patterns by in situ hybridization, the probe was prepared by amplifying the coding region with the following primers: forward primer (5′-TAATACGACTCACTATATAGGGGGGCATATATGGCAAACAGGAGTCAG-3′) and reverse primer (5′-GGCAACTAGACAAATAAGGCAAATG-3′). The PCR product was subjected to in vitro transcription using an SP6/T7 transcription kit (Sigma-Aldrich) to obtain an RNA probe labeled with digoxigenin. For the ABL2 probe, template DNA for in vitro transcription was obtained by PCR with the following primer set: forward primer (5′-CGGGCTGCAGGAATTCATAGCTCCTTCTCGACTTGGAC-3′) and reverse primer (5′-CGGTATCGATAAGCTTTGTTCATCCACACAGCAGCAG-3′). PCR products were cloned into pBSK(−) and transcribed to generate RNA probes after digestion by SmaI. The regions used for the experiments contained the 3′ UTR and it was verified that they had no more than 20 bp of complete matches on non-target sequences in rice cDNA by BLAST. The OSH1 probe was prepared from the coding region of OSH1. In situ hybridization using alkaline phosphatase was performed following the methods described by Kouchi and Hata (1993). Fixation of samples were performed as mentioned above for sectioning.

The target site of ABL2/OSHB3 (TGGACGCCGGGAAGTACGTCCGG) was selected using the CRISPR-P program (http://cbi.hzau.edu.cn/crispr/) (Lei et al., 2014) (Fig. 7). Only one potential off-target site with three mismatches to the on-target sequence was detected in the first exon of the other HD-ZIP III gene, OSHB2/LOC_Os10g33960. The single-guide RNA (sgRNA) cloning vector (pZK_gRNA) and all-in-one binary vector (pZH_ OsU6gRNA_MMCas9) harboring sgRNA, Cas9 and NPTII were provided by Masaki Endo (Mikami et al., 2015). The pZH_ OsU6gRNA_MMCas9 vector including target-guide RNA for ABL2/OSHB3 was constructed as described previously (Mikami et al., 2015). The constructs were introduced into Agrobacterium tumefaciens strain EHA105 and transformed into cultivar Taichung-65 (T-65) calli via Agrobacterium-mediated transformation. Mutations and transgenes in each transformant were confirmed by sequencing and PCR-based detection (Fig. S8, Table S1).

For the yeast complementation experiments, the following materials were kindly provided by Dr Tatsuo Kakimoto at Osaka University: Escherichia coli strain DH10B, yeast vector p415 with CYC promoter, yeast ΔSLN1 mutant strain (TM182), and a clone of the Arabidopsis CRE1 gene.

Full-length cDNAs were cloned into p415 by using the TaKaRa DNA Ligation Kit LONG (Takara Bio Inc.). Transformation of the yeast mutant TM182 was performed by electroporation and selection on a −Ura/galactose selection medium consisting of 0.67% (w/v) Difco TM Yeast Nitrogen base without amino acids (Thermo Fisher Scientific), 0.067% (w/v) −Ura/−Leu/−Trp DO supplement (Takara Bio Inc.) and 2% (w/v) galactose. After using a Bio-Rad Micro Pulser for electroporation, the yeast cells were cultured in a 1 M sorbitol/−Ura/−Leu/galactose selective medium.

The ΔSLNs yeast strain was transformed with constructs expressing ABL1 cDNA derived from the WT (Akitakomachi) abl1-1d, abl1-2d and abl1-3d mutants, and was cultured in −Ura/−Leu/galactose selective medium. Yeast cells were harvested by centrifugation (3000 g for 5 min), diluted to 1×10³ cells/10 µl, and were resuspended in the medium mentioned above. The growth of yeast cells was evaluated after 3 days of incubation at 30°C.

For first-strand cDNA synthesis, 500 µg of RNA was used, with the ReverTra Ace qPCR RT Master Mix with gDNA Remover (TOYOBO). The cDNA was diluted 20 times and used for real-time PCR. For quantification of the genes, KOD SYBR qPCR Mix (TOYOBO) was used. The expression level of each sample was normalized to that of an internal control, UBIQUITIN5 (UBQ5). The primers for the detection of ABL1, ABL2, OsRR21, OsRR22, OsRR24 and UBQ5 are listed in Table S6.

We thank Dr Tatsuo Kakimoto for his generous gift of the strains and clones to perform the yeast histidine kinase complementation test. We also thank Dr Hajime Sakai, Dr Emil Orozco and Anjalie Chakravertti for critical reading of manuscript, and Dr Kentaro Yasuda and Tomonori Ito for their assistance in cultivating rice plants in Agri-Innovation Education and Research Center in Akita. We also acknowledge the Biotechnology center in Akita Prefectural University for sequencing.