Mutations in the Rice OsCHR4 Gene, Encoding a CHD3 Family Chromatin Remodeler, Induce Narrow and Rolled Leaves with Increased Cuticular Wax

Leaf blade width, curvature, and cuticular wax are important agronomic traits of rice. Here, we report the rice Oschr4-5 mutant characterized by pleiotropic phenotypes, including narrow and rolled leaves, enhanced cuticular wax deposition and reduced plant height and tiller number. The reduced leaf width is caused by a reduced number of longitudinal veins and increased auxin content. The cuticular wax content was significantly higher in the Oschr4-5 mutant, resulting in reduced water loss rate and enhanced drought tolerance. Molecular characterization reveals that a single-base deletion results in a frame-shift mutation from the second chromodomain of OsCHR4, a CHD3 (chromodomain helicase DNA-binding) family chromatin remodeler, in the Oschr4-5 mutant. Expressions of seven wax biosynthesis genes (GL1-4, WSL4, OsCER7, LACS2, LACS7, ROC4 and BDG) and four auxin biosynthesis genes (YUC2, YUC3, YUC5 and YUC6) was up-regulated in the Oschr4-5 mutant. Chromatin immunoprecipitation assays revealed that the transcriptionally active histone modification H3K4me3 was increased, whereas the repressive H3K27me3 was reduced in the upregulated genes in the Oschr4-5 mutant. Therefore, OsCHR4 regulates leaf morphogenesis and cuticle wax formation by epigenetic modulation of auxin and wax biosynthetic genes expression.


Introduction
In plants, leaves are the major organ used for photosynthesis and transpiration, and are considered to be the main source of dry matter accumulation in plants. Blade width and curvature are key factors of leaf morphology. As reported, blade width, particularly that of the first and second uppermost leaves, can significantly influence the rice grain yield, whereas appropriately rolled leaves can optimize canopy light transmission, increase the effective leaf area per unit of land and, consequently, the canopy photosynthesis efficiency [1].
To date, a number of narrow and/or rolled leaf genes have been reported in rice. The majority of narrow leaf genes are related to auxin signaling. For example, a mutant of Nal1 (encoding a plant-specific and inhibition of meristematic genes in carpel tissue [33,34]. In rice, the closest homolog with PICKLE is OsCHR702, while both T-DNA insertion mutation and RNAi of OsCHR702 did not produce any morphological phenotypes [35,36]. OsCHR4 is another CHD member characterized in rice. The Oschr4 mutant shows defective chloroplasts in the adaxial mesophyll cells due to blocked proplastid growth and thylakoid membrane formation [37]. As an allelic gene of OsCHR4, OsCHR729 was revealed to encode a bifunctional chromatin regulator associated with gene transcription regulation through modulation of histone H3K4 and H3K27 methylation during plant development [36]. Meanwhile, the Oschr729 mutant induced by T-DNA insertion shows pleiotropic phenotypes, including small and rolled leaves, reduced stem elongation and chlorophyll contents, and absence of secondary panicle branches [36]. One study reported that CHR729 regulates seedling development by affecting contents of gibberellin acid (GA) in rice, and the transcript level of GA biosynthesis genes was altered in the corresponding mutant line t483 [38]. In addition, CRL6, an allelic gene of CHR729, was speculated to regulate crown root development via auxin signaling pathway due to decreased expression of most OsIAA genes in crl6 compared with the wild type (WT) [39]. Recently, rfs, another allelic mutant of OsCHR4/OsCHR729, was characterized by the chlorotic and cell death phenotypes due to accumulation of reactive oxygen species (ROS). Both transcript and H3K4me3 levels of five ROS-related genes declined in the rfs mutant [40].
In this study, three additional allelic mutants of OsCHR4 were reported, namely Oschr4-5, Oschr4-6, and Oschr4-7, which exhibited pleiotropic phenotypes such as narrow and rolled leaf, reduced plant height, tiller number and seed size at mature stage. Anatomical investigations revealed that the reduced leaf width was due to a decreased number of major and minor veins while the cell size remained unchanged. The auxin content was significantly higher in Oschr4-5, which may explain the altered venation pattern and leaf size at physiological level. Interestingly, more cuticular wax was deposited on the surface of the Oschr4-5 leaves, which reduced the water-loss rate and enhanced the drought tolerance. Expression levels of several auxin and wax biosynthesis genes were up-regulated in the Oschr4-5 mutant. Consistent with the role of CHD3 protein in chromatin remodeling, histone modifications H3K4me3 and H3K27me3 in the upregulated genes were increased and reduced in Oschr4-5, respectively. Our data reveal novel functions of OsCHR4 in the epigenetic regulation of auxin and wax biosynthesis-related genes, leaf development and cuticle wax formation in rice.

Pleiotropic Phenotype of the Oschr4-5 Mutant
A rice mutant characterized by narrow and rolled leaves was isolated from the gamma ray mutagenized Indica rice cultivar Indica 9, which was named as Oschr4-5 based on the gene identity revealed in the following study. At the seedling stage, slight dwarfism and albino on the adaxial side of leaves were observed in Oschr4-5 ( Figure 1A,B). At the mature stage, the plant height and tiller number of Oschr4-5 was significantly reduced by 45% and 76% when compared with that of WT (p < 0.01) ( Figure 1C, Table S1). The length and width of the leaves in Oschr4-5 was significantly reduced by 38% and 50% when compared with that of the WT (p < 0.01) ( Figure 1D,F, Table S1). Meanwhile, smaller seeds were also observed in Oschr4-5 ( Figure 1G). Later on, two allelic mutants, namely Oschr4-6 andOschr4-7, were isolated from the Ethyl methane sulfonate (EMS) mutagenized japonica cultivar Nipponbare, which exhibited the same pleiotropic phenotypes, including narrow and rolled leaves with albino on the adaxial side, significantly reduced leaf length, plant height, tiller number and seed size ( Figure S1, Table S1).

Anatomical and Physiological Basis of Reduced Leaf Width
To reveal the underlying cellular basis of narrow leaves in the Oschr4-5 mutant, the venation pattern of leaf was first observed. The number of large veins and small veins of Oschr4-5 was significantly reduced by 36% and 42%, respectively, of that of the WT (p < 0.01) (Figure 2A-D,G-H). Then the epidermal cell profile on the abaxial side of leaf blade was examined. Neither the number of cell column between the adjacent small vein nor the cell width at the horizontal orientation showed significant difference between Oschr4-5 and WT ( Figure 2E,F,I,J). These results indicate that the decreased leaf width in Oschr4-5 may be caused by reduction in the number of veins. Since auxin is required for specification of the venation pattern in leaves [3,41] and auxin overproduction usually leads to narrow and epinastic leaves in Arabidopsis [42,43], the amount of free indole acetic acid (IAA), the major auxin, in Oschr4-5 and WT was measured. As a result, the content of IAA in Oschr4-5 was significantly higher than that of WT (86.39 ± 4.55 vs. 24.39 ± 0.76 ng/g, p < 0.01) ( Figure  2L).

Anatomical and Physiological Basis of Reduced Leaf Width
To reveal the underlying cellular basis of narrow leaves in the Oschr4-5 mutant, the venation pattern of leaf was first observed. The number of large veins and small veins of Oschr4-5 was significantly reduced by 36% and 42%, respectively, of that of the WT (p < 0.01) (Figure 2A-D,G-H). Then the epidermal cell profile on the abaxial side of leaf blade was examined. Neither the number of cell column between the adjacent small vein nor the cell width at the horizontal orientation showed significant difference between Oschr4-5 and WT ( Figure 2E,F,I,J). These results indicate that the decreased leaf width in Oschr4-5 may be caused by reduction in the number of veins. Since auxin is required for specification of the venation pattern in leaves [3,41] and auxin overproduction usually leads to narrow and epinastic leaves in Arabidopsis [42,43], the amount of free indole acetic acid (IAA), the major auxin, in Oschr4-5 and WT was measured. As a result, the content of IAA in Oschr4-5 was significantly higher than that of WT (86.39 ± 4.55 vs. 24.39 ± 0.76 ng/g, p < 0.01) ( Figure 2L). Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 5 of 18 . Significance of data is tested by student's t test. ** p < 0.01.

Increased Cuticular Wax on the Surface of Leaves with Enhanced Drought Tolerance in Oschr4-5
Scanning electron microscopy was used to examine the surface of leaves. Interestingly, the platelet-like wax crystals deposited on the leaves' surface were larger and denser in Oschr4-5 compared with WT ( Figure 3A,B). Then the compositions and contents of cuticular wax were measured by gas chromatography-mass spectrometry (GC-MS). As shown in Figure 3C, the wax components of fatty acids, aldehydes, alcohols and alkanes were all significantly increased in Oschr4-5 than WT (p < 0.05). . Significance of data is tested by student's t test. ** p < 0.01.

Increased Cuticular Wax on the Surface of Leaves with Enhanced Drought Tolerance in Oschr4-5
Scanning electron microscopy was used to examine the surface of leaves. Interestingly, the platelet-like wax crystals deposited on the leaves' surface were larger and denser in Oschr4-5 compared with WT ( Figure 3A,B). Then the compositions and contents of cuticular wax were measured by gas chromatography-mass spectrometry (GC-MS). As shown in Figure 3C, the wax components of fatty acids, aldehydes, alcohols and alkanes were all significantly increased in Oschr4-5 than WT (p < 0.05).  . Significance of data is tested by student's t test. ** p < 0.01.
As the content and structure of cuticular wax could strongly influence cuticle permeability, its effects on transpiration rate were first evaluated. Indeed, the transpiration rate of Oschr4-5 was significantly lower than that of WT (p < 0.01) ( Figure 3D). Secondly, measurement of cuticle membrane permeability as demonstrated by chlorophyll-leaching rate showed that chlorophyll leaches slower from Oschr4-5 than WT at different time points ( Figure 3E). Furthermore, a dehydration-rehydartion treatment revealed that drought tolerance of Oschr4-5 was significantly enhanced. The Oschr4-5 seedlings showed significantly higher recovery rate than WT (60% vs. 8%, p < 0.01) ( Figure 4A-D). The water loss rate was also measured to evaluate the water retention capacity of detached leaves. The data showed that water loss rates of Oschr4-5 were significantly slower than in WT ( Figure 4E). The possible effects of stomatal density and leaf thickness on the differential water loss rate might be ruled out because no significant difference was detected in the stomatal density ( Figure 2E,F,K) and leaf blade thickness (WT: 107.43 ± 7.51 μm ; Oschr4-5: 108.17 ± 7.23 μm) between Oschr4-5 and WT. These results prove that Oschr4-5 exhibited lower cuticle permeability and water loss because of higher cuticular wax content on the surface of leaves compared to WT. Significance of data is tested by student's t test. * p < 0.05, ** p < 0.01. (D,E) The transpiration rate (D) and chlorophyll leaching rate (E) of leaf blades from wild type and Oschr4-5 mutant. Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test. ** p < 0.01.
As the content and structure of cuticular wax could strongly influence cuticle permeability, its effects on transpiration rate were first evaluated. Indeed, the transpiration rate of Oschr4-5 was significantly lower than that of WT (p < 0.01) ( Figure 3D). Secondly, measurement of cuticle membrane permeability as demonstrated by chlorophyll-leaching rate showed that chlorophyll leaches slower from Oschr4-5 than WT at different time points ( Figure 3E). Furthermore, a dehydration-rehydartion treatment revealed that drought tolerance of Oschr4-5 was significantly enhanced. The Oschr4-5 seedlings showed significantly higher recovery rate than WT (60% vs. 8%, p < 0.01) ( Figure 4A-D). The water loss rate was also measured to evaluate the water retention capacity of detached leaves. The data showed that water loss rates of Oschr4-5 were significantly slower than in WT ( Figure 4E). The possible effects of stomatal density and leaf thickness on the differential water loss rate might be ruled out because no significant difference was detected in the stomatal density ( Figure 2E,F,K) and leaf blade thickness (WT: 107.43 ± 7.51 µm; Oschr4-5: 108.17 ± 7.23 µm) between Oschr4-5 and WT. These results prove that Oschr4-5 exhibited lower cuticle permeability and water loss because of higher cuticular wax content on the surface of leaves compared to WT. Recovery rate from dehydration of wild type and Oschr4-5 mutant in (C). Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test, ** p < 0.01. (E) Water loss rate of detached leaf blades between wild type and the Oschr4-5 mutant. Data are the means ± standard errors (n = 3).

Characterization of the Molecular Lesions in the Oschr4 Mutant
To find out the molecular lesion responsible for the pleiotropic phenotypes of the Oschr4-5 mutant, three F2 populations were generated (Oschr4-5 × 9311; Oschr4-5 × Nipponbare; Oschr4-5 × 02428). Within these F2 populations, a mendelian segregation ratio of 3:1 (χ 2 0.05 < 3.84) was calculated, which indicated the mutant phenotype was controlled by a single recessive gene. Then a large F2 population generated by Oschr4-5 × 02428 was used to map the causative locus. In primary mapping, 116 F2 mutant individuals located this gene between two Indel markers ha1 and ha10 on the long arm of chromosome 7. Fine mapping using 7 new Indel markers and 2000 F2 recessive individuals further localized this gene within 50 kb region delimited by Indel markers ha5 and ha6 ( Figure 5A).
According to the MSU (Michigan State University) Rice Genome Annotation Release 7 (http://rice.plantbiology.msu.edu), a total of 5 predicted open reading frames (ORFs) were revealed within the 50 kb fine mapping interval, including 3 ORFs with known biochemical functions (ORF 2: phosphate-induced protein 1; ORF 4: OsCHR4/CHR729; ORF 5: peptide-Nasparagine amidase), 1 ORF encoding expressed hypothetical protein (ORF 3) and 1 transposon (ORF 1) ( Figure 5B). Then genomic DNA of all these five ORFs in the Oschr4-5 mutant was sequenced and compared with that . Significance of data is tested by student's t test, ** p < 0.01. (E) Water loss rate of detached leaf blades between wild type and the Oschr4-5 mutant. Data are the means ± standard errors (n = 3).

Characterization of the Molecular Lesions in the Oschr4 Mutant
To find out the molecular lesion responsible for the pleiotropic phenotypes of the Oschr4-5 mutant, three F 2 populations were generated (Oschr4-5 × 9311; Oschr4-5 × Nipponbare; Oschr4-5 × 02428). Within these F 2 populations, a mendelian segregation ratio of 3:1 (χ 2 0.05 < 3.84) was calculated, which indicated the mutant phenotype was controlled by a single recessive gene. Then a large F 2 population generated by Oschr4-5 × 02428 was used to map the causative locus. In primary mapping, 116 F 2 mutant individuals located this gene between two Indel markers ha1 and ha10 on the long arm of chromosome 7. Fine mapping using 7 new Indel markers and 2000 F 2 recessive individuals further localized this gene within 50 kb region delimited by Indel markers ha5 and ha6 ( Figure 5A).
A single-base deletion at position 3688 was found in the ORF4 in Oschr4-5 (created through gamma ray radiation), whereas no mutation was found in other four ORFs. Sequence analysis of other two allelic mutants Oschr4-6 and Oschr4-7 created through EMS mutagenesis revealed a C4382T and G5190A substitution in ORF4, respectively ( Figure 5C). Therefore, ORF4 encoding OsCHR4/CHR729 was found to be the target gene, which is quite large with 11 exons and 10 introns. of WT. A single-base deletion at position 3688 was found in the ORF4 in Oschr4-5 (created through gamma ray radiation), whereas no mutation was found in other four ORFs. Sequence analysis of other two allelic mutants Oschr4-6 and Oschr4-7 created through EMS mutagenesis revealed a C4382T and G5190A substitution in ORF4, respectively ( Figure 5C). Therefore, ORF4 encoding OsCHR4/CHR729 was found to be the target gene, which is quite large with 11 exons and 10 introns.

Alignment and Phylogenetic Analysis
OsCHR4 encodes a member of the CHD3 subfamily of chromatin remodeling factors [36,37]. OsCHR4 has 2259 amino acids in length with several putative domains, including one PHD domain, two chromodomains, one SNF2 related helicase/ATPase domain, one Helicase C and one DNA-binding domain from amino terminus to carboxyl terminus ( Figure 5D). In the Oschr4-5 mutant, the single-base deletion results in a frame-shift mutation from the second chromodomain. In Oschr4-6, the amino acid substitution L830F (CTC to TTC) occurred within the SNF2 related helicase/ATPase domain. In Oschr4-7, the amino acid substitution R1099K (AGA to AAA) occurred within the Helicase C domain ( Figure 5D).
Multiple sequence alignment of the OsCHR4 homologous proteins in eight species was performed. As shown in Figure S2A, the substitution mutation sites in Oschr4-6 and Oschr4-7 were highly conserved in these species, which highlights the important role of these residuals in CHR4. Then the evolutionary relationship among CHD families in rice and Arabidopsis was evaluated. As a result, CHR705 was revealed to belong to the member of subfamily I (CHD1), and the other CHR proteins belonged to subfamily II (CHD3). There were no subfamily III members found in rice or Arabidopsis. Meanwhile, PICKLE in Arabidopsis was found to be close to CHR702 in rice, and the closest gene with OsCHR4 in Arabidopsis was CHR4 (PKR1) ( Figure S2B).

Alignment and Phylogenetic Analysis
OsCHR4 encodes a member of the CHD3 subfamily of chromatin remodeling factors [36,37]. OsCHR4 has 2259 amino acids in length with several putative domains, including one PHD domain, two chromodomains, one SNF2 related helicase/ATPase domain, one Helicase C and one DNA-binding domain from amino terminus to carboxyl terminus ( Figure 5D). In the Oschr4-5 mutant, the single-base deletion results in a frame-shift mutation from the second chromodomain. In Oschr4-6, the amino acid substitution L830F (CTC to TTC) occurred within the SNF2 related helicase/ATPase domain. In Oschr4-7, the amino acid substitution R1099K (AGA to AAA) occurred within the Helicase C domain ( Figure 5D).
Multiple sequence alignment of the OsCHR4 homologous proteins in eight species was performed. As shown in Figure S2A, the substitution mutation sites in Oschr4-6 and Oschr4-7 were highly conserved in these species, which highlights the important role of these residuals in CHR4. Then the evolutionary relationship among CHD families in rice and Arabidopsis was evaluated. As a result, CHR705 was revealed to belong to the member of subfamily I (CHD1), and the other CHR proteins belonged to subfamily II (CHD3). There were no subfamily III members found in rice or Arabidopsis. Meanwhile, PICKLE in Arabidopsis was found to be close to CHR702 in rice, and the closest gene with OsCHR4 in Arabidopsis was CHR4 (PKR1) ( Figure S2B).

Expression of Several Wax and Auxin Biosynthesis Related Genes Were Upregulated in Oschr4-5
As a chromatin remodeling factor, the loss of OsCHR4 function was identified to influence the rice plant growth and development in various ways [36][37][38][39][40], but the underlying molecular mechanisms need to be further analyzed. To understand why the content of cuticular wax increased in the Oschr4-5 mutant, we examined the expression levels of 16 genes encoding enzymes involved in wax biosynthesis and three genes (ROC4, WR1, WR2) encoding transcription factors that activate the expression of wax biosynthetic genes [26][27][28]44,45]. Quantitative real-time polymerase chain reaction (RT-PCR) using RNA samples prepared from leaves of 30-day-old plants revealed that transcript levels of six wax biosynthetic genes (GL1-4, WSL4, OsCER7, LACS2, LACS7, BDG) and one transcription factor gene (ROC4) [26][27][28]44,45] were significantly upregulated in Oschr4-5, compared with WT ( Figure 6A). This result suggests that loss of function of CHR4 affects the expression of important genes involved in the wax biosynthesis. mechanisms need to be further analyzed. To understand why the content of cuticular wax increased in the Oschr4-5 mutant, we examined the expression levels of 16 genes encoding enzymes involved in wax biosynthesis and three genes (ROC4, WR1, WR2) encoding transcription factors that activate the expression of wax biosynthetic genes [26][27][28]44,45]. Quantitative real-time polymerase chain reaction (RT-PCR) using RNA samples prepared from leaves of 30-day-old plants revealed that transcript levels of six wax biosynthetic genes (GL1-4, WSL4, OsCER7, LACS2, LACS7, BDG) and one transcription factor gene (ROC4) [26][27][28]44,45] were significantly upregulated in Oschr4-5, compared with WT ( Figure 6A). This result suggests that loss of function of CHR4 affects the expression of important genes involved in the wax biosynthesis.
The YUCCA gene, which encodes a flavin monooxygenase-like enzyme, is known to be involved in Trp-dependent auxin biosynthesis [46]. The TAA/TAR genes, which encode putative Trp aminotransferases, have been proposed to act in the same pathway with YUCCA genes [47]. To explain the increased IAA level in the Oschr4-5 mutant, we measured the expression levels of 13 genes from the YUCCA and TAA/TAR family between three-day-old seedlings of WT and Oschr4-5. The transcript levels of four YUCCA related genes including YUC2, YUC3, YUC5 and YUC6 were upregulated significantly in the Oschr4-5 mutants ( Figure 7A). However, the other 9 genes showed no significant alteration in the Oschr4-5 mutants. These results suggested that OsCHR4 may influence the expression of several wax-related and IAA-related genes, and the increased gene expressions in the Oschr4-5 mutants conduces to higher wax and IAA content in Oschr4-5 than WT. Significance of data is tested by student's t test. * p < 0.05, ** p < 0.01. Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test. * p < 0.05, ** p < 0.01. (B,C) ChIP-qPCR analysis for H3K4me3 (B) and H3K27me3 (C) levels on 7 up-regulated wax genes in Oschr4-5 relative to the WT. Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test. * p < 0.05, ** p < 0.01.
The YUCCA gene, which encodes a flavin monooxygenase-like enzyme, is known to be involved in Trp-dependent auxin biosynthesis [46]. The TAA/TAR genes, which encode putative Trp aminotransferases, have been proposed to act in the same pathway with YUCCA genes [47]. To explain the increased IAA level in the Oschr4-5 mutant, we measured the expression levels of 13 genes from the YUCCA and TAA/TAR family between three-day-old seedlings of WT and Oschr4-5. The transcript levels of four YUCCA related genes including YUC2, YUC3, YUC5 and YUC6 were upregulated significantly in the Oschr4-5 mutants ( Figure 7A). However, the other 9 genes showed no significant alteration in the Oschr4-5 mutants. These results suggested that OsCHR4 may influence the expression of several wax-related and IAA-related genes, and the increased gene expressions in the Oschr4-5 mutants conduces to higher wax and IAA content in Oschr4-5 than WT. Significance of data is tested by student's t test. ** p < 0.01, *** p < 0.001. (B,C) ChIP-qPCR analysis showed variations of H3K4me3 (B) and H3K27me3 (C) levels on 4 up-regulated IAA genes in Oschr4-5 compared to WT. Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test. * p < 0.05,*** p < 0.001.

Histone Modifications of the Wax and Auxin Biosynthesis-Related Genes Were Affected in Oschr4-5
Histone modification is recognized as an important mechanism in the epigenetic regulation of gene expression. For example, trimethylation of histone H3 lysine 4 (H3K4me3) is generally correlated with gene activation, whereas trimethylation of histone H3 lysine 27 (H3K27me3) is generally correlated with gene repression [48,49]. As reported, the rice CHD3 protein OsCHR4/CHR729 is a chromatin remodeler and can modulate H3K4me3 and H3K27me3 level to regulate the genes expression [35]. In order to test whether increased expression of wax and auxin biosynthesis related genes in Oschr4-5 could be attributed to altered histone modification, a ChIP analysis with antibody recognizing H3K4me3 and H3K27me3 was performed. Of the seven wax biosynthetic genes, three genes (WSL4, CER7, BDG) showed increased H3K4me3 and decreased H3K27me3 simultaneously, four genes showed either increased H3K4me3 (GL1-4, LACS2) or decreased H3K27me3 (LACS7, ROC4) ( Figure 6B,C). Similarly, of the four auxin biosynthetic genes, one gene (YUC3) showed increased H3K4me3 and decreased H3K27me3 simultaneously, three genes showed either increased H3K4me3 (YUC2) or decreased H3K27me3 (YUC5, YUC6) ( Figure  7B,C). LACS1 and TAR4, two control genes which were not differentially expressed between Oschr4-5 and WT, did not show significant changes in level of H3K4me3 and H3K27me3 (Figure 6;  Figure 7). Taken together, OsCHR4 affects modifications of histone protein associated with wax and auxin biosynthesis related genes and regulates their expression epigenetically. Significance of data is tested by student's t test. ** p < 0.01, *** p < 0.001. (B,C) ChIP-qPCR analysis showed variations of H3K4me3 (B) and H3K27me3 (C) levels on 4 up-regulated IAA genes in Oschr4-5 compared to WT. Data are the means ± standard errors (n = 3). Significance of data is tested by student's t test. * p < 0.05,*** p < 0.001.

Histone Modifications of the Wax and Auxin Biosynthesis-Related Genes Were Affected in Oschr4-5
Histone modification is recognized as an important mechanism in the epigenetic regulation of gene expression. For example, trimethylation of histone H3 lysine 4 (H3K4me3) is generally correlated with gene activation, whereas trimethylation of histone H3 lysine 27 (H3K27me3) is generally correlated with gene repression [48,49]. As reported, the rice CHD3 protein OsCHR4/CHR729 is a chromatin remodeler and can modulate H3K4me3 and H3K27me3 level to regulate the genes expression [35]. In order to test whether increased expression of wax and auxin biosynthesis related genes in Oschr4-5 could be attributed to altered histone modification, a ChIP analysis with antibody recognizing H3K4me3 and H3K27me3 was performed. Of the seven wax biosynthetic genes, three genes (WSL4, CER7, BDG) showed increased H3K4me3 and decreased H3K27me3 simultaneously, four genes showed either increased H3K4me3 (GL1-4, LACS2) or decreased H3K27me3 (LACS7, ROC4) ( Figure 6B,C). Similarly, of the four auxin biosynthetic genes, one gene (YUC3) showed increased H3K4me3 and decreased H3K27me3 simultaneously, three genes showed either increased H3K4me3 (YUC2) or decreased H3K27me3 (YUC5, YUC6) ( Figure 7B,C). LACS1 and TAR4, two control genes which were not differentially expressed between Oschr4-5 and WT, did not show significant changes in level of H3K4me3 and H3K27me3 (Figure 6; Figure 7). Taken together, OsCHR4 affects modifications of histone protein associated with wax and auxin biosynthesis related genes and regulates their expression epigenetically.
The closest homolog gene with OsCHR4 in Arabidopsis is considered to be AtCHR4 (PICKLE RELATED1, PKR1) ( Figure S2B), while their functions in plants seem to be different. Homozygous mutants of neither PICKLE homologs PKR1 nor PKR2 exhibited significant phenotypic differences to WT plants under standard growth conditions [51]. However, OsCHR4 was revealed to be closely related with the development of rice leaves in this study. It is noteworthy that the function of OsCHR4 is more similar to CHD3-related PICKLE. PICKLE was initially isolated from the Arabidopsis pickle mutant characterized by green tuberous root phenotype [52]. It has been reported that PICKLE functions in repression of ectopic stipules and meristems in leaf tissue and represses meristematic genes in carpel tissue [51,53]. PICKLE is not only involved in organ or cell polarity, but also plays an important role in cell differentiation [52]. Loss-of-function mutations in PICKLE could always influence cell transition from embryonic to vegetative development, thereby inducing pleiotropic phenotypes in the root, leaves and even the whole plants [54,55]. These phenotypes were just consistent with those of Oschr4-5 isolated in this study and indicated a similar function between OsCHR4 and PICKLE. However, T-DNA insertion mutation and RNAi of the closest homolog gene with PICKLE in rice (OsCHR702) did not produce any morphological phenotype [36]. Therefore, we suspect that the function of CHR proteins may be diversified between Arabidopsis and rice due to specific species difference.
Cuticular waxes are complex mixtures of very-long-chain fatty acids (VLCFAs) and their derivatives and provide effective protection for plants to confront various environmental stresses [28]. In this study, we discovered that OsCHR4 negatively regulates the wax biosynthesis and more cuticle wax accumulated on the adaxial side of leaves in Oschr4-5 mutant compared to the WT. Previous reports demonstrate that excessive deposition of cuticular wax can decrease nonstomatal water loss in plants [24,56]. We also proved that increased wax content in Oschr4-5 results in reduced water loss and improved drought tolerance. As a result of increased wax deposition in Oschr4-5, the mutant Oschr4-5 exhibits less chlorophyll leaching rate compared with WT. Furthermore, increased transcript levels of genes involved in wax biosynthesis in Oschr4-5 were also in support of the regulation of OsCHR4 in wax synthesis. Among the seven up-regulated genes in Oschr4-5 ( Figure 6A), WSL4 and its homolog gene OsCER7 encode β-Ketoacyl-coenzyme A synthase (KCS) which is the key enzyme in the first step of VLCFA elongation [45,57]. OsGL1-4 encodes the fatty aldehyde decarbonylase, which has been identified to positively regulate various steps of wax biosynthesis [21][22][23][24][25]. LACS2 and LACS7 encode long-chain acyl-CoA synthase that preferentially modifies VLCFAs for wax biosynthesis [58]. BDG (BODYGUARD) encodes an α/β-hydrolase fold protein important for cuticle development [28]. ROC4 encodes a HD-ZIP IV transcription factor which positively regulates wax deposition in rice by activating wax biosynthetic gene such as BDG [28]. This result suggests multiple steps of wax biosynthesis pathway are affected by the loss of function of CHR4 and is consistent with the quantification data that many wax components were significantly increased in Oschr4-5 ( Figure 3C).
It has been demonstrated in previous research that OsCHR4 is essential for various aspects of plant development via influencing the gibberellin pathway and ROS homeostasis [38,40]. In this study, we found that the content of IAA in Oschr4-5 was significantly higher than that of WT. As reported, many narrow leaves genes were interrelated with IAA. Auxin overproduction is well known to lead to narrow and epinastic leaves. Ectopic overexpression of either YUCCA genes (encoding flavin monooxygenases) in Arabidopsis [46] or the Agrobacterium auxin biosynthetic gene iaaM (encoding tryptophan monooxygenase) and petunia FLOOZY gene (ortholog of YUCCA) in petunia resulted in narrow and rolled leaves with high levels of IAA [42,43]. In Arabidopsis, narrow and rolled leaves were also caused by elevated IAA levels by overexpressing cytochrome P450 CYP79B2 which converts tryptophan (Trp) to indole-3-acetaldoxime [59]. A similar case was reported in rice, where OsYUCCA8/NAL7 also affects leaf width via altered IAA content [6]. In this study, increased IAA content was also detected in the narrow leaf mutant Oschr4-5. Auxin is principally converted from tryptophan via the TAA family of aminotransfrases and the YUCCA (YUC) family of flavin-containing monooxygenases [60,61]. For further verification of the altered IAA concentration in Oschr4-5, we detected the expression of TAA and YUC family genes and found that expression levels of four YUC genes increased in Oschr4-5 mutants compared with WT. The finding is consistent with increased IAA concentration in Oschr4-5, which strengthens the evidence that OsCHR4 might regulate auxin biosynthetic process in rice. Interestingly, 31 OsAux/IAA genes involved in the auxin signaling pathway was significantly downregulated in crl6, an allelic mutant of Oschr4-5 [39]. The decreased transcript level of OsIAA genes might be due to higher IAA content as revealed in this study. As early-response genes and negative regulators of auxin signaling, Aux/IAA transcripts tend to be rapidly induced and then repressed by elevated IAA content [62].
Histone modification was an important mechanism in regulation of DNA transcription and other activities. As reported, trimethylated histone H3 lysine 27 (H3K27me3) was associated with gene silencing and repression during plant development, while trimethylated histone H3 lysine 4 (H3K4me3) is generally correlated with gene activation [63,64]. In the chr729 mutant, 56% and 23% of marked genes lost H3K27me3 and H3K4me3, while 754 and 724 genes gained ectopic H3K27me3 and H3K4me3 [36]. Meanwhile, PICKLE was revealed to be able to reduce the level of H3K27me3 on root stem cells, thereby promoting the expression of meristem marker genes [65]. To evaluate the histone modification states in the Oschr4-5 mutant, ChIP was performed on several differently expressed genes involved in wax and auxin biosynthesis in this study. As a result, the increased level of H3K4me3 and/or reduced level of H3K27me3 in the Oschr4-5 mutant were significantly correlated with increased expression of genes related to wax or IAA biosynthesis. Consistent with the genome-wide ChIP-Seq analysis showing that 724 genes were detected to have gained ectopic H3K4me3 in the chr729 mutant [36], H3K4me3 was also increased in several auxin and wax biosynthesis genes in the Oschr4-5 mutant in our study. The underlying mechanism for the increased H3K4me3 at some specific loci still remains elusive [36]. As a bifunctional chromatin regulator that recognizes and modulates both H3K4 and H3K27 methylation, the function of OsCHR4 might be determined by sequence-specific transcription factors that associates with their targets and/or by the recruitment of different histone modification enzymes that catalyze methylation or demethylation.
In conclusion, our results suggest that OsCHR4 not only regulates the accumulation of cuticle wax on leaf surface, but also plays an essential role in the modulation of IAA biosynthesis. Expression levels of several genes involved in the two pathways were increased accompanied with up-regulated H3K4me3 or down-regulated H3K27me3. Epigenetic regulation by OsCHR4 is required for multiple aspects of plant development.

Plant Materials and Growth Condition
The rice mutant Oschr4-5 (requested from Dale Bumpers National Rice Research Center, Stuttgart, AR, USA) was isolated from the indica cultivar Indica 9 treated by gamma ray radiation. The other two allelic mutants Oschr4-6 and Oschr4-7 were isolated from the japonica cultivar Nipponbare treated with EMS in the Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing, China. All rice plants were cultivated in the experimental field of Shandong Rice Research Institute (Jining, China) under natural growing conditions, and the climatic conditions during the rice growing season were listed in Table S2. The mutant phenotype inherited stably after continuous cultivation. Field data of the mutant and WT plants, including plant height, tiller number, leaf width and leaf length were measured at the heading stage. The transpiration rate of plants was detected in field using LI-6400XT Portable Photosynthesis System.

Microscopic Observation
At the heading stage, the top first leaf of the mutant and WT were cut into pieces (2 cm) in the widest part and divided into three groups. One group of samples were used for microscopic observation of leaf vein and transverse cross section. Another group of samples were air-pumped in 2% cellulase R-10 solution for 30 min and digested for at least 2 days. After washing with water, mesophyll cells were peeled and the abaxial epidermis was observed under microscope. The cell width was measured using the Image J software (https://imagej.nih.gov/ij/). Stomatal density (mm −2 ) was counted between two lateral veins in the center region of five individual leaves. Finally, the last group of samples were fixed in fixative solution (75% ethanol, 5% acetic acid, 5% glycerol, 5% formaldehyde and 10% deionized water) for at least one day, and then critical point dried, sputter-coated with gold and observed with a scanning electron microscope (UANTA 200).

Measurement of Chlorophyll and Indole Acetic Acid (IAA)
The top first leaf of plants at heading stage was used for chlorophyll measurement. Briefly, 0.05 g leaf samples were cut into small pieces and incubated in 80% acetone overnight in dark. Then the chlorophyll contents were measured and calculated by absorption spectrophotometry. For the measurement of IAA, 0.2 g freeze dried sample of the whole shoots of 3-day-old seedlings was harvested for each replicate. Free IAA extraction and measurement was performed as described previously [66].

Measurement of Cuticular Waxes
The content of cuticular waxes in the leaf of mutant and WT was measured as described previously [24]. Briefly, cuticular wax samples were first extracted from the top first leaf of plants at the heading stage by immersion in chloroform (60 • C, 30 s). A total of 20 µg n-tetracosane (C24, SUPELCO, Sigma, Saint Louis, USA) was added to these samples as an internal standard. Then the solutions were soaked with 100 mm 3 bis-N,N-(trimethylsilyl) trifluoroacetamide (BSTFA, SUPELCO, Sigma) and 100 mm 3 pyridine for 1 h at 70 • C. After evaporation and filtration (WHATMAN, PTFE, 13 mm × 0.22 µm), the compositions of cuticular waxes were analyzed using a capillary gas chromatograph equipped with an HP-1MS column (30 m length, inner diameter 0.32 mm, film thickness 0.25 µm) attached to a mass spectrometer (GCMS-QP2010, Kyoto, Japan). The GC-MS protocol included injection at 250, 50 • C for 2 min, ramped to 200 at 20 • C min −1 , 2 min at 200 • C, ramped to 320 at 2 • C min −1 , 14 min at 320 • C with supplied at 1.2 cm 3 min −1 as the carrier gas. A flame ionization detector was used for quantitative analyses, and the quantification of each component was calculated by the equivalent ratio of mass to peak area between the component and internal standard.

Water Loss and Chlorophyll-Leaching Assays
Leaf water loss rate was measured using detached leaves of the same parts from 6 week-old seedlings cultured in climate chambers. Each leaf was kept in a box without lid, and weighed every 15 min [28]. The water loss rate was calculated by dividing the lost weight of leaf at different time points by the initial weight. In the assay of chlorophyll leaching, the leaves (0.25 g) were cut into 2 cm lengths and immersed in 25 cm 3 80% ethanol. 1 cm 3 chlorophyll leaching solution was taken for spectrophotometry at wave lengths of 664 and 647 nm every 30 min. 1 cm 3 80% ethanol was used as control. The solution should be collected and poured back in the same tube after each measurement. The chlorophyll concentration was calculated by the formula of 7.939 A664 × 19.539 A647 [24]. The chlorophyll-leaching rate was given by dividing the concentration monitored at different time points by the chlorophyll concentration measured after 24 h of immersion. The experiment comprised three replicates.

Mapping and Sequencing
An F 2 segregating population generated from a cross between Oschr4-5 and 02428 (a polymorphic japonica cultivar) was used for mapping and cloning of the target gene. For the primary mapping, a DNA pool composed of DNA from 10 individual mutant plants was used for bulked segregation analysis (BSA) [67]. Approximately 200 polymorphic indel markers distributed evenly throughout the rice genome were used for the BSA analysis. For fine mapping, new indel markers were designed by utilizing genomic sequence information from Indica and Japonica (http://rgp.dna.affrc.go.jp/) (specific primers were shown in Table S3). DNA was extracted from fresh leaves of each plant. PCR program included 5 min at 95 • C followed by 35 cycles of 30 s at 94 • C, 30 s at 55 • C, 30 s at 72 • C, and a final extension of 10 min at 72 • C. PCR products were analyzed on a polyacrylamide gel stained with silver nitrate. The genomic region of the candidate gene was divided into several overlapping 1.5-kb fragments, and amplified using high-fidelity PrimeSTAR DNA polymerase (TaKaRa Bio, DaLian, China). The PCR program included 3 min at 94 • C followed by 40 cycles at 98 • C for 30 s, 55 • C for 5 s, 72 • C for 1.5 min, and a final extension at 72 • C for 10 min. PCR products were sequenced directly.

Real-Time Polymerase Chain Reaction (RT-PCR)
Total RNA was first extracted from rice plants with Trizol method, and then first-strand cDNA was synthesized using a Superscript III Reverse Transcription Kit (Invitrogen, Carlsbad, CA, USA). Semiquantitative PCR was performed using LA Taq DNA polymerase (TaKaRa) with the rice ACTIN1 gene serving as an internal control. The PCR program included 1 min at 94 • C followed by 30 cycles at 94 • C for 30 s, 60 • C for 30 s, 72 • C for 30 s, and a final extension at 72 • C for 10 min. Quantitative PCR was performed using a SYBR Premix Ex Taq2 kit (TaKaRa) on ABI PRISM 7900HT under the following conditions: 10 s denaturing at 95 • C, 30 s annealing at 60 • C, 40 cycles. The mRNA amount relative to ACTIN1 was finally calculated. Specific primers were shown in Table S4.

Sequence and Phylogenetic Analyses
Gene prediction was performed using online search software (http://rice.plantbiology.msu.edu/ cgi-bin/gbrowse/rice). Exon/intron structures were identified by alignment of coding sequences (CDS) and genomic DNA sequences. Multiple sequence alignments were conducted using CLUSTALX software, and a phylogenetic tree was built using MEGA 4 software [68].

Chromatin Immunoprecipitation (ChIP) Analysis
The ChIP experiment was performed based on a previously published protocol [69]. Simply, 0.5 g of 3-day-old or four-leaf-old seedlings of Oschr4-5 and WT were first harvested and crosslinked in 1% formaldehyde under vacuum for 15 min or 30 min, respectively. Then the isolated chromatin complex was fragmented to 200-500 bp by sonication, and 1% of the fragmented chromatin from each tube was kept as input DNA. The chromatin modification states of several target genes were analyzed using antibodies recognizing H3K4me3 (ab8580, Abcam, Cambridge, UK) and H3K27me3 (ab6002, Abcam) and the pre-immune serum (02-6502, Invitrogen) as a negative control. Then the immune complexes were collected with Protein A Agarose beads (P3476, Sigma). After reversing the crosslink, the precipitated and input DNAs were detected by real-time PCR, and the enrichment of H3K4me3 and H3K27me3 of each genes was quantified by normalizing the threshold cycle (Ct) of the ChIP sample with that of the input with 2 (Ct of input − Ct of sample ChIP) . Specific primers used in ChIP-PCR are shown in Table S4.

Accession Number
Sequence data from this article can be found in the GenBank databases under the following accession numbers: OsCHR4 genomic DNA, MK765112; OsCHR4 cDNA, MK765113.