Regulation of the plastochron by three many-noded dwarf genes in barley

The plastochron, the time interval between the formation of two successive leaves, is an important determinant of plant architecture. We genetically and phenotypically investigated many-noded dwarf (mnd) mutants in barley. The mnd mutants exhibited a shortened plastochron and a decreased leaf blade length, and resembled previously reported plastochron1 (pla1), pla2, and pla3 mutants in rice. In addition, the maturation of mnd leaves was accelerated, similar to pla mutants in rice. Several barley mnd alleles were derived from three genes—MND1, MND4, and MND8. Although MND4 coincided with a cytochrome P450 family gene that is a homolog of rice PLA1, we clarified that MND1 and MND8 encode an N-acetyltransferase-like protein and a MATE transporter-family protein, which are respectively orthologs of rice GW6a and maize BIGE1 and unrelated to PLA2 or PLA3. Expression analyses of the three MND genes revealed that MND1 and MND4 were expressed in limited regions of the shoot apical meristem and leaf primordia, but MND8 did not exhibit a specific expression pattern around the shoot apex. In addition, the expression levels of the three genes were interdependent among the various mutant backgrounds. Genetic analyses using the double mutants mnd4mnd8 and mnd1mnd8 indicated that MND1 and MND4 regulate the plastochron independently of MND8, suggesting that the plastochron in barley is controlled by multiple genetic pathways involving MND1, MND4, and MND8. Correlation analysis between leaf number and leaf blade length indicated that both traits exhibited a strong negative association among different genetic backgrounds but not in the same genetic background. We propose that MND genes function in the regulation of the plastochron and leaf growth and revealed conserved and diverse aspects of plastochron regulation via comparative analysis of barley and rice.

The plastochron, the time interval between the formation of two successive leaves, is an important determinant of plant architecture. We genetically and phenotypically investigated many-noded dwarf (mnd) mutants in barley. The mnd mutants exhibited a shortened plastochron and a decreased leaf blade length, and resembled previously reported plastochron1 (pla1), pla2, and pla3 mutants in rice. In addition, the maturation of mnd leaves was accelerated, similar to pla mutants in rice. Several barley mnd alleles were derived from three genes-MND1, MND4, and MND8. Although MND4 coincided with a cytochrome P450 family gene that is a homolog of rice PLA1, we clarified that MND1 and MND8 encode an Nacetyltransferase-like protein and a MATE transporter-family protein, which are respectively orthologs of rice GW6a and maize BIGE1 and unrelated to PLA2 or PLA3. Expression analyses of the three MND genes revealed that MND1 and MND4 were expressed in limited regions of the shoot apical meristem and leaf primordia, but MND8 did not exhibit a specific expression pattern around the shoot apex. In addition, the expression levels of the three genes were interdependent among the various mutant backgrounds. Genetic analyses using the double mutants mnd4mnd8 and mnd1mnd8 indicated that MND1 and MND4 regulate the plastochron independently of MND8, suggesting that the plastochron in barley is controlled by multiple genetic pathways involving MND1, MND4, and MND8. Correlation analysis between leaf number and leaf blade length indicated that both traits exhibited a strong negative association among different genetic backgrounds but not in the same genetic background. We propose that MND genes function in the regulation of the plastochron and leaf growth and revealed conserved and diverse aspects of plastochron regulation via comparative analysis of barley and rice. PLOS

Introduction
The spatiotemporal pattern of leaf initiation is a major contributor to the formation of plant architecture. The temporal pattern of leaf initiation is termed the plastochron; that is, the time interval between the initiation of two successive leaf primordia. The spatial pattern of leaf initiation along the shoot axis is referred to as phyllotaxy. Both patterns of leaf initiation are determined by the activity of the shoot apical meristem (SAM), which is the source of leaf primordia [1,2]. Although the plastochron and the phyllotaxy are sometimes regulated by shared genetic components operating at the SAM, some genes are specific to one or the other of the patterns [1,3]. PLASTOCHRON1 (PLA1) specifically regulates the plastochron in rice (Oryza sativa) [4,5]. Loss of function of PLA1 causes rapidly emerging small leaves, resulting in more than twice the number of leaves compared to the wild type (WT). PLA1 encodes a plant-specific cytochrome P450 family protein, CYP78A11 [5]. Arabidopsis KLU is also a CYP78A-family member and its loss-of-function mutant exhibits accelerated leaf initiation and produces small organs [6]. PLA2 is another plastochron-regulating factor in rice [7]. PLA2 encodes a MEI2-like RNA-binding protein, which is an ortholog of TERMINAL EARl (TE1) in maize (Zea mays) [8]. Although TE1 plays a role in the regulation of phyllotaxy, accelerated leaf initiation in the loss-of-function mutant is shared between te1 and pla2. PLA3 has also been reported to regulate the plastochron. A loss-of-function mutant of PLA3 exhibits not only a shortened plastochron but also pleiotropic phenotypes such as embryonic defects [9]. PLA3 encodes a homolog of glutamate carboxypeptidase, which dissociates glutamate from small peptides [9]. PLA3 is the rice ortholog of Arabidopsis ALTERED MERISTEM PROGRAM1 (AMP1) [10] and maize VIVIPAROUS8 [11]. Consequently, in rice, loss-of-function mutants of three PLA genes share the phenotypes of rapid leaf production, small leaf size, and aberrant inflorescence structure [4,7,9].
SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) genes are plant-specific transcription factors, some of which are negatively regulated by miR156 [12][13][14][15]. SPL genes are also involved in plastochron regulation. A double loss-of-function mutant of AtSPL9 and AtSPL15 in Arabidopsis exhibited a short plastochron [6,16]. Conversely, the expression of an miR156-resistant form of SPL9 caused a prolonged plastochron [6]. In grass species such as rice and maize, SPL genes and miR156 have conserved functions in plastochron regulation. Accumulation of OsSPL transcripts caused a prolonged plastochron in rice and loss-of-function mutants of several SPL genes and plants overexpressing miR156 exhibited a short plastochron in both rice and maize [17][18][19][20][21][22][23].
The big embryo1 (bige1) mutant in maize exhibits an increased leaf and seminal root number in addition to a large embryo [24]. BIGE1 encodes a MATE-type transporter that likely plays a role in the secretion of an unidentified small molecule in the trans-Golgi. BIGE1 function is conserved between maize and Arabidopsis, because both single and double mutants of BIGE1 homologs in Arabidopsis produced an increased number of leaves, and the introduction of BIGE1-GFP fusion genes into an Arabidopsis bige1 mutant partially complemented the mutant phenotype [24].
Despite the identification of genes involved in the plastochron, the relationships among these genes and genetic pathways in plastochron regulation are unclear. In Arabidopsis, KLUH/CYP78A5 and AtSPL9/miR156 affect the plastochron and organ size in parallel genetic pathways [6]. In rice, a pla1 and pla2 double mutant exhibited enhanced phenotypes compared with each single mutant, suggesting that PLA1 and PLA2 function in independent pathways [7]. Although maize BIGE1 is involved in the feedback regulation of a CYP78A pathway, the phenotypic effect of the interaction on the plastochron has not been elucidated [6]. In addition to the genetic pathways, knowledge of functional conservation of plastochron-related genes among plant species is fragmentary.
Barley (Hordeum vulgare) is an important cereal crop for which considerable genetic resources are available, including collections of morphological and developmental mutants [25]. In addition, high-quality genome sequence information is available [26]. Thus, barley is an alternative model for grass molecular genetics with less gene redundancy and is suitable for comparative studies with other grass species. Here, we genetically and phenotypically characterized many-noded dwarf (mnd) mutants in barley, which were originally described in the 1920s as dwarf mutants with many nodes [27]. All mnd mutants exhibited a shortened plastochron, comparable to pla mutants in rice. We identified the genes responsible for mnd mutants and showed that the plastochron and leaf length are independently controlled by three genetic factors-MND1, MND4, and MND8. Here, we propose to assign a new locus name mnd8 to a mutation that was found at a locus different from the seven previously described mnd loci in barley. We revealed that MND8 encodes a MATE transporter-family protein, which is an ortholog of maize BIGE1. MND1 encodes an N-acetyltransferase-like protein that reportedly regulates phase changes [28]. MND4 had been reported as a cytochrome P450-family gene and a homolog of rice PLA1 [29]. Our phenotypic and genetic analyses of the mnd mutants suggested functions for the three MND genes and the existence of complex genetic interactions among them. Furthermore, our comparative analysis of rice and barley clarified the diversity and conservation of plastochron genetic pathways.

Plastochron of mnd mutants
The mnd mutants used in this study are listed in Table 1. Allelism tests revealed that these mutants carried alleles derived from at least three independent genes, MND1, MND4, and MND8 (Table 1). MND4 is also known as HvMND [29]. Among the mutants, three-OUM165, OUM169, and OUX051-which are presumed to contain the alleles mnd8 OUM165 , mnd4 OUM169 , and mnd1 OUX051 , respectively, and Akashinriki, an original cultivar of OUM165 and OUM169, were subjected to detailed genetic and phenotypic analyses.
First, we evaluated the vegetative phenotypes. The three mnd mutants produced more leaves than Akashinriki at all stages of vegetative development (Fig 1A-1D). However, the extent of leaf emergence varied among the mutants and growth stages. In the WT and mnd mutants, leaf emergence was slow in the early vegetative phase and rapid in the late vegetative phase (Fig 1E). At 40 days after germination, the number of leaves that emerged in mnd8 OUM165 and mnd1 OUX051 was twofold that in the WT, and that in mnd4 OUM169 was intermediate between those of the WT and other mutants, indicating that the three mutants have an increased leaf production rate ( Fig 1E).
Two indices represent the temporal pattern of leaf production, the plastochron and the leaf emergence rate which is also referred to as the phyllochron [30]. The plastochron is the time interval between two successive leaves produced at the SAM, and the leaf emergence rate / the phyllochron is the time interval between two successive leaves emerging from the top of the former leaf sheath. Although the plastochron and the leaf emergence rate are equal in rice, the plastochron is shorter than the leaf emergence rate in most other cereal crops [30]. To calculate the plastochron, we observed shoot apices of the WT and three mnd mutants at 7 and 14 days after germination (DAG) (S1 Fig). In most shoot apices of Akashinriki at 7 DAG, the P1 primordium was the fifth leaf protruding from the SAM (S1A Fig). At 14 DAG, the P1 primordium was the seventh leaf in most Akashinriki shoot apices (S1E Fig). Therefore, two new leaf primordia were produced from the SAM over 1 week in Akashinriki. In the mnd mutants, most of the P1 primordia were fifth leaves at 7 DAG, but eighth leaves at 14 DAG (S1B-S1D and S1F-S1H Fig). The plastochrons of the three mnd mutants were shorter than that of the WT ( Table 2). We also calculated the leaf emergence rates for the four growth periods. For the first period from 7 to 14 DAG, the leaf emergence rate was longer than the plastochron in the WT and mnd mutants. Although the leaf emergence rates in the WT and mnd mutants became shorter in the later periods, the difference between the WT and mnd mutants increased, suggesting that the leaf production rates of mnd mutants increased at later developmental stages. These results suggested that the three mnd mutants were defective in plastochron control, although the plastochron was shorter than the leaf emergence rate in both the wild-type and mnd mutants. In addition to the vegetative development, the three mnd mutants showed abnormal panicle morphology. In the normal panicle of barley, a non-branched single axis of the panicle is produced, but diverged and multiple panicles were observed in mnd8 OUM165 and mnd1 OUX051 (S2 Fig). A panicle abnormality was also recognized in mnd4 OUM169 , which exhibited the weakest vegetative-development phenotype among the mnd mutants.

Growth and cell-division pattern of mnd leaf primordia
An increased leaf production rate may alter the spatial relationship among leaf primordia in the mnd mutants. It is possible that the mnd mutants have more leaf primordia inside a leaf primordium than wild type at the same developmental stage. To assess this, we examined the shoot apices of the mnd mutants at 14 DAG under an electron microscope (Fig 2). The   relationships between successive leaf primordia in the mnd mutants were similar to that in the WT. Namely, when the P1 leaf primordium protruded from the flank of the SAM, the P4 leaf primordium gradually enclosed the inner leaf primordia in the WT (Fig 2A and 2B), and this relationship was conserved in the three mnd mutants (Fig 2C-2F).
To understand the growth pattern of leaf primordia in the mnd mutants, cell division around the shoot apex was investigated by in situ hybridization using the cell-division biomarker histone H4 (Fig 2G-2J). Patterns of density of histone H4 signals in the mnd4 and mnd8 mutants were similar to that in the WT; namely, cell division activity was relatively low at P1 to P4 and enhanced in P5-stage leaf primordia in the WT, mnd4, and mnd8. The number of histone H4 signals in P2 and P3 primordia did not significantly differ among Akashinriki, mnd4 and mnd8, although those of the mnd1 were increased in P3 (Figs 2G-2J and S3). Therefore, stage-specific developmental events in leaf primordia in the mnd4 and mnd8 are likely unchanged, although leaf stage progressions could be more accelerated in the mnd1. Thus, the barley mnd mutants exhibited rapid leaf production but maintained leaf stage-specific development by rapid leaf maturation, as do pla mutants in rice [7].

Molecular identification of three MND genes
The three mnd mutants exhibited similar phenotypes, and the affected traits were comparable to those of pla mutants in rice. Accordingly, the three MND genes in barley could be counterparts of the three rice PLA genes [5,7,9].
To identify the three MND genes, we first determined the nucleotide sequences of MND4/ HvMND (HOR5Hr1G081060) in OUM169, a homolog of rice PLA1 [29] We detected a single nucleotide polymorphism between Akashinriki and mnd4 OUM169 . The mutation was a G-to-A single base change (G425A) causing a Gly-to-Asp amino-acid substitution in the first exon of MND4 (Fig 3A and Table 1). This amino acid is conserved among angiosperms, suggesting it to be responsible for the phenotype of mnd4 OUM169 (S5 Fig). We detected an identical mutation in the other two mnd mutants, mnd4 GSHO2135 and mnd4 GSHO1798 , that caused an Arg-to-Lys amino-acid substitution, which has been reported previously as the mnd4.e allele [29] (Figs 3A and S5 and Table 1). Because we found multiple mnd4 alleles linked to a shortened plastochron phenotype, MND4/HvMND regulates the plastochron in barley.
Next, to identify candidates for MND1 and MND8, we tested the barley ortholog of PLA2: HvPLA2 (HORVU3Hr1G091930). Rice pla2 exhibits a phenotype similar to those of barley mnd mutants [7]. We determined the nucleotide sequences of HvPLA2 in various mnd mutant and WT pairs. A database search showed that HvPLA2 is located in barley chromosome 3H, which is syntenic to rice chromosome 1 carrying PLA2, but our mapping showed that MND8 and MND1 are located on chromosomes 7H and 5H, respectively (S6 Fig). We found no mutation causing an amino-acid change or frameshift in HvPLA2 in mnd8 OUM165 and mnd1 OUX051 , A barley ortholog of rice pla3 was next investigated because of its similar phenotype [9]. The barley ortholog HvPLA3 (HORVU5Hr1G103900) of PLA3 in rice chromosome 3, is located on syntenic chromosome 5H. We mapped MND8 to the distal region of the long arm of barley chromosome 5H (S6 Fig), but this position deviated distally from the HvPLA3 candidate interval. No mutation was detected in the nucleotide sequence of HvPLA3 in mnd8 OUM165 . Sequencing of HvPLA3 in mnd1 OUX051 revealed two point mutations causing amino-acid substitutions, and a six-base deletion causing two amino-acid deletions in HvPLA3 in mnd1 OUX051 . Although one of the point mutations and the deletion mutation did not change the amino acids conserved among angiosperms, a C-to-A point mutation (C773A) in the first exon caused a Pro-to-Gln amino-acid substitution (P258Q) at a conserved position (S7 and S8 Figs). However, we did not find nucleotide polymorphisms between Akashinriki and the other mnd mutants with mnd1 alleles, i.e., mnd1 GSHO253 and mnd1 SM6 . Based on these inconsistent mapping and sequencing results, HvPLA3 is not responsible for mnd1.
Because it is unlikely that MND1 and MND8 are HvPLA2 or HvPLA3, we searched for other candidate genes that regulate the plastochron based on reports in species other than rice. Rough mapping in barley showed that mnd8 is located 7.2 cM distal to the barley EST marker k08652 (syn. HORVU5Hr1G112990), in the distal region of the long arm of chromosome 5H according to the EST map [31] (S6 Fig). Around the mapped region, we found a candidate gene, a homolog of BIGE1, which regulates not only embryo size but also the leaf initiation rate in maize [24]. We subsequently examined the nucleotide sequence of the BIGE1 homolog (HORVU5Hr1G103900), which encodes a MATE transporter-family protein. We found a Tto-C single base change (T3022C) mutation in the second exon of the gene in mnd8 OUM165 that caused a Leu-to-Pro amino-acid substitution (L404P) (Figs 3B and S9 and Table 1). The nucleotide sequence of the gene in four mnd mutants-OUM166, OUM168Ku, NBG114547, and NGB117205-revealed three independent mutations causing amino-acid substitutions (G106E, W278R, and G298N) in the coding region. Accordingly, we concluded that this MATE transporter-like gene is MND8. A phylogenetic analysis indicated that MND8 is an ortholog of maize BIGE1 (S4 Fig). Next, we mapped mnd1 using publicly available simple sequence repeat (SSR) markers in barley [32] (S6 Fig). Barley mnd1 was located in the distal region of the long arm of barley chromosome 7H, flanked by the barley EST markers, k09432 (syn. HOR-VU7Hr1G109770) and k04741 (syn. HORVU7Hr1G115050). Using this map information, we searched for mnd1 candidate genes. We identified as a strong candidate for MND1 (HORVU7Hr1G113480), a homolog of HOOKLESS1 (HLS1), which encodes an N-acetyltransferase-like protein. Although there are no reports on HLS1 homologs regulating the plastochron in Arabidopsis, multiple Arabidopsis mutants of HLS1 homologs exhibited excess leaf production and a small leaf size [33][34][35]. One HLS1 homolog in barley was located near the distal arm region of chromosome 7HL, where mnd1 was mapped. Sequence analysis of this gene revealed that mnd1 OUX051 carries an eight-nucleotide deletion in the second exon starting at position 464, causing a frameshift and introducing a premature stop codon (Figs 3C and S10 and Table 1). Sequencing of four mnd mutants with the mnd1 alleles (mnd1 SM5 , mnd1 SM6 , mnd1 GSHO253 , and mnd1 GSHO2038 ) and one other mnd mutant, NGB114540, revealed an eight-base insertion, two independent one-base deletions, and one amino-acid substitution in the coding region (Figs 3C and S10 and Table 1). In addition, among the mutations in MND1, the eight-base insertion in mnd1 OUX051 , mnd1 GSHO253 , and mnd1 GSHO2038 was identical to that was recently reported in mnd1.a [28]. Therefore, we concluded that this HLS1-like gene is MND1. A phylogenetic analysis indicated that MND1 is an ortholog not of HLS1 but of GW6a, a quantitative trait locus that regulates grain weight in rice [36] (S4 Fig).
Accordingly, the three MND genes-MND4, MND8, and MND1-encode CYP78A family, MATE transporter-family, and N-acetyltransferase-like proteins, respectively. The MND8 is newly identified by this study.

Expression pattern of the three MND genes
To evaluate their regulation of the plastochron and leaf growth, the transcript accumulation patterns of the three MND genes around the shoot apex were investigated using in situ hybridization (Fig 4). No signals were detected in longitudinal sections hybridized with sense probes for MND4, MND8 and MND1 (S11 Fig). MND4 expression was observed both in the SAM and leaf primordia (Fig 4A). The expression pattern in the SAM was complex, with expression detected in small patches of the SAM flank in the longitudinal sections. One patch was observed at the boundary between the SAM and P1 leaf primordia. Cross sections revealed that the patch corresponded to a semicircular band in the SAM (Fig 4D). Strong MND4 expression was also observed in the proximal part of the adaxial and abaxial sides of P1-P4 leaf primordia, including the boundary between successive leaf primordia (Fig 4A and 4G). Expression of MND8 was weak in the SAM and young leaf primordia, and was observed faintly in the young stem (Fig 4B, 4E and 4H).
The expression pattern of MND1 was similar to that of MND4; that is, expression was observed in small patches of the SAM and the proximal part of leaf primordia (Fig 4C). However, the area of expression was different from that of MND4. First, MND1 expression was detected not at the boundary between the SAM and P1 leaf primordia but in the inner region of the SAM (Fig 4C and 4F). Second, gene expression at the boundary between successive leaf primordia was marked for MND4, but not for MND1. Finally, MND1 expression in the proximal part of leaf primordia was shifted distally and medially compared to MND4 (Fig 4C and  4I).
These observations suggest that MND4 and MND1 regulate the plastochron and leaf growth by means of their expression in limited but distinct regions of the SAM and leaf primordia.

Effect of mnd mutations on the expression level of the three MND genes
To examine whether MND genes regulate other MND genes, we compared the expression levels of MND genes among three mnd mutant backgrounds and WT Akashinriki and Mesa, with the former the original cultivar of mnd4 OUM169 and mnd8 OUM165 and the latter that of mnd1 OUX051 (Fig 5). Quantitative real-time PCR revealed that the expression of MND8 was slightly decreased in mnd4 OUM169 (Fig 5A). This indicates that MND4 positively regulates MND8 expression. The expression level of MND4 was slightly upregulated in mnd8 OUM165 and significantly upregulated in mnd1 OUX051 , suggesting that MND1 negatively regulates MND4 expression (Fig 5B). The expression level of MND1 was upregulated in mnd8 OUM165 and downregulated in mnd1 OUX051 , implying that MND8 negatively regulates MND1 expression and that mnd1 OUX051 affects the accumulation of MND1 mRNA (Fig 5C). In addition, we examined the expression of HvPLA2 and HvPLA3 (Fig 5D and 5E). Although HvPLA3 expression was not significantly altered in the three mnd mutants (Fig 5E), HvPLA2 expression was downregulated in the mnd mutants (Fig 5D). This suggests that the three MND genes positively regulate HvPLA2 expression.
Accordingly, the expression levels of the three MND genes in addition to that of HvPLA2 were affected by functional defects in other MND genes.

Genetic interactions among MND genes
To examine the genetic interactions among MND genes, we generated double mutants between mnd8 OUM165 and mnd4 OUM169 as well as mnd8 OUM165 and mnd1 OUX051 (Fig 6A and  6B). We measured the leaf number and leaf blade length in the F 3 population at 2 months after germination and determined the genotypes (Fig 6C and 6D). Among the segregants of the F 3 population of mnd8 OUM165 and mnd4 OUM169 , double homozygotic plants for mnd8 OUM165 and mnd4 OUM169 produced the largest number of leaves among the nine genotypes (Fig 6A and  6C). Although the difference in leaf number between single mnd8 OUM165 homozygotic plants and double homozygotic plants was slight, mature double-mutant plants exhibited an enhanced phenotype relative to the mnd8 OUM165 single mutant (S12 Fig), indicating that the mnd8 OUM165 allele has an additive effect with mnd4 OUM169 . Among the segregants of the F 3 population produced from mnd8 OUM165 and mnd1 OUX051 , an additive effect between the two alleles on leaf number was evident; that is, the mnd8 OUM165 mnd1 OUX051 double mutant had the largest number of leaves among the genotypes (Fig 6B and 6D).
The mnd alleles had different effects on the increase in leaf number depending on their genotype (Fig 6C and 6D). For example, mnd8 OUM165 increased leaf numbers in the WT and mnd4 OUM169 backgrounds by 94% and 73%, respectively. Similarly, mnd4 OUM169 increased leaf numbers in the WT and mnd8 OUM165 backgrounds by 21% and 8%, respectively (Fig 6C). The same tendency was observed for combinations of mnd8 OUM165 and mnd1 OUX051 (Fig 6D). Accordingly, the effect of a single mnd allele on the increase in leaf number was diminished by the accumulation of other mnd alleles.
Therefore, MND8 regulates the plastochron independently of MND4 and MND1. In addition, there are genetic or developmental mechanisms that moderately affect the leaf production rate in the short plastochron background.

Relationship between the plastochron and leaf-blade length in the mnd mutant
The shortened plastochron in most plastochron-related mutants is reportedly accompanied by the production of small leaves [4,6,7,24]. However, the correlation between leaf production rate and leaf length is unknown. Therefore, the leaf number at 2 months after germination and the length of the second leaf blade of F 3 segregants produced from mnd8 OUM165 and mnd4 OUM169 and from mnd8 OUM165 and mnd1 OUX051 were measured (Fig 7). A strong negative correlation between leaf number and leaf-blade length was observed in all F 3 segregants for both mnd8 OUM165 and mnd4 OUM169 crosses and mnd8 OUM165 and mnd1 OUX051 crosses ( Fig  7A and 7B). This correlation was caused by the strength of the allelic effect on leaf production (Fig 6C and 6D). By contrast, no negative correlations were detected among the most of groups of segregants with identical genotypes (S13 Fig). In addition, among various WT and mnd mutant strains, no correlation between leaf number and the maximum length of the leaf blade was observed (S14 Fig and S2 Table).
These observations indicate that the plastochron and leaf-blade length are positively regulated by MND genes. However, the correlations between the plastochron and leaf length in the mnd mutants were slightly incomplete, suggesting that MND may partially acts on pathways that regulate the plastochron and leaf-blade length independently.

Effect of rice MND orthologs on plastochron regulation
Our results demonstrated that MND8 and MND1 regulate the plastochron in barley, but whether the rice homologs of these genes are involved in plastochron regulation is unknown. Therefore, we constructed knockout mutants of rice MND8 and MND1 orthologs using CRISPR/Cas9 (S15 Fig). A phylogenetic analysis indicated that two orthologs of barley MND8 and MND1 are present in rice (S4 Fig; two rice orthologs, OsBE1/Os03g0839200 and OsBE2/ Os12g0552600, for barley MND8 and two rice orthologs, OsGNAT1/Os06g0650300 and OsG-NAT2/Os02g0180400, for barley MND1). We generated single and double mutants of the rice orthologs and calculated the plastochron (Fig 8A-8G). The number of leaves increased in osg-nat1, osbe1, osgnat1 osgnat2, and osbe1 osbe2 compared to the WT, but to a lesser extent compared to pla1-4 (Figs 8G and S16). This indicates that rice OsBE1 and OsGNAT1 are involved in plastochron regulation, as are their orthologs in barley, but their contribution to the plastochron is of lesser magnitude than that of MND in barley.
In situ hybridization showed that OsBE1 and OsGNAT1 were expressed around SAMs, particularly at the base of young leaf primordia (Fig 8H-8M), whereas OsBE2 and OsGNAT2 expression was not observed. We next compared the expression patterns of PLA1, OsBE1, and OsGNAT1. Although PLA1 and OsBE1 expression domain overlapped, PLA1, but not OsBE1, was expressed at the base of the P1-related region in the SAM (Fig 8H and 8I). In addition, OsBE1, but not PLA1, expression was detected in the inner tissue of P3 and the stem tissue ( Fig  8K and 8L). OsGNAT1 expression was strongest at the base of P3 and the boundaries of leaf primordia (Fig 8J and 8M). Expression of the three genes partially overlapped, but none was expressed in the vascular bundle (Fig 8K-8M). By contrast, there were differences in the expression patterns of the rice and barley orthologs, although MND4 and PLA1 expression was similar. Punctate expression of MND1 was observed in the SAM (Fig 4C), whereas OsGNAT1 expression was detected in the P3 leaf primordium but not in the SAM (Fig 8J and 8M). MND8 did not exhibit a specific expression pattern (Fig 4B), but OsBE1 was expressed around the shoot apex (Fig 8I and 8L).
These results indicate functional conservation between the barley and rice orthologous genes (MND8 versus OsBE1 and MND1 versus OsGNAT1).

Three MND genes regulate the plastochron via similar developmental pathways, but have unrelated functions and genetic pathways
Our results indicate that the many-noded phenotype of mnd mutants results from the rapid production of leaf primordia, which is caused by loss-of function of three independent genes, MND4, MND8, and MND1. Although the magnitude of the plastochron reduction differed among the mutant alleles, the overall phenotypes of all mnd mutants during vegetative development were similar. In addition, the panicle-development phenotypes were comparable among the mutants. Therefore, common abnormalities were exhibited during vegetative and reproductive development, indicating that the three MND genes have similar roles in barley development. Despite their similar biological functions, the three MND genes encode unrelated proteins. MND4 encodes a CYP78A monooxygenase whose substrates are unknown [29]. Although CYP78A genes are important for development in several species [29,[37][38][39][40][41], the synthetic or metabolic pathways in which CYP78A is involved are unknown. MND1 encodes an N-acetyltransferase-like protein and is the closest homolog of GW6a in rice [36]. Because GW6a exhibits histone H4 acetyltransferase activity, MND1 also likely regulates the transcription of downstream genes by acetylating histone H4. Furthermore, MND8 is an ortholog of maize BIGE1, a transporter implicated in the secretion of an unidentified small molecule [24]. Accordingly, it is assumed that MND4, MND1, and MND8 are involved in the synthesis or metabolism of unknown factors, transcriptional regulation of downstream genes, and transportation of unidentified molecules, respectively. Although a close relationship between the CYP78A pathway and BIGE1 has been proposed [24], there is no direct evidence that BIGE1 is associated with the transportation of CYP78A-related molecules so far. In fact, although we did not investigate the relationship between MND4 and MND1, our genetic analysis indicates that MND4 and MND1 regulate the plastochron independently of MND8. Accordingly, at least two different genetic pathways regulate the plastochron in barley. This is also the case for plastochron regulation by three PLA genes in rice. The pla1, pla2, and pla3 mutants have a short plastochron and a small leaf size, and conversion of the primary rachis into a shoot occurs in all three, but it has been proposed that PLA1, PLA2, and PLA3 regulate the plastochron independently [7,9]. On the other hand, Arabidopsis AMP1 and CYP78A5/7, which are orthologs of PLA3 and PLA1, respectively, act on a common downstream process [42].
Therefore, the developmental program controlling leaf initiation, leaf growth, and panicle development in rice and barley is regulated by multiple and independent genetic pathways involving PLA and MND genes.

Interactions of MND genes
At least two independent genetic pathways seem to regulate the plastochron, but complex genetic interactions were also suggested by the expression analysis. In short, MND8 negatively affects the expression level of MND1, MND1 negatively affects that of MND4, and MND4 positively affects that of MND8. It was also noted that MND4 expression was slightly upregulated, but not significantly, in mnd8. Such a regulatory relationship between CYP78A and a MND8-related transporter has also been reported in maize; i.e., the expression of some CYP78A family genes was upregulated in the bige1 mutant background [24]. In maize, BIGE1 is required for feedback regulation of CYP78A family genes. If this is also the case in barley, the upregulation of MND4 and MND1 might have been the result of a deficiency in feedback regulation caused by the loss of MND8 function. By contrast, overexpression of rice GW6a, an ortholog of MND1, and the mnd1.a mutation affected the expression levels of thousands of genes in rice and barley, respectively [28,36]. Because GW6a is predicted to be a transcriptional regulator that modulates chromatin status, a change in MND4 expression in mnd1 could be a direct or indirect effect of loss of histone acetyltransferase activity by MND1. Although MND4 and MND8 do not directly affect gene transcription, they modulate that of other genes by influencing downstream events. In addition to the three MND genes, expression of HvPLA2 was downregulated in the three mnd mutants. This indicates that HvPLA2 is involved in plastochron regulation downstream of the three MND genes.
Phenotypic analysis of double mutants revealed that the effect of MND genes on plastochron regulation was dependent on the number of mnd mutant alleles present. In addition, the increase in leaf number in the double mutants was much less than when the effects of the two mutant alleles on leaf number were equally additive. There are two possible explanations for this phenomenon. One is that genetic pathways regulating plastochron are not completely independent between MND8 and MND4/MND1. In fact, expression levels of MND4 and MND1 were increased in the mnd8 background, which might weaken the short plastochron phenotype in the single and double mutants. Another is that a developmental restriction would exist, for example, responses of cells (e.g. division activity) in the SAM by the loss of each MND function do not proportionally affect leaf production rate. In any case, complex genetic and developmental interactions are implicated in the regulation of the plastochron.

MND genes function in leaf production and growth
In shortened-plastochron mutants, rapid leaf production is accompanied by small leaves [4,6,7,24]. In pla1 and pla2 mutants in rice, a model of the effect of leaf production on leaf size was proposed. According to the model, rapid leaf production by pla mutants is not the result of the loss of PLA function in the SAM, but an indirect effect of accelerated leaf maturation in leaf primordia [7]. The model was based on the observation that PLA genes were expressed in leaf primordia but not in the SAM, although the mechanism by which the change in leaf-primordia maturation affects leaf production is unknown.
Our barley analysis basically supports the above notion. The relative developmental stage and cell-division activity of successive leaf primordia were maintained or increased in the mnd mutants. Therefore, leaf production and leaf-primordia maturation were accelerated, as for pla mutants in rice. In addition, maize PLA1 genes control the duration of cell division in the leaf primordia [41], which is consistent with the rapid leaf maturation observed in the mnd4 mutant in this study. Accordingly, MND genes maintain the leaf-maturation schedule. However, if the leaf production rate is affected only by leaf maturation, as suggested by the rice model, the correlation between leaf number and leaf-blade length should be solid. However, we showed that leaf number and leaf-blade length were not always correlated in plants of identical genotype, implying that the plastochron and leaf size are differentially regulated. In fact, expression of MND4 and MND1 was observed both in the SAM and leaf primordia. MND expression in the SAM may be required for the suppression of leaf initiation independently of expression in the leaf primordia, which is necessary for the suppression of leaf maturation. Based on our findings, we propose a model that not only MND function in the leaf primordia but also in the SAM would be important for plastochron regulation in barley. This is supported by the fact that enhanced expression of the maize PLA1 ortholog in leaf primordia affected leaf size but not leaf number [41], whereas overexpression of PLA1 via the introduction of an increased gene copies resulted in a prolonged plastochron and an increased leaf size [43].
The mechanisms by which MND genes regulate leaf initiation in the SAM and leaf growth in the leaf primordia are unknown. Auxin may be associated with control of the duration of cell-division activity by maize PLA1 [41]. It is also probable that auxin triggers leaf initiation via MND-mediated regulation, because the local auxin concentration is important for leaf initiation in the SAM [1,44]. It is possible that the limited MND1 and MND4 expression domain in the SAM is important for the auxin flow or concentration required for proper leaf initiation. Analysis of auxin dynamics in the mutant SAM may provide insight into the role of auxin in the temporal regulation of leaf initiation by MND genes.

Conservation and diversity of genetic pathways in rice and barley
Although several genetic factors regulating the plastochron have been identified in rice, maize, Arabidopsis, and other species, our understanding of their functional conservation among plant species is inadequate. We characterized mnd mutants in barley and identified the responsible genes. It was somewhat surprising that two of the three MND genes were not homologs of three PLA genes in rice. At least five genes are involved in developmental pathways underpinning plastochron regulation in grasses. Moreover, three of the five orthologs perform similar functions in rice and barley. However, there was a considerable difference in the magnitude of their effects between the two species. For example, loss-of-function of MND1 and MND8 markedly shortened the plastochron in barley, but that of the rice orthologs OsGNAT1 and OsBE1 resulted in a slight effect, even for double loss-of-function of closely related paralogs. This might have been caused by differences in genetic redundancy and functional diversification of gene families between the two species. Namely, other homologs in different clades compensate for the function of plastochron regulation in rice. In fact, the functional ortholog of MND4 in rice is PLA1, but it is not the closest homolog among the CYP78A members [29]. It is predicted that Os09g0528700 is a phylogenetic ortholog of MND4 in rice, which is specifically expressed in roots and the embryo (http://ricexpro.dna.affrc.go.jp), unlike PLA1. In terms of the functions of MND8-related MATE transporter-family genes, BIGE1 in maize regulates both the leaf initiation rate and embryo size [24]. However, mnd8 in barley did not exhibit a phenotype change in terms of embryo size (S17 Fig). In addition, OsBE1, a rice ortholog of MND8, exhibited specific expression around the shoot apex, unlike MND8. Accordingly, the phenotypic effect and expression pattern of the closest homologs of MND genes differ among grass species.
Our understanding about a conserved function of plastochron-related genes among grass species is still insufficient. Identifications of HvPLA2 and HvPLA3 mutants in barley and comparative analyses with their rice counterparts would advance an understanding of the conservation and diversity of plastochron-related genes.

Plant materials and growth conditions
The barley mnd mutant and WT strains used in this study were listed in Table 1. Mutant and wild-type seeds were sown in soil and the seedlings were transplanted into the pots under conditions in the greenhouse of the University of Tokyo, and the plants were sampled and their phenotypes evaluated at predetermined timepoints (S1 Table). For the double mutant analysis, the genotypes of the F 3 plants were identified using PCR-based genotyping.
The pla1-4 mutant and WT rice plants were grown on soil or MS medium containing 3% sucrose and 1% agar at 28˚C under continuous light. Transgenic plants were grown in a biohazard greenhouse with temperatures of 30˚C in the daytime and 25˚C at night.

Histological and morphological analysis
Samples of the mutant and WT plants were fixed with FAA (formaldehyde:glacial acetic acid:50% ethanol, 2:1:17) for 24 h at 4˚C for histological analysis, or with PFA (4% [w/v] paraformaldehyde and 1% Triton X in 0.1 M sodium phosphate buffer) for 48 h at 4˚C for in situ hybridization. They were then dehydrated in a graded ethanol series, after which the ethanol was substituted with 1-butanol, and the samples were embedded in Paraplast Plus (McCormick Scientific). The samples were sectioned at a thickness of 8 μm using a rotary microtome. The sections were stained with hematoxylin for histological analysis. After staining, the sections were mounted with Poly-Mount (Polysciences, Inc.) and observed under a light microscope.
For scanning electron microscopy, plant materials were fixed in PFA for 24 h and dehydrated in an ascending ethanol series, which was then gradually replaced with 3-methyl-butylacetate. Samples were critical-point dried, sputter-coated with platinum, and observed under a scanning electron microscope (S-4800, Hitachi) at an accelerating voltage of 10 kV.

In situ hybridization
Paraffin sections were prepared as described above. For digoxigenin-labeled antisense RNA probes, those for histone H4 (HvH4: HORVU5Hr1G086620), MND4, MND8, and MND1 in barley and OsBE1 and OsGNAT1 in rice were prepared using cDNAs and specific primers (S3 Table). In addition, antisense RNA probes for MND4, MND8, and MND1 were also prepared. The PLA1 antisense probes were prepared as described previously [5]. In situ hybridization and immunological detection using alkaline phosphatase were performed according to the methods of Kouchi and Hata [45]. The experiments were performed using at least five samples per each probe.
For quantification of cell division activity in the P2 and P3 primordia in WT and mnd mutants, the number of histone H4 spots per area of the leaf primordia were calculated using three independent samples of each genotype (S1 Table). Five serial sections in a sample were used for measurements. The area of the leaf primordium was measured with IMAGEJ (http:// rsb.info.nih.gov/ij/).
For phylogenic analyses, the amino-acid sequences of MND4, MND8, and MND1 homologs in various plant species were obtained from the Phytozome database. The amino-acid alignment was carried out using GENETYX software (Genetyx), and the phylogenic tree was constructed based on the neighbor-joining method with 1,000-replicate bootstrapping using Molecular Evolutionary Genetics Analysis (MEGA X; [46]).

Quantitative real-time PCR
RNA was extracted from 1 cm of the most basal part of the shoot apices of barley seedlings at 14 DAG using TRIzol reagent (Invitrogen). The extracted RNA was treated with Recombinant DNase I (TaKaRa), and cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Life Technologies). Quantitative real-time PCR was performed with the StepOne Real-Time PCR System (Life Technologies) using TaqMan Fast Universal PCR Master Mix and FAM-labeled TaqMan probes for each gene. ADP-ribosylation factor 1-like protein (ADP) was used as the internal standard [47]. In all experiments, we analyzed three technical and five biological replicates (S1 Table). The primers and TaqMan probes for MND1, MND4, MND8, ADP, HvPLA2, and HvPLA are listed in S4 Table. Generation of knockout alleles of rice MND orthologs using the CRISPR/ Cas9 system The CRISPR/Cas9 system was used to generate knockout alleles for OsBE1, OsBE2, OsGNAT1, and OsGNAT2, which are orthologs of MND1 and MND8 in rice. The target sites were selected using the CRISPR-P program (http://cbi.hzau.edu.cn/crispr/) [48] (S15 Fig). The single-guide RNA (sgRNA) cloning vector (pZK_gRNA) and all-in-one binary vector (pZH_OsU6gR-NA_MMCas9) harboring sgRNA, Cas9, and NPTII were provided by Masaki Endo [49]. The pZH_OsU6gRNA_MMCas9 vector including target-guide RNA for MND orthologs was constructed as described previously [49].
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 PCRbased detection (S15 Fig). We performed rough observations of morphological phenotypes with at least three or more loss-of-function alleles. And then, we used only transgene-free genome edited plants (null segregants) for each gene for analyzing in detail and generating double mutants in our experiments.