Arabidopsis ERF012 Is a Versatile Regulator of Plant Growth, Development and Abiotic Stress Responses

The AP2/ERF transcription factors are widely involved in the regulation of plant growth, development and stress responses. Arabidopsis ERF012 is differentially responsive to various stresses; however, its potential regulatory role remains elusive. Here, we show that ERF012 is predominantly expressed in the vascular bundles, lateral root primordium and vein branch points. ERF012 overexpression inhibits root growth, whereas it promotes root hair development and leaf senescence. In particular, ERF012 may downregulate its target genes AtC4H and At4CL1, key players in phenylpropanoid metabolism and cell wall formation, to hinder auxin accumulation and thereby impacting root growth and leaf senescence. Consistent with this, exogenous IAA application effectively relieves the effect of ERF012 overexpression on root growth and leaf senescence. Meanwhile, ERF012 presumably activates ethylene biosynthesis to promote root hair development, considering that the ERF012-mediated root hair development can be suppressed by the ethylene biosynthetic inhibitor. In addition, ERF012 overexpression displays positive and negative effects on low- and high-temperature responses, respectively, while conferring plant resistance to drought, salinity and heavy metal stresses. Taken together, this study provides a comprehensive evaluation of the functional versatility of ERF012 in plant growth, development and abiotic stress responses.


Introduction
Transcription factors (TFs), which are central to gene expression regulation, play critical roles in regulating major developmental processes in the plant life cycle, including seed germination, seedling growth, tissue morphogenesis, reproduction and senescence. Meanwhile, accumulating evidence implicates the importance of various TFs, such as NAC (no apical meristem (nam), arabidopsis transcription activation factor (ataf) and cupshaped cotyledon (cuc)), WRKY, MYB, bHLH (basic helix-loop-helix), bZIP (basic-region leucine zipper) and ERF (ethylene-responsive element binding), in mediating plant growth, development and responses to environmental stresses [1]. For example, the MYC2 TF, a master regulator of the JA signaling pathway, is essentially involved in plant growth, development and defense responses to biotic and abiotic stresses [2][3][4].

Tissue-Specific Expression of ERF012 in Arabidopsis
To understand the tissue-specific expression of ERF012, we performed qRT-PCR analysis of ERFF012 transcription levels in Arabidopsis tissues. Strong expression levels were detected in the root, leaves, stem and pod, whereas they were significantly reduced in the bud and flower ( Figure 1A). To obtain a high-resolution image of the ERF012 expression in tissues, the pERF012:GUS (β-glucuronidase) transgenic lines were established ( Figure 1B). Based on the GUS staining, the tissue-specificity was detailed. The whole cotyledon was stained, showing that true leaves possessed margin labeling and scattered spots ( Figure 1C,D). Interestingly, these spots were specifically localized on the vein branch points. In the root, no GUS staining was observed in the primary and lateral root tip ( Figure 1H,I). However, the mature regions, especially the lateral emerged and emerging cells that included the lateral root primordium and vascular cells, had elevated expression ( Figure 1E-G). These findings were contrary to a previous report [25] that only strong GUS activity was observed in the cotyledon. We further performed a transactivation assay of ERF012 in the Arabidopsis protoplast. Expression of GAL4-ERF012 activated GAL4-LUC activity largely compared with the GAL4-BD control ( Figure 1J), demonstrating that ERF012 can regulate downstream gene transcription. These results indicate that the ERF012 transcription factor may play an important role in the growth and development of Arabidopsis. Thirty-five-day-old Arabidopsis was used for qRT-PCR analysis. Letters indicate significant differences between different tissues: Duncan's test (p < 0.05). (B-I) Expression pattern analysis of ERF012 in seedlings. Twelve-day-old seedling expressing pERF012:GUS was used. A close-up was indicated by a black frame in (C). Scale bar = 100 μm in (C-I). (J) Transactivation assay of ERF012 genes in protoplast. GAL4-ERF012 is the effector and GAL4-LUC is the reporter. GAL4-DB is the control. The values are means ± SD, n = 5. Asterisks indicate significant differences between different treatments: Student's t-test, *** p < 0.001.

ERF012 Regulates Plant Growth and Development
To explore the effect of ERF012 on plant growth and development, the ERF012-overexpressing (OE) transgenic lines were established ( Figure 2E) by transforming a 35S:ERF012 plasmid in Col-0. In parallel, the erf012 mutants were established using a gene-editing system (CRISPR-Cas9) ( Figure 2F). Under normal conditions (1/2 MS medium), Col-0 and erf102 mutant lines developed comparable primary root lengths and fresh weights, whereas the primary lengths and fresh weights of ERF012-OE plants were reduced by 58-68% and 50-70% compared to Col-0, respectively (Figure 2A-C). As shown in Figure 2A,D, ERF012-OE lines obviously had a lower lateral root number than Col-0 and erf012 mutants. We noticed that the root tips of ERF012-OE plants were significantly shorter than that of Col-0 plants, including the meristem zone (MZ) and elongation zone (EZ) ( Figure 3A,B). Microscopic statistics showed that cell numbers, but not the cell lengths, of MZ and EZ in ERF012-OE roots were less than the Col-0 roots ( Figure  3C,D). These results suggest that ERF012 expression might downregulate cell division. To this end, we compared the cyclin-dependent protein kinase (CYCB1;1) activity in Col-0 and ERF012-OE backgrounds. CYCB1;1 functions as an effector of growth control at G2/M, thus indicating the cell division rate. We speculated that CYCB1;1 has higher activity in Col-0 than in ERF012-OE plants ( Figure 3E). As expected, pCYCB1;1:GUS showed stronger expression and activity in the meristem zone of Col-0 than in that of Thirty-five-day-old Arabidopsis was used for qRT-PCR analysis. Letters indicate significant differences between different tissues: Duncan's test (p < 0.05). (B-I) Expression pattern analysis of ERF012 in seedlings. Twelve-day-old seedling expressing pERF012:GUS was used. A close-up was indicated by a black frame in (C). Scale bar = 100 µm in (C-I). (J) Transactivation assay of ERF012 genes in protoplast. GAL4-ERF012 is the effector and GAL4-LUC is the reporter. GAL4-DB is the control. The values are means ± SD, n = 5. Asterisks indicate significant differences between different treatments: Student's t-test, *** p < 0.001.

ERF012 Regulates Plant Growth and Development
To explore the effect of ERF012 on plant growth and development, the ERF012overexpressing (OE) transgenic lines were established ( Figure 2E) by transforming a 35S:ERF012 plasmid in Col-0. In parallel, the erf012 mutants were established using a gene-editing system (CRISPR-Cas9) ( Figure 2F). Under normal conditions (1/2 MS medium), Col-0 and erf102 mutant lines developed comparable primary root lengths and fresh weights, whereas the primary lengths and fresh weights of ERF012-OE plants were reduced by 58-68% and 50-70% compared to Col-0, respectively (Figure 2A-C). As shown in Figure 2A,D, ERF012-OE lines obviously had a lower lateral root number than Col-0 and erf012 mutants. We noticed that the root tips of ERF012-OE plants were significantly shorter than that of Col-0 plants, including the meristem zone (MZ) and elongation zone (EZ) ( Figure 3A,B). Microscopic statistics showed that cell numbers, but not the cell lengths, of MZ and EZ in ERF012-OE roots were less than the Col-0 roots ( Figure 3C,D). These results suggest that ERF012 expression might downregulate cell division. To this end, we compared the cyclin-dependent protein kinase (CYCB1;1) activity in Col-0 and ERF012-OE backgrounds. CYCB1;1 functions as an effector of growth control at G2/M, thus indicating the cell division rate. We speculated that CYCB1;1 has higher activity in Col-0 than in ERF012-OE plants ( Figure 3E). As expected, pCYCB1;1:GUS showed stronger expression and activity in the meristem zone of Col-0 than in that of ERF012-OE plants ( Figure 3F).

ERF012 Downregulates Gene Expression of Cell Wall Formation and Phenylpropanoid Metabolism Pathway
To interpret the growth alterations of ERF012-OE plants, we performed the interaction gene analysis of ERF012 (https://bar.utoronto.ca/eplant/ accessed on 1 April 2022). Many genes were predicted to be the targets of ERF012 (Supplementary Figure S1A). A part of cell wall formation involves genes such as At1g27440, AT2G34710, AT3G18660, AT5G15630 and AT5G17420 encoding the cell wall formation-related proteins [26]. In addition, the upstream regulatory genes of cell wall formation, At2g30490 and At1g51680, which are involved in the phenylpropanoid metabolism pathway, were listed in the target candidates. At2g30490 and At1g51680 encode a cinnamate 4-hydroxylase (AtC4H) [27] and a 4-coumarate:coenzyme A ligase (At4CL1) [28], respectively, and AtC4H1 directly functions upstream of At4CL1. All predicted targets were downregulated in the ERF012-OE plants (Supplementary Figure S1B). The phenylpropanoid metabolism makes an important contribution to the coordination of plant development and plant-environment interaction via cell wall formation containing diverse metabolic routes; thus, the disruption of phenylpropanoid metabolism severely limits plant growth, development and stress responses [29]. Compared with Col-0, the expression levels of AtC4H in the shoot and root of ERF012-OE plants and At4CL1 in the shoot of ERF012-OE plants were significantly reduced ( Figure 4A), demonstrating that ERF012 overexpression disrupts the phenylpropanoid metabolism. To validate the interaction of ERF012 with AtC4H and At4CL1, a yeast one-hybrid assay was conducted. The results showed that both promoters of AtC4H and At4CL1 can directly interact with the ERF012 protein ( Figure 4B). These results suggest that ERF012 expression downregulate gene expression in cell wall formation and the phenylpropanoid metabolism pathway, which might lead to limited growth.  with AtC4H and At4CL1, a yeast one-hybrid assay was conducted. The results showed that both promoters of AtC4H and At4CL1 can directly interact with the ERF012 protein ( Figure 4B). These results suggest that ERF012 expression downregulate gene expression in cell wall formation and the phenylpropanoid metabolism pathway, which might lead to limited growth.

Auxin and Ethylene Involvement in the Root Growth and Lateral Root Formation in ERF012-OE Plants
Loss of lateral roots was obvious in the ERF012-OE plants ( Figure 2A). It was reported that the Atc4h/ref3 mutant reduced lateral roots because blocking of the AtC4H1 function largely hinders the accumulation of lignin and causes slow, phloem-mediated auxin transport [30]. Indeed, the auxin level in the root tip of ERF012-OE plants was lower than the Col-0, evidenced by DR5:GFP signals ( Figure 4C). Plant hormone auxin is a key factor that controls lateral root formation [31][32][33] due to the auxin-dependent signaling. To investigate whether auxin is involved in lateral root formation in ERF012-OE lines, exogenous IAA was used to treat Col-0 and ERF012-OE lines under normal growth conditions. Five-day-old seedlings were transferred to normal growth conditions supplemented with 0.1 μM IAA for a further 6 days of growth. With the IAA addition, the lateral roots were formed in large quantities in ERF012-OE lines ( Figure 4D,E). This result demonstrates that ERF012 expression limits the formation of lateral roots.
Short root tips and high root hair density are the typical characteristics of excess ethylene in plant roots. Compared to Col-0, the roots of ERF012-OE plants developed shorter root tips and dense root hair ( Figure 5A), indicating the high ethylene level in ERF012-OE roots. To investigate whether ethylene is involved in the root growth of ERF012-OE plants, we compared the expression levels of the ethylene precursor and bi-

Auxin and Ethylene Involvement in the Root Growth and Lateral Root Formation in ERF012-OE Plants
Loss of lateral roots was obvious in the ERF012-OE plants (Figure 2A). It was reported that the Atc4h/ref3 mutant reduced lateral roots because blocking of the AtC4H1 function largely hinders the accumulation of lignin and causes slow, phloem-mediated auxin transport [30]. Indeed, the auxin level in the root tip of ERF012-OE plants was lower than the Col-0, evidenced by DR5:GFP signals ( Figure 4C). Plant hormone auxin is a key factor that controls lateral root formation [31][32][33] due to the auxin-dependent signaling. To investigate whether auxin is involved in lateral root formation in ERF012-OE lines, exogenous IAA was used to treat Col-0 and ERF012-OE lines under normal growth conditions. Five-day-old seedlings were transferred to normal growth conditions supplemented with 0.1 µM IAA for a further 6 days of growth. With the IAA addition, the lateral roots were formed in large quantities in ERF012-OE lines ( Figure 4D,E). This result demonstrates that ERF012 expression limits the formation of lateral roots.
Short root tips and high root hair density are the typical characteristics of excess ethylene in plant roots. Compared to Col-0, the roots of ERF012-OE plants developed shorter root tips and dense root hair ( Figure 5A), indicating the high ethylene level in ERF012-OE roots. To investigate whether ethylene is involved in the root growth of ERF012-OE plants, we compared the expression levels of the ethylene precursor and biosynthetic gene 1-aminocyclopropane-1-carboxylic acid (ACC) synthase (ACS) between Col-0 and ERF012-OE plants. Two ERF012-OE lines had significantly higher expression levels of ACS7 and ACS11 than in Col-0 ( Figure 5B), suggesting ethylene accumulation in ERF012-OE lines. We further used an ethylene biosynthesis inhibitor to estimate the involvement of ethylene in the root growth of ERF012-OE lines. AVG treatment largely improved the root growth of ERF012-OE lines ( Figure 5C). The root tips were elongated and root hair density was dramatically reduced in the AVG-treated ERF012-OE lines ( Figure 5B-D). These results demonstrate that ERF012 expression enhances ethylene biosynthesis, leading to root growth inhibition.
osynthetic gene 1-aminocyclopropane-1-carboxylic acid (ACC) synthase (ACS) between Col-0 and ERF012-OE plants. Two ERF012-OE lines had significantly higher expression levels of ACS7 and ACS11 than in Col-0 ( Figure 5B), suggesting ethylene accumulation in ERF012-OE lines. We further used an ethylene biosynthesis inhibitor to estimate the involvement of ethylene in the root growth of ERF012-OE lines. AVG treatment largely improved the root growth of ERF012-OE lines ( Figure 5C). The root tips were elongated and root hair density was dramatically reduced in the AVG-treated ERF012-OE lines ( Figure 5B-D). These results demonstrate that ERF012 expression enhances ethylene biosynthesis, leading to root growth inhibition.

ERF012 Responds Quickly to Temperature Dynamics
To estimate the responses of ERF012 to environmental stresses, temperature dynamics were first tested. Col-0 seedlings grown under normal conditions (22 °C) were transferred to high temperatures (37 °C) or low temperatures (4 °C) for several hours. Time-course qRT-PCR analysis of ERF012 expression in the shoots and roots was performed. The rapid responses (within 2 h) to dynamics were observed for Col-0 shoots, whereas no significant changes occurred in the roots ( Figure 6A,B). The consistent results were confirmed by pERF012:GUS lines treated at different temperatures ( Figure 6C), suggesting that ERF012 plays a critical role in shoots under temperature dynamics. To validate the involvement of ERF012 expression in shoot response to temperature dynamics, Col-0, erf012 mutants and ERF012-OE lines were phenotypically compared at

ERF012 Responds Quickly to Temperature Dynamics
To estimate the responses of ERF012 to environmental stresses, temperature dynamics were first tested. Col-0 seedlings grown under normal conditions (22 • C) were transferred to high temperatures (37 • C) or low temperatures (4 • C) for several hours. Time-course qRT-PCR analysis of ERF012 expression in the shoots and roots was performed. The rapid responses (within 2 h) to dynamics were observed for Col-0 shoots, whereas no significant changes occurred in the roots ( Figure 6A,B). The consistent results were confirmed by pERF012:GUS lines treated at different temperatures ( Figure 6C), suggesting that ERF012 plays a critical role in shoots under temperature dynamics. To validate the involvement of ERF012 expression in shoot response to temperature dynamics, Col-0, erf012 mutants and ERF012-OE lines were phenotypically compared at different temperatures. At normal conditions (22 • C), ERF012-OE shoots had lower chlorophyll content than the Col-0 and erf012 mutants, corresponding to visible chlorosis ( Figure 6D,E). High temperature (37 • C) largely reduced the chlorophyll content in all plants with chlorosis, although the ERF012-OE lines still had the lowest levels ( Figure 6D,E). Interestingly, the chlorosis of ERF012-OE lines at 22 • C and all plants at 37 • C can be lightened largely by the IAA supply ( Figure 7A,B). In contrast, the low-temperature treatment (4 • C) did render growth retardation ( Figure 6D) and the erf012 mutants accumulated the highest anthocyanin content and ERF012-OE lines accumulated the lowest anthocyanin content ( Figure 6D,F). Anthocyanin accumulation is usually an adaptation strategy of plants to low temperatures [34]. Thus, ERF012 expression disturbed anthocyanin biosynthesis, leading to reduced low-temperature tolerance. These results indicate that ERF012 expression negatively regulates Arabidopsis adaptation of temperature dynamics.  Figure 7A,B). In contrast, the low-temperature treatment (4 °C) did render growth retardation ( Figure 6D) and the erf012 mutants accumulated the highest anthocyanin content and ERF012-OE lines accumulated the lowest anthocyanin content ( Figure 6D,F). Anthocyanin accumulation is usually an adaptation strategy of plants to low temperatures [34]. Thus, ERF012 expression disturbed anthocyanin biosynthesis, leading to reduced low-temperature tolerance. These results indicate that ERF012 expression negatively regulates Arabidopsis adaptation of temperature dynamics.

ERF012 Overexpression Reduces Plant Sensitivity to Abiotic Stresses
To further dissect the possible function of ERF012 in plant responses otic stresses, we performed a phenotypical evaluation among Col-0, erf01 ERF012-OE lines under various stresses including drought (100 mM man (100 mM NaCl) and arsenite (10 μM NaAsO2) and cadmium (15 μM C Drought stimulation limited plant growth to a large extent ( Figure 8Awith normal conditions (CK), the relative root lengths (RRLs) and relati weights (RSFWs) of erf012 mutants were lower and greater, respectivel RRLs and RSFWs of ERF012-OE plants did not alter ( Figure 8G), suggestin expression is not sensitive to drought stress. Consistent with this, expre drought stress-responsive genes [35][36][37] were comparable in Col-0 and ERF

ERF012 Overexpression Reduces Plant Sensitivity to Abiotic Stresses
To further dissect the possible function of ERF012 in plant responses to various abiotic stresses, we performed a phenotypical evaluation among Col-0, erf012 mutants and ERF012-OE lines under various stresses including drought (100 mM mannitol), salinity (100 mM NaCl) and arsenite (10 µM NaAsO 2 ) and cadmium (15 µM CdSO 4 ) toxicity. Drought stimulation limited plant growth to a large extent ( Figure 8A-C). Compared with normal conditions (CK), the relative root lengths (RRLs) and relative shoot fresh weights (RSFWs) of erf012 mutants were lower and greater, respectively, whereas the RRLs and RSFWs of ERF012-OE plants did not alter ( Figure 8G), suggesting that ERF012 expression is not sensitive to drought stress. Consistent with this, expression levels of drought stress-responsive genes [35][36][37] were comparable in Col-0 and ERF012-OE plants (Supplementary Figure S2A). Except that the #7 ERF012-OE line showed worse growth, a similar result was observed due to the salt stress, including the plant growth changes and gene responses to salt stress [38] (Figures 8A-D,G and S2B). Cd toxicity was indicated by the shoot chlorosis in all plants with a reduction of fresh weights and primary root length ( Figure 8A,E). Compared with normal conditions (CK), the RSFWs and the RRLs of ERF012-OE plants were greater and lower, respectively ( Figure 8G). Both upregulation of NRAMP3 and NRAMP4 and the downregulation of GSH2 and PCS2 [39] in ERF012-OE plants would reduce the ability of Cd tolerance in the root (Supplementary Figure S2C). Intriguingly, stress largely caused growth retardation of Col-0 and erf012 plants, whereas the ERF012-OE plants showed tolerance accompanied by RRLs and RSFWs (Figure 8A,F,G), suggesting that ERF012 overexpression reduces the plant sensitivity to As toxicity. The downregulation of NIP1;1, an arsenite channel [40], might contribute to the As tolerance of ERF012-OE plants (Supplementary Figure S2D). In summary, because of the strong inhibitory effect of ERF012 overexpression on plant growth and development at normal conditions, the low sensitivity of ERF012-OE plants to various abiotic stress was evident.  Figure S2D). In summary, because of the strong inhibitory effect of ERF012 overexpression on plant growth and development at normal conditions, the low sensitivity of ERF012-OE plants to various abiotic stress was evident.

Discussion
The AP2/ERF superfamily is one of the biggest TF families in higher plants and regulates various biological processes of plant growth and development, as well as the response to biotic and abiotic stresses. Most proteins in this superfamily, characterized by a single AP2 domain and introns, are assigned to the ERF family, whereas the proteins characterized by a tandem repetition of two AP2 domains and a small number of proteins with a single AP2 domain are assigned to the AP2 subfamily [6]. It was proposed that AP2 TFs are primarily involved in developmental programs, whereas ERF TFs are mainly responsible for environmental stimuli or hormones [8].
A general DNA element A/GCCGAC usually exists in many ABA-, drought-and cold-responsive genes and can be bound by the DREB proteins. For example, the one AP2 domain containing the CBF1/DREB1B TF was demonstrated to bind to A/GCCGAC in gel-shift assays and showed rapid cold response within 2 h, but did not change appreciably at 24 h [41]. The ERF012 protein also contains one AP2 domain and belongs to the group II, A5 subfamily corresponding to the ERF subfamily in Nakano's classification [6] and the dehydration-responsive element binding proteins (DREBs) group in Sakuma's classification [42]. It has been shown that ERF012 gene expression in the root of the 12-dayold seedlings was not responsive to the temperature dynamics [25], which is consistent with our observation (Figure 6). However, rapid responses for ERF012 were detected in seedling shoots within 2 h of low-temperature (4 • C) or high-temperature (37 • C) treatments ( Figure 6A,B). Thus, ERF012 and CBF1/DREB1B have a similar response to low temperature, suggesting that both might play a role in cold stress adaptation. Indeed, the overexpression of CBF1 in Arabidopsis induces the expression of several COR (coldregulated) genes and increases the freezing tolerance of plants [43]. In agreement with the function of CBF1, the higher anthocyanin content in ERF012-OE plants and lower anthocyanin content in erf012 mutants and Col-0 plants suggest that the expression of ERF012 plays a positive role in cold adaptability. By contrast, at 22 • C and 37 • C, the ERF012-OE plants showed severe chlorosis accompanied by reduced chlorophyll content as compared with Col-0 and erf012 plants (Figures 6 and 7), suggesting that ERF012 plays a negative role in photosynthesis at normal-temperature and high-temperature conditions and promoting the leaf senescence process. Leaf senescence represents the functional transition from nutrient assimilation to nutrient remobilization, a process involving complex genetic programs caused by various stresses such as phytohormone dynamics, drought, heat, heavy metal and nutrient deficiency, etc. The ERF012-OE plants showed leaf senescence under normal conditions ( Figure 6E), which may be caused by phytohormones or nutrient deficiency. The underdeveloped roots of ERF012-OE plants would significantly reduce nutrient uptake such as of magnesium (Mg) and iron (Fe). However, the Mg concentration in the shoots of ERF012-OE plants was higher than in that of Col-0, and no different Fe concentrations were observed in the shoots of both plants (Supplementary Figure S3), excluding their contribution to leaf senescence of ERF012-OE plants. It has been reported that elevated auxin levels, by expressing an auxin biosynthetic enzyme called the YUCCA6, which mutation can significantly delay the leaf senescence [44]. Indeed, exogenous application of IAA largely delayed the leaf senescence and accompanied the elevation of chlorophyll content in all plants at 22 • C and 37 • C conditions (Figure 7), although it was still lower in the ERF012-OE plants. The increased root hair and the enhanced expression of ethylene biosynthetic genes ACS7 and ACS11 in ERF012-OE plants ( Figure 5A-C) indicate that there is more ethylene in ERF012-OE plants. This does not exclude the ethylene, which plays an important role in the onset of leaf senescence. For example, overexpression of AtERF4 positively regulates leaf senescence and loss of function of AtERF4 delays leaf senescence in an EIN3-dependent manner [45,46]. These results indicate that expression of ERF012 might repress endogenous auxin accumulation, which at least partially contributes to leaf senescence.
The auxin-deficient growth was also evident in the ERF012-OE roots with few lateral roots (Figures 2 and 4). The root system is crucial for plant growth and the lateral roots contribute, essentially, to ensuring anchorage acquisition of water and nutrients from the soil. In Arabidopsis, lateral roots are initiated at the protoxylem poles, called founder cells, by high levels of auxin stimuli. A series of auxin components have been widely identified, for example, in a study that evaluated auxin and found it causes the degradation of IAA/AUX and increases the auxin response factors, ARF7 and ARF19, which directly regulate the auxin-mediated transcription of lateral organ boundaries-domain16/asymmetric leaves2-like18 (LBD16/ASL18) and/or LBD29/ASL16 to activate lateral root formation in roots [47]. ERF TFs regulate auxin signaling or are regulated by auxin in lateral root formation. Arabidopsis ERF109 binds directly to the promoters of auxin biosynthesis, ASA1 and YUC2, to evaluate the auxin level in lateral root primordium [48] and auxin induces degradation of the ERF13 and releases fatty acid elongase KCS16 to initiate lateral root emergence [49]. Exogenous IAA applications largely restored the lateral root formation ( Figure 4D), suggesting that ERF012 expression represses auxin accumulation in roots as validated by the DR5:GFP signals between Col-0 and ERF012-OE roots ( Figure 4C). The auxin reduction in ERF012-OE roots may also be attributed to the disturbed phenylpropanoid metabolism pathway. In ERF012-OE plants, the key enzyme gene AtC4H1 of the phenylpropanoid metabolism pathway was downregulated and directly interacted with the ERF012 protein ( Figure 4A). Mutation of the AtC4H1 gene severely impairs the synthesis of lignin and cell wall formation, leading to slow, phloem-mediated auxin transport [30]. Certainly, many cell wall formation genes were downregulated in ERF012-OE seedlings (Supplementary Figure S1B), implying that ERF012 expression coordinated multilayered regulation to ensure auxin decline in the root. Besides, ERF012 expression displayed a negative effect on the root tip growth and development, which may be due to ethylene accumulation. Ethylene accumulation dramatically inhibits root elongation and promotes root hair lushness [50,51]. Overexpression of ERF012 activated the ethylene biosynthetic genes ACS7 and ACS11 to cause root growth retardation and root hair increase ( Figure 5A-C), which could be partially counteracted by the AVG, an ethylene biosynthetic inhibitor ( Figure 5B,C). On the other hand, overexpression of ERF012 in plants did appear to lose sensitivities to various environmental stresses such as drought, salinity and heavy metals, although some of the related genes were still regulated by ERF012 directly or indirectly (Figures 8 and S2), suggesting that ERF012 has more effects on growth and development than on environmental stresses. Taken together, our findings provide a comprehensive evaluation of the functional versatility of the ERF012 transcription factor in Arabidopsis growth, development and abiotic stress responses.

Plant Materials and Growth Conditions
For the solid medium growth, seeds were sterilized with 75% alcohol (v/v) for 1 min and then immediately immersed with 1% (w/v) NaClO for 10 min. The seeds were then washed with pure water 5 times. Then, vernalization of the washed seeds was conducted and they were incubated at 4 • C for 2 days. The 1/2 strength MS medium supplemented with 1% sucrose and 1% (w/v) gellan gum (Wako Pure Chemicals, Osaka, Japan) was used for normal conditions. The solid plates were grown in a growth chamber with a 16 h light/8 h dark cycle at 22 • C. For the IAA and AVG treatments, 5-day-old seedlings grown on a normal growth medium were transferred to the medium supplemented with IAA (1 µM) or aminoethoxyvinyl glycine (AVG) (1 µM) for another 6 days of growth. In the temperature stress and the abiotic stress experiments, the 6-day-old seedlings were treated at 4 • C or 37 • C, or with mannitol (100 mM), sodium chloride (NaCl, 100 mM), sodium arsenite (NaAsO 2 , 10 µM) or cadmium chloride (CdCl 2 , 15 µM) for another 6 days of growth, respectively. Plant growth parameters were measured by using Image J software (National Institutes of Health, and LOCI University of Wisconsin; USA), with images obtained from an Olympus SZX16 stereoscopic microscope (Olympus, Tokyo, Japan) and a digital camera.

Vector Generation and Plant Transformation
To construct 35S:ERF012 transgenic plants, the open reading frame (ORF) sequences of ERF012 were amplified by polymerase chain reaction (PCR) using specific primers (Supplemental Table S1). The ERF012 CRISPR vector was generated according to the method of Yang [52]. To construct the pERF012:GUS vector, the promoter sequence of ERF012 was amplified by PCR using the specific primers, then fused with the Sma I restriction endonuclease-digested DX2181 using a 5× infusion kit (Clontech, Takara, Beijing, China). The Agrobacterium GV3101-mediated flower-dip method was used to generate the transgenic lines [53]. For the transcription activation assay of ERF012, the effector vectors were constructed according to the methods [45,54]. The ERF012 CDS sequence fragment was amplified by PCR, then inserted into the binary vector GAL4DB [55] using the EcoR I restriction endonuclease. For the yeast one-hybrid assay, the promoter of 4CL and C4H were inserted into a pHis2 vector using Sma I by the 5× infusion kit (Clontech, Takara, Beijing, China), and the CDS sequences of ERF012 were inserted into the pGADT7-rec2 vector using Xma I and Xho I with specific primers. All the specific primers are listed in Supplementary Table S1.

Transient Expression in Arabidopsis Protoplasts
The Arabidopsis protoplasts' preparation, cotransfection and the expression analysis were carried out by the modified method, as previous described [56]. We used 3 µg of the GAL4-LUC reporter plasmid and GALDB-ERF012 effector plasmid, respectively. In addition, to normalize each transfection value, 1 µg of plasmid Ubi-Rennila LUC was used as an internal control [49]. The Dual-Luciferase Reporter Assay System was used to detect the luciferase assays through a luminescence reader (TECAN Infinite M200, Switzerland; Promega, Madison, WI, USA). The data were from at least three independent experiments with consistent results.

β-Glucuronidase (GUS) Histochemical Staining and Quantification of GUS Activity
Seedlings of pERF012:GUS lines were incubated in a staining solution prepared using the GUS Histochemical Kit (bioshap, China, Cat. BL622A). The samples were observed under an Olympus SZX16 stereomicroscope. GUS activity was assayed according to the method [57].

Anthocyanin Measurement
To analyze anthocyanin content, 0.3 mg shoots were harvested and frozen with liquid nitrogen immediately. Anthocyanins were extracted with 1 mL methanol containing 1% (v/v) hydrochloric acid and then were shaken at room temperature (50 R/min) with a shaker for 18 h. After centrifuging at 14,000 rpm for 10 min, 0.4 mL of the suspension was transferred to 0.6 mL methanol containing 1% (v/v) hydrochloric acid in a 1.5 mL tube. The absorbance of diluted supernatants was measured using a microplate reader (TECAN Infinite M200, Switzerland) at 530-675 nm. The anthocyanin calculation formula is QAnthocyanins = (ODA530-0.25×ODA657)/sample fresh weight (g).

Quantification Chlorophyll Concentration
The quantification of chlorophyll concentration was performed following the modified protocol that was described by Lichtenthaler and Buschmann [58]. About 100 mg fresh weight of leaves in seedlings was ground with 2 mL of 80% isopropyl alcohol (v/v), which contained 10 mg CaCO 3 . After centrifuging at 12,000 rpm for 10 min at 4 • C, then the suspension was transferred to a 1.5 mL tube. The diluted supernatants were measured using a microplate reader (TECAN Infinite M200, Switzerland) at the absorbance wavelengths of 663.2 and 646.8 nm. The concentrations for chlorophyll a (Ca) and chlorophyll b (Cb) were calculated according to the formula: Chlorophyll = (20.29A646.8 + 8.05 A663.2) × v/w × 1000.

RNA Extraction and qRT-PCR ANALYSIS
The shoots and roots of the Col-0 and OE ERF012 grown under indicated growth conditions were sampled with three independent biological replicates for the analysis of the ERF012 expression. Total RNA was extracted using the Trizol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Then, 1 µg total RNA in a 10 µL reaction volume was used for reverse transcription into a single cDNA. Once the reverse transcription reaction was finished, the cDNA was diluted 20 times. A quantitative reverse transcription-PCR (qRT-PCR) for detecting the relative expression of genes was performed using the SYBR Green Real-Time PCR Master Mix Kit (TOYOBO, Osaka, Japan) and the CFX96 TM Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The composition (4 µL cDNA templates, 0.2 µL 10 µM primers, 5 µL SYBR Green mix, 0.4 µL sterile ddH2O) were based on 10 µL reaction volume per well for 96-well plates. Arabidopsis actin and UBQ5 were used as two endogenous controls for sample normalization. The gene-specific primer sequences used are listed in Supplementary Table S1. Expression data were normalized with the expression level of the AtActin and AtUBQ5 by the 2 −∆∆CT method [59].

Statistical Analysis
Statistical analysis was performed using SPSS16.0 for Windows software (SPSS 249 Inc., Chicago, IL, USA). Significant differences among treatments were analyzed by analysis of variance and Duncan's test at the p < 0.05 level, and Student's t-test at the * p < 0.05, ** p < 0.01 and *** p < 0.001 level.