The Type-B Cytokinin Response Regulator ARR1 Inhibits Shoot Regeneration in an ARR12-Dependent Manner in Arabidopsis

One-sentence summary: ARR1, a type-B ARR is defined as an important inhibitor of in vitro shoot regeneration that modulates the expression of WUS and CLV3 in an ARR12-dependent manner, and directly activates IAA17. ABSTRACT Exogenous cytokinin is critical for in vitro shoot regeneration. Proteins involved in the cytokinin signal transduction pathway, including type-B ARABIDOPSIS RESPONSE REGULATORs (ARRs), participate in shoot regeneration in Arabidopsis thaliana . Some type-B ARRs (e.g., ARR1 and ARR12) promote shoot regeneration by directly activating WUSCHEL ( WUS ) expression; however, it is unclear how type-B ARRs inhibit shoot regeneration. Here, we show that ARR12 is a central enhancer of callus formation and shoot regeneration, whereas ARR1 is a strong inhibitor of this process that counteracts the positive effect of ARR12. ARR1 indirectly represses CLAVATA3 expression in an ARR12-dependent manner via competing with ARR12 for binding to the CLV3 promoter, which contributes to its ARR12-dependent inhibitory effect on callus formation and shoot regeneration. In parallel, ARR1 inhibits shoot regeneration through transcriptional activation of INDOLE-3-ACETIC ACID INDUCIBLE17 ( IAA17 ), an auxin response repressor gene, and the consequent indirect repression of WUS expression. Thus, type-B ARRs have diverse effects on callus formation and shoot regeneration. Our study reveals novel molecular pathways linking cytokinin signaling, the CLV3 regulator, and auxin signaling, and sheds light on the mechanism underlying cytokinin-regulated shoot regeneration.


INTRODUCTION
In contrast to most animal cells, plant cells typically maintain totipotency, allowing regeneration of a wide variety of explant types into whole plants through judicious in vitro culture conditions (Birnbaum and Sanchez Alvarado, 2008). In Arabidopsis (Arabidopsis thaliana), shoots normally regenerate from explants in a two-step process (Che et al., 2006); first, formation of a callus is induced by incubating root or hypocotyl explants in an auxin-rich callus-inducing medium (CIM); second, shoot regeneration is induced by transferring the callus to a cytokinin-rich shoot-inducing medium (SIM) (Che et al., 2006).
Exogenous cytokinin is critical for in vitro shoot regeneration: growth media with a high cytokinin/auxin ratio are required for shoot regeneration (Skoog and Miller, 1957;Valvekens et al., 1988;Che et al., 2006;Gordon et al., 2007). Response to cytokinins is mediated by type-B ARABIDOPSIS RESPONSE REGULATORs (ARRs), which bind to the cytokinin response element and activate the expression of cytokinin-responsive genes, including type-A ARRs, which are negative feedback regulators of cytokinin signaling (Mason et al., 2004;Muller and Sheen, 2007;To et al., 2007;Ramireddy et al., 2013). Type-B ARRs participate in a variety of developmental processes in a largely functionally redundant manner (Mason et al., 2005;Moubayidin et al., 2010;Zhao et al., 2010). For example, ARR1 4 Here, we demonstrate that ARR1, a type-B ARR, is an essential inhibitor of shoot regeneration during in vitro culture that modulates the expression of WUS and CLV3 in an ARR12-dependent manner and directly activates IAA17, thereby blocking shoot regeneration.

ARR1 is a Negative Regulator of Shoot Regeneration
To identify the role of ARR1 in shoot regeneration, we subjected root explants to regeneration assays via a two-step process. We evaluated the callus formation ability of root explants derived from the arr1-4 mutant, a transgenic line overexpressing ARR1 (AOE), and the arr1-4 mutant complemented with ARR1pro:ARR1-GFP (hereafter referred to as the RS [rescued] line) on CIM. After 21 d of culture, the arr1-4 mutant had formed ~70% more calli than the wild type, the RS line as many calli as the wild type, and the AOE line slightly fewer calli than the wild type ( Figures 1A and 1B). After 7 d of culture, callus formation was more extensive in arr1-4 than in the wild type ( Figures 1C and 1D), but not in the RS line, as expected (Figures 1C and 1D). These data indicate that ARR1 reduces the capacity for callus formation.
After incubation on SIM for 14 d, root explants obtained from arr1-4 plants regenerated more shoots than those obtained from wild-type plants (Figures 1E,1F and 1H). Root explants from the AOE line produced fewer shoots than those derived from the wild type, while root explants derived from the RS line had as many shoots as the wild type (Figures 1E,1F and 1H;Supplemental Figure 1). No shoots formed from any of the root explants within 7 d of transfer to SIM ( Figure 1G), although arr1-4 explants had already formed more dense cell masses than wild-type or RS explants, whereas explants obtained from the AOE line had formed fewer dense cell masses (Figures 1G and 1I). Thus, the abundance of ARR1 transcript in root explants negatively correlated with the number of regenerated shoots ( Figure 1E to 1I; Supplemental Figure 1), suggesting that ARR1 represses shoot regeneration.
Next, we transferred CIM-induced calli derived from root explants of arr1-4, AOE, and wild-type plants onto fresh SIM supplemented with various concentrations of N6-(2-isopentenyl) adenine (2-ip) and 0.15 mg/L Indole-3-acetic acid (IAA). More shoots regenerated from arr1-4 calli than from wild-type calli, except in the presence of 0.05 mg/L 2-ip, which blocked shoot regeneration in all the explants (Figures 1J to 1S). Fewer shoots regenerated from AOE calli than from wild-type calli at 2-ip concentrations of both 1.0 and 2.0 mg/L ( Figures 1N to 1Q). Hypocotyl explants of arr1-4, AOE, and wild-type lines responded similarly to the presence of various concentrations of 2-ip and 0.15 mg/L IAA (Supplemental Figure 2). These findings suggest that the repressive effect exerted by ARR1 on shoot regeneration is independent of the local cytokinin concentration. In addition, explants derived from the other strong allele arr1-3 generated more calli and shoots than the wild type, which is similar to the phenotype of arr1-4 (Supplemental Figure 3), further supporting the inhibitory effect of ARR1.

ARR1 Inhibits Shoot Regeneration in an ARR12-Dependent Manner
We previously showed that ARR12 promotes shoot regeneration during the two-step culture process (Dai et al., 2017). To understand the roles of different type-B ARR genes, we subjected arr1, arr10, arr11, arr12, and arr18 explants and the corresponding double mutants to regeneration assays.
After incubation on SIM for 14 d, arr1 explants contained three times as many regenerated shoots than the wild type, arr12 explants produced only one quarter as many regenerated shoots as the wild type, and arr10 contained slightly fewer regenerated shoots (Supplemental Figures 4A to 4F and 4Q). arr12containing double mutants all produced a similar or smaller number of regenerated shoots than the arr12 single mutant, regardless of which other gene was mutated (Supplemental Figures 4G to 4J and 4Q). The arr11 arr18 double mutant formed as many shoots as the wild type (Supplemental Figures 4K,4L and 4Q). In addition, the introduction of ARR12pro:ARR12 in the arr12 mutant complemented its shoot regeneration defect (Supplemental Figures 4M to 4Q). These results indicate that there is functional diversity among type-B ARRs in regulating shoot regeneration in vitro, with ARR12 acting as a central enhancer and ARR1 as a strong inhibitor of this process.
To explore the relationship between ARR1 and ARR12, we evaluated the callus and shoot regeneration abilities of the wild type and the arr1, arr12, and arr1 arr12 mutants. The arr1 single mutant generated more calli ( 6D). By contrast, expression of the 35Spro:ARR1 transgene in the arr1 arr12 double mutant background 6 resulted in more shoots than in explants derived from the arr1 arr12 double mutant (Supplemental Figures   6E to 6H). These data suggest that when cytokinin signal transduction is blocked (as in the arr1 arr12 double mutant), overexpression of ARR1 slightly enhances shoot regeneration; however, in the wild-type background, overexpression of ARR1 inhibits shoot regeneration, possibly by altering the function of ARR12 in explants.
The transcription levels of ARR12 in arr1 explants was slightly lower than that in wild-type explants after culture on CIM for 2 d and on SIM for 4 d (Supplemental Figures 5C and 5D), whereas it was slightly higher in arr1 and 35Spro:ARR1 seedlings than in the wild type (Supplemental Figure 5E). We observed no significant difference in ARR1 transcription levels between wild-type and arr12 explants or seedlings (Supplemental Figures 5F to 5H), whereas ARR1 transcript levels were slightly higher in 35Spro:ARR12 seedlings compared to the wild type (Supplemental Figure 5H). These results suggest that ARR1 and ARR12 slightly influence the transcription of the other gene during shoot regeneration.

Functions and Expression Patterns
To explore whether the different effects of ARR1 and ARR12 on shoot regeneration are determined by distinct protein functions or expression patterns of their encoding genes, we evaluated the shoot regeneration abilities of explants with different levels of ARR1 and ARR12. The ranked shoot regeneration ability of the explants was arr1 arr12 < arr12 < arr12 + ARR12pro:ARR1 < Col-0 < arr1 < arr1 +  (Kurepa et al., 2014). Thus, the phosphorylation status of ARR1 contributes to its inhibitory effect on shoot regeneration.

Localization and Accumulation Pattern of ARR1 and ARR12 During Shoot Regeneration
We next used ARR1pro:ARR1-GFP and ARR12pro:ARR12-GFP translational reporter lines to evaluate the localization of ARR1 and ARR12 during shoot regeneration. We detected ARR1 signal in the transition zone of primary roots (Supplemental Figure Figure 9D to 9I). After culturing root explants on CIM for 5 d, most of the ARR1 signal present was associated with stele cells and the cells surrounding the stele ( Figure 4C). The level of ARR1 signal increased in the inner layer of cells after culturing root explants on SIM for 2, 4, and 7 d ( Figures 4D to 4F). We observed no specific GFP signal at 7 d, when nascent SAMs formed ( Figure 4F). After 10 d on SIM, we detected ARR1 throughout the explant, including in the mature SAMs ( Figure 4G).
After culturing root explants on CIM for 0 and 2 d, we detected ARR12 signal specifically in stele cells (Supplemental Figures 10A to 10D). After culturing root explants on CIM for 5 d, most of the 8 ARR12-GFP signal was associated with stele cells and the cells around the stele, similar to the ARR1-GFP signal (Supplemental Figure 10E). After culturing root explants on SIM for 2, 4, and 7 d, the ARR12 signal level was higher than before transferring the explants to SIM, and the sites of GFP signal coincided with those of nascent SAMs (Supplemental Figures 10F to 10H). After 10 d on SIM, ARR12 localized specifically to newly formed SAMs (Supplemental Figure 10I).

Dependent Manner
We measured the expression level of various genes associated with shoot regeneration in wild-type (Col-0), arr1, and AOE explants. There was no significant difference in WOX5 expression between Col-0 and arr1 mutant explants on CIM (Supplemental Figures 11A and 11C), while there was a higher level of WOX5 expression in arr1 explants than in Col-0 after culture on SIM for 4, 7, and 10 d (Supplemental Figure 11A). SCR expression in arr1 explants was slightly higher in arr1 explants than in Col-0 after culture on CIM for 5 d and on SIM for 2 and 4 d (Supplemental Figure 11B and 11C). IAA3 and IAA14 expression was higher in arr1 explants than in Col-0 explants after culture on SIM for 10 d (Supplemental Figures 11D and 11E). ARR5, ARR7, and ARR15 expression in arr1 explants was lower than in Col-0 after culture on CIM for 5 d (Supplemental Figure 11F).
CLV3 is an established stem cell regulator in the SAM (Schoof et al., 2000). CLV3 expression in Col-0 explants gradually increased when cultured on CIM and peaked after culture on SIM for 4 d, with widespread distribution of CLV3 transcript in the explants (Figures 5A and 5D; Supplemental Figure   12A). CLV3 transcript distribution then decreased and concentrated in certain areas after culture on SIM for 7 d and was finally located in the newly formed SAMs after culture on SIM for 10 d ( Figure 5A; Supplemental Figure 12A). CLV3 transcript was significantly more abundant in root explants derived from the arr1 single mutant than in those from Col-0, both on CIM and SIM (Figures 5A,5B and 5D;Supplemental Figures 12A,12B,12E and 12F). In both arr12 and arr1 arr12 explants, CLV3 was transcribed at a much lower level than in 5C and 5D;Supplemental Figures 12A and 12D). Furthermore, the GUS signal of ProCLV3:GFP-GUS was weaker in AOE explants than in the wildtype background (Supplemental Figures 12A and 12C). These data suggest that ARR12 is a critical positive regulator of CLV3 expression, that ARR1 is a negative regulator of CLV3 transcription in explants 9 during CIM and SIM culture, and that the significant repression of CLV3 transcription by ARR1 requires the presence of ARR12.
Evaluation of the CLV3 promoter sequence revealed 29 copies of the core cytokinin response motif 5'-GAT(T/C) within the 2,091-bp upstream sequence of the start codon (Ramireddy et al., 2013) ( Figure   5E). A chromatin immunoprecipitation (ChIP) assay conducted on transgenic ARR1pro:ARR1-GFP and ARR12pro:ARR12-GFP explants at the nascent SAM formation stage (SIM 7 d) showed that ARR1 and ARR12 specifically bind to the core cytokinin response motif-containing region of the CLV3 promoter ( Figure 5E). A transient expression assay in Arabidopsis protoplasts showed that the luciferase signal was stronger in both the CLV3pro:LUC + 35Spro:ARR1 and ProCLV3:LUC + 35Spro:ARR12 samples compared to the CLV3pro:LUC + empty vector ( Figures 5F and 5G). When 18 copies of the core cytokinin response motif (within the 1,400-bp upstream sequence of the CLV3 start codon) were mutated (mCLV3) (Supplemental Table 1), the activation of mCLV3pro:LUC by 35Spro:ARR1 and 35Spro:ARR12 was reduced compared to the wild-type CLV3 promoter ( Figure 5F and 5G). These data suggest that ARR1 and ARR12 bind to the same sites of the CLV3 promoter and modulate its expression. However, we observed an Together, these data support the hypothesis that ARR1 competes with ARR12 for binding to the CLV3 promoter region and represses its expression indirectly.
The observation that ARR1 represses CLV3 expression suggested that CLV3 functions downstream of ARR1. To test this hypothesis, we compared the ability of arr1-4, clv3-7, arr1-4 clv3-7, and wild-type root explants to generate calli and shoots. clv3-7 root explants generated fewer calli and shoots than the other genotypes ( Figures 6A, 6B, 6E, 6F, 6G and 6J), confirming that CLV3 promotes callus formation and shoot regeneration. The arr1-4 clv3-7 double mutant generated an intermediate number of calli and shoots compared to arr1-4 and clv3-7 single mutants ( Figure 6B to 6E and 6G to 6J), suggesting that the loss of CLV3 function partially rescued the arr1-4 phenotype and that CLV3 acts downstream of ARR1 in regulating shoot regeneration. 10 WUS is considered to be a key regulator of shoot regeneration (Gordon et al., 2007;Meng et al., 2017;Zhang et al., 2017b). We therefore examined whether ARR1 or ARR12 regulates WUS expression. The distribution of GUS signal in transgenic WUSpro:GFP-GUS explants (Supplemental Figure 13A) (Cui et al., 2015) was similar to that in in vitro hypocotyl explants reported previously (Zhang et al., 2017b). We also confirmed the expression pattern of WUS in transgenic WUSpro:GFP-GUS root explants (Supplemental Figure 13B Figures 16B and 16C). We observed no synergistic effect on luciferase signal when both 35Spro:ARR1 and 35Spro:ARR12 were co-transformed compared to transformation of 35Spro:ARR1 or 35Spro:ARR12 alone, suggesting that ARR1 and ARR12 regulate WUS expression by competitively binding to the same sites of the WUS promoter.

ARR1 Inhibits Shoot Regeneration Through Transcriptional Activation of IAA17
A comparison of the transcriptomes of arr1, AOE, and wild-type root explants at the nascent SAM formation stage by deep sequencing (RNASeq) revealed many differentially transcribed genes (DTGs).
In the transcriptomes of arr1 versus wild-type explants, we detected 177 upregulated and 260 downregulated DTGs, and in the AOE versus wild-type transcriptomes, we detected 129 and 6 DTGs, respectively ( Figure 7A). Pathways involved in hormone signal transduction and indole alkaloid synthesis were well represented among the DTGs in arr1 versus the wild type ( Figure 7B). These DTGs also included genes involved in auxin transport and signaling, as well as IAA17, encoding an Aux/IAA repressor that is downregulated in arr1 root explants ( Figure 7C). IAA17 expression in Col-0 explants slightly increased when cultured on CIM, and highly induced and peaked after culture on SIM for 4 d ( Figure 7D). IAA17 transcript levels then decreased after culture on SIM for 7 and 10 d ( Figure 7D). IAA17 transcript levels were lower in arr1 than in Col-0 root explants during both CIM and SIM incubation ( Figure 7D). In transgenic root explants harboring IAA17pro:GFP and analyzed at different stages of shoot regeneration, GFP signals were all weaker in arr1 than in the wild type ( Figures 7E and 7F). Together, these results suggest that ARR1 promotes IAA17 expression. GFP explants at the nascent SAM formation stage showed that ARR1 specifically binds to the core cytokinin response motif-containing region of the IAA17 promoter ( Figure 7G). When IAA17pro:LUC or mIAA17pro:LUC was transiently expressed in Arabidopsis protoplasts harboring 35Spro:ARR1, the signal level was lower in the case of the mutated promoter ( Figure 7H). An electrophoretic mobility shift assay (EMSA) confirmed that ARR1 binds to the IAA17 promoter in vitro ( Figure 7I). Furthermore, transgenic plants harboring mIAA17pro:GFP expressed GFP less effectively than plants harboring IAA17pro:GFP ( Figures 7J and 7K). Together, these results suggest that ARR1 promotes IAA17 expression by directly binding to its promoter region.
The transcript abundance of IAA17 in arr12 and wild-type root explants was similar during CIM incubation, whereas IAA17 transcript was more abundant in arr12 than in wild-type root explants after being cultured on SIM for 7, 10, and 14 d (Supplemental Figure 17A). However, 35Spro:ARR12 activated the IAA17pro:LUC reporter in Arabidopsis protoplasts, but not mIAA17pro:LUC, and co-transformation of 35Spro:ARR1 and 35Spro:ARR12 did not have a synergistic effect on IAA17pro:LUC expression 12 (Supplemental Figure 17B), suggesting that ARR1 and ARR12 regulate IAA17 expression by competitively binding to its promoter.
We then compared the ability of root explants of the four genotypes to regenerate shoots: AOE, 35Spro:VP16-IAA17mImII, the double overexpressor, and the wild type. AOE-derived explants formed fewer shoots than the wild type, whereas 35Spro:VP16-IAA17 explants produced more shoots (Figures 8F and 8G). The AOE 35Spro:VP16-IAA17mImII double overexpressor line generated as many shoots as the 35Spro:VP16-IAA17 line ( Figures 8F and 8G), suggesting that the 35Spro:VP16-IAA17mImII transgene completely rescued the AOE phenotype.
We next compared the shoot regeneration capacity of axr3-3, arr1, axr3-3 arr1 double mutant, and wild-type root explants. Explants derived from axr3-3 arr1 failed to produce shoots, similar to explants derived from axr3-3 ( Figures 8H and 8I). These results indicate that axr3-3 rescued the phenotype of arr1 and that IAA17 acts downstream of ARR1 during in vitro shoot regeneration.

DISCUSSION
Shoot regeneration from an explant is prompted by exogenous cues that elicit endogenous signals.
Cytokinin has long been known to be a critical exogenous cue (Skoog and Miller, 1957;Che et al., 2006;Gordon et al., 2007); therefore, genes that regulate cytokinin signal transduction, including those encoding type-B ARRs, are likely involved in the regeneration process. In this study, we revealed that ARR12 is a central enhancer of shoot regeneration, whereas ARR1 is a strong inhibitor of this process by modulating WUS and CLV3 expression in an ARR12-dependent manner. Furthermore, we showed that a regulatory cascade involving ARR1 and the auxin signaling repressor IAA17 inhibits shoot regeneration in cultured explants. These findings provide new evidence for direct links between cytokinin and shoot meristem regulators or auxin signaling and shed light on the mechanism underlying cytokinin-regulated shoot regeneration.

ARR1 Inhibits in vitro Shoot Regeneration
ARRs are components of the cytokinin signal transduction pathway. Loss-of-function mutants of the type-A ARR genes ARR7 and ARR15 both exhibit an enhanced capacity for shoot regeneration; it has therefore been suggested that type- In the two-step regeneration system for Arabidopsis, pericycle cells are induced to form callus, then shoots are generated efficiently on SIM (Che et al., 2006;Gordon et al., 2007). A 2 d preincubation period on CIM is needed for explants to acquire competence to form shoots, and this competence continues to increase for 2 more d (Che et al., 2007). Auxin is thought to activate callus formation on CIM, and cytokinin is responsible for shoot primordium formation in this system (Gordon et al., 2007). Here, we determined that ARR1 not only inhibits SIM-induced shoot regeneration, but also suppresses CIMinduced callus formation (Figure 1; Supplemental Figure 3). These observations suggest that ARR1 inhibits shoot regeneration by restricting the formation of cells competent for shoot regeneration. 15 ARR1 was previously reported to function in cytokinin-mediated protuberance formation and greening of Arabidopsis hypocotyl explants (Sakai et al., 2001). Overexpression of ARR1 increased cytokinin sensitivity, and its loss of function reduced sensitivity to cytokinin compared to wild-type explants (Sakai et al., 2001). Protuberances formed from a 35Spro:ARR1 explant, but not from wild-type or arr1 explants, and occasionally produced shoots when provided with 2,4-D and kinetin (Sakai et al., 2001). In the previous study, hypocotyl explants were only incubated on cytokinin-rich medium without CIM preincubation. In our study, we carried out similar experiments, and in most cases no shoots formed when root and hypocotyl explants were directly incubated on SIM, regardless of the concentration of cytokinin provided or the genotype (wild type, arr1, or AOE) of the explant (Supplemental Figures 20A to 20J). In the presence of 0.5 mg/L 2-ip, AOE hypocotyl explants produced many calli and occasionally produced shoots (Supplemental Figure 20C). When explants were directly incubated on cytokinin-rich medium, the lack of CIM-induced competent cells led to a high DNA methylation level and the repression of WUS expression in explants, thus suppressing shoot regeneration (Supplemental Figures 20K to 20O). Compared to the wild type, explants harboring 35Spro:ARR1 were more sensitive to exogenous cytokinin (Sakai et al., 2001), and cell division was overactivated following treatment with a certain concentration of cytokinin (Supplemental Figure   20C).
Thus, the occasional shoot regeneration in explants harboring 35Spro:ARR1 was driven by the accelerated proliferation of protuberances and the subsequent demethylation of the WUS promoter, which might partly compensate for the lack of CIM-induced competent cells (Supplemental Figures 20C and   20N). Although ARR1 enhanced the cytokinin response in explants (Supplemental Figure 21), it inhibited callus formation and shoot regeneration in the two-step culture system.

Regeneration
The type-B ARR family contains multiple members, and several studies based on double or triple mutants have revealed functional redundancy among these members in a variety of developmental processes These results support the notion that the production of ARR1 and ARR12 proteins is responsible for the different abilities of these explants to undergo shoot regeneration. Indeed, on medium containing 0.5 μM of the auxin 1-naphthaleneacetic acid, no shoots formed from wild-type, arr1, or ARR1overexpressing explants, whereas explants overexpressing a phosphomimetic form of ARR1 generated shoots (Kurepa et al., 2014), indicating that phosphorylation of ARR1 protein promotes shoot regeneration. In this study, the 35Spro:ARR1 transgene inhibited shoot regeneration in wild-type explants 17 (Supplemental Figures 8A and 8B), whereas the expression of 35Spro:mARR1D94E (encoding a phosphomimetic mutant version of ARR1 protein) dramatically increased shoot regeneration (Supplemental Figures 8A, 8C to 8E), suggesting that the phosphorylation status of ARR1 protein contributes to its function in shoot regeneration.
The ARR1 locus could not compensate for the loss of ARR12 function, while the ARR12pro:ARR1 transgene partially rescued the arr12 phenotype ( Figures 3C and 3D; Supplemental Figures 7A and 7B), indicating that under the control of different promoters, ARR1 has slightly different effects on shoot regeneration, suggesting that the ARR1 and ARR12 promoters also contribute to the different effects of these proteins on shoot regeneration. The differential effects of ARR1 and ARR12 proteins on shoot regeneration ( Figures 3C and 3F) were much more apparent than those of the ARR1 and ARR12 promoters ( Figures 3B and 3C). Thus, the different shoot regeneration phenotypes of arr1 and arr12 are determined by both the protein functions and expression patterns of ARR1 and ARR12, with the former dominating.
Although the functional redundancy of type-B ARRs in various developmental processes has been reported, the specific relationships between individual type-B family members remain unclear. In this study, we demonstrated the functional dependency of ARR1 on ARR12 during shoot regeneration. In the presence of ARR12, ARR1 inhibited shoot regeneration, whereas in the absence of ARR12, ARR1 slightly enhanced shoot regeneration (Figure 2). ARR1 inhibited shoot regeneration by counteracting the positive effect of ARR12 on shoot regeneration (Figure 2; Supplemental Figures 5A and 5B). ARR1 had a slight effect on ARR12 transcription (Supplemental Figures 5C to 5H), suggesting that ARR1 counteracts the positive effect of ARR12 on shoot regeneration through ways other than repressing its transcription.
Type-B ARRs play redundant roles in regulating the transcription of target genes. For example, ARR1, ARR10, and ARR12 participate in axillary meristem formation and shoot regeneration through directly activating WUS expression (Meng et al., 2017;Wang et al., 2017;Zhang et al., 2017b). Here, we show that ARR1 and ARR12 exert complex regulatory effects on CLV3 and WUS during shoot regeneration.
ARR12 strongly activates the transcription of CLV3 in explants cultured on both CIM and SIM, which enhances callus formation and shoot regeneration (Figures 2, 5, 10A and 10B). However, ARR1 competes with ARR12 for binding to the same CLV3 promoter sites ( Figures 5H and 10A), partly blocks the stimulatory effect of ARR12 on CLV3 expression, and thus represses CLV3 transcription indirectly ( Figures 5D and 10B), which inhibits callus formation and shoot regeneration. ARR12 is critical for the widespread distribution of WUS in the explants and for the concentrated WUS expression in the newly formed SAMs, which enhances shoot regeneration (Supplemental Figures 14, 22A and 22C). However, 18 ARR1 competes with ARR12 for binding to the same WUS promoter sites and induces the widespread distribution of WUS in the explants in the presence of ARR12 (Supplemental Figures 16D, 22B and 22D), which delays the concentration of WUS transcripts and therefore inhibits shoot regeneration (Supplemental Figures 16). Thus, ARR1 modulates CLV3 and WUS expression in an ARR12-dependent manner, which contributes to its ARR12-dependent inhibitory effect on shoot regeneration.
Cytokinin-induced shoot regeneration requires the integration of multiple regulatory steps controlling CLV3 and WUS expression mediated by ARR1 and ARR12. In addition, ARR1 and ARR12 directly activate CLV3, WUS, and IAA17 transcription in Arabidopsis protoplasts ( Figure 5G In the two-step regeneration model of Arabidopsis, WUS expression is specifically induced in the explant on SIM (Supplemental Figures 13A to 13D) and is critical for in vitro stem cell niche specification (Gordon et al., 2007;Meng et al., 2017;Zhang et al., 2017b). The specific functions of the shoot meristem regulator CLV3 during shoot regeneration have long been elusive. Here, we demonstrated that, in contrast to observations for WUS, CLV3 transcript was detected in explants cultured on both CIM and SIM ( Figure   5A to 5D), which is consistent with a previous study (Atta et al., 2009). Loss of function of CLV3 resulted in a slight decrease in the ability to form calli and regenerate shoots ( Figures 6A, 6B, 6E, 6F, 6G and 6J), and mutation of CLV3 in the arr1-4 mutant line partly rescued both the callus formation and shoot regeneration phenotype (Figures 6B to 6E and 6G to 6J), indicating that CLV3 promotes both callus formation and shoot regeneration and functions in the pathway downstream of ARR1 during the two-step regeneration process.

Regeneration
The interaction between auxin and cytokinin influences many aspects of plant development (Moubayidin et al., 2009). In the primary root meristem, cytokinin reduces the auxin response by inducing SHORT HYPOCOTYL2 (SHY2), thereby promoting cell differentiation (Moubayidin et al., 2010). The negative effect of cytokinin on PIN-FORMED (PIN) -dependent auxin distribution prevents the ectopic initiation of lateral roots (Marhavy et al., 2014). The auxin-mediated repression of type-A ARR genes ARR7 and ARR15 is required for embryonic stem cell niche formation and SAM maintenance (Muller and Sheen, 2008;Zhao et al., 2010). Under in vitro conditions, ARF3 promotes shoot regeneration by activating IPT5 (Cheng et al., 2013). Cytokinin and auxin play critical roles in shoot regeneration (Ikeuchi et al., 2016).
However, the full nature of the interaction between auxin and cytokinin signaling during shoot regeneration remains obscure.
Here, we revealed a new link between auxin and cytokinin signaling involving the interaction between ARR1 and IAA17. ARR1 binds to the IAA17 promoter and promotes its expression, especially at SIM incubation stage (Figures 7 and 10C), which indirectly represses the expression of WUS (Figures 9 and   10C) and inhibits shoot regeneration ( Figure 10C). Combined with the direct activation of WUS expression by ARR1 and ARR12 (Supplemental Figures 16 and 22), these results further support the previous finding that type-B ARRs regulate shoot regeneration via dual regulation of WUS (Meng et al., 2017) ( Figure 10C). ARR1 promotes the expression of IAA17, resulting in a decreased auxin response ( Figure 7; Supplemental Figure 23). Root explants derived from loss-of-function iaa17 mutants regenerated more shoots than the wild type (Supplemental Figure 18). In addition, the 35Spro:VP16-IAA17mImII transgene fully restored the phenotype of the AOE line, and axr3-3 rescued the phenotype of arr1 (Figure 8). These observations suggest that IAA17 acts downstream of ARR1.
During shoot regeneration from a cultured explant, cytokinin is thought to help determine cell fate, whereas the presence of cytokinin in the SAM is critical for stem cell maintenance (Zhao et al., 2010). Therefore, cytokinin homeostasis is thought to be closely associated with shoot regeneration, possibly via the activation of SAM-related genes (Jasinski et al., 2005;Yanai et al., 2005). Recent

Shoot Regeneration and Callus Greening in vitro
We harvested root and hypocotyl explants from seedlings cultured on solid half-strength MS medium for photons m−2 s−1) at 22 ± 1ºC. We then re-plated calli on SIM (MS medium supplemented with Gamborg's B-5 vitamins plus 10 g/L sucrose, 0.5 g/L MES, 6 g/L agar, 1 mg/L N6-(2-isopentenyl) adenine (Sigma, USA), and 0.15 mg/L IAA (Sigma, USA), pH 5.8) in constant light (110 µmol photons m−2 s−1) at 22 ± 1ºC for 2 to 21 d. For callus greening in vitro, we transferred root and hypocotyl explants directly to SIM containing the indicated concentrations of N6-(2-isopentenyl) adenine (Sigma, USA) for 28 d. All regeneration experiments were performed in glass dishes. We carried out statistical treatment with Microsoft Excel software. Statistically significant differences were calculated using two-sided statistical tests at P-value < 0.05 (Supplemental Data Set 2). Three biological replicates with separately cultured explant samples were conducted.

Vector Construction, Transgenesis, and Identification of T-DNA Insertion Mutants
The 2,073 bp coding region of ARR1 was inserted into the pROK2 plant expression vector (http://signal.salk.edu/tdna_protocols.html) under the control of the CaMV 35S promoter to generate the transgene 35Spro:ARR1. Plant transformation was carried out using the floral dip method (Clough and Bent, 1998). Lines carrying the 35Spro:ARR1 transgene are referred to as AOE. To generate phosphomimetic ARR1-overexpressing plants (35Spro:mARR1D94E), we used fusion PCR to introduce a C-to-G mutation (thus D to E substitution at position 94) into the coding region of ARR1 with the primer pairs ARR1-pB2GW7-F/mARR1-R and mARR1-F/ARR1-pB2GW7-R (sequences in Supplemental Data Set 1). We cloned the mutated coding region of ARR1 (mARR1) into pDONR221 (Invitrogen, USA) using BP Clonase (Invitrogen, USA), and recombined into the pB2GW7 binary vector using LR Clonase enzyme mix (Invitrogen, USA). To generate the ARR1pro:ARR1-GFP vector, we inserted the 2,073 bp ARR1 coding region and its 2,196 bp upstream sequence into the pGFPGUSplus vector (Vickers et al., 2007). The arr1 mutant was transformed with the ARR1pro:ARR1-GFP vector to obtain the RS transgenic lines. The construction of the 35Spro:ARR12 and ARR12pro:ARR12-GFP vectors was described previously (Dai et al., 2017). To construct the ARR12pro:ARR1 vector, we inserted the 3,441 bp ARR12 upstream sequence and 2,073 bp coding region of ARR1 into the pGFPGUSplus vector (Vickers et al., 2007) via In-Fusion cloning method (Clontech, USA). To construct the ARR1pro:ARR12 vector, we inserted the 3,246 bp ARR1 upstream sequence and 1,791 bp coding region of ARR12 into the pGFPGUSplus vector via In-Fusion cloning method (Clontech, USA). To generate the ARR12pro:ARR12 vector, we inserted the 4,539 bp ARR12 upstream sequence, 2,472 bp genomic region of ARR12 and the 2,094 bp ARR12 downstream sequence into the pGFPGUSplus vector via In-Fusion cloning method (Clontech, USA). The 22 mutated upstream sequences of CLV3 and WUS were synthesized by the Beijing Genomics Institute (http://www.bgitechsolutions.com/). We then inserted the wild-type and the mutated forms of the CLV3 promoter into pDONR221 (Invitrogen, USA) using BP Clonase (Invitrogen, USA), and recombined into the pBGWFS7 binary vector using LR clonase enzyme mix (Invitrogen, USA). We inserted the wild-type and the mutated form of WUS into pGreenII 0800-LUC vector (Hellens et al., 2005) to obtain the plasmids WUSpro:LUC, mWUS1pro:LUC or and mWUS2pro:LUC, respectively. We generated mutated sites in the IAA17 promoter via fusion PCR. The 1,211 nt upstream sequence of the wild-type and the mutated form of IAA17 were inserted into either pGFPGUSplus vector (Vickers et al., 2007) or pGreenII 0800-LUC vector (Hellens et al., 2005) to obtain the plasmids PIAA17pro:GFP and mIAA17pro:GFP or IAA17pro:LUC and mIAA17pro:LUC, respectively. The protein expression vector pGST-ARR1-DBD was generated as

Microscopy
We cleared explants cultured on CIM or SIM following an established protocol (Malamy and Benfey, 1997). We detected GUS activity as previously described (Liu et al., 2013). At least 20 samples per line were analyzed. We captured images with a BX51 or SZX16 microscope (Olympus, Japan). To acquire fluorescence images, we mounted explants on a glass slide in water. The GFP was excited at 488 nm and the signal was collected at 495 to 550 nm using an LSM 700 confocal laser scanning microscope (Zeiss, Germany). We mounted roots in 50 mg/mL propidium iodide (Sigma, USA), which was excited at 593 nm and the signal collected at 610-680 nm.

RT-qPCR, RNA-Seq Analysis, DNA Methylation Assay and RNA in situ Hybridization
For gene transcript levels determination, we collected explant samples and isolated total RNA using TRIzol reagent (Invitrogen, USA). We treated a 3 μg aliquot of total RNA for each sample with RNasefree DNase I (Takara, China) to remove contaminating gDNA and reverse-transcribed using a FastQuant RT Kit (TIANGEN Biotech, Beijing, China) in a 40 μL reaction, following the manufacturer's protocol.
Three replicates of each reaction were included in the subsequent RT-qPCR, which was based on a Bio-Rad MyiQTM2 detection system (Bio-Rad, USA) with SYBR Green I master mix (Roche, Switzerland).
The WUS antisense and sense probes, as well as the in situ hybridization protocol were as described (Xin et al., 2017).

ChIP, EMSA, and Transient Expression Analyses
We chose the protuberance from root explants at the nascent SAM formation stage (SIM 7 d) as the experimental material for ChIP analysis. Our procedure followed a protocol described elsewhere ( (w/v) PEG as previously described (Yoo et al., 2007). We used empty pGreenII 0800-LUC or pROK2 vector d as negative control. We tested LUC/REN activity using the dual-luciferase reporter assay system (Promega, USA). Three independent biological replicates with separately prepared mesophyll protoplasts were performed.

Supplemental Data
Supplemental Figure  Supplemental Figure 18. The shoot regeneration capacity of the wild type and IAA17 mutant from root explants.
Supplemental Figure 20. Greening and shoot regeneration capacity of calli derived from hypocotyl and root explants without pre-incubation on CIM.
Supplemental Figure 21. ARR1 modulates the cytokinin response during shoot regeneration.
Supplemental Figure 22. ARR1 inhibits shoot regeneration by competing with ARR12 for binding to the WUS promoter.
Supplemental Figure 23. ARR1 and IAA17 modulate the auxin response during shoot regeneration in root explants.
Supplemental     (G) The enrichment of IAA17 promoter fragments following ChIP.
(I) ARR1 binds to the promoter region of IAA17 (indicated by arrows) in an EMSA.
(J-K) Temporal and spatial expression patterns of IAA17pro:GFP and mIAA17pro:GFP in root explants at the nascent SAM formation stage. Black boxes refer to the cytokinin response motif 5'-(A/G)GAT(T/C) and black boxes marked with "×" denote Ato-C and G-to-C point mutations. **: means differ at P < 0.01 (Student's t-test). Scale Bar: 50 μm. Two independent transgenic lines were analyzed for each genetic background, with similar expression patterns (C, F, J, K).