The MYB36 transcription factor orchestrates Casparian strip formation

Significance

Casparian strips play a critical role in sealing endodermal cells in the root to block uncontrolled extracellular uptake of nutrients and water. Building Casparian strips requires the construction of extracellular lignin structures that encircle cells within the cell wall and that are anchored to the plasma membranes of adjacent cells to form tight seals between them. The transcription factor we have discovered, and the set of genes it regulates, now provides us with the detailed “parts list” necessary to build Casparian strips. This finding has clear implications for better understanding the nature of tight cellular junctions in biology and also has practical implications of agricultural, offering the potential for improved water and nutrient use efficiencies and enhanced resistance to abiotic stresses.

Abstract

The endodermis in roots acts as a selectivity filter for nutrient and water transport essential for growth and development. This selectivity is enabled by the formation of lignin-based Casparian strips. Casparian strip formation is initiated by the localization of the Casparian strip domain proteins (CASPs) in the plasma membrane, at the site where the Casparian strip will form. Localized CASPs recruit Peroxidase 64 (PER64), a Respiratory Burst Oxidase Homolog F, and Enhanced Suberin 1 (ESB1), a dirigent-like protein, to assemble the lignin polymerization machinery. However, the factors that control both expression of the genes encoding this biosynthetic machinery and its localization to the Casparian strip formation site remain unknown. Here, we identify the transcription factor, MYB36, essential for Casparian strip formation. MYB36 directly and positively regulates the expression of the Casparian strip genes CASP1, PER64, and ESB1. Casparian strips are absent in plants lacking a functional MYB36 and are replaced by ectopic lignin-like material in the corners of endodermal cells. The barrier function of Casparian strips in these plants is also disrupted. Significantly, ectopic expression of MYB36 in the cortex is sufficient to reprogram these cells to start expressing CASP1–GFP, correctly localize the CASP1–GFP protein to form a Casparian strip domain, and deposit a Casparian strip-like structure in the cell wall at this location. These results demonstrate that MYB36 is controlling expression of the machinery required to locally polymerize lignin in a fine band in the cell wall for the formation of the Casparian strip.

Plant roots are able to selectively take up both essential nutrients and water from the soil. This selectivity is provided by the endodermis, the innermost cell layer encircling the vascular system. However, to perform this function, it is critical that the Casparian strip that encircles endodermal cells works to block extracellular diffusion. Because of the vital importance of these structures for endodermal function, Casparian strips are one of the primary features of endodermal differentiation.

Casparian strips are made of a lignin polymer that is deposited as a fine band in the anticlinal cell wall, encircling endodermal cells to seal the cell wall space between them (1). This precisely situated lignin polymerization is mediated through the oxidation of monolignols by localized Peroxidase 64 (PER64) and a Respiratory Burst Oxidase Homolog F (2). This biosynthetic machinery is placed at the Casparian strip deposition site by association with Casparian strip domain proteins (CASPs). CASPs are specifically expressed in the endodermis and localize in the plasma membrane in a region in the middle of the anticlinal endodermal cell wall (3), guiding where the Casparian strip forms. Enhanced Suberin 1 (ESB1) also localizes to the Casparian strip domain, where it is required for the correct deposition of lignin and stabilization of CASPs (4). Expression of these Casparian strip-associated genes—the toolkit for the formation of Casparian strips—is regulated in both time and space during root development and marks the differentiation of the endodermis.

Here, we present our discovery of the transcriptional regulator MYB36 that orchestrates the developmentally and spatially coordinated expression of the genes necessary to position and build Casparian strips in the root endodermis. Strikingly, ectopic expression of MYB36 is sufficient to reprogram cells to both express the genetic machinery required to synthesize Casparian strips and to locate and assemble this machinery, such that the strips develop in the correct cellular location, even though they are in cell types that do not normally form Casparian strips.

Results and Discussion

Through two different forward genetic screens using Arabidopsis thaliana, we isolated plants that have mutations in MYB36. First, in a screen of fast neutron mutagenized plants to identify genes involved in mineral nutrient and trace element homeostasis (i.e., the ionome), we identified mutant 11250 (5), now termed myb36-1. This mutant has multiple changes to its leaf ionome, including elevated concentrations of sodium, magnesium, and zinc and decreased calcium, manganese, and iron (Fig. S1A). The myb36-1 leaf ionome is also similar to the other known Casparian strip mutants, esb1-1 and casp1;casp3 (4), illustrated here by using principal component analyses to display the full multielement ionomic phenotypes (Fig. 1A). To determine which tissues (root or shoot) are responsible for the observed alterations in the leaf ionome in myb36-1, we performed reciprocal grafting experiments. Plants were grafted at the 5-d-old seedling stage and allowed to grow for 5 wk before the leaf ionome was measured. The leaves of grafted plants with wild-type shoots and myb36-1 roots had a similar ionome to that of both self-grafted and nongrafted myb36-1 plants. However, leaves from grafted plants with myb36-1 shoots and wild-type roots had ionomes that were indistinguishable from self-grafted or nongrafted wild-type plants (Fig. S1B). These results show that the leaf ionomic phenotype of myb36-1 is caused by a defective root function. Second, we performed an independent screen to identify genes involved in the formation of the Casparian strip. We screened ethyl methanesulfonate-mutagenized plants for individuals with no visible accumulation of CASP1–GFP when CASP1–GFP was expressed from the native CASP1 promoter. Using this screen, we isolated the mutants myb36-3 and -4 (Fig. 1B). We observed that these myb36 mutants have longer root hairs than wild-type (Fig. 1B), but exhibited no other obvious visible phenotypes.

Fig. S1.

The leaf ionomic phenotype and the causal gene of the mutant. (A) Leaf elemental concentrations of soil-grown plants. Leaves of 5-wk-old plants were harvested and subjected to ICP-MS analysis. Raw data are available from the iHUB (http://www.ionomicshub.org/home/PiiMS) for tray 1,975. Two-tailed Student’s t test was performed with Benjamini and Hochberg’s FDR correction for multiple testing (n = 12). (B) Leaf ionomic phenotype of myb36-1 is determined by the roots. Reciprocal grafting was performed between Col-0 and myb36-1. Grafted seedlings were transferred to soil. After 4 wk, leaves were harvested for ionomic analysis. (C) Top 10 down-regulated genes in myb36-1. Microarray analysis was performed by using roots from 5-wk-old soil-grown myb36-1 plants. Within the region identified by linkage mapping (Chr. 5, 22.359178–23.645590 Mb) At5g57620 (MYB36) was the only gene with reduced expression compared with Col-0. (D) qPCR analysis of MYB36 in myb36-1 and -2. The plants were grown on agar-solidified medium for 2 wk (n = 4). Bars indicate mean ± SD. (E) Leaf ionome of myb36-1 and -2. Principal component analysis was performed based on 20 elements quantified in leaves from 2-wk-old plants grown on agar-solidified medium (n = 15 in Col-0; n = 18 in myb36-1; n = 17 in myb36-2). (F) Allelism test between myb36-1 and -2. Principal component analysis was performed based on 20 elements quantified in leaves of 5-wk-old plants grown on soil (n = 12 in Col-0, myb36-1, and -2; n = 6 in F1 cross between myb36-1 and -2).

Fig. 1.

Disruption of MYB36 alters the leaf ionome and CASP1 expression. (A) Principal component analysis based on the concentration of 20 elements in shoots (n = 15). (B) Mutants identified by measuring level of accumulation of CASP1–GFP expressed from the CASP1 native promoter. (C) Mutation sites in the myb36 mutants. The myb36-1 mutant may have a large change in the promoter region because we were unable to amplify this region; myb36-2 contains a T-DNA insertion in the second exon; myb36-3 contains an Arg-to-Trp substitution in the second MYB domain repeat, and myb36-4 contains a splice-site mutation after the second intron. (D–G) Endodermal localization of MYB36 using a MYB36–GFP fusion protein expressed from the MYB36 native promoter. Magenta [propidium iodide (PI)], cell wall. E and G were taken with the same microscope settings. The dotted box in E is shown in F with the brightness of the GFP signal artificially enhanced. Arrowheads (E and F) point to examples of MYB36–GFP in the endodermal cell layer. [Scale bars: 100 µm (B and D); 50 µm (E and G).]

To identify the causal gene in myb36-1, we performed genetic mapping using bulk segregant analysis (BSA-seq). We analyzed the leaf ionome of several hundred individual F2 plants from a cross between myb36-1 in the Columbia-0 (Col-0) background and the Landsberg erecta accession. Based on the ionomic phenotype of these F2 plants, the mutant locus was determined to be recessive. We generated two pools of plants, each containing 28 individuals with either wild-type or mutant phenotypes. The ionomic phenotype of these 56 F2 plants was confirmed in the F3 generation. DNA from these two pools was extracted and sequenced on an Applied Biosystems SOLiD next-generation DNA sequencer. Short-read sequence data were aligned to the Col-0 reference genome sequence, and analysis of genome-wide heterozygosity identified a region of the genome enriched in Col-0 genotypes, which placed the causal mutation within a 22.4- to 23.6-Mb interval on chromosome V. Because the myb36-1 mutant was generated by fast-neutron mutagenesis, which is known to cause deletion that can alter gene expression, we reasoned that genes with altered expression within our 1-Mb mapping interval would be good candidate for the causal gene in myb36-1. We therefore performed a microarray analysis (Affymetrix ATH1 array) to assess expression in roots of genes in our BSA-seq–mapping interval. MYB36 (At5g57620), encoding a transcription factor, was the only gene in this interval with lower expression levels than wild-type and which is normally highly expressed in roots (Fig. S1 C and D). We were unable to identify any mutations in MYB36 between the start codon and the next gene downstream (At5g57625). However, we were also unable to amplify the promoter region of MYB36 in myb36-1 using several different sets of primers, suggesting the existence of a large rearrangement in the genome in this promoter region. To confirm MYB36 as the causal gene in myb36-1 we obtained a T-DNA insertional allele (GK-543B11) of MYB36 (named myb36-2) (Fig. 1C). This T-DNA allele showed a similar leaf ionomic phenotype to myb36-1, and F1 plants from a cross between myb36-1 and -2 had the mutant phenotype, demonstrating that these two mutants are allelic (Fig. S1 E and F) and confirming MYB36 as the causal gene. The myb36-3 and -4 alleles were also crossed with myb36-1 and shown to be allelic as well, and DNA sequencing revealed mutations in MYB36 in both these alleles (Fig. 1C).

To identify the cell type in which the MYB36 protein is accumulated, a GFP fusion construct was introduced into myb36-1. GFP was fused to the C terminus of the MYB36 genomic sequence, which starts from 3,976 bp upstream of the start codon and extends to the end of the coding sequence. In these transgenic lines, GFP fluorescence was clearly visible in endodermal cells from the late elongation zones to the differentiation zone (Fig. 1 D and G). Much weaker fluorescence was also observed in endodermal cells of the meristematic zone (Fig. 1 E and F). The leaf ionomic phenotype of myb36-1 was partially rescued by this genomic sequence fused with GFP (Fig. S2 A and B). Further, the myb36-1 mutant also displayed decreased expression of genes known to be involved in Casparian strip development, including CASP1, PER64, and ESB1, along with defective Casparian strips and an enhanced leak into the stele of propidium iodide (PI) (Fig. S2 C–E). The MYB36 genomic sequence fused with GFP also partially rescued these phenotypes. This result suggests that the GFP signal observed in myb36-1 transformed with the MYB36–GFP construct likely reflects the endogenous localization of MYB36.

Fig. S2.

Complementation of myb36-1 with a MYB36 genome–GFP construct. Four independent homozygous lines were used for each analysis. (A) MYB36 genome–GFP complements the ionome phenotype of myb36-1. Leaves of 2-wk-old plants grown on agar-solidified medium were analyzed for their elemental content by using ICP-MS. Principal component analysis was performed based on the concentration of 20 elements. (B) As a representative element altered in myb36-1, calcium (Ca) concentration is shown. Different characters indicate significant differences by Tukey’s HSD (P < 0.05). Bars represent means ± SD (n = 5). (C) Expression of MYB36, CASP1, PER64, and ESB1 are shown as relative to Col-0. Roots of 2-wk-old plants grown on agar-solidified medium were used for the expression analysis (n = 4). Note that expression of MYB36 is shown with a log₁₀ value. Different characters indicate significant differences by Tukey’s HSD (P < 0.05). (D) Casparian strip deposition patterns were observed in the protoxylem and metaxylem regions by using autofluorescence. Spiral structures at the center of roots are xylem vessels. (E) PI penetration assay was performed as in Fig. 2D. Different characters indicate significant differences by Tukey’s HSD (P < 0.05) (n = 8). Bars indicate mean ± SD.

Existing evidence suggests that MYB36 expression is directly regulated by SCARECROW (SCR), as part of the differentiation program controlled by SHORT-ROOT (6–10). Considering the key role that Casparian strip development plays in marking endodermal differentiation, and the specific localization of MYB36–GFP to the endodermis (Fig. 1), it seemed plausible that MYB36 may be directly controlling Casparian strip formation. We used autofluorescence of the lignin within Casparian strips (1) to observe them in cleared roots of the myb36 mutants. Autofluorescence in myb36-1 was stronger than wild-type and more irregular in intensity than either wild-type or the known Casparian strip mutant esb1-1 (Fig. 2A). To further identify the lignin deposition site in myb36-1 we simultaneously visualized lignin and cell wall by treating cleared roots with PI, which stains lignin, and Calcofluor White, which stains cellulose. In myb36, lignin deposition was completely lacking at the endodermal cell–cell contact site, where it normally occurs to form the Casparian strip in wild-type (Fig. 2 B and C). Instead, lignin-like material accumulated exclusively in the cell corners of endodermal and cortical cells on the cortex side of the endodermis. These results indicate that MYB36 is essential for the correct localized lignin biosynthesis required to form Casparian strips. This finding contrasts with the esb1-1 mutant in which lignin deposition still occurs at the endodermal cell–cell contact site, but the development of a continuous central lignin ring is disrupted (4) (Fig. 2 B and C).

Fig. 2.

Loss of Casparian strip and disruption of the apoplastic barrier in myb36 mutants. (A) Z-stack confocal image of Casparian strip autofluorescence. Spiral structures in the center of the root are xylem. (B) Lignin (yellow) deposition site in longitudinal section. Boxed region are enlarged in Right. Cleared roots were stained with PI (yellow; lignin) and Calcofluor White (blue; cell wall). Although both of these dyes stain cell walls, PI primarily interacts with lignin and Calcofluor White with cellulose. Cor, cortex; End, endodermis; Epi, epidermis. (C) Schematic diagram of lignin deposition sites (magenta) in roots. Front and top views of roots are shown. (D) Casparian strip functionality was quantified by PI penetration. Asterisks in Left indicate the 15th endodermal cell from the onset of elongation. (E) Suberin accumulation detected with fluoral yellow 088. Left shows merged bright-field and fluorol yellow fluorescence (yellow; suberin) imaged around the 14th endodermal cell from the onset of elongation. The number of endodermal cells at which PI penetration into stele was blocked (D) or suberin accumulation first appeared (E) were counted from the onset of elongation. Different characters indicate significant differences by Tukey’s HSD (P < 0.05) (D, Right) and Steel–Dwass test (P < 0.05) (E, Right). Data represent means ± SD (n = 16 in Col-0, n = 8 in mutants). [Scale bars: 50 µm (A, D, and E); 25 µm (B).]

Using PI as an apoplastic tracer, we evaluated the presence of an apoplastic barrier in the myb36 mutants. To quantify this barrier function, we counted the number of endodermal cells from the onset of elongation to the point where PI fluorescence was no longer observed in the stele-facing cell wall of the endodermis. We found that blockage of PI penetration into the stele in the myb36 mutants was delayed compared with wild-type and was similar to the delay observed in esb1-1 (Fig. 2D). This result indicates that the loss of the centrally located Casparian strip in myb36 eliminates the apoplastic barrier in that region of the root. Furthermore, the ectopic lignin-like material deposited in the corners of myb36 endodermal cells is not able to form an effective barrier to apoplastic transport. However, the diffusional barrier in myb36 is recovered in the more mature region of the root, where suberin is normally deposited in wild-type (1).

Similar to esb1-1 and casp1;casp3 (4), the myb36 mutants also showed early accumulation of suberin in the endodermis between the plasma membrane and the cell wall (Fig. 2E). Interestingly, this early accumulation of suberin was not observed in mutants made between esb1-1 or casp1;casp3 and schengen3 (sgn3), suggesting that SGN3, which encodes a leucine-rich receptor like kinase, may mediate this suberin accumulation (11). By reducing transmembrane transport and enhancing apoplastic diffusion across the endodermis, the early suberin accumulation and defective Casparian strips of myb36 could be responsible for the altered leaf ionome of this mutant (Fig. S1A). Suberin deposition would be expected to reduce movement of ions across the endodermal plasma membrane, but to not affect the ions that move symplastically via plasmodesmata. Further, depending on the concentration gradient for a particular ion across the endodermis, loss of functional Casparian strips could lead to either enhanced diffusion into the stele or increased leakage back out of the stele. Such processes would be expected to give rise to the complex ionomic changes observed in myb36.

To identify the genes regulated by MYB36, we performed a microarray analysis of genome-wide gene expression in the roots of two myb36 alleles (Arabidopsis Gene 1.0 ST array). In roots from both myb36-1 and -2, the expression of a common set of 39 genes was reduced, and 38 genes increased [false discovery rate (FDR) < 0.05; |log₂ fold change| > 1] (Table S1). To narrow this gene set to the targets of MYB36, we limited our selection to genes normally expressed in the endodermis (12, 13) (Fig. 3 A and B). Further, to eliminate those genes whose expression in myb36 is pleiotropically affected by the loss of functional Casparian strips and ectopic suberin and lignin deposition, we eliminated genes whose expression is also altered in esb1-1, because esb1-1 also lacks functional Casparian strips and develops ectopic suberin and lignin, but is not a transcriptional regulator (4) (Fig. 3A). Using quantitative PCR (qPCR), we confirmed that 30 genes, normally expressed in the endodermis, have reduced expression in myb36 (Fig. 3C and Fig. S3). After subtracting genes showing reduced expression in esb1-1, a final set of 23 genes positively regulated by MYB36 was identified (Fig. 3). This set of 23 genes includes all of the CASPs, six ESB-like genes including ESB1, and PER64, representing many of the major genes identified as players in Casparian strip formation (2–4). In addition to CASPs, ESBs, and PER64, this gene set also contains uncharacterized protein kinase and LRR-RLK, as well as other uncharacterized proteins predicted to be localized to the extracellular space (Table S1). Together, these genes are likely to define a critical gene set required for Casparian strip formation, giving us important clues to the molecular mechanism of Casparian strip biogenesis.

Table S1.

Microarray analysis of genome-wide gene expression in roots of two myb36 alleles

AGI	myb36-1			myb36-2		Annotation	Other name	SUBAcon*
	FDR	log2 FC	FDR		log2 FC
Down-regulated in myb36 mutants
AT3G11550.1	0	−4.471	0		−4.77	Uncharacterized protein family (UPF0497)	CASP2	Vacuole
AT2G40113.1	0	−3.205	0		−3.8	Pollen Ole e 1 allergen and extensin family protein		Extracellular
AT3G24020.1	0	−3.11	0		−4.8	Disease resistance-responsive (dirigent-like protein) family protein	ESB6	Extracellular
AT2G39430.1	0	−3.066	0		−3.8	Disease resistance-responsive (dirigent-like protein) family protein	ESB3	Plasma membrane
AT2G27370.1	0	−2.678	0		−3.59	Uncharacterized protein family (UPF0497)	CASP3	Plasma membrane
AT1G71740.1	0	−2.646	0		−3.9	—		Mitochondrion
AT2G36100.1	0	−2.46	0		−4.65	Uncharacterized protein family (UPF0497)	CASP1	Vacuole
AT5G52790.1	0	−2.379	0		−2.28	CBS domain-containing protein with a domain of unknown function (DUF21)		Plasma membrane
ATMG00030.1	0	−2.287	0		−2.28	—		Mitochondrion
AT1G30750.1	0	−2.201	0		−2.82	—		Extracellular
AT3G55230.1	0	−2.198	0		−5.38	Disease resistance-responsive (dirigent-like protein) family protein	ESB4	Extracellular
AT5G15290.1	0	−2.171	0		−1.91	Uncharacterized protein family (UPF0497)	CASP5	Plasma membrane
AT5G42180.1	0	−2.097	0		−3.96	Peroxidase superfamily protein	PER64	Extracellular
AT2G28670.1	0	−2.013	0		−3.12	Disease resistance-responsive (dirigent-like protein) family protein	ESB1	Extracellular
AT1G43020.1	0	−1.855	0		−1.83	Protein of unknown function, DUF547		Plastid
AT2G32300.1	0	−1.799	0		−1.97	Uclacyanin 1		Extracellular
AT5G51680.1	0	−1.756	0		−2.07	hydroxyproline-rich glycoprotein Family protein		Nucleus
AT1G53940.1	0	−1.732	4E-04		−1.51	GDSL-motif lipase 2		Extracellular
AT5G65530.1	0	−1.649	0		−1.81	Protein kinase superfamily protein		Nucleus
AT2G28470.1	0	−1.53	3E-04		−1.42	Beta-galactosidase 8		Extracellular
AT1G44970.1	0	−1.449	9E-04		−1.26	Peroxidase superfamily protein	PER9	Extracellular
AT1G07740.1	0	−1.446	3E-04		−1.46	Tetratricopeptide repeat (TPR)-like superfamily protein		Mitochondrion
AT4G13580.1	0	−1.429	0		−1.87	Disease resistance-responsive (dirigent-like protein) family protein	ESB5	Extracellular
AT4G02090.1	0	−1.396	0		−1.55	—		Nucleus
AT5G06200.1	0	−1.362	0		−2.03	Uncharacterized protein family (UPF0497)	CASP4	Plasma membrane
AT2G28210.1	0	−1.351	0.002		−1.18	Alpha carbonic anhydrase 2		Cytosol
AT5G42655.1	0	−1.337	0		−1.58	Disease resistance-responsive (dirigent-like protein) family protein	ESB like	Mitochondrion
AT3G48346.1	0	−1.325	0		−1.55	—		Extracellular
AT5G43540.1	0	−1.301	0		−1.63	C2H2 and C2HC zinc fingers superfamily protein		Nucleus
AT1G61590.1	0	−1.238	3E-04		−1.4	Protein kinase superfamily protein		Plastid
AT4G30110.1	0	−1.228	3E-04		−1.38	Heavy metal atpase 2		Plasma membrane
AT2G36255.1	0.003	−1.17	8E-04		−1.28	Defensin-like (DEFL) family protein		Extracellular
AT4G33020.1	0.001	−1.156	0.005		−1.03	ZIP metal ion transporter family		Vacuole, plasma membrane
AT1G14160.1	0.001	−1.138	0.001		−1.13	Uncharacterized protein family (UPF0497)	CASPL1A1	Plasma membrane
AT4G21340.1	6E-04	−1.129	8E-04		−1.24	Basic helix–loop–helix (bHLH) DNA-binding superfamily protein		Nucleus
AT5G33355.1	0.003	−1.063	1E-03		−1.13	Defensin-like (DEFL) family protein		Extracellular
AT4G30270.1	0.003	−1.029	0.009		−1.02	Xyloglucan endotransglucosylase		Extracellular
AT5G54270.1	0.003	−1.011	6E-04		−1.36	Light-harvesting chlorophyll B-binding protein 3		Plastid
AT2G15300.1	0.006	−1.003	0.001		−1.19	Leucine-rich repeat protein kinase family protein		Plasma membrane
Up-regulated in myb36 mutants
AT5G05340.1	0	2.848	0		2.374	Peroxidase superfamily protein
AT5G05390.1	0	2.625	0		2.547	Laccase 12
AT1G65500.1	0	2.338	0		2.534	—
AT4G02520.1	0	2.258	0		1.952	GST PHI 2
AT5G52390.1	0	2.183	0		2.841	PAR1 protein
AT1G49570.1	0	2.05	0		2.175	Peroxidase superfamily protein
AT4G12480.1	0	2.024	0		1.849	Bifunctional inhibitor
AT2G29220.1	0	1.946	0		2.604	Con A-like lectin protein kinase family protein
AT1G30760.1	0	1.656	4E-04		1.409	FAD-binding Berberine family protein
AT5G38940.1	0	1.583	0.005		1.254	RmlC-like cupins superfamily protein
AT4G00910.1	0	1.582	7E-04		1.411	Aluminum activated malate transporter family protein
AT1G66783.1	0	1.57	0		2.004	MIR157A; miRNA
AT1G51920.1	0	1.52	0		1.787	—
AT4G03540.1	0	1.447	4E-04		1.422	Uncharacterized protein family (UPF0497)
AT2G41100.4	0	1.444	0		1.582	Calcium-binding EF hand family protein
AT3G55090.1	0	1.444	0.003		1.245	ABC-2 type transporter family protein
AT4G12490.1	0	1.437	0.002		1.326	Bifunctional inhibitor
AT1G69920.1	0	1.357	6E-04		1.368	GST TAU 12
AT1G34910.1	0.001	1.338	0		1.948	—
AT5G39670.1	0	1.303	0		2.114	Calcium-binding EF-hand family protein
AT5G64120.1	0	1.278	0.006		1.181	Peroxidase superfamily protein
AT1G77380.1	0	1.261	0		1.634	Amino acid permease 3
AT1G63560.1	0	1.256	0		1.596	Receptor-like protein kinase-related family protein
AT4G22710.1	4E-04	1.239	6E-04		1.409	Cytochrome P450, family 706, Subfamily A, polypeptide 2
AT5G36140.1	4E-04	1.219	0.01		1.11	Cytochrome P450, family 716, subfamily A, polypeptide 2
AT1G21120.1	0.002	1.151	0		1.718	O-methyltransferase family protein
AT4G38401.1	0.002	1.144	0		1.473	—
AT2G33950.1	0.008	1.131	7E-04		1.557	Pre-tRNA
AT1G02930.1	0.002	1.122	0.009		1.061	GST 6
AT4G22470.1	0.003	1.088	0.009		1.169	Protease inhibitor
AT1G72060.1	0.005	1.087	0.007		1.244	Serine-type endopeptidase inhibitors
AT2G27550.1	0.003	1.072	0.01		1.118	Centroradialis
AT5G36130.1	0.003	1.064	0.011		1.023	Cytochrome P450 superfamily protein
AT1G67148.1	0.003	1.057	0.009		1.043	—
AT1G21110.1	0.005	1.033	0		1.5	O-methyltransferase family protein
AT1G29860.1	0.005	1.029	0.007		1.148	WRKY DNA-binding protein 71
AT5G09290.1	0.005	1.017	0.01		1.096	Inositol monophosphatase family protein
AT1G02920.1	0.005	1.015	0.005		1.192	GST 7

FDR < 0.05; |log₂ FC| > 1. FC, fold change.

^*Subcellular localization was predicated with SUBAcon (suba3.plantenergy.uwa.edu.au/).

Fig. 3.

MYB36 regulates Casparian strip associated genes. (A) Strategy to identify MYB36-target genes. (B) Heatmap of z-score–normalized expression of genes with FDR < 0.05 and |log₂ fold change| > 1 in both myb36 mutants mapped onto their radial expression pattern (12, 13). Gene IDs boxed in blue and yellow indicate genes with reduced and increased expression, respectively. End, endodermis; GT, ground tissue. Both GT and End include endodermally expressed genes. See also Table S1. (C) Heatmap showing gene expression levels in myb36-1 and esb1-1 relative to wild type based on qPCR results (Fig. S3). Gene IDs highlighted with an asterisk indicate genes not on the ATH1 array. Magenta shows CASPs, ESBs, and PER64. (D) ChIP assays using anti-GFP antibody. Red lines below the gene structure with numbers mark the location of amplicons amplified in the ChIP-qPCR. EIF4A, negative control. n = 3 from two independent experiments (Exp. 1 and 2). Bars represent mean ± SD.

Fig. S3.

qPCR analysis of endodermally expressed genes with altered expression in myb36-1. Based on our microarray analysis, we selected genes in myb36 with down-regulated expression that in wild type have expression enriched in the endodermis (Fig. 3B) and also genes that are not on the ATH1 array. Expression of these selected genes was analyzed by qPCR in roots of 2-wk-old plants grown on agar-solidified medium. Magenta characters show genes known or suggested to be involved in Casparian strip formation. Gene IDs with a gray background are not on the ATH1 array. Black frames are the CASP family of genes, among which only At1g14160 was down both in myb36-1 and esb1-1. Orange frames are the ESB family of genes. Sky blue frames are other genes down-regulated in myb36-1. Green frames are genes down-regulated both in myb36-1 and esb1-1. Yellow frames are genes not altered in the mutants. Purple frames are up-regulated genes. Values are relative to Col-0. Different characters indicate significant differences by Tukey’s HSD (P < 0.05) (n = 4). Bars indicate mean ± SD.

To investigate whether MYB36 directly regulates known Casparian strip associated genes by binding to their promoters, we performed chromatin immunoprecipitation (ChIP)-qPCR against CASP1, PER64, and ESB1 using the MYB36 genome-GFP/myb36-1 line (Fig. 3D). MYB36 binding was found to be consistently enriched in the promoter region of these three genes relative to the wild type. Furthermore, no enrichment was observed for binding to the promoter region (1,123–1,289 bp upstream from start codon) of a ubiquitously expressed negative control gene, EUKARYOTIC TRANSLATION INITIATION FACTOR 4A (EIF4A). These results indicate that MYB36 exerts its regulatory functions by associating directly with the CASP1, PER64, and ESB1 promoters.

Casparian strip formation requires precise localization of the lignin-polymerizing machinery, which is directed by the CASPs in the plasma membrane (3). However, the molecular mechanisms that locate the CASPs to the Casparian strip formation site are unknown. To probe whether MYB36-regulated genes are involved in localization of the CASPs, we expressed CASP1–mCherry in the endodermis of myb36-1 mutants, using the endodermally active SCR promoter (3). As expected in wild type, CASP1–mCherry was found to be localized throughout the plasma membrane of the endodermis in the meristematic zone (Fig. S4). Further, as the root matures, CASP1–mCherry localization in the plasma membrane becomes restricted to a central band encircling the cell, where the Casparian strip is formed (Fig. S4 and Fig. 4A). In contrast, in the myb36-1 mutant, CASP1–mCherry fluorescence does not localize into a band, but, rather, remains localized throughout the plasma membrane and also accumulates inside cells (Fig. 4A). Thus, CASP1–mCherry in myb36 behaved in a similar manner to that previously observed for CASP1–GFP when ectopically expressed in nonendodermal cells in wild-type (Fig. 4A) (3). This result demonstrates that MYB36 in the endodermis not only regulates expression of the CASP genes, but also regulates expression of endodermal genes required for CASP1 localization to the plasma membrane, a critical step in marking the site for Casparian strip deposition.

Fig. S4.

Restricted localization of CASP1–mCherry was observed around the 15th cell from the onset of elongation in wild-type Col-0. (A) CASP1–mCherry was expressed in wild-type Col-0 by using the endodermis-expressing promoter of SCR. Root tip (Lower) and the region where restricted CASP1–mCherry localization to Casparian strip deposition site was observed (Upper) are shown. The asterisk indicates the 20th endodermal cell from the onset of elongation. Yellow arrowheads indicate restricted localization of CASP1–mCherry. (Scale bar: 100 µm.) (B) Restricted CASP1–mCherry localization as observed in A (arrowheads) was observed around the 15th cell from the onset of elongation in wild-type Col-0 (n = 11). Horizontal bar indicates the median.

Fig. 4.

MYB36 is sufficient for Casparian strip formation. (A) CASP1–mCherry localization in Col-0 and myb36-1. The confocal image was taken around the 20th endodermal cell (asterisks) from the onset of elongation. The images in Lower show radial optical sections taken from the dashed line shown in Upper. (B) Ectopic formation of Casparian strips in the TRANSPLANTA β-estradiol inducible line. Although both of these dyes stain cell walls, PI primarily interacts with lignin (yellow) and Calcofluor White with cellulose (blue). (C and D) CASP1–GFP (green) localization in the root of β-estradiol–treated line. Shown are longitudinal (C) and radial (D) optical sections (taken at the dashed line in C). Magenta (PI), cell wall. Cor, cortex; End, endodermis; Epi, epidermis. (Scale bars: 50 µm.)

We have shown that MYB36 is necessary for the targeted deposition of lignin for the formation of Casparian strips. Next, we determined whether its expression was sufficient for Casparian strip formation. To test this hypothesis, we used transgenic lines expressing MYB36 under the control of the β-estradiol–inducible promoter (14), with MYB36 expression expected in all tissues. After β-estradiol treatment in these lines, lignin deposition was observed in the cortex, in the middle of the anticlinal cell wall in a band encircling the cells, precisely where Casparian strips would form in the endodermis (Fig. 4B and Fig. S5 A and B). We confirmed that these ectopic lignin structures observed in the cortex are Casparian strip-like by demonstrating that they contain CASP1–GFP (Fig. 4 C and D). Here, CASP1–GFP was expressed from the CASP1 native promoter, which is normally only active in the endodermis and is never observed in the cortex or epidermis (3). The strong and specific signal generated by CASP1–GFP also allowed us to observe Casparian strip-like patterns of CASP1–GFP accumulation in the epidermis after β-estradiol induction (Fig. 4 C and D). Ectopic expression from the 35S promoter of CASP1–GFP in the cortex or epidermis has previously been shown to not be sufficient to cause the accumulation of CASP1–GFP into a Casparian strip-like domain (3). However, we show that expression of MYB36 is sufficient to drive expression of both CASP1–GFP from its native promoter and the genes required for localization of CASP1–GFP to a Casparian strip-like domain in both the cortex and the epidermis. Interestingly, the Casparian strip-like structures in the cortex have a discontinuous pattern similar to that of the sgn3 mutant (11) (Fig. S5C). We show that SGN3 appears to not be a target for regulation by MYB36 based on our microarray data. Because SGN3 is not normally expressed in the cortex (11), this discontinuous pattern of the Casparian strip-like structure we observed when MYB36 was ectopically expressed in the cortex may be due to a lack of SGN3.

Fig. S5.

Ectopic lignin deposition in β-estradiol–inducible MYB36 line. After mock treatment with DMSO or treatment with β-estradiol, roots were cleared and subjected to confocal microscopy to observe lignin autofluorescence. (A and B) Longitudinal (A) and radial (B) optical section of the roots are shown. Lignin deposition associated with Casparian strips was only observed in the endodermis of DMSO-treated root. However, in β-estradiol–treated roots, lignin deposition was also observed in the cortex (yellow arrowheads). (C) A z-stack image of the region between the dashed lines (nine images) in the β-estradiol–treated roots is shown, depicting the autofluorescence pattern in the cortex. (Scale bars: 50 µm.)

Here, we demonstrated that MYB36 regulates expression of genes critical for the localized polymerization of lignin required for the formation of Casparian strips, in both a developmentally and cell-type–specific manner. Our identification of the genes regulated by MYB36 now provides the “list of parts” needed for localizing and building Casparian strips. Further analysis of these genes should allow us to understand how these MYB36-regulated parts come together and function to overcome the engineering challenges of building Casparian strips.

Methods

Plant Materials and Growth Conditions.

The Col-0 A. thaliana accession was used throughout the experiments. T-DNA insertion alleles of MYB36 (myb36-2: GK-543B11) and β-estradiol–inducible TRANSPLANTA lines (N2102512 and N2102513) (14) were obtained from the Nottingham Arabidopsis Stock Centre (NASC). Homozygous lines were established for GK-543B11 by using the PCR primers described in Table S2. Plants were grown on MGRL medium solidified with 1.2% (wt/vol) agar supplemented with 1% sucrose (15). After incubation for 2 d at 4 °C, plates were placed vertically, and plants were grown at 22 °C under a 16-h light/8-h dark photoperiod. For observation of Casparian strips, suberin staining by fluoral yellow 088, and PI blockage, 6-d-old seedlings were used. For β-estradiol induction, TRANSPLANTA lines were grown on MGRL medium for 5 d, and seedlings were transferred to MGRL medium containing 10 µM β-estradiol (Sigma-Aldrich; E8875) and grown for a further 2 d.

Table S2.

Primers

Primer name	Sequence	Note
myb36-R1	CGGCTTCCAATGCTAATGTAG	T-DNA check of GK-543B11
myb36-F2	ATGGGAAGAGCTCCATGCTG	T-DNA check of GK-543B11
JL202	CATTTTATAATAACGCTGCGGACATCTAC	T-DNA check of GK-543B11 (T-DNA left border)
myb36-F9-Kpn	GGGGTACCTAACACGCGCTAATGACGTC	MYB36 genome cloning (KpnI)
myb36-R3-Xho	CCGCTCGAGGCAACACTGTGGTAGCTCATCTG	MYB36 genome cloning (XhoI)
EF1a-F	CCTTGGTGTCAAGCAGATGA	qPCR (Internal control)
EF1a-R	TGAAGACACCTCCTTGATGATTT	qPCR (Internal control)
AT1G07730-F	CGGTCTTCCCGTGGCTAAC	qPCR
AT1G07730-R	TCGTTGTCCATCACCGTCAA	qPCR
AT1G14160-F	ACGCATCTCGGGATCTCTCA	qPCR
AT1G14160-R	TGGCGAGAGAGGGAGATAGC	qPCR
AT1G30750-F	GTTGAAGGTCTCCGGCACAA	qPCR
AT1G30750-R	ATGTGTAGGGTTCAGTTCCTGTCA	qPCR
AT1G43020-F	CGTCCATGCCATTCACCAT	qPCR
AT1G43020-R	CAACGCGTAAGTGTGTCTACCACTA	qPCR
AT1G44970-F	AGATTGTTATGACGGTGCTCGAA	qPCR
AT1G44970-R	CGAAGCAGTCGTGGAAGTGA	qPCR
AT1G61590-F	TGTTGTTCCGTTGATGACCAA	qPCR
AT1G61590-R	AAGAAGGAAGTGGTCCGAGATG	qPCR
AT1G71740-F	CGAAGCCGCTTACAATTTTCA	qPCR
AT1G71740-R	CCCAACGACGAAGCTTCCT	qPCR
AT2G15300-F	CAACGCCAGTGAGCTTCGTAT	qPCR
AT2G15300-R	TGCAGACTAGCCACGTTGCT	qPCR
AT2G27370-F	CGCTTCCGCAGCCATAGTT	qPCR
AT2G27370-R	GCTCGTTCCTTGGCAAAAGT	qPCR
AT2G28210-F	GCATTCTCCCTCTGAACATACTATGA	qPCR
AT2G28210-R	TGACTACAGCCAAACTTCCGTTAA	qPCR
AT2G28670-F	ATGTCCCTTTCCTCGTTGGA	qPCR, ChIP-qPCR
AT2G28670-R	GCCACTAGCAACAGGGAAACC	qPCR, ChIP-qPCR
AT2G32300-F	AGCTCTAGCAGCACAACATCCA	qPCR
AT2G32300-R	AGAGTCTGCGCCAGCAAGA	qPCR
AT2G36100-F	GAGGTGGTGCCAAGAGAGGTT	qPCR, ChIP-qPCR
AT2G36100-R	TCGGCGGTGTACATGACAGA	qPCR, ChIP-qPCR
AT2G36255-F	CGGTAGCCCAAACATGCTTTA	qPCR
AT2G36255-R	CCCTGCAAAGAAGGTCACACA	qPCR
AT2G39430-F	AAACCTCCACCGTCATCCAA	qPCR
AT2G39430-R	GAAGGTTCCCGGCAGTTACA	qPCR
AT2G40113-F	GAATTTGTCATCCATCTTCCTTCTC	qPCR
AT2G40113-R	GTGCTTAGGCACATGTATTGGTTT	qPCR
AT3G11550-F	TAGCACACAGTGGTAACCAGAACA	qPCR
AT3G11550-R	AGACAACAGCTCCGCTGGAT	qPCR
AT3G24020-F	TGCACCTGAAGACCCCATTT	qPCR
AT3G24020-R	CCGAGCAAACCCGTGATG	qPCR
AT3G48346-F	TTTATTGATTACTTGCGTTCATGGA	qPCR
AT3G48346-R	CATACAACAAACTAAGCAAACAAGCA	qPCR
AT3G55230-F	CCCATAGCTGCCCTCCAA	qPCR
AT3G55230-R	GATCGCAGAGCCCAACTCAT	qPCR
AT4G02090-F	CGCAAAAACGGTGAAGAGAAA	qPCR
AT4G02090-R	CTTTTGGCGGGAAGCACTAA	qPCR
AT4G13580-F	AGAACGCTTGCGGATTTGC	qPCR
AT4G13580-R	CAATGATCCTCAACAGCATCTCA	qPCR
AT4G21340-F	CCATTAAAAAGGCCAAGACTTGA	qPCR
AT4G21340-R	AAGCGCGGTAATTCGATCTC	qPCR
AT5G06200-F	TCTTTGTCGTTGCCATTGCA	qPCR
AT5G06200-R	CGACGGCTAGAGGACGAACA	qPCR
AT5G15290-F	CTGCCTTTCTTTACTCAATTCATACG	qPCR
AT5G15290-R	TTGCAACCACGAAAAACGTTAA	qPCR
AT5G42180-F	ACCCAACACTAAACCCCTCATTC	qPCR, ChIP-qPCR
AT5G42180-R	CATCCATGTTCGATCCAGCAT	qPCR, ChIP-qPCR
AT5G42655-F	GACGTTGGAGACGCCTGAGT	qPCR
AT5G42655-R	CCCACTACCTCCATAACCTCTTTC	qPCR
AT5G43540-F	CTTGGAGGCTACGAGCAAGTAGA	qPCR
AT5G43540-R	TGTAGCCATCGAGCCGATTC	qPCR
AT5G51680-F	CGATATTTTTGATGATGCAAGCA	qPCR
AT5G51680-R	ACAGCACCGGATATCGAAACC	qPCR
AT5G52790-F	CTTCTCCACCAGCGCAATCT	qPCR
AT5G52790-R	TCCTCAACACCCTACTAGGAGACAA	qPCR
AT5G57620-F	ACAGAGCCCTTTCTCTAGTTTCTACAA	qPCR
AT5G57620-R	TTGAGATGCAAGCTTAGGGTTTT	qPCR
AT5G65530-F	CCGAGACATCAAAGCTTCCAA	qPCR
AT5G65530-R	TGCTCTGGGAGCCACTTAGC	qPCR
ATMG00030-F	TCCATGGATCACAGACGGTATC	qPCR
ATMG00030-R	CCGGGCATTGAGAAGGAA	qPCR
CASP1_F1	GTTCTCCAAAGAAATCGATGTAGC	ChIP-qPCR
CASP1_R1	ATGCAGATTAAATGGCTAGGATCAG	ChIP-qPCR
CASP1_F2	GGGGAAATGATTATGGGTTGAA	ChIP-qPCR
CASP1_R2	TAGCGCCAGCCCAAACAA	ChIP-qPCR
CASP1_F3	TTGCCTACGAAGGGATACCAA	ChIP-qPCR
CASP1_R3	CGGGAAGAGGAGTTGTTGCT	ChIP-qPCR
PER64_F5	GGATCTAGACTCGCATACCTCCA	ChIP-qPCR
PER64_R5	GTTCCACAGCATCCTCTGTTTG	ChIP-qPCR
PER64_F6	CGTGGCTTTCGCTTTTCTC	ChIP-qPCR
PER64_R6	CAACCTACCAAATACTGTCGAAACA	ChIP-qPCR
PER64_F7	CCACCACAACCAAAATTAAACG	ChIP-qPCR
PER64_R7	TTGGGAGATGGAGTTGTTAGTGAG	ChIP-qPCR
ESB1_F9	GCCATTTTCTTCTCTCCTTCCA	ChIP-qPCR
ESB1_R9	CGGTCCATAAAACCCAAACC	ChIP-qPCR
ESB1_F10	AAATCCGCGAAATATGCAAG	ChIP-qPCR
ESB1_R10	ACTGACTGTTACGTGTTCCGTGTT	ChIP-qPCR
ESB1_F11	TGCTTTGCCAGATTCCACA	ChIP-qPCR
ESB1_R11	GACTTACACAATCCCACCTCCA	ChIP-qPCR
ESB1_F12	CAACCAACGATCGTGATTTACA	ChIP-qPCR
ESB1_R12	AAAAGGTCGCTGATTGACAGAG	ChIP-qPCR
EIF4A-FW	TGTTTTGCTTCGTTTCAAGGA	ChIP-qPCR
EIF4A-RV	GCATTTTCCCGATTACAAC	ChIP-qPCR

Expression Analysis.

For the microarray and qPCR analysis, root samples from 2-wk-old seedlings were used. Total RNA was prepared by using a PureLink RNA Mini Kit (Life Technologies). For the qPCR analysis, RNA was converted to cDNA by using SuperScript III (Life Technologies). The cDNA was diluted 10-fold and used for qPCR by using StepOnePlus (Life Technologies) and SYBR Select Master Mix (Life Technologies). Results of qPCR are from two independent experiments with biological duplicates. The primer sequences used are listed in Table S2. Microarray analysis was performed by NASC using the Arabidopsis 1.0 ST Array (Affymetrix). The microarray analysis was performed for both myb36-1 and -2 by using three biological replicates for each genotype. Statistical analysis of microarray data were performed by using the R bioconductor (www.bioconductor.org/). After normalization using robust multiarray average, the rank products method was performed by using the R package RankProd (16, 17). The genes satisfying the criteria (FDR < 0.05; |log₂ fold change| > 1) in both of the myb36 mutants compared with Col-0 were selected, and the cell types in which they are normally expressed were determined from the published dataset of radial expression patterns (12, 13).

Plasmid Construction and Transformation.

For MYB36 localization, a MYB36 genomic DNA fragment was amplified by PCR (see Table S2 for primers), and the DNA fragment was digested and cloned into the KpnI and XhoI site of the pENTR2B dual selection vector (Life Technologies) and transferred to the destination vector pMDC107 (18) by using LR clonase (Life Technologies). For CASP1–-mCherry localization in myb36-1, pSCR–CASP1–mCherry (3) was introduced into myb36-1. All genetic transformations were performed by using the Agrobacterium-mediated floral dip method. Homozygous transgenic lines were used for all experiments.

Microscope Observations.

For GFP localization experiments, the FV1000 (Olympus) confocal microscope was used. Excitation and emission wavelengths were as follow: GFP, 488 and 485–545 nm; PI and mCherry, 559 and 570–670 nm. The GFP and mCherry signal was confirmed in at least five independent plants, and representative images are shown. For observation of the Casparian strip, a clearing treatment was performed as described (3, 19, 20), and cleared roots were stored in 50% (vol/vol) glycerol at 4 °C before use. Cleared roots were observed with the same settings as used for GFP. One-micrometer step-size images were taken, and z-stack images were constructed with Fiji, a distribution of ImageJ (fiji.sc/Fiji). For visualization of Casparian strips and cell wall, PI and Calcofluor White M2R (Fluorescent Brightener 28; Sigma-Aldrich; F3543) staining were performed as follows. Cleared roots (3, 19, 20) were stained with 10 µg/mL PI in 50% (vol/vol) glycerol for 10 min and then transferred to 0.001% Calcofluor White M2R in 50% (vol/vol) glycerol. After 10 min of staining, roots were transferred to 50% (wt/vol) glycerol for observation. Confocal microscope setting for Calcofluor White M2R was excitation 405 nm, and emission was 425–475 nm. In z-stack images, 1-µm step-size images were taken, and radial optical sections were constructed with Fiji. The autofluorescence of Casparian strip, Calcofluor White, and PI staining experiments were confirmed in at least five plants from three independent experiments, and representative images are shown. For quantification of Casparian strips as an apoplastic barrier, the PI penetration assay was performed as described (20). The “onset of elongation” was defined as the point at which an endodermal cell in a median optical section was clearly more than twice its width (20). For suberin observation, staining by Fluoral yellow 088 was performed as described (21).

Ionomic Analysis.

Ionomic analysis of plants grown on nutrient medium solidified with agar was performed as described (4). Briefly, plants were grown on agar-solidified medium. After 2 wk, shoots were harvested, dried at 88 °C for 20 h, and digested with concentrated nitric acid with an indium internal standard. Digested samples were diluted with 18 MΩ water and analyzed by using inductively coupled plasma (ICP)-MS (Elan DRC II; PerkinElmer) equipped with an Apex sample introduction system (Elemental Scientific). Twenty elements (Li, B, Na, Mg, P, S, K, Ca, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Mo, and Cd) were monitored.

ChIP.

ChIP was performed by following the protocol as described (22) with modifications as follows. Roots (100 mg fresh weight) from 11-d-old plants were cross-linked by using 4 mL of the buffer (10 mM PBS, pH 7.0, 50 mM NaCl, 0.1 M sucrose, and 1% formaldehyde) for 1 h at room temperature with the application of three cycles of vacuum infiltration (10 min under vacuum and 10 min of vacuum release). Glycine was added to a final concentration of 0.1 M to stop the cross-linking reaction, and the samples were incubated for a further 10 min. After being washed with tap water, the samples were ground to a fine powder by using a Multibeads Shocker (Yasui Kikai) at 1,500 rpm for 30 s. The powder was suspended with 2 mL of Lysis buffer [50 mM Tris⋅HCl, pH 7.5, 100 mM NaCl, 1% Triton X-100, 1 mM EDTA, EDTA-free Complete protease inhibitor (Roche)] and sonicated by using a Bioruptor UCD-250 (Cosmo Bio) with the following setting: mild intensity, 45 cycles (30 s ON and 30 s OFF) at 4 °C. A 100-µL sample of the chromatin sheared to between 200 and 1,500 bp was stored as the input fraction, and the rest (1.9 mL) was mixed with Dynabeads Protein G (Life Technologies) bound with anti-GFP antibody (ab290; Abcam) and incubated for 2 h at 4 °C. The beads were washed with Lysis buffer, twice with high-salt buffer [50 mM Tris⋅HCl, pH 7.5, 400 mM NaCl, 1% Triton X-100, 1 mM EDTA, and EDTA-free Complete protease inhibitor (Roche)], and then with Lysis buffer. After Elution buffer (50 mM Tris⋅HCl, pH 8.0, 10 mM EDTA, and 1% SDS) and proteinase K (0.5 mg/mL) were added to the beads, the beads were incubated overnight at 65 °C. The DNA was purified with NucleoSpin Gel and PCR Clean-up (Macherey-Nagel) with Buffer NTB (Macherey-Nagel). Eluted solutions were used for qPCR. EIF4A (At3g13920) was used as a negative control, as is often used (23). The primer sequences used are listed in Table S2. Two independent experiments were performed with three biological replicates for each.

Statistical Analysis.

Replicates were biological replicates from separate plants. Data in all bar graphs represent the mean ± SD. Statistical analysis was performed by using Microsoft Excel or R. No statistical methods were used to predetermine the sample size. No samples were excluded from data analysis except for ICP-MS data. For the ICP-MS data, the Sumirnov–Grubb test (P < 0.01) was used to remove outliers as contaminations of several elements, such as Ni and Zn, which can be derived from the ICP-MS instrument. For qPCR analysis, we assumed the data came from a normally distributed population and used Tukey’s honest significant difference (HSD). For the counting experiment with PI staining and suberin accumulation (Fig. 2 D and E), Bartlett’s test was used, followed by Tukey’s HSD (Fig. 2D) and Steel–Dwass test (Fig. 2E).