**Establishment of the Embryonic Shoot Meristem Involves Activation of Two Classes of Genes with Opposing Functions for Meristem Activities**

**Mitsuhiro Aida 1,\* , Yuka Tsubakimoto <sup>2</sup> , Satoko Shimizu <sup>2</sup> , Hiroyuki Ogisu <sup>2</sup> , Masako Kamiya <sup>2</sup> , Ryosuke Iwamoto <sup>2</sup> , Seiji Takeda 2,3 , Md Rezaul Karim 2,4 , Masaharu Mizutani <sup>5</sup> , Michael Lenhard <sup>6</sup> and Masao Tasaka <sup>2</sup>**


Received: 31 July 2020; Accepted: 12 August 2020; Published: 15 August 2020

**Abstract:** The shoot meristem, a stem-cell-containing tissue initiated during plant embryogenesis, is responsible for continuous shoot organ production in postembryonic development. Although key regulatory factors including *KNOX* genes are responsible for stem cell maintenance in the shoot meristem, how the onset of such factors is regulated during embryogenesis is elusive. Here, we present evidence that the two *KNOX* genes *STM* and *KNAT6* together with the two other regulatory genes *BLR* and *LAS* are functionally important downstream genes of *CUC1* and *CUC2*, which are a redundant pair of genes that specify the embryonic shoot organ boundary. Combined expression of *STM* with any of *KNAT6*, *BLR*, and *LAS* can efficiently rescue the defects of shoot meristem formation and/or separation of cotyledons in *cuc1 cuc2* double mutants. In addition, *CUC1* and *CUC2* are also required for the activation of *KLU*, a cytochrome P450-encoding gene known to restrict organ production, and *KLU* counteracts *STM* in the promotion of meristem activity, providing a possible balancing mechanism for shoot meristem maintenance. Together, these results establish the roles for *CUC1* and *CUC2* in coordinating the activation of two classes of genes with opposite effects on shoot meristem activity.

**Keywords:** shoot meristem; embryogenesis; stem cell; boundary; transcription factor; cytochrome P450; *CUC*; *STM*; *LAS*; *BLR*; *KNAT6*; *KLU*; *CYP78A5*

#### **1. Introduction**

Primary growth in plant shoots depends on the activity of stem-cell-containing tissue called the shoot meristem, which is located at the tip of the stem [1]. The shoot meristem is initially formed during embryogenesis and is activated upon germination to produce shoot organs such as leaves, stems, and floral organs, while it maintains an undifferentiated stem cell population at its center. Once activated, the shoot meristem keeps the balance between cell proliferation and differentiation to maintain an appropriate size of the stem cell population within it; factors essential for this process have been identified [2,3]. Although activation of these maintenance factors is associated with shoot meristem initiation during embryogenesis, how the process is coordinated is unknown.

Several key regulators for shoot meristem initiation have been reported [4–11]. Among them, the NAM/CUC3 type of NAC-domain transcription factors represents a class of regulators that are required for specification of shoot organ boundaries, which are sites for shoot meristem formation in embryonic and postembryonic development [12–16]. In *Arabidopsis thaliana*, the three NAM/CUC3 genes *CUC1*, *CUC2*, and *CUC3* are expressed in cells along the boundary between two cotyledon primordia and promote shoot meristem formation and the separation of cotyledons [5,6,17]. As shoot meristem formation proceeds, expression of these genes is downregulated from the meristem center and becomes restricted to the adaxial and lateral boundaries of cotyledons. In postembryonic development, the three *CUC* genes are expressed at the adaxial and lateral boundaries of leaf primordia and are required for the formation of axillary shoot meristem as well as for the separation of leaves [6,18].

Several genes whose expression is dependent on *CUC* gene activities have been identified. Expression of the two *KNOTTED1-like homeobox* (*KNOX*) genes *SHOOT MERISTEMLESS* (*STM*) and *KNOTTED1-like from Arabidopsis thaliana 6* (*KNAT6*), which are required for shoot meristem maintenance, is absent from the *cuc1 cuc2* double mutant [17,19] and ectopic expression of the *CUC* genes induces *STM* expression [5,20,21]. The *LIGHT-DEPENDENT SHORT HYPOCOTYLS* (*LSH*) genes *LSH3* and *LSH4*, which encode nuclear proteins of the *Arabidopsis* LSH1 and *Oryza* G1 (ALOG) family, have been identified as direct transcriptional targets of the CUC1 protein and their overexpression induces ectopic shoot meristem formation [22]. Genome-wide mapping of protein–DNA interactions among boundary-enriched genes has identified the *GRAS* family gene *LATERAL SUPPRESSOR* (*LAS*) and the microRNA gene *MIR164C* as direct transcriptional targets of CUC2 [23]. *LAS* encodes a putative transcriptional regulator and is required for axillary shoot meristem formation [24]. Together, these analyses indicate that *CUC* genes regulate multiple genes involved in shoot meristem activity or boundary specification, or both. However, the functional relationship between these downstream genes and *CUC* gene activity remains elusive.

Here, we selected a set of genes whose expression is dependent on *CUC1* and *CUC2* during embryogenesis and demonstrated that the combined activities of *STM*, *KNAT6*, *BLR*, and *LAS* are important for promoting shoot meristem formation and cotyledon separation downstream of *CUC1* and *CUC2*. Moreover, *CUC1* and *CUC2* are also required for the expression of *KLUH* (*KLU*)/*CYP78A5*, a cytochrome P450-encoding gene involved in the rate of shoot organ production and organ size [25,26]. Genetic analysis indicates that *KLU* restricts shoot meristem activity and counteracts *STM* function. Our results thus indicate that the activation of two classes of genes with opposing functions, one positively and the other negatively affecting meristem activity, is an important step for shoot meristem formation.

#### **2. Results**

#### *2.1. Selection of Candidate CUC1 and CUC2 Downstream Genes*

It has been reported that the two *KNOX* genes *STM* and *KNAT6*, the GRAS gene *LAS*, and the two *ALOG* genes *LSH3* and *LSH4* show overlapping expression patterns to those of *CUC1* and *CUC2* in the boundary region of cotyledons, and their expression is absent in the corresponding region of *cuc1 cuc2* double-mutant embryos [5,6,17,19,22]. We identified six additional candidate downstream genes positively regulated by *CUC1* and/or *CUC2* from microarray-based screening combined with quantitative real-time polymerase chain reaction (qRT-PCR) and in situ hybridization experiments (Supplementary Text S1; Supplementary Tables S1–S3). These genes were expressed in the boundary region that overlapped with the *CUC* gene expression domain [5,17] and were downregulated embryogenesis.

specifically in the corresponding region of *cuc1 cuc2* embryos (Figure 1A–D; Supplementary Figure S1), indicating the dependence of their expression on *CUC1* and *CUC2* activities during embryogenesis. plants, which do not express CUC1‐GR, none of the genes were upregulated by DEX treatment in the presence or absence of CHX (Figure 1F), indicatingthat the induction of the downstream genes was not a secondary effect of DEX or CHX.

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 3 of 16

S1), indicating the dependence of their expression on *CUC1* and *CUC2* activities during

To gain insight into how CUC1 regulates expression of these candidate genes, we used the glucocorticoid receptor (GR) system, in which the activity of CUC1 is induced by the exogenous application of dexamethasone (DEX) [27]. Using this system, we previously found evidence for the direct activation of the *LSH3*, *LSH4*, and *STM* genes by CUC1 [22,28]. Among the remaining eight genes, we found that only *LAS* and *PAN* were significantly upregulated upon DEX treatment alone in *CUC1‐GR* plants (Figure 1E, left panel). By contrast, treating with both DEX and the protein synthesis inhibitor cycloheximide (CHX), which blocks secondary transcriptional responses caused by genes directly activated by CUC1, significantly upregulated not only *LAS* and *PAN*, but also four additional genes among the eight tested. Treatment with CHX alone did not alter their expression levels (Figure 1E, right panel). Together, the results suggest that the six genes are under direct transcriptional regulation by the CUC1‐GR protein, but that the action of CUC1 is counteracted by a CHX‐sensitive negative factor with respect to activation of four of the genes (*KLUH*, *KNAT6*, *UFO*,

**Figure 1.** Regulation of candidate downstream genes by *CUC1* and *CUC2*. (**A**–**D**) In situ hybridization of newly identified candidate downstream genes. Four of the six candidates are shown. Three serial longitudinal sections of wild‐type L*er* (**left**) and *cuc1‐1 cuc2‐1* double‐mutant (**right**) embryos at the late heart stage. Arrowheads indicate the position of the cotyledon boundary region. Brackets in A, B, and D indicate the position of expression outside the boundary region. Bars = 50 μm. (**E**,**F**) Transcriptional responses of candidate downstream genes upon dexamethasone (DEX) treatment in *CUC1‐GR* (**E**) and non‐transgenic (**F**) plants in the absence (**left**) or presence (**right**) of the protein synthesis inhibitor cycloheximide (CHX). Three biological replicates of 7‐day‐old seedlings. Single and double asterisks indicate *p* < 0.05 and *p* < 0.01, respectively, in comparisons between samples with and without DEX (Welch's *t*‐test). **Figure 1.** Regulation of candidate downstream genes by *CUC1* and *CUC2*. (**A**–**D**) In situ hybridization of newly identified candidate downstream genes. Four of the six candidates are shown. Three serial longitudinal sections of wild-type L*er* (**left**) and *cuc1-1 cuc2-1* double-mutant (**right**) embryos at the late heart stage. Arrowheads indicate the position of the cotyledon boundary region. Brackets in (**A**), (**B**), and (**D**) indicate the position of expression outside the boundary region. Bars = 50 µm. (**E**,**F**) Transcriptional responses of candidate downstream genes upon dexamethasone (DEX) treatment in *CUC1-GR* (**E**) and non-transgenic (**F**) plants in the absence (**left**) or presence (**right**) of the protein synthesis inhibitor cycloheximide (CHX). Three biological replicates of 7-day-old seedlings. Single and double asterisks indicate *p* < 0.05 and *p* < 0.01, respectively, in comparisons between samples with and without DEX (Welch's *t*-test).

To gain insight into how CUC1 regulates expression of these candidate genes, we used the glucocorticoid receptor (GR) system, in which the activity of CUC1 is induced by the exogenous application of dexamethasone (DEX) [27]. Using this system, we previously found evidence for the direct activation of the *LSH3*, *LSH4*, and *STM* genes by CUC1 [22,28]. Among the remaining eight genes, we found that only *LAS* and *PAN* were significantly upregulated upon DEX treatment alone in *CUC1-GR* plants (Figure 1E, left panel). By contrast, treating with both DEX and the protein synthesis inhibitor cycloheximide (CHX), which blocks secondary transcriptional responses caused by genes directly activated by CUC1, significantly upregulated not only *LAS* and *PAN*, but also four additional genes among the eight tested. Treatment with CHX alone did not alter their expression levels (Figure 1E, right panel). Together, the results suggest that the six genes are under direct transcriptional regulation by the CUC1-GR protein, but that the action of CUC1 is counteracted by a CHX-sensitive negative factor with respect to activation of four of the genes (*KLUH*, *KNAT6*, *UFO*, and *SAI-LLP1*). Another possibility is that DEX treatment alone indirectly promotes expression of genes that negatively affect CUC1-dependant activation of the four genes. In control non-transgenic plants, which do not express CUC1-GR, none of the genes were upregulated by DEX treatment in the presence or absence of

CHX (Figure 1F), indicating that the induction of the downstream genes was not a secondary effect of DEX or CHX.

#### *2.2. Combined Expression of STM with LAS, BLR, and KNAT6 is Su*ffi*cient to Rescue the Embryonic Shoot Phenotypes of cuc1 cuc2*

To examine the functional significance of the candidate genes in processes downstream of *CUC1* and *CUC2*, we expressed each gene in the *cuc1 cuc2* double-mutant background under the control of the *CUC2* promoter (*ProCUC2*)*,* which drives expression in the boundary region in both wild-type and double-mutant embryos (Figure 2A,B). In wild type, seedlings develop a shoot that continuously produces leaves immediately after germination and have two completely separated cotyledons (Figure 2C). By contrast, the *cuc1 cuc2* double mutant has two cotyledons fused along their margins and fails to form a shoot, and this phenotype is fully penetrant (Figure 2D) [4]. When the coding sequence of *CUC2* was used as a positive control (*ProCUC2:CUC2*), 41.7% of the T1 seedlings showed a strongly rescued phenotype with no or slight delay in shoot formation and with completely separated cotyledons, 50.0% showed a mildly rescued phenotype with delayed or no shoot formation and half-separated cotyledons, and the remaining 8.3% showed non-rescued phenotype identical to that of *cuc1 cuc2* (Table 1, Supplementary Figure S2A,B).

Among the 10 downstream genes, the *ProCUC2:STM* and *ProCUC2:LAS* transgenes were able to mildly rescue the *cuc1 cuc2* phenotype, resulting in the occasional formation of a functional shoot that can produce leaves as well as in the partial separation of cotyledons (Table 1, Figure 2E, and Supplementary Figure S2C–E). In cleared seedlings, wild type has the dome-shaped shoot meristem with a few leaf primordia, whereas *cuc1 cuc2* plants lack either structure (Figure 2F,G) [4]. On the other hand, plants partially rescued by *ProCUC2*:*STM* showed variable phenotypes: some lacked a shoot meristem and leaf primordia, some developed small undifferentiated tissue, and the other produced the shoot meristem and leaf primordia (Figure 2H). These results indicate that *STM* and *LAS* play prominent roles in shoot meristem formation and cotyledon separation and that their individual activities can partially bypass the requirements for *CUC1* and *CUC2* for embryonic shoot meristem formation and cotyledon separation.

The rescue of the *cuc1 cuc2* mutant phenotype by the *STM* or *LAS* transgene alone was only mild and partial, thus we next tested their combined activities. In the F<sup>2</sup> generation of the cross between the lines with the *STM* and *LAS* transgenes, only plants carrying both showed a rescued phenotype (Table 2). These rescued plants showed either partial (Figure 2J) or complete separation of cotyledons (Figure 2K), with the latter forming leaves with only a slight delay compared with the timing in the wild type, indicating that combined expression of *STM* and *LAS* is sufficient to compensate for the loss of *CUC1* and *CUC2* activities.

We next selected five other genes encoding transcription factors or transcriptional co-regulators, and tested their ability to rescue the *cuc1 cuc2* phenotype in combination with the *STM* transgene (Table 3). *BLR* and *KNAT6* were able to achieve rescue when combined with *STM* (Figure 2K,L), while the rest failed to do so. Plants expressing both *STM* and *BLR* produced nearly normal shoots with completely separated cotyledons (Figure 2K), indicating that, similarly to *LAS*, *BLR* can efficiently support the ability of *STM* to promote shoot meristem formation and cotyledon separation in the absence of *CUC1* and *CUC2*. By contrast, plants expressing both *STM* and *KNAT6* only rescued the cotyledon phenotype, but not that of shoot formation (Figure 2L), indicating that *KNAT6* can support the *STM* activity only in the limited developmental pathway downstream of the *CUC* genes.

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 5 of 16

**Figure 2.** Rescue of *cuc1 cuc2* phenotype by candidate downstream genes. (**A**) Schematic diagram of rescue experiments. *CUC2* promoter(*ProCUC2*), cDNA of downstream gene (*DG*), and nos terminator (*ter*). (**B**) *CUC2* promoter activity detected by β‐glucuronidase (GUS). Longitudinal views of GUS‐ stained wild‐type Col (**top**) and *cuc1‐5 cuc2‐3* (**bottom**) embryos in three different optical sections, showing expression in the boundary region. (**C**,**D**) Seedlings of wild‐type Col (**C**) and *cuc1‐5 cuc2‐3* (**D**), 7 days after germination (dag). Wild type has two separated cotyledons (co) and develops a shoot between them (arrowhead), whereas *cuc1‐5 cuc2‐3* has cotyledons (co) fused along their margins. (**E**) A *cuc1‐5 cuc2‐3* seedling mildly rescued by the *STM* transgene at 9 dag (**left**) and 16 dag (**right**). Cotyledons (co) are partially fused on one side. Note that the shoot is only visible at 16 dag (arrowhead). (**F**,**G**) Shoot apices in cleared seedlings (11 dag) of wild type (**F**) and *cuc1‐5 cuc2‐3* (**G**). Wild type develops the shoot meristem (open circle) and leaf primordia (asterisks), whereas *cuc1‐5 cuc2‐3* lacks these structures at the corresponding postion (arrow). (**H**) Shoot apices in cleared seedlings (11 dag) of *cuc1‐ 5 cuc2‐3* rescued by *STM*, showing variable phenotypes: no visible shoot meristem and leaf primordia (**left**, arrow), small undifferentiated tissue (**middle**, closed circle), and shoot meristem with leaf primordia (**right**, open circle, and asterisks). (**I**–**L**) *cuc1‐5 cuc2‐3* seedlings (9 dag) rescued by combined transgenes: *STM* and *LAS* (**I**,**J**), *STM* and *BLR* (**K**), and *STM* and *KNAT6* (**L**). Arrowheads represent emerging shoot. co, cotyledon. Bars = 50 μm (**B**,**F**–**H**); 1 mm (**C**–**E**,**I**–**L**). **Figure 2.** Rescue of *cuc1 cuc2* phenotype by candidate downstream genes. (**A**) Schematic diagram of rescue experiments. *CUC2* promoter (*ProCUC2*), cDNA of downstream gene (*DG*), and nos terminator (*ter*). (**B**) *CUC2* promoter activity detected by β-glucuronidase (GUS). Longitudinal views of GUS-stained wild-type Col (**top**) and *cuc1-5 cuc2-3* (**bottom**) embryos in three different optical sections, showing expression in the boundary region. (**C**,**D**) Seedlings of wild-type Col (**C**) and *cuc1-5 cuc2-3* (**D**), 7 days after germination (dag). Wild type has two separated cotyledons (co) and develops a shoot between them (arrowhead), whereas *cuc1-5 cuc2-3* has cotyledons (co) fused along their margins. (**E**) A *cuc1-5 cuc2-3* seedling mildly rescued by the *STM* transgene at 9 dag (**left**) and 16 dag (**right**). Cotyledons (co) are partially fused on one side. Note that the shoot is only visible at 16 dag (arrowhead). (**F**,**G**) Shoot apices in cleared seedlings (11 dag) of wild type (**F**) and *cuc1-5 cuc2-3* (**G**). Wild typedevelops the shoot meristem (open circle) and leaf primordia (asterisks), whereas *cuc1-5 cuc2-3* lacks these structures at the corresponding postion (arrow). (**H**) Shoot apices in cleared seedlings (11 dag) of *cuc1- 5 cuc2-3* rescued by *STM*, showing variable phenotypes: no visible shoot meristem and leaf primordia (**left**, arrow), small undifferentiated tissue (**middle**, closed circle), and shoot meristem with leaf primordia (**right**, open circle, and asterisks). (**I**–**L**) *cuc1-5 cuc2-3* seedlings (9 dag) rescued by combined transgenes: *STM* and *LAS* (**I**,**J**), *STM* and *BLR* (**K**), and *STM* and *KNAT6* (**L**). Arrowheads represent emerging shoot. co, cotyledon. Bars = 50 µm (**B**,**F**–**H**); 1 mm (**C**–**E**,**I**–**L**).

The rescue of the *cuc1 cuc2* mutant phenotype by the *STM* or *LAS* transgene alone was only mild and partial, thus we next tested their combined activities. In the F2 generation of the cross between the lines with the *STM* and *LAS* transgenes, only plants carrying both showed a rescued phenotype (Table 2). These rescued plants showed either partial (Figure 2J) or complete separation of cotyledons


**Table 1.** Rescue of *cuc1 cuc2* seedling phenotype by downstream gene expression under the control of *CUC2* regulatory sequence.

<sup>a</sup> Cotyledons were fused along both sides with only a small split at their tips and no shoot was formed. <sup>b</sup> Cotyledons were fused along one or both sides with more than half of their margins split. No shoot was formed or a shoot became visible only after 9 dag. <sup>c</sup> Cotyledons were completely separated and a shoot was visible by 9 dag. <sup>d</sup> One seedling had flat and round green tissue on top of the hypocotyl instead of a cup-shaped cotyledon. This is classified as "No rescue", as neither shoot formation nor cotyledon separation occurred.



<sup>a</sup> Cotyledons were fused along both sides with only a small split at their tips and no shoot was formed. <sup>b</sup> Cotyledons were fused along one or both sides with more than half of their margins split. No shoot was formed or a shoot became visible only after 9 dag. <sup>c</sup> Cotyledons were completely separated and a shoot was visible by 9 dag. <sup>d</sup> Different letters indicate statistically significant differences (*p* < 0.01, Fisher's exact test with Holm multiple testing correction).


**Table 3.** Effect of combined expression of *STM* and other downstream genes on the phenotype of *cuc1 cuc2*.

<sup>a</sup> Cotyledons were fused along both sides with only a small split at their tips and no shoot was formed. <sup>b</sup> Cotyledons were fused along one or both sides with more than half of their margins split. No shoot was formed or a shoot became visible only after 9 dag. <sup>c</sup> Cotyledons were completely separated and a shoot became visible by 9 dag. Asterisks indicate significant differences in the ratio of the phenotypes of plants carrying the transgene B alone and those carrying both transgenes A and B (\* *p* < 0.05, \*\* *p* < 0.01; Fisher's exact test).

#### *2.3. Combined Loss of Function of STM, LAS, BLR, and KNAT6 Severely Impairs Shoot Meristem Formation and Cotyledon Separation 2.3. Combined Loss of Function of STM, LAS, BLR, and KNAT6 Severely Impairs Shoot Meristem Formation and Cotyledon Separation*

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 7 of 16

We then tested the combined effect of the loss-of-function mutations in the four genes (*STM*, *LAS*, *BLR*, and *KNAT6*) in young seedlings. Among the single mutants of these genes, only *stm* shows defects in shoot development with complete penetrance. In the case of the strong allele *stm-1C* [29], the mutant typically stops leaf formation after producing the first two leaves (Figure 3A–D). It has been reported that mutations in *BLR* or *KNAT6* alone do not cause visible phenotypes in seedlings, but enhance the defects of *stm* mutants [19,30]. Consistent with this, we found that the *stm-1C blr* double mutant produced fewer leaves than *stm-1C* (Figure 3A,C,D) and showed delayed first-leaf growth (Figure 3C,E). In addition, *stm-1C blr* showed a higher frequency of cotyledon fusion than *stm-1C* (Table 4). The *stm-1C knat6* double mutant showed strong cotyledon fusion and the lack of leaf production and both phenotypes were observed with complete penetrance (Figure 3F). These results confirmed the previous results that *BLR* and *KNAT6* are required for shoot meristem formation and cotyledon separation. We then tested the combined effect of the loss‐of‐function mutations in the four genes (*STM*, *LAS*, *BLR*, and *KNAT6*) in young seedlings. Among the single mutants of these genes, only *stm* shows defects in shoot development with complete penetrance. In the case of the strong allele *stm‐1C* [29], the mutant typically stops leaf formation after producing the first two leaves (Figure 3A–D). It has been reported that mutations in *BLR* or *KNAT6* alone do not cause visible phenotypes in seedlings, but enhance the defects of *stm* mutants [19,30]. Consistent with this, we found that the *stm‐1C blr* double mutant produced fewer leaves than *stm‐1C* (Figure 3A,C,D) and showed delayed first‐leaf growth (Figure 3C,E). In addition, *stm‐1C blr* showed a higher frequency of cotyledon fusion than *stm‐1C* (Table4). The *stm‐1C knat6* double mutant showed strong cotyledon fusion and the lack of leaf production and both phenotypes were observed with complete penetrance (Figure 3F). These results confirmed the previous results that *BLR* and *KNAT6* are required for shoot meristem formation and cotyledon separation.

**Figure 3.** Genetic interactions of *stm, las, blr*, and *knat6* mutants. (**A**) The *las* mutation enhances the shoot production phenotype of *stm‐1C* and *stm‐1C blr* mutants. Shoot apex at 9 dag. Asterisks indicate developing leaves. (**B**) Examples of shoot phenotype measurements in the wild‐type Col (**left**) and *stm‐1C* (**right**). (**C**) Change in leaf number (**left**) and first leaf width (**right**). (**D**,**E**) Box plots of leaf number (**D**) and first leaf width (**E**) at 11 dag. (**F**) *las* enhances the cotyledon fusion phenotype of *stm‐ 1C knat6* and *stm‐1C knat6 blr* mutants. Seedlings at 7 dag. (**G**) Quantification of cotyledon fusion. (**H**) Box plot showing extent of cotyledon fusion at 7 dag in each genotype. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Steel–Dwass method for **D**; *p* < 0.05, Tukey– Kramer method for **E** and **H**). Sample size is 24, 23, 17, and 23 for **C** (**left**) and **D**; 24, 12, 13, and 2 for **C** (**right**) and **E**; and 11, 23, 12, and 19 for **H**. Bars in **A** and **E**, 2 mm. **Figure 3.** Genetic interactions of *stm, las, blr*, and *knat6* mutants. (**A**) The *las* mutation enhances the shoot production phenotype of *stm-1C* and *stm-1C blr* mutants. Shoot apex at 9 dag. Asterisks indicate developing leaves. (**B**) Examples of shoot phenotype measurements in the wild-type Col (**left**) and *stm-1C* (**right**). (**C**) Change in leaf number (**left**) and first leaf width (**right**). (**D**,**E**) Box plots of leaf number (**D**) and first leaf width (**E**) at 11 dag. (**F**) *las* enhances the cotyledon fusion phenotype of *stm-1C knat6* and *stm-1C knat6 blr* mutants. Seedlings at 7 dag. (**G**) Quantification of cotyledon fusion. (**H**) Box plot showing extent of cotyledon fusion at 7 dag in each genotype. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Steel–Dwass method for (**D**); *p* < 0.05, Tukey–Kramer method for (**E**) and (**H**)). Sample size is 24, 23, 17, and 23 for (**C**) (**left**) and (**D**); 24, 12, 13, and 2 for (**C**) (**right**) and (**E**); and 11, 23, 12, and 19 for (**H**). Bars in (**A**) and (**E**), 2 mm.


**Table 4.** Frequency of cotyledon fusion in *stm* mutant combined with *blr* and *las* mutants.

\* Different letters indicate statistically significant differences (*p* < 0.01, Fisher's exact test with Holm multiple testing correction).

Similar to *blr* and *knat6* single mutants, young seedlings of the *las* single mutant reportedly show a normal appearance, except for a very small proportion of plants with fused cotyledons (0.26%) [6]. However, we found that the *las* mutation enhanced defects in shoot meristem activity when combined with the *stm-1C* single or*stm-1C blr* double mutant (Figure 3A–E). In addition, the *las* mutation enhanced the cotyledon fusion phenotype of both *stm-1C knat6* and *stm-1C blr knat6* mutants (Figure 3F–H). These results show that the *LAS* gene contributes to embryonic shoot meristem formation and cotyledon separation independently of *STM*, *KNAT6*, and *BLR*. The *blr knat6 las* triple mutant was phenotypically normal. Together, our results demonstrate that the four transcription factor-encoding genes *STM*, *KNAT6*, *BLR*, and *LAS* play key roles for embryonic shoot meristem formation and cotyledon separation downstream of CUC1 and CUC2.

#### *2.4. The KLUH Gene Restricts the Embryonic Shoot Meristem and Counteracts STM*

Among the downstream target genes, *KLU*/*CYP78A5* encoding a cytochrome P450 enzyme of the *CYP78A* family plays postembryonic roles in shoot organ size and organ production rate [25,26], raising the possibility that this gene also affects shoot meristem activity. Moreover, the mutation in rice *PLA1*, a member of the same family, causes an enlarged shoot meristem phenotype [31,32]. Indeed, we found that the two independent *klu* insertion alleles (*klu-019348* and *klu-4*) were associated with precocious leaf initiation in young seedlings (Figure 4A–F), suggesting enhanced shoot meristem activity. In addition, the width of the shoot meristem was significantly greater in *klu* than in the wild type (Figure 4G). Furthermore, embryos of the *klu* mutants displayed an enlarged cotyledon boundary region (Figure 4H–J) and this phenotype was associated with an enlarged expression domain of the shoot stem cell marker *CLV3::GUS* [33] (Figure 4K,L). These results indicate that the *KLU* gene is required for restricting shoot meristem size and activity during embryogenesis and postembryonic development.

The identification of *KLU* as a downstream gene of *CUC1* and *CUC2* was unexpected because *KLU* negatively affects shoot meristem activity, whereas the *CUC* genes are positive regulators of shoot meristem formation. To further investigate the relationship between *KLU* and the *CUC* genes, we crossed the *klu* mutant with the *cuc1 cuc2* double mutant and examined their genetic interactions. In single mutants of *klu*, *cuc1*, and *cuc2*, all seedlings produced a functional shoot as the wild type, and except for a small fraction of *cuc1* mutants with partially fused cotyledons, their cotyledons were completely separated (Table 5; Figure 5A). Similarly, seedlings of *klu cuc1* and *klu cuc2* double mutants all produced a fully functional shoot and most of them developed completely separated cotyledons, whereas small fractions had partially fused cotyledons (Table 5; Figure 5B). Importantly, the *klu cuc1 cuc2* triple mutant was indistinguishable from *cuc1 cuc2* double mutants in that they formed strongly fused cup-shaped cotyledons and lacked a shoot meristem (Table 5; Figure 2D,G and Figure 5C–E). These results show that the *cuc1 cuc2* double mutations are epistatic to *klu* and indicate that the *KLU* gene can affect shoot meristem activity only when the functional *CUC1* and *CUC2* genes are present.

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 9 of 16

**Figure 4.** Mutations in *KLU* cause shoot meristem enlargement and precocious leaf formation. (**A**–**F**) Shoot apex of wild type (**A**,**D**), *klu‐019348* (**B**,**E**), and *klu‐4* (**C**,**F**) in cleared seedlings at 3 dag. (**D**–**F**) are close‐up views of (**A**–**C**), respectively. Optical sections are photographed at a position slightly off‐ center to reveal the precociously formed third‐leaf primordia in the *klu* mutants (arrows). lp, the first two leaf primordia; m, shoot meristem; st, stipule. (**G**) Shoot meristem width of wild type and two *klu* mutant alleles at 3 dag. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Tukey–Kramer method). Sample size is 34, 28, and 42 for **G**. (**H**–**J**) Torpedo‐stage embryos of wild type (**H**), *klu‐019348* (**I**), and *klu‐4* (**J**). Arrowheads indicate the position of the cotyledon boundary region. (**K**,**L**) *CLV3::GUS* expression in wild‐type (**K**) and *klu‐4* (**L**) torpedo‐stage embryos. Scale bar = 50 μm. **Figure 4.** Mutations in *KLU* cause shoot meristem enlargement and precocious leaf formation. (**A**–**F**) Shoot apex of wild type (**A**,**D**), *klu-019348* (**B**,**E**), and *klu-4* (**C**,**F**) in cleared seedlings at 3 dag. (**D**–**F**) are close-up views of (**A**–**C**), respectively. Optical sections are photographed at a position slightly off-center to reveal the precociously formed third-leaf primordia in the *klu* mutants (arrows). lp, the first two leaf primordia; m, shoot meristem; st, stipule. (**G**) Shoot meristem width of wild type and two *klu* mutant alleles at 3 dag. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Tukey–Kramer method). Sample size is 34, 28, and 42 for **G**. (**H**–**J**) Torpedo-stage embryos of wild type (**H**), *klu-019348* (**I**), and *klu-4* (**J**). Arrowheads indicate the position of the cotyledon boundary region. (**K**,**L**) *CLV3::GUS* expression in wild-type (**K**) and *klu-4* (**L**) torpedo-stage embryos. Scale bar = 50 µm.


**Table 5.** Genetic interactions among *klu‐4*, *cuc1‐5*, and *cuc2‐3*. **Table 5.** Genetic interactions among *klu-4*, *cuc1-5*, and *cuc2-3*.

*klu cuc2* 97.2 2.8 0 217 *cuc1 cuc2* 0 0 100 50 *klu cuc1 cuc2* 0 0 100 54 <sup>a</sup> Cotyledons were completely separated and a shoot was formed immediately after germination. <sup>b</sup> Cotyledons were partially fused and a shoot was formed immediately after germination. <sup>c</sup> Cotyledons were strongly fused, forming a cup shape, and no shoot was formed after two weeks of observation. \* A very small fraction (1.7%) had three cotyledons instead of two.

<sup>a</sup> Cotyledons were completely separated and a shoot was formed immediately after germination. <sup>b</sup> Cotyledons were partially fused and a shoot was formed immediately after germination. <sup>c</sup> Cotyledons were strongly fused, forming a cup shape, and no shoot was formed after two weeks of observation.

\* A very small fraction (1.7%) had three cotyledons instead of two.

Next, we examined the genetic interaction of *KLU* with *STM*, which represents a functionally important class of *CUC* downstream genes that positively affects shoot meristem activity. Young seedlings of the *stm klu* double mutant produced more leaves and had an enlarged shoot meristem

indicate that the *KLU* gene activity counteracts that of *STM* in postembryonic shoot development.

**Figure 5.** *KLU* acts downstream of *CUC1* and *CUC2*, and counteracts *STM* in the regulation shoot meristem activity. (**A**–**C**) Seven‐day‐old seedlings of *klu‐4* (**A**), *klu‐4 cuc2‐3* (**B**), and *klu‐4 cuc1‐5 cuc2‐ 3* (**C**). (**D**,**E**) Shoot apices of the progeny from *klu‐4 cuc1‐5 cuc2‐3*/*+* parent plants. A sib seedling with the shoot meristem (**D**) and a *klu cuc1 cuc2* seedling without it (**E**). (**F**,**G**) Nine‐day‐old seedlings of the strong *stm* allele *stm‐1C* (**F**) and *stm‐1C klu‐4* (**G**). (**H**,**I**) Shoot apex of the weak *stm* allele *bum1‐3* (**H**) and *bum1‐3 klu‐4* (**I**) at 4 dag. (**J**) Leaf number at 11 dag. (**K**) Shoot meristem width at 4 dag. (**L**,**M**) Twenty‐nine‐day‐old plants of *stm‐1C* (**L**) and *stm‐1C klu‐4* (**M**). Four plants are grown in each pot. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Steel–Dwass method for **J**; *p* < 0.05, Tukey–Kramer method for **K**). Sample size is 55, 30, 24, and 14 for **J**; 14, 18, 15, and 9 for **K**. Arrowheads indicate the shoot meristem. lp, leaf primordia. lp, leaf primordia. Scale bar, 1 mm for **A** to **C**, **F**, and **G**; 50 μm for **D**, **E**, **H**, and **I**; 10 mm for **L** and **M**. **Figure 5.** *KLU* acts downstream of *CUC1* and *CUC2*, and counteracts *STM* in the regulation shoot meristem activity. (**A**–**C**) Seven-day-old seedlings of *klu-4* (**A**), *klu-4 cuc2-3* (**B**), and *klu-4 cuc1-5 cuc2-3* (**C**). (**D**,**E**) Shoot apices of the progeny from *klu-4 cuc1-5 cuc2-3*/+ parent plants. A sib seedling with the shoot meristem (**D**) and a *klu cuc1 cuc2* seedling without it (**E**). (**F**,**G**) Nine-day-old seedlings of the strong *stm* allele *stm-1C* (**F**) and *stm-1C klu-4* (**G**). (**H**,**I**) Shoot apex of the weak *stm* allele *bum1-3* (**H**) and *bum1-3 klu-4* (**I**) at 4 dag. (**J**) Leaf number at 11 dag. (**K**) Shoot meristem width at 4 dag. (**L**,**M**) Twenty-nine-day-old plants of*stm-1C* (**L**) and *stm-1C klu-4* (**M**). Four plants are grown in each pot. Different letters in box plots indicate statistically significant differences (*p* < 0.01, Steel–Dwass method for (**J**); *p* < 0.05, Tukey–Kramer method for (**K**)). Sample size is 55, 30, 24, and 14 for (**J**); 14, 18, 15, and 9 for (**K**). Arrowheads indicate the shoot meristem. lp, leaf primordia. lp, leaf primordia. Scale bar, 1 mm for (**A**) to (**C**,**F**,**G**); 50 µm for (**D**,**E**,**H**,**I**); 10 mm for (**L**) and (**M**).

To clarify the mechanism by which *CUC1* and *CUC2* regulate *KLU* gene expression, we examined the expression of reporter genes containing *cis*‐regulatory sequences of *KLU*. A reporter construct that carried regions 2 kb upstream and 0.6 kb downstream (*Pro2kb*) showed activities in the cotyledon boundary region, cotyledon margins, and root pole (Figure 6A,B). This expression pattern resembled that of *KLU* mRNA detected by in situ hybridization, except that the reporter activity in cotyledons was broader and that in the root pole was detected for a prolonged time (compare Figure 6B with Figure 1A and Figure S1A), indicating that the *Pro2kb* reporter contained a set of *cis*‐ Next, we examined the genetic interaction of *KLU* with *STM*, which represents a functionally important class of *CUC* downstream genes that positively affects shoot meristem activity. Young seedlings of the *stm klu* double mutant produced more leaves and had an enlarged shoot meristem compared with *stm* single mutants (Figure 5F–K). Moreover, whereas the strong *stm-1C* mutant allele typically arrested shoot growth after producing a few leaves (Figure 5L), the *stm-1C klu-4* double mutant showed prolonged vegetative shoot growth (Figure 5M). Taken together, these results indicate that the *KLU* gene activity counteracts that of *STM* in postembryonic shoot development.

regulatory elements sufficient for driving the native expression pattern of the gene at least in the cotyledon boundary region. When this construct was introduced into the *cuc1 cuc2* double‐mutant background, its expression disappeared specifically in the cotyledon boundary region (Figure 6C), which corresponds to the region where the *CUC* genes act in normal development. Moreover, deletion of the region between 1 and 2 kb upstream of the gene resulted in the loss of expression in the cotyledon boundary region as well as in cotyledon margins (Figure 6A,D; *Pro1kb*). These results indicate that this 1 kb region contains *cis*‐regulatory elements required for CUC1‐ and CUC2‐ dependent transcription in the shoot apex. To clarify the mechanism by which *CUC1* and *CUC2* regulate *KLU* gene expression, we examined the expression of reporter genes containing *cis*-regulatory sequences of *KLU*. A reporter construct that carried regions 2 kb upstream and 0.6 kb downstream (*Pro2kb*) showed activities in the cotyledon boundary region, cotyledon margins, and root pole (Figure 6A,B). This expression pattern resembled that of *KLU* mRNA detected by in situ hybridization, except that the reporter activity in cotyledons was broader and that in the root pole was detected for a prolonged time (compare Figure 6B with Figure 1A and Figure S1A), indicating that the *Pro2kb* reporter contained a set of *cis*-regulatory elements sufficient for driving the native expression pattern of the gene at least in the cotyledon boundary region. When this construct was introduced into the *cuc1 cuc2* double-mutant background, its expression

disappeared specifically in the cotyledon boundary region (Figure 6C), which corresponds to the region where the *CUC* genes act in normal development. Moreover, deletion of the region between 1 and 2 kb upstream of the gene resulted in the loss of expression in the cotyledon boundary region as well as in cotyledon margins (Figure 6A,D; *Pro1kb*). These results indicate that this 1 kb region contains *cis*-regulatory elements required for CUC1- and CUC2-dependent transcription in the shoot apex. *Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 11 of 16

**Figure 6.** *KLU* expression is regulated by *CUC1* and *CUC2* via a specific promoter region. (**A**) Schematic diagram of the *KLU* reporter genes. (**B**–**D**) Expression of *KLU* reporter genes in embryos. *Pro2kb* in wild type L*er* (**B**), *Pro2kb* in *cuc1‐1 cuc2‐1* (**C**), and *Pro1kb* in wild type L*er* (**D**). Arrows indicate the position of the cotyledon boundary region of wild type (**B**,**D**) and the corresponding region of *cuc1‐1 cuc2‐1* (**C**). Scale bar, 50 μm. **Figure 6.** *KLU* expression is regulated by *CUC1* and *CUC2* via a specific promoter region. (**A**) Schematic diagram of the *KLU* reporter genes. (**B**–**D**) Expression of *KLU* reporter genes in embryos. *Pro2kb* in wild type L*er* (**B**), *Pro2kb* in *cuc1-1 cuc2-1* (**C**), and *Pro1kb* in wild type L*er* (**D**). Arrows indicate the position of the cotyledon boundary region of wild type (**B**,**D**) and the corresponding region of *cuc1-1 cuc2-1* (**C**). Scale bar, 50 µm.

#### **3. Discussion 3. Discussion**

In this work, through functional analyses of genes acting downstream of *CUC1* and *CUC2*, we found that the combined activities of *STM* with *LAS*, *BLR*, and *KNAT6* were important for shoot meristem formation and cotyledon separation. The abilities of these genes to rescue the *cuc1 cuc2* mutant phenotypes when expressed under the boundary‐specific promoter, together with the strong shoot meristem and cotyledon phenotypes observed in the quadruple mutant, support a model in which the *CUC* genes promote shoot meristem formation and cotyledon separation mainly through the activation of these four genes. It was previously shown that *STM* is a direct transcriptional target of *CUC1* [28]*.* The experiments using the DEX‐inducible CUC1‐GR plants and CHX treatments in our current work indicate that *LAS* and *KNAT6* are additional direct targets of the CUC1 protein, whereas the regulation of *BLR* by CUC1 may be indirect. In this work, through functional analyses of genes acting downstream of *CUC1* and *CUC2*, we found that the combined activities of *STM* with *LAS*, *BLR*, and *KNAT6* were important for shoot meristem formation and cotyledon separation. The abilities of these genes to rescue the *cuc1 cuc2* mutant phenotypes when expressed under the boundary-specific promoter, together with the strong shoot meristem and cotyledon phenotypes observed in the quadruple mutant, support a model in which the *CUC* genes promote shoot meristem formation and cotyledon separation mainly through the activation of these four genes. It was previously shown that *STM* is a direct transcriptional target of *CUC1* [28]. The experiments using the DEX-inducible CUC1-GR plants and CHX treatments in our current work indicate that *LAS* and *KNAT6* are additional direct targets of the CUC1 protein, whereas the regulation of *BLR* by CUC1 may be indirect.

Our analysis highlights the importance of the activation of *STM* expression by the *CUC* genes in shoot meristem formation and cotyledon separation. *STM* encodes a KNOX transcription factor [34] whose activity is continuously required for shoot meristem maintenance in postembryonic development through regulating various aspects of shoot meristem properties including pluripotency, self‐maintenance, promotion of cell cycle, repression of differentiation, and hormone metabolism [28,35–37]. How *STM* promotes cotyledon separation is currently unknown, but its ability to repress growth and promote leaf dissection, when ectopically expressed, may be involved in the repression of growth at the cotyledon boundary [38]. Our analysis highlights the importance of the activation of *STM* expression by the *CUC* genes in shoot meristem formation and cotyledon separation. *STM* encodes a KNOX transcription factor [34] whose activity is continuously required for shoot meristem maintenance in postembryonic development through regulating various aspects of shoot meristem properties including pluripotency, self-maintenance, promotion of cell cycle, repression of differentiation, and hormone metabolism [28,35–37]. How *STM* promotes cotyledon separation is currently unknown, but its ability to repress growth and promote leaf dissection, when ectopically expressed, may be involved in the repression of growth at the cotyledon boundary [38].

The possible molecular mechanisms by which *STM* acts in concert with the rest of the four genes may vary. *KNAT6* encodes a KNOX protein closely related to STM, so its ability to enhance the The possible molecular mechanisms by which *STM* acts in concert with the rest of the four genes may vary. *KNAT6* encodes a KNOX protein closely related to STM, so its ability to enhance the rescuing

coexpression of *BLR* with *STM* provides a sufficient amount of BLR–STM complex to the nucleus, thereby efficiently promoting shoot meristem formation and cotyledon separation. In axillary shoot meristem formation, another BEL protein, ATH1, which is functionally redundant with BLR, forms a heterodimer with STM and directly promotes the transcription of *STM*, thus forming a self‐activation

activity of *STM* can simply be explained by functional redundancy [19,39]. By contrast, *BLR* encodes a BEL-class homeodomain protein, which physically interacts with STM [30], and nuclear localization of STM requires its interaction with BLR [40–42]. These results indicate that the coexpression of *BLR* with *STM* provides a sufficient amount of BLR–STM complex to the nucleus, thereby efficiently promoting shoot meristem formation and cotyledon separation. In axillary shoot meristem formation, another BEL protein, ATH1, which is functionally redundant with BLR, forms a heterodimer with STM and directly promotes the transcription of *STM*, thus forming a self-activation loop [43]. It is also possible that the formation of BLR–STM heterodimer during embryogenesis is critical for initiating the *STM* self-activation loop, allowing self-maintenance of the shoot meristem.

Our functional analyses also demonstrate that the *LAS* gene contributes to embryonic shoot meristem formation and cotyledon separation downstream of CUC1 and CUC2. *LAS* encodes a putative transcriptional regulator of the GRAS family and acts as a central hub in the gene regulatory network for axillary meristem formation downstream of CUC2 [23,24]. The precise mechanisms by which *LAS* regulates these processes remain elusive; however, the mutation in the *LAS* ortholog in tomato affects the levels of hormones involved in shoot meristem activity, such as auxin, gibberellin (GA), and cytokinin [44,45]. Recently, it has been shown that the LAS protein binds to the promoter of *GA2ox4*, which encodes a GA deactivation enzyme, and promotes its expression [46]. These results raise the possibility that *LAS* reduces the levels of GA in the boundary region to promote shoot meristem activity, thereby contributing independently of *STM*, *KNAT6*, and *BLR* to the process downstream of the *CUC* genes.

In contrast to the above four genes, which are positive regulators of shoot meristem activity, *KLU*/*CYP78A5* represents the functionally opposite class of downstream genes regulated by *CUC1* and *CUC2*, as shown by the enhanced shoot meristem size and activity in the *klu* mutant as well as its genetic interaction with the *stm* mutant. Our data thus indicate that, by activating the two classes of genes with opposing functions during embryogenesis, the *CUC* genes create an optimal microenvironment for the shoot meristem to maintain its appropriate size and to produce organs at an appropriate rate. It has been well established that the balance between the self-renewal of stem cells and differentiation of their progeny is critical for postembryonic shoot meristem and that this balance is maintained by the WUS–CLV3 feedback system, which is supported by multiple transcription factors and plant hormones [3,47]. Our results provide an additional level of regulation for the balancing mechanism of shoot stem cell maintenance. Detailed functional analysis of the *KLU* gene as well as its relationship to previously known stem cell regulators will further improve our understanding of shoot meristem regulation.

#### **4. Materials and Methods**

#### *4.1. Plant Materials*

*Arabidopsis thaliana* accessions Columbia (Col) and Landsberg *erecta* (L*er*) were used as the wild type. *CUC1-GR* was established in the L*er* background using the same construct as described previously [22]. *CLV3::GUS* was reported previously [33]. Expression analyses of genes downstream of *CUC* were carried out in *cuc1-1 cuc2-1* [4,5]. For rescue experiments, *cuc1-5 cuc2-3* was used [6]. Loss-of-function mutants of the downstream genes were as follows: *klu-019348* (Sallk\_019348) and *klu-4* for *KLU* [25]; *knat6-2* for *KNAT6* [19]; *las-101* for *LAS* [6]; *stm-1C* for *STM* [29]; and *pny-40126* for *BLR* [48]. Details of the mutants are described in Supplementary Table S4.

#### *4.2. Constructs*

For *ProCUC2:LAS*, we amplified the *LAS* coding sequence derived from Col with the PCR primers BamHI-LAS-F (50 -TATCTGGATCCATGCTTACTTCCTTCAAATC-30 ) and EcoRI-LAS-R (50 -TTCTCGAATTCTCATTTCCACGACGAAACGG-30 ) and placed it upstream of the *35S* terminator in a modified UAS cassette vector [49] using the EcoRI and BamHI sites. The BamHI-NotI fragment

containing *LAS* and the terminator was then placed downstream of the *CUC2* promoter of *pBS-gC2*, which contains a 5.9 kb fragment of the *CUC2* genomic sequence [50], yielding *ProCUC2:LAS BS*. The SalI-NotI fragment of *ProCUC2:LAS BS* was then inserted into pBIN50, a modified binary vector carrying a kanamycin resistance gene [29]. To obtain *ProCUC2:BLR*, cDNA derived from L*er* was amplified using the primers BLRcDNAfull-F (50 -TTTCCCATGGCTGATGCATA-30 ) and BLRcDNAfull-R (50 -TCAACCTACAAAATCATGTA-30 ), cloned into pCR™-Blunt II-TOPO (Invitrogen, [Waltham, MA, USA]), and the resulting EcoRI fragment was then placed upstream of the 35S terminator of the modified UAS cassette. Fusion with the *CUC2* promoter and transfer to a binary vector was carried out in the same manner as that for *ProCUC2:LAS*, except that pBIN60, a modified binary vector with a sulfadiazine resistance gene, was used. To obtain the other chimeric constructs of the CUC2 promoter and downstream genes, the 3.1 kb SalI-BglII fragment of the CUC2 promoter was blunt-ended and cloned into the blunt-ended HindIII site of the gateway destination vector pGWB1 carrying kanamycin and hygromycin resistance genes [51], resulting in *ProCUC2 pGWB1*. To generate entry clones, cDNAs were first amplified with gene-specific primers and then with the *attB* adaptor primers listed in Supplementary Table S5 and cloned into pDONR221 by BP reaction. The inserts were then transferred to *ProCUC2 pGWB1* by LR reaction. For *ProCUC2:GUS*, the *GUS* gene fragment was transferred from the entry clone pENTR-gus (Invitrogen [Waltham, MA, USA]) to *ProCUC2 pGWB1* via LR reaction. Plant transformation was carried out by the floral dip method [52]. The *Pro2kb* and *Pro1kb* reporter constructs of the *KLU* gene contain −2066 to +16 and −1061 to +16 sequences from the start codon, respectively, at their 50 end of *vYFPer*, an ER-localized version of Venus, and −85 to +338 sequence from the stop codon at their 30 end and are cloned in the binary vector pBarMAP [53]. Both constructs were transformed to L*er*.

#### *4.3. Rescue Experiments*

Double-homozygous *cuc1-5 cuc2-3* plants were seedling lethal and did not produce flowers, thus each construct for the rescue experiments was transformed to *cuc1-5 cuc2-3*/+ plants via the floral dip method. T<sup>1</sup> seeds were selected for drug resistance on Murashige–Skoog plates [54] and the phenotypes of the resistant seedlings were scored. Subsequently, double-homozygous plants were identified by PCR-based genotyping. For analyses of the combined effect of *STM* with *LAS*, one transgenic line containing the *ProCUC2:STM* transgene was maintained and T<sup>2</sup> plants with the *cuc1-5 cuc2-3*/+ genotype were crossed with two independent lines of *ProCUC2:LAS* with the *cuc1-5 cuc2-3*/+ genotype. The seedling phenotype was first scored in the F<sup>2</sup> generation and the genotype of the *CUC2* locus as well as the presence of each transgene was subsequently analyzed. The same *ProCUC2:STM* line was used for analysis of the combined effect of *STM* with the rest of the genes and was crossed with two independent transgenic lines for each gene. Scoring of the seedling phenotype and subsequent genotyping were carried out in the F<sup>1</sup> generation. To classify the phenotypes into the mild and strong categories, we examined the presence or absence of visible shoot under a binocular. The absence of shoot production was confirmed at 14 dag or later.

#### *4.4. Histological Analysis*

In situ hybridization was carried out as described previously [17]. As a template for the *UFO* probe, we used pDW221.1 [55]. For generating templates for other probes, gene-specific fragments were PCR-amplified using the primers listed in Supplementary Table S6 and cloned into pCR™-Blunt II-TOPO. For GUS detection, embryos were dissected and immediately stained in staining solution with 5 mM ferricyanide and ferrocyanide [56]. Embryos and seedling apices were visualized after clearing as described previously [17]..

#### *4.5. DEX Induction and qRT-PCR*

Induction of CUC1-GR with DEX and expression analysis by qRT-PCR were carried out as described previously [22]. Primers used for qRT-PCR are listed in Supplementary Table S7.

*Int. J. Mol. Sci.* **2020**, *21*, 5864

**Supplementary Materials:** Supplementary Materials can be found at http://www.mdpi.com/1422-0067/21/16/ 5864/s1. Table S1: List of 52 Genes Downregulated in cuc1 cuc2 Embryos in Microarray Experiments. Table S2: Quantitative RT-PCR of 51 Selected Genes in Heart Stage Embryos. Table S3: Quantitative RT-PCR of 51 Selected Genes in Bending Cotyledon Stage Embryos. Table S4: Mutants used in this analysis. Table S5: Gene-Specific Primers for Rescue Constructs. Table S6: Primers used for cloning probe templates for in situ hybridization. Table S7: Primers used for qRT-PCR experiments. Figure S1: Expression of candidate downstream genes in wild-type Ler and cuc1-1 cuc2-1 embryos. Figure S2: Examples of rescued and non-rescued plants. Text S1: Screening of additional genes regulated by CUC1 and CUC2.

**Author Contributions:** Conceptualization, M.A. and M.T.; Transcriptome analysis and qRT-PCR verification, Y.T.; In situ hybridization, M.A.; Expression analysis of *CUC1-GR*, S.S.; Rescue experiments and loss-of-function analysis of *CUC* downstream genes, H.O., M.K., M.A., S.T., M.R.K. and M.M.; Genetic analysis of *KLU* and *STM*, M.A.; Reporter analysis of *KLU*, R.I., M.A. and M.L.; Original draft preparation, M.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Ministry of Education, Culture, Sports, Science, and Technology of Japan (grant numbers 23012031, 2411400901, 18H04842, 20H04889 to M.A.); the Japan Society for the Promotion of Science (Grant No. 23370023, 16K07401 to M.A.); and Takeda Science Foundation (to M.A.).

**Acknowledgments:** We thank Detlef Weigel for pDW221.1, Chiyoko Machida for the *knat6-2* allele, Ayako Yamaguchi and *Arabidopsis* Biological Resource Center (ABRC) for the *klu* alleles, Rüdiger Simon for *CLV3::GUS*, and RIKEN Biological Resource Center (BRC) for RAFL cDNAs (pda12509 and pda11315). We also thank Mizuki Yamada for critical reading and Edanz (www.edanzediting.co.jp) for editing the English text of a draft of this manuscript. Finally, we thank Etsuko Habe, Seiko Ishihara, Naoko Fujihara, Eriko Tanaka, Maki Niidome, Mie Matsubara, and Kazuko Onga for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Synergistic Interaction of Phytohormones in Determining Leaf Angle in Crops**

**Xi Li 1,2, Pingfan Wu 1,2, Ying Lu 1,2, Shaoying Guo 1,2, Zhuojun Zhong 2,3, Rongxin Shen 2,3,\* and Qingjun Xie 1,2,\***


Received: 17 June 2020; Accepted: 15 July 2020; Published: 17 July 2020

**Abstract:** Leaf angle (LA), defined as the angle between the plant stem and leaf adaxial side of the blade, generally shapes the plant architecture into a loosen or dense structure, and thus influences the light interception and competition between neighboring plants in natural settings, ultimately contributing to the crop yield and productivity. It has been elucidated that brassinosteroid (BR) plays a dominant role in determining LA, and other phytohormones also positively or negatively participate in regulating LA. Accumulating evidences have revealed that these phytohormones interact with each other in modulating various biological processes. However, the comprehensive discussion of how the phytohormones and their interaction involved in shaping LA is relatively lack. Here, we intend to summarize the advances in the LA regulation mediated by the phytohormones and their crosstalk in different plant species, mainly in rice and maize, hopefully providing further insights into the genetic manipulation of LA trait in crop breeding and improvement in regarding to overcoming the challenge from the continuous demands for food under limited arable land area.

**Keywords:** Leaf angle; Phytohormones; crop yield; BR; Crosstalk

#### **1. Introduction**

To overcome the challenge of ever-increasing global demands for food, feedstock, and bioenergy products, breeders have been forced to select and breed cultivars with a key feature that can be planted at higher densities in order to increase grain yield with the limited availability of arable land area [1,2]. To this end, genetic improvement of crops with ideal plant architecture is considered as one of the most powerful strategies for addressing this issue. The key components of ideal plant architecture in crop generally include plant height, grain architecture, and leaf angle (LA) [3]. Since the 1960s, genetic engineering of decreasing plant height in crops has dramatically boosted the crop yield and definitely benefited millions of people worldwide, which is remarked as the "Green Revolution". The great achievement of Green Revolution is attributed to the utilization of two "Green Revolution" genes, mutant allelic *Semi-dwarf1* (*Sd1*) in rice, which encodes a key enzyme in the gibberellic pathway GA20ox2, and *Reduced height-1* (*Rht-1*) in wheat that encodes a key repressor in gibberellin signaling pathway called DELLA, are responsible for gibberellin metabolism in rice and wheat [4,5], respectively. However, excessive application of nitrogen fertilizer for ensuring the

yield and productivity of the semidwarf varieties has brought severe contamination on environment. Recently, a novel gibberellin-GIBBERELLIN INSENSITIVE DWARF1 (GID1)-NITROGEN-MEDIATED TILLER GROWTH RESPONSE 5 (NGR5) signaling pathway has been stated [6], which can be effectively used to boost yield of semi-dwarf crop by simultaneously enhancing the tiller number and nitrogen use efficiency (NUE) in next-generation Green Revolution. Besides plant height, the LA trait has also been substantially selected for high-yield varieties breeding, particularly in maize and rice, due to its vital role for optimal light interception and competition between neighboring plants in natural settings. and productivity of the semidwarf varieties has brought severe contamination on environment. Recently, a novel gibberellin-GIBBERELLIN INSENSITIVE DWARF1 (GID1)-NITROGEN-MEDIATED TILLER GROWTH RESPONSE 5 (NGR5) signaling pathway has been stated [6], which can be effectively used to boost yield of semi-dwarf crop by simultaneously enhancing the tiller number and nitrogen use efficiency (NUE) in next-generation Green Revolution. Besides plant height, the LA trait has also been substantially selected for high-yield varieties breeding, particularly in maize and rice, due to its vital role for optimal light interception and competition between

Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 2 of 19

A typical grass leaf is consisted of distal blade, proximal sheath, and a boundary called ligular region (also called lamina joint) that separates blade and sheath into distinct parts, all of which contain epidermal, ground, and vascular tissues that are continuous with each other but distinct in cell types and patterns [7,8]. LA, defined as the angle between the stem and adaxial side of the blade (Figure 1), is one of the most important architecture traits selected for crop yield improvement [9]. Crops with architecture of smaller LA and more upright leaves can facilitate higher plant density and enhance the photosynthetic efficiency, thus elevating yield [10–12]. The ligular region, an annular structure outside the joint of the leaf blade and leaf sheath, is the pivotal structure for determining LA in grasses. Accumulating evidences have been implicated that the formation of LA is mediated by the shape of the lamina joint, differences in cell numbers/size at adaxial/abaxial region and distinctive mechanical tissue strength [9,13–15], and thus any influence on them could alter the LA. For example, failure of longitudinal elongation of the adaxial cells in the lamina joint may result in erect leaves [16], whereas excessive expansion of the adaxial cells would increase the LA [17,18]. neighboring plants in natural settings. A typical grass leaf is consisted of distal blade, proximal sheath, and a boundary called ligular region (also called lamina joint) that separates blade and sheath into distinct parts, all of which contain epidermal, ground, and vascular tissues that are continuous with each other but distinct in cell types and patterns [7,8]. LA, defined as the angle between the stem and adaxial side of the blade (Figure 1), is one of the most important architecture traits selected for crop yield improvement [9]. Crops with architecture of smaller LA and more upright leaves can facilitate higher plant density and enhance the photosynthetic efficiency, thus elevating yield [10–12]. The ligular region, an annular structure outside the joint of the leaf blade and leaf sheath, is the pivotal structure for determining LA in grasses. Accumulating evidences have been implicated that the formation of LA is mediated by the shape of the lamina joint, differences in cell numbers/size at adaxial/abaxial region and distinctive mechanical tissue strength [9,13–15], and thus any influence on them could alter the LA. For example, failure of longitudinal elongation of the adaxial cells in the lamina joint may result in erect leaves [16], whereas excessive expansion of the adaxial cells would increase the LA [17,18].

Figure 1. Structural composition of leaf angle in rice and maize. Leaf angle is defined by the angle between the plant stem and leaf adaxial side of the blade, indicating as the α in the figure. This leaf morphology is affected by the development of ligule and auricle as well. **Figure 1.** Structural composition of leaf angle in rice and maize. Leaf angle is defined by the angle between the plant stem and leaf adaxial side of the blade, indicating as the α in the figure. This leaf morphology is affected by the development of ligule and auricle as well.

Given the rapid developments of plant functional genomics, a number of genes controlling LA have been cloned and the relevant regulatory network underlying the development of LA in grasses Given the rapid developments of plant functional genomics, a number of genes controlling LA have been cloned and the relevant regulatory network underlying the development of LA in grasses

has been well characterized. These studies revealed that phytohormones, such as brassinosteroids (BRs), auxin, and gibberellins (GAs), comprehensively participate in regulating LA by orchestrating the homeostais of their biosynthesis and the expression of signaling transcription factors (TFs). Though each phytohormone and TF has been found to contribute quite similar traits/phenotypes in term of LA, how they interplay with each other is still far beyond understood. This review is an attempt to highlight the regulation mechanism of LA and the genetic interactions among phytohormones in regulating LA formation with particular emphasis on maize and rice. Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 3 of 19 has been well characterized. These studies revealed that phytohormones, such as brassinosteroids (BRs), auxin, and gibberellins (GAs), comprehensively participate in regulating LA by orchestrating the homeostais of their biosynthesis and the expression of signaling transcription factors (TFs). Though each phytohormone and TF has been found to contribute quite similar traits/phenotypes in term of LA, how they interplay with each other is still far beyond understood. This review is an attempt to highlight the regulation mechanism of LA and the genetic interactions among phytohormones in regulating LA formation with particular emphasis on maize and rice.

#### **2. Regulation of Lamina Joint Bending by Brassinosteroid (BR)**

BRs are a group of steroid phytohormones involved in many important biological processes, and thus play a vital role in regulating several important agronomic traits, such as leaf angle, plant height and stress resistance. The regulatory pathways of BR biosynthesis, metabolism, and signal transduction have been well established in rice [19–21]. Since LA is a grass-species-specific trait, the role of BR in regulating LA has only been characterized in maize and rice rather than Arabidopsis, indicating that BR is a positive regulator of LA (Figure 2). 2. Regulation of Lamina Joint Bending by Brassinosteroid (BR) BRs are a group of steroid phytohormones involved in many important biological processes, and thus play a vital role in regulating several important agronomic traits, such as leaf angle, plant height and stress resistance. The regulatory pathways of BR biosynthesis, metabolism, and signal transduction have been well established in rice [19–21]. Since LA is a grass-species-specific trait, the role of BR in regulating LA has only been characterized in maize and rice rather than Arabidopsis, indicating that BR is a positive regulator of LA (Figure 2).

Up-regulation of the BR content within lamina joint region or enhanced BR signaling pathway by boosting the expression of BR related regulators resulted in increasing LA (Table 1). For instance, many researches in rice have elaborated that leaf inclination is closely associated with biosynthesis or signaling of BR [19,22]. Loss-of-function of BR biosynthetic genes, such as *Dwarf 2* (*D2)*, perturbed the endogenous level of BR, eventually resulting in erect leaf phenotype [11,23–26]. Besides, other BR biosynthetic regulators, such as the Cytochrome P450 family proteins, have also been documented to function in the development of LA. For example, *BR-deficient Dwarf1* (*OsBRD1*) is cytochrome P450 protein and encodes a key enzyme (BR C-6 oxidase) catalyzing BR biosynthesis. Disruption of *OsBRD1* causes pleiotropic effects, including severe dwarf phenotype, tortuous leaves, short panicles, small seeds, etc. [27,28]. Another two Cytochrome P450 proteins, OsDWARF4 and OsDWARF11, also catalyze the rate-limiting reaction (c-22 hydroxylation) of BR biosynthesis. *OsDWARF4* is highly expressed in leaf blade and root, which is inhibited by BR but increased in BR-insensitive or -deficient mutants, indicating there is a feedback regulation on *OsDWARF4* expression. Depletion of *OsDWARF4* caused mild phenotype with erect leaves but not any detrimental effect on the development of leaf, inflorescence, and seed relative to wild-type plant [11,24]. Distinct from *Osdwarf4* mutant, *Osdwarf11* mutant showed a much more severe phenotype, including dwarfism in plants, erection of leaves, pollen abortion, and small grains. Further investigation revealed that *OsDWARF4* and *OsDWARF11* function redundantly in BRs biosynthesis, but *OsDWARF11* performs a major role in BR biosynthesis pathway while *OsDWARF4* plays the complementary role [25], which explained the different effects of them in term of plant growth and development, as well as the LA. It is worthy to mention that the *Osdwarf4* mutant with erect leaf simultaneously promoted biomass production and higher yields than wild type at different planting densities, even without additional fertilization, indicating that this gene/allele is a potential candidate for sustainability increasing crop yield in limited land area [11]. Taken together, these studies have clearly illustrated that BR metabolism is responsible for the LA formation in crop.

In the BR signaling transduction pathway, BR is perceived by extracellular domain of BRASSINOSTEROID INSENSITIVE 1 (BRI1), a single transmembrane leucine-rich repeat receptor-like protein kinase (LRR-RLK), and its co-receptor BRI1-ASSOCIATED KINASE 1 (BAK1). In the absence of BR, BRI10 s kinase domain is deactivated by the negative factor BRI1 KINASE INHIBITOR 1 (BKI1); while BR is present, the BR compound binds to the extracellular domain of the BRI1 and BAK1. Subsequently, BRI1 phosphorylates BKI1, which results in the release of BKI1 from the plasma membrane, and then induce the phosphorylation of the kinase domain of BRI1 and BAK1, finally activating the initiation of BRI1-mediated signaling transduction [29]. The activation of BRI1 in turn phosphorylates downstream BR-SIGNALING KINASE1 (BSK1), CONSTITUTIVE DIFFERENTIAL GROWTH 1 (CDG1), and some of their homologs. Both BSK1 and CDG1/CDL1 phosphorylates BRI1-SUPPRESSOR 1 (BSU1) and subsequently activates BSU1. Activated BSU1 dephosphorylates and inactivates BRASSINOSTEROID INSENSITIVE 2 (BIN2) to release the suppression of BRASSINAZOLE-RESISTANT 1 (BZR1) and BRI1-EMS-SUPPRESSOR 1 (BES1), which are two key downstream transcription factors positively mediating BR responses. BZR1 and BES1 are rapidly dephosphorylated by Protein Phosphatase 2A (PP2A) family, leading to their nuclear accumulation and their regulation of thousands of BR-responsive genes expression [30]. Besides, in the absence of BR, the interaction of phosphorylated BZR1 and BES1 with 14-3-3 proteins, a group of conserved phosphopeptide-binding proteins, can lead to their cytoplasm retention and degradation, while in the presence of BR, dissociated BKI1 in cytoplasm can competitively bind to 14-3-3 proteins to inhibit this process [31]. Otherwise, the PPA2 dephosphorylates BRI1, BZR1 and BES1, relieving the binding of BZR1 and BES1 to 14-3-3 proteins, which regulates downstream genes in response to BR and induce diverse BR responses [32]. In addition to the BR biosynthetic genes mentioned above, BR signaling genes also play an essential role in regulating the development of LA in rice. For example, OsBRI1 is an ortholog of Arabidopsis BRI protein in rice. *OsBRI1* participates in regulating various aspects of growth and development processes in rice, including the bending

of lamina joint, the intercalary meristem formation, and the longitudinal elongation of internode cells. Therefore, depletion of *OsBRI1* significantly perturbed the internode elongation and bending of the lamina joint [33]. *OsBAK1* (a homologous gene to *Arabidopsis BAK1*) encodes SERK family receptor kinase protein and acts as co-receptor kinase for OsBRI1 to mediate BR signal transduction. Down-regulating the expression level of *OsBAK1* produced a rice variety with erect leaf and normal reproduction so that considered as a promising target for improving rice grain yield [34]. Nevertheless, suppression of *OsBZR1* expression by RNA interference (RNAi) in rice greatly altered the expression of BR-responsive genes and resulted in plant dwarfism and erect leaf, suggesting that *OsBZR1* plays an central downstream role in BR signaling [35]. The rice 14-3-3 proteins have been proven to interact with and retain OsBZR1 in cytoplasm instead of nucleus, ultimately leading to inhibiting the function of *OsBZR1*. However, BR treatment can dissociate their interactions and activate *OsBZR1*, in turn causing the erect leaf phenotype [36]. Similarly, a rice zinc finger transcription factor, *LEAF* and *TILLER ANGLE INCREASED CONTROLLER* (*OsLIC*)*,* also antagonized *OsBZR1* to repress BR signaling in rice. Overexpression of *OsLIC* resulted in erect leaves by eliminating BR response, indicating that *OsLIC1* negatively modulates leaf inclination in rice. *OsLIC* directly regulated the *INCREASED LEAF INCLINATION 1* (*ILI1*), a positive regulator in lamina inclination, to oppose the action of BZR1 [36]. In addition, recent studies have identified an APETALA2 (AP2)/ERF (ethylene-responsive element binding factor) family transcription factor, *Reduced Leaf Angle 1* (*RLA1*) that is identical to the *SMALL ORGAN SIZE 1* (*SMOS1*), as a positive regulator of BR signaling, which physically interacted with OsBZR1 to enhance its transcriptional activity to enlarge LA in rice [37,38]. *BU1* encodes a helix-loop-helix protein and acts as a novel BR positive regulator. Overexpressing *BU1* in rice led to enhanced bending of the lamina joint, increased grain size, and resistance to BR, while repression of *BU1* and its homologs in rice displays erect leaves [39]. Furthermore, a pair of antagonizing HLH/bHLH factors, *OsILI1* that is the homology of *Arabidopsis thaliana Paclobutrazol Resistance 1* (*PRE1*) and *ILI1 binding bHLH 1* (*OsIBH1*), functioned as downstream factors of *OsBZR1* to regulate leaf angle. *OsILI1* positively regulated BR-mediated cell elongation, whereas OsIBH1 directly interacted with OsILI1 and performed an opposite role [40]. In addition, another bHLH transcription factor, *BRASSINOSTEROID-RESPONSIVE LEAF ANGLE REGULATOR 1* (*OsBLR1*), has also been implicated to participate in LA regulation though the BR pathway in rice. Over-expressing *OsBLR1* simultaneously increased leaf angle, grain length and sensitivity to BR, whereas mutation of *OsBLR1* resulted in erect leaf and shorter grain [41]. Similarly, gain-of-function of *OsbHLH079* also caused wide LA, longer grain and hypersensitive to BR in rice, while *OsbHLH079*-RNAi lines showed opposite phenotype [42]. Collectively, these findings suggested that *bHLH* family transcription factors may be broadly involved in BR-mediated LA regulation. Previously, a plant-specific gene family transcription factor, *OsGRAS19* (*GA INSENSITIVE (GAI), REPRESSOR OF GAI (RGA),* and *SCARECROW (SCR) 19*) was suggested to be involved in regulating BR signaling, because the knockdown lines of *OsGRAS19* displayed less sensitivity to the 24-epi-brassinolid (BL) treatment as compared to WT in rice. Higher expression of *OsGRAS19* caused larger LA, narrow leaf and thin culm and panicle in rice [43]. Recently, a novel mutant of *OsGRAS19*, *D26*, has been identified in rice, which displayed typical BR-mediated phenotypes, including semidwarf, wider, and shorter leaf in addition to the erect leaf, as well as longer grain [44]. Taken together, these results further elaborated that the BR signaling is extensively involved in regulating LA in rice.

Until now, the molecular regulation of BR underlying LA in maize has been rarely investigated. Previous reports illustrated that suppression of *ZmBRI1* or knock-out of *Dwarf* and *Irregular Leaf 1* (*ZmDIL1*) displayed dwarf plant and erect leaves [45,46], suggesting a conserved function of BR for plant architecture in monocots. Recently, a SQUAMOSA-PROMOTER BINDING PROTEIN-LIKE (SPL) family protein LIGULELESS 1 (LG1), which is recognized as a conserved key factor in the formation of ligule and auricle in both rice and maize [47,48], confers an important role in controlling the leaf angle through regulating BR and auxin signaling pathway in maize and wheat. In maize, *ZmLG1* activated a B3-domain containing transcription factor *ZmRAVL1*, which sequentially activated *ZmBRD1* that was

also designated as *Upright Plant Architecture 1* (*UPA1*), eventually resulting in the alternation of leaf angle. *UPA2* functioned as a distant *cis* element to regulate *ZmRAVL1* expression, which was controlled by the *Drooping Leaf 1* (*DRL1*). This *UPA2*-*ZmRAV1*-*UPA1* module fine-tuned the BR pathway and finally determined the plant architecture and LA in maize [49]. In wheat, the *LG1* ortholog, *TaSPL8*, directly bound to the promoter of an auxin response factor *TaARF6* and the BR biosynthesis gene *CYP90D2* (*TaD2*), and subsequently activated their expressions, leading to enhancing lamina joint development [50]. Recently, *ZmILI1,* an ortholog of *OsILI1* that plays an important role in LA in rice, was found to directly bind to the *ZmLG1* and *CYP90D1* promoters to affect the BR biosynthesis and signal, eventually changing the LA [51]. Collectively, these studies supported a notion that components of BR signaling pathway play a dominant and conserved role in the formation of leaf angle in multiple crops and would be the potential targets for genetic improvement of crop architecture and yield.

#### **3. Regulation of Lamina Joint Bending by Indoleacetic Acid (IAA)**

Indoleacetic acid (IAA) is another crucial hormone, regulating leaf inclination mainly through patterning the adaxial/abaxial cell growth of leaves. In contrast to the positive role of BR in LA regulation, IAA functioned as a negative regulator since eliminating IAA content resulted in increased leaf inclination while increasing IAA content caused reduction of leaf inclination and upright leaves [52].

Auxin signal transduction is a sophisticated pathway, which is consisted of several key components, including the F-box TRANSPORT INHIBITOR RESPONSE 1/AUXIN SIGNALING F-BOX PROTEIN (TIR1/AFB) auxin co-receptors, the Auxin/INDOLE-3-ACETIC ACID (Aux/IAA) transcriptional repressors, and the AUXIN RESPONSE FACTOR (ARF) transcription factors and a ubiquitin-dependent protein degradation system. When IAA is deficient, Aux/IAA proteins form heterodimers with ARFs and block the function of ARFs to inhibit the auxin signal. Presence of auxin promotes the interaction between TIR1/AFB and Aux/IAA proteins, and then triggers a proteasome-mediated degradation of Aux/IAA by 26S Proteasome, eventually resulting in the release of ARF transcriptional activity. Subsequently, activation of ARF induces the changes of auxin-mediated gene expression pattern and growth responses [53–56]. This auxin signaling transduction model in rice has been implicated to associate with the regulation of lamina joint bending (Figure 1).

For instance, *FISH BONE* (*FIB*) encodes an orthologous of TAA protein, which plays a negative effect on leaf inclination. Loss-function of *FIB* caused a reduction of IAA level and altered auxin polar transport activity, thus producing small leaves with enlarged lamina joint angle [57]. *LEAF INCLINATION1* (*LC1*) encoding an IAA amino synthetase, termed as OsGH3.1 in rice, maintained auxin homeostasis by catalyzing excess IAA binding to various amino acids. A gain-of-function mutant *lc1-D* in rice displayed a reducing content of free IAA and increasing leaf angle due to the promotion of cell elongation in the adaxial surface of lamina joint [58]. In addition, two *OsMIR393a*/*b* targeted auxin receptors, *OsTIR1* and *AUXIN SIGNALING f-box 2* (*OsAFB2*), have also been proven to be involved in the regulation of leaf angle. Overexpression of *OsmiR393a*/*b* repressed the expression of *OsTIR1* and *OsAFB2*, which led to greater leaf angle in rice [59]. Further study demonstrated that BR promoted *OsTIR1* and *OsAFB2* to trigger the degradation of OsIAA1, which resulted in de-suppression of OsARF11 and OsARF19 proteins and consequently caused enlarge LA in rice [60]. Furthermore, a SPOC domain-containing transcription suppressor Leaf inclination 3 (LC3) regulated leaf inclination through interacting with a HIT zinc finger domain-containing protein, LC3-interacting protein 1 (LIP1). Meanwhile, LC3 could also directly bind to the promoter regions of *OsIAA12* and *OsGH3.2* to regulate auxin signal transduction and auxin homeostasis, finally influencing the formation of leaf inclination [61].

The maize auxin efflux carrier P-glycoprotein (ZmPGP1) was an adenosine triphosphate (ATP) binding cassette (ABC) transporter, which was involved in the polar transport of auxin and associated with the LA in maize [62]. Notably, multiple haplotypes of *ZmPGP1* were present in various landraces, teosintes, and inbred accessions [63], suggesting a domesticated selection and improvement due to the demand of higher density planting.

#### **4. Regulation of Lamina Joint Bending by Gibberellins (GA)**

GA signaling pathway has also been implicated to participate in the regulation of the LA through the BR signaling dependent manner. Previously, transcriptomic profiling identified the co-expression pattern between GA and BR associated genes, however, it was less understood whether these genes were coordinated to regulate the LA. Recently, several researches have verified that certain GA genes interact with BR genes to modulate the LA. For instance, knockdown of the rice *SPINDLY* (*OsSPY*) caused the eliminated expression of BR biosynthesis genes *D11*, *D2*, and *OsDWARF*/*BRD1*, but increased the *OsDWARF4* expression that functions downstream of the GA gene, *SLENDER RICE 1* (*SLR1*), ultimately resulting in the enhanced leaf angle [64]. A recent study found that the Arabidopsis O-fucosyltransferase SPY could mono-O-fucosylate the DELLA protein, leading to higher affinity of interaction between DELLA and BZR1 [65]. From this point of view, it is supposed that OsSPY also could activate the OsSLR1 encoding a DELLA protein to interact with OsBZR1, thereby affecting LA in rice. The rice GA-stimulated transcript gene (*OsGSR1*) induced by GA is another downstream gene of *SLR1*, and it directly binds to BRD2 to enhance the BR biosynthesis, which subsequently altered the leaf angle [64,66]. In addition to these GA genes, other regulators involved in GA pathway have also been identified to participate in LA regulation. *OsDCL3a* encodes a Dicer-like endoribonuclease involved in generating siRNA. Down-regulation of the *OsDCL3a* resulted in increased leaf angle by modulating the expression of GA and BR associated genes, including *OsGSR1* and *BRD1* [67]. Recently, *OsmiR396d* has been reported to regulate the LA in rice, which was promoted by OsBZR1 and then regulated the expression of BR responsive genes by targeting the *GROWTH REGULATING FACTOR 4* (*OsGRF4*) [68]. Alternatively, as another target of *OsmiR396d*, the *OsGRF6* participated in GA biosynthesis and signal transduction but was not directly involved in BR signaling only modulated the plant height rather than LA [68]. Taken together, it is supposed that GA likely regulates LA dependent on the BR pathway, and thus identification of much more GA components involved in response to BR would extend our knowledge to this issue.

#### **5. Regulation of Lamina Joint Bending by Crosstalk among Various Phytohormones**

Recent studies have shown that crosstalk between IAA and BR cooperatively regulates the development of leaf angle. OsIAA1 is a key negative regulator for auxin signal transduction, which is induced by auxin and BR. Over-expression of *OsIAA1 i*n rice resulted in dwarfism and increased leaf angle with decreased sensitivity to auxin treatment but increased sensitivity to BR treatment. *OsARF1* is an auxin signal positive regulator which is inhibited by *OsIAA1* in the absence of auxin. Further analysis showed that mutation of *OsARF1* reduced sensitivity to BR treatment, resembling the phenotype of *OsIAA1*-overexpression plants, which indicated that BR may interact with auxin through the OsIAA1-OsARF1 module to regulate LA in rice [69,70]. *OsARF19* acted as another coordinator of auxin and BR by positively regulating the expression of *OsGH3.5* to reduce the content of free IAA on the one hand and activating *OsBRI1* to stimulate BR signal cascades on the other hand, thus resulting in increased leaf angle [71]. *RLA1*/*SMOS1* functioned downstream of the auxin signaling pathway, and enhanced the transcriptional activity of *OsBZR1* by interacting with OsBZR1, suggesting that *RLA1*/*SMOS1* integrated BR and IAA signal pathway to regulate the development of leaf angle [37,38]. In summary, the above studies showed that the auxin antagonized BR by interfering both BR metabolism and signaling to negatively regulate the leaf angle in rice.

Additionally, the rice *d1* mutant with null function of an α subunit of G-protein (Gα), Rho GTPase activating protein 1 (RGA1), exhibited a dwarfism phenotype with erect leaves, and reduced sensitivity to GA and BR, indicating that *RGA1* was involved in both GA and BR responses [29,72]. Further studies showed that D1/RGA1 interacted with TUD1 to induce *BU1* expression, resulting in the increased leaf inclination [73,74]. BR can also act upstream of GA by modulating GA metabolism to regulate cell elongation. BR activated *OsBZR1* and induced the expression of *D18*/*GA3ox-2*, one of the GA biosynthetic genes, leading to increased bioactive GA levels in rice seedlings. In contrast, GA extensively inhibited BR biosynthesis and the BR response with a feedback mechanism, so that GA

leaf while ethylene leads to flat leaf.

treatment decreased the enlarged leaf angles in plants by attenuated BR biosynthesis or signaling [75]. These results showed that BR and GA were intertwined to regulate the leaf angle in rice. Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 8 of 19

#### **6. Other Phytohormones Involved in Regulation of Lamina Joint** Phytohormones such as ethylene, strigolactones (SLs), jasmonic acid (JA), and abscisic acid

Phytohormones such as ethylene, strigolactones (SLs), jasmonic acid (JA), and abscisic acid (ABA) were also involved in regulating leaf inclination of rice (Figure 3). An early report demonstrated the interaction of ethylene and BR to regulate leaf inclination of rice, but the underlying mechanism remaining to be elusive [18]. Currently, a study has validated that altering the C terminus of 1-Aminocyclopropane-1-carboxylate (ACC) synthase 7 (ZmACS7) responsible for ethylene biosynthesis in maize led to the stability of this protein and the accumulation of ACC and ethylene contents, as well as the up-regulation of ethylene responsive genes, which finally reduced plant height and increased leaf angle [76]. Similar to *ZmACS7*, overexpression of its closest paralog *ZmACS2* also resulted in flatter leaves [76], further suggesting that ethylene positively regulates LA. (ABA) were also involved in regulating leaf inclination of rice (Figure 3). An early report demonstrated the interaction of ethylene and BR to regulate leaf inclination of rice, but the underlying mechanism remaining to be elusive [18]. Currently, a study has validated that altering the C terminus of 1-Aminocyclopropane-1-carboxylate (ACC) synthase 7 (ZmACS7) responsible for ethylene biosynthesis in maize led to the stability of this protein and the accumulation of ACC and ethylene contents, as well as the up-regulation of ethylene responsive genes, which finally reduced plant height and increased leaf angle [76]. Similar to ZmACS7, overexpression of its closest paralog ZmACS2 also resulted in flatter leaves [76], further suggesting that ethylene positively regulates LA.

Figure 3. Crosstalk of phytohormones in determining leaf angle. Abscisic acid (ABA), CK, jasmonic acid (JA), and strigolactones (SLs) negatively regulate BR, thereby inhibiting leaf angle, whereas gibberellins (GA) positively coordinates BR to increase leaf angle. Application of auxin results in erect **Figure 3.** Crosstalk of phytohormones in determining leaf angle. Abscisic acid (ABA), CK, jasmonic acid (JA), and strigolactones (SLs) negatively regulate BR, thereby inhibiting leaf angle, whereas gibberellins (GA) positively coordinates BR to increase leaf angle. Application of auxin results in erect leaf while ethylene leads to flat leaf.

SLs negatively regulate leaf inclination at seedling stage [77]. Interestingly, recent report showed that SLs also mediated leaf inclination in response to nutrient deficiencies in rice [78]. Similar to SLs, JA also showed a negative role in leaf inclination. JA treatment decreased lamina joint inclination by repressing the expression of BR biosynthesis-related genes which thus decreased endogenous BRs levels. Besides, inactivation of a negative regulator of BR signaling, GSK3-like kinase, partly rescued the inhibited effect of JA on lamina joint inclination, indicating that JA may disturb both BR SLs negatively regulate leaf inclination at seedling stage [77]. Interestingly, recent report showed that SLs also mediated leaf inclination in response to nutrient deficiencies in rice [78]. Similar to SLs, JA also showed a negative role in leaf inclination. JA treatment decreased lamina joint inclination by repressing the expression of BR biosynthesis-related genes which thus decreased endogenous BRs levels. Besides, inactivation of a negative regulator of BR signaling, GSK3-like kinase, partly rescued the inhibited effect of JA on lamina joint inclination, indicating that JA may disturb both BR biosynthesis and BR signaling pathway to limit lamina joint inclination [79].

biosynthesis and BR signaling pathway to limit lamina joint inclination [79]. Previous research has revealed that BR antagonized with ABA in regulating seed germination and hypocotyl elongation in Arabidopsis. The study showed that the ABSCISIC ACID INSENSITIVE5 (ABI5) directly interacts with BRASSINOSTEROID INSENSITIVE2 (BIN2), and then Previous research has revealed that BR antagonized with ABA in regulating seed germination and hypocotyl elongation in Arabidopsis. The study showed that the ABSCISIC ACID INSENSITIVE5 (ABI5) directly interacts with BRASSINOSTEROID INSENSITIVE2 (BIN2), and then was phosphorylated and stabilized by BIN2 upon ABA treatment [80]. Additionally, the BES1 also physically interacted

was phosphorylated and stabilized by BIN2 upon ABA treatment [80]. Additionally, the BES1 also physically interacted with ABI5 to hinder the expression of ABI5-targeted EARLY METHIONINE-

Interestingly, the ABI1 and ABI2 can interact with and dephosphorylate BIN2 to attenuate the BR signaling in Arabidopsis [82], indicating a complicated crosstalk between ABA and BR. Recently, with ABI5 to hinder the expression of ABI5-targeted *EARLY METHIONINE-LABELED 1* (*EM1*) and *EM6* [81], eventually facilitating the seed germination in Arabidopsis. Interestingly, the ABI1 and ABI2 can interact with and dephosphorylate BIN2 to attenuate the BR signaling in Arabidopsis [82], indicating a complicated crosstalk between ABA and BR. Recently, there is research which further revealed that ABA also antagonized BR to regulate the lamina joint inclination in rice by targeting the BR biosynthesis gene *D11* and BR signaling genes *GSK2* and *DLT* [83], and it therefore raises an issue that whether the rice homologs of Arabidopsis *ABI1* and *ABI2* may also interfere the BR-mediated LA. Investigation of the LA phenotype of the *Osabi1* and *Osabi2*, as well as overexpression lines of these two genes, may provide the answer for this hypothesis. Another research preprinted in bioRxiv demonstrated that a transcriptional repressor *ZmCLA4* (the ortholog of *LAZY1* in rice and Arabidopsis) responsible for multiple phytohormone mediated pathways negatively regulated LA by altering mRNA accumulation. Further analysis showed that ZmCLA4 could directly bind to two key components of BR signaling *ZmBZR3* and *14-3-3*, and two important responsive transcription factors of ABA *ZmWRKY4* and *ZmWRKY72* respectively, thus mediating the crosstalk between BR and ABA in LA regulation [84]. In maize, a bHLH transcription factor *ZmIBH1-1* is a negative regulator of LA in maize. Transcriptome analysis suggested that the *ZmIBH1-1*-mediated LA in the leaf ligular region was highly correlated with cytokinin (CK), JA, and ethylene synthesis and signal transduction pathways associated genes, in particular the two CK responsive genes, GRMZM2G145280 (*BBC1*) and GRMZM2G149952 (*ZmAS1*) that were tightly correlated with cell division, implying that CK may modulate LA through the control of cell profile [85]. Notably, it has been demonstrated that CK indirectly interacted with BR through auxin pathway. For example, BR regulated the development of root primordia through increasing the *PIN* genes expression while the CK inhibited the root primordia by repressing *PIN* genes [86]. However, it was also found that CK was accumulated in the young seedling of wheat with BR treatment [87], whereas overexpression of the rice *Isopentyl Transferase* (*IPT*) driven by the promoter of stress- and maturation-inducible gene, *Senescence-associated Receptor Kinase* (*SARK*), resulted in up-regulation of BR genes, including *DWF4*, *BRI1*, and *BZR1* etc., ultimately leading to higher rice yield during water-stress [88]. These studies further suggest that CK interplays with BR in regulating plant growth and stress response. However, it is still unclear whether and how they cooperated in the LA regulation. As mentioned above, the *bHLH* family generally integrated BR signaling to regulate the LA, such as the *OsILI1* and *OsBLR1*, and thus we proposed that the CK may interplay with BR to regulate LA through the *bHLH*-*BBC1*/*AS1* module mediated BR signaling.

In summary, almost all of the regulators modulated leaf angle through the phytohormones-dependent manner (Table 1), however, whether there are unknown components independent on phytohormones, the pathway still remains elusive.


**Table 1.** Cloned genes associated with leaf angle in rice and maize.

#### **Table 1.** *Cont.*



#### **Table 1.** *Cont.*


**Table 1.** *Cont.*

<sup>1</sup> Role of the corresponding gene in regulating leaf angle in rice or maize. NA, none available.

#### **7. Future Perspectives**

Leaf angle directly influences the shape of plant architecture, consequentially affecting yield. To feed the ever-increasing global population, demand of higher crop yield has triggered the breeders to breed new cultivars that maintain sustainable productivity and yield by dense planting with limited arable lands. Therefore, how to genetically manipulating leaf angle has become one of the most important tasks to be tackled in crop genetic improvement. Until now, the regulatory mechanism underlying leaf angle has been extensively elucidated. However, a few issues are still beyond understanding, preventing the application of specific gene resource in term of crop improvement. To address these issues, the relationship among these genes still needs to be further uncovered, in particular the interaction network. On another hand, knockout or overexpression of these genes generally caused pleiotropic effects on plant growth and development in addition to the LA, such as plant dwarfism and smaller grain. Therefore, identification of interest single nucleotide polymorphisms (SNPs) and/or haplotypes of LA genes could not only be an efficient strategy to extent our knowledge about the relevant regulatory network, but also provide suitable alleles for marker selection breeding. The studies of *ZmLG1* provide an excellent case for bridging the basic research and application. *ZmLG1* was initially identified as a key factor for formation of ligule and auricle in maize and rice, and then regard to be an important player during rice domestication [47,90,91]. Further large-scale genetic analysis revealed that *ZmLG1* was the major QTL controlling leaf angle in maize [92–95]. Based on these studies, a recent study demonstrated that genetic manipulation of *ZmLG1* gene can significantly increase photosynthetic efficiency and maize yield at higher planting density [96]. Therefore, genome resequencing of larger accessions in crop indeed facilitates deciphering interest haplotypes of LA regulators. Alternatively, generation of novel alleles by CRISPR-mediated gene editing is also a powerful and efficient approach regarding this issue. Nevertheless, it would be fascinating to ask if there are novel regulators involved in LA regulation independent on the phytohormones pathways, which might only modulate LA without pleiotropic effects on other traits. On the other hand, it is also attractive to see whether and how LA coordinates or integrates with other agronomic traits in promoting yield, since yield consists of numerous components, such as plant height and tiller number. Furthermore, manipulating the spatial and temporal expression pattern of LA genes may be also important for timely shaping the plant architecture to achieve optimal photosynthesis, as well as eliminating shade avoidance and neighbor interference.

In conclusion, BR signaling pathway performs along with other plant hormones to form a complex signaling crosstalk to coordinate LA architecture under regular growth condition and in response to environmental stimuli. The well-established regulatory network for LA would provide vast promising

targets to be manipulated, no matter through traditional molecular marker-assisted breeding or the gene editing technology, for crop architecture improvement to pursue high production.

**Author Contributions:** Writing—original draft preparation, X.L., R.S. and Q.X.; writing—review and editing, X.L., P.W., Y.L., Z.Z., R.S. and Q.X.; visualization, X.L., P.W., Y.L., Z.Z. and S.G.; supervision, Q.X.; project administration, S.G. and Q.X.; funding acquisition, R.S. and Q.X. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Major Program of Guangdong Basic and Applied Research (2019B030302006), the National Natural Science Foundation of China (31971920) and the support from the "Top Young Scientist of the Pearl River Talent Plan" (No. 20170104) to Qingjun Xie, as well as the National Natural Science Foundation of China (31771739) and the Natural Science Foundation of Guangdong Province-Guangzhou City Collaborative Key Project (2019B1515120061) to Rongxin Shen.

**Acknowledgments:** We are grateful to the supports of experimental platform and funding from the Guangdong Provincial Key Laboratory of Plant Molecular Breeding (No. GPKLPMB201804). We apologize in advance to colleagues whose valuable work was not cited due to article length considerations.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*
