CITF1 Functions Downstream of SPL7 to Specifically Regulate Cu Uptake in Arabidopsis

Copper (Cu) is one of the most indispensable micronutrients, and proper Cu homeostasis is required for plants to maintain essential cellular functions. Plants activate the Cu uptake system during Cu limitation. Although SPL7 (SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 7) and CITF1 (Cu-DEFICIENCY INDUCED TRANSCRIPTION FACTOR 1) are two transcription factors in Cu homeostasis, it remains unclear how SPL7 and CITF1 control the Cu uptake system. Here, we reveal that overexpression of CITF1 causes the enhanced tolerance to Cu deficiency and the elevated expression of Cu uptake genes COPT2, FRO4 and FRO5. Electrophoretic mobility shift assays (EMSA) and transient expression assays indicate that SPL7 directly binds to and activates the promoter of CITF1. The overexpression of CITF1 partially rescues the sensitivity of spl7-1 to Cu deficiency. Transcriptome data suggest that SPL7 and CITF1 coregulate the Cu-homeostasis-signaling network, and CITF1 has its own independent functions. Moreover, both SPL7 and CITF1 can directly bind to and activate the promoters of three Cu uptake genes COPT2, FRO4 and FRO5. This work shows the functions of CITF1 in the Cu-homeostasis-signaling network, providing insights into the complicated molecular mechanism underlying Cu homeostasis.


Introduction
Copper (Cu) is one of the mineral nutrients in plants and acts as a redox active cofactor in a wide variety of plant proteins such as plastocyanin, cytochrome c oxidase, Cu/Zn superoxide dismutase (CSD), ethylene receptors, laccases, ascorbate and amine oxidases, plantacyanin, polyphenol oxidases, etc. These proteins are essential for fundamental biological processes in plants, including photosynthetic and respiratory electron transport, oxidative stress protection, cell wall metabolism, ethylene perception, pathogen defense, and molybdenum cofactor biosynthesis [1][2][3]. Therefore, Cu is a vital trace element for the growth and development of plants.
In soil, Cu ions are prone to binding soil components and can be absorbed onto the surfaces of clays and Fe or Mn oxides, precipitated with carbonates and phosphates, or sequestered in silicate minerals [4]. Thus, plants often face Cu deficiency. When suffering from Cu deficiency, plants develop Cu-deficiency symptoms such as curled leaf margins, reduced fertility, abnormal fruit or seed formation, and yield damage [1,[5][6][7][8]. Although Cu is indispensable for plants, excessive Cu reacts in cells with oxygen and generates noxious reactive oxygen species that are deleterious for plant growth and development [9]. Consequently, plants have to maintain Cu levels in a proper range, namely Cu homeostasis.
Certain plants have evolved sophisticated mechanisms to efficiently acquire and utilize Cu. In response to Cu limitation, these plants economize Cu by modifying Cu allocation for the efficient utilization of that limited resource. For example, Arabidopsis plants switch their types of superoxide dismutase (SOD) to adapt to Cu deficiency; the

CITF1 Specifically Affects the Expression of Cu Uptake Genes COPT2, FRO4 and FRO5
To further investigate how CITF1 regulates Cu uptake, the homozygous citf1-1 mutant [19] was isolated and the complementation lines (COM1 and COM2) of citf1-1 ( Figure S1) were constructed. Although no visible phenotypic difference was observed between citf1-1 and the wild type under +Cu (1µM CuSO4) conditions, citf1-1 developed shorter roots than the wild type under −Cu (Cu-free) conditions ( Figure S1A,B). Meanwhile, the complementation lines grew as well as those of the wild type. Next, the rosettes from plants grown in soil were harvested and used for Cu-concentration analysis. Significantly, the Cu concentration of citf1-1 was lower than that of the wild type and the complementation lines ( Figure S1C), consistent with the previous report [19].
Next, we asked whether the elevated expression of CITF1 would enhance the Cudeficiency response. Thus, we constructed transgenic plants (CITF1-OE) constitutively expressing HA-tagged CITF1. Subsequently, we tested their growth on +Cu and −Cu media, respectively. The growth of CITF1-OE plants was similar to the wild type when grown on +Cu medium. In contrast, compared with the wild type, the CITF1-OE plants displayed an enhanced tolerance to Cu deficiency, as shown by the longer roots ( Figure 1A,B). Correspondingly, we also measured Cu concentration, finding that the CITF1-OE plants had higher Cu concentration than the wild type ( Figure 1C). These results suggest that the overexpression of CITF1 promotes the Cu accumulation. Seven-day-old seedlings were analyzed. Three biological replicates were conducted. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained at least 15 plants). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (C) Cu concentration. Leaves from plants grown in soil for 3 weeks were used for Cu measurement. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained 100 mg pooled dry leaves). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (D) Expression of Cu-deficiency responsive genes. Seedlings were grown on +Cu or −Cu medium for 7 days. Roots were used for RNA extraction. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).

CITF1 Is a Direct Target of SPL7
It was reported that CITF1 is induced by Cu deficiency [12,19]; however, it is unclear if SPL7 directly regulates the expression of CITF1. As an SBP-domain-containing protein, SPL7 recognizes GTAC cis elements in its target promoters [15,20]. Sequence analysis indicated that there are eight GTAC boxes in the CITF1 promoter ( Figure 2A). We proposed that SPL7 directly binds to the CITF1 promoter by recognizing the GTAC boxes. To confirm this hypothesis, we conducted electrophoretic mobility shift assays (EMSAs). Since the SBP domain of SPL7 is responsible for DNA binding, the SPL7 N-terminal containing the SBP domain was fused with the 6×His tag and expressed in and purified from Escherichia coli. A DNA fragment with two GTAC boxes was used as the probe in EMSAs. The recombinant His-SPL7 protein dramatically reduced the electrophoretic mobility of biotin-labeled probes, resulting in a super-shift band. The unlabeled wild-type probes (Cold- Figure 1. Overexpression of CITF1 enhances Cu uptake. (A) Seedlings were grown on +Cu or −Cu medium for 7 days. (B) Root length. Seven-day-old seedlings were analyzed. Three biological replicates were conducted. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained at least 15 plants). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (C) Cu concentration. Leaves from plants grown in soil for 3 weeks were used for Cu measurement. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained 100 mg pooled dry leaves). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (D) Expression of Cu-deficiency responsive genes. Seedlings were grown on +Cu or −Cu medium for 7 days. Roots were used for RNA extraction. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).

CITF1 Is a Direct Target of SPL7
It was reported that CITF1 is induced by Cu deficiency [12,19]; however, it is unclear if SPL7 directly regulates the expression of CITF1. As an SBP-domain-containing protein, SPL7 recognizes GTAC cis elements in its target promoters [15,20]. Sequence analysis indicated that there are eight GTAC boxes in the CITF1 promoter ( Figure 2A). We proposed that SPL7 directly binds to the CITF1 promoter by recognizing the GTAC boxes. To confirm this hypothesis, we conducted electrophoretic mobility shift assays (EMSAs). Since the SBP domain of SPL7 is responsible for DNA binding, the SPL7 N-terminal containing the SBP domain was fused with the 6×His tag and expressed in and purified from Escherichia coli. A DNA fragment with two GTAC boxes was used as the probe in EMSAs. The recombinant His-SPL7 protein dramatically reduced the electrophoretic mobility of biotin-labeled probes, resulting in a super-shift band. The unlabeled wild-type probes (Cold-Probe) reduced the binding of biotin-labeled probes with SPL7, whereas the cold mutated probes (Cold-Probem) with the mutated GTAC boxes did not ( Figure 2B).
To further determine whether SPL7 activates the expression of CITF1, we performed transient expression assays. The promoter region of CITF1 was used to drive a nuclear localization signal (NLS)-fused GFP gene as the reporter. The full-length coding sequences of SPL7 and SPL9 were respectively driven by the 35S promoter as the effectors, and SPL9 was used as the control effector ( Figure 2C). The reporter was co-expressed with the effectors Pro35S:SPL9 and Pro35S:SPL7, respectively, in tobacco leaves, and then the expression of GFP was examined. Compared with the control effector Pro35S:SPL9, Pro35S:SPL7 significantly activated the promoter of CITF1 ( Figure 2C). Taken together, our data suggest that SPL7 directly activates the expression of CITF1. EMSA showing that SPL7 directly binds to the promoter of CITF1. Either His-SPL7 or 6×His was incubated with the biotin-labeled probes. Biotin-probe, biotin-labeled probe; cold-probe, unlabeled probe; cold-probe-m, unlabeled mutated probe with mutated GTAC-boxes. (C) Activation of CITF1 promoter by SPL7. The CITF1 promoter was used to drive the NLS-GFP as the reporter. SPL7 and SPL9 driven respectively by the CaMV 35S promoter were used as the effectors. The SPL9 was used as a negative control effector. The GFP/NPT II ratio represents the GFP transcript abundance relative to the internal control NPT II transcript. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).

Genetic Interaction of CITF1 and SPL7
Given that CITF1 is involved in the regulation of Cu homeostasis and is a target of SPL7, we sought to test whether the overexpression of CITF1 could rescue the defective response of spl7-1 to Cu-deficiency. The CITF1 overexpression construct was introduced into the spl7-1 mutant, and the transcript abundance of CITF1 was examined ( Figure S3). Subsequently, plants were grown on -Cu medium for phenotype analysis. As expected, the overexpression of CITF1 partially rescued the sensitivity of spl7-1 to Cu deficiency ( Figure 3A,B). Meanwhile, the expression of COPT2, FRO4, and FRO5 was dramatically increased in the spl7-1/CITF1-OE plants compared with the spl7-1 mutant ( Figure 3C). These results suggest the partial function complementation of spl7-1 by the overexpression of CITF1. (B) EMSA showing that SPL7 directly binds to the promoter of CITF1. Either His-SPL7 or 6×His was incubated with the biotin-labeled probes. Biotin-probe, biotin-labeled probe; cold-probe, unlabeled probe; cold-probe-m, unlabeled mutated probe with mutated GTAC-boxes. (C) Activation of CITF1 promoter by SPL7. The CITF1 promoter was used to drive the NLS-GFP as the reporter. SPL7 and SPL9 driven respectively by the CaMV 35S promoter were used as the effectors. The SPL9 was used as a negative control effector. The GFP/NPT II ratio represents the GFP transcript abundance relative to the internal control NPT II transcript. Data represent arithmetic means ± SD (n = 3). * p < 0.01.
To further determine whether SPL7 activates the expression of CITF1, we performed transient expression assays. The promoter region of CITF1 was used to drive a nuclear localization signal (NLS)-fused GFP gene as the reporter. The full-length coding sequences of SPL7 and SPL9 were respectively driven by the 35S promoter as the effectors, and SPL9 was used as the control effector ( Figure 2C). The reporter was co-expressed with the effectors Pro 35S :SPL9 and Pro 35S :SPL7, respectively, in tobacco leaves, and then the expression of GFP was examined. Compared with the control effector Pro 35S :SPL9, Pro 35S :SPL7 significantly activated the promoter of CITF1 ( Figure 2C). Taken together, our data suggest that SPL7 directly activates the expression of CITF1.

Genetic Interaction of CITF1 and SPL7
Given that CITF1 is involved in the regulation of Cu homeostasis and is a target of SPL7, we sought to test whether the overexpression of CITF1 could rescue the defective response of spl7-1 to Cu-deficiency. The CITF1 overexpression construct was introduced into the spl7-1 mutant, and the transcript abundance of CITF1 was examined ( Figure S3). Subsequently, plants were grown on −Cu medium for phenotype analysis. As expected, the overexpression of CITF1 partially rescued the sensitivity of spl7-1 to Cu deficiency ( Figure 3A,B). Meanwhile, the expression of COPT2, FRO4, and FRO5 was dramatically increased in the spl7-1/CITF1-OE plants compared with the spl7-1 mutant ( Figure 3C). These results suggest the partial function complementation of spl7-1 by the overexpression of CITF1.
x FOR PEER REVIEW 5 of 15 Seven-day-old seedlings were analyzed. Three biological replicates were conducted. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained at least 15 plants). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (C) Expression of Cu-deficiency responsive genes. Seedlings were grown on +Cu or -Cu medium for 7 days. Roots were used for RNA extraction. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).
To further uncover the genetic relationship of SPL7 and CITF1 in Cu-deficiency response, we constructed the citf1-1 spl7-1 double mutants by crossing the corresponding citf1-1 and spl7-1 mutants ( Figure S4A-C). When grown on +Cu medium, neither the single mutants nor the double mutant showed any obvious phenotypic differences compared with wild type plants. In contrast, when exposed to -Cu medium, all mutants displayed Cu-deficient symptoms, and the double mutant had the severest phenotypes ( Figure S4D). Similarly, when grown in hydroponic solution, the growth of citf1-1 spl7-1 was severely retarded ( Figure S4E). We then determined the expression of their common target genes, COPT2, FRO4, and FRO5. In consistence with the phenotypes, the citf1-1 spl7-1 mutant had the lowest levels of COPT2, FRO4, and FRO5 compared with citf1-1 and spl7-1 ( Figure  4). Taken together, our data suggest that CITF1 functions downstream of SPL7, and it plays an additive role in the regulation of Cu uptake genes. Seven-day-old seedlings were analyzed. Three biological replicates were conducted. Data represent arithmetic means ± SD (n = 3 biological replicates, each of which contained at least 15 plants). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01). (C) Expression of Cu-deficiency responsive genes. Seedlings were grown on +Cu or −Cu medium for 7 days. Roots were used for RNA extraction. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).
To further uncover the genetic relationship of SPL7 and CITF1 in Cu-deficiency response, we constructed the citf1-1 spl7-1 double mutants by crossing the corresponding citf1-1 and spl7-1 mutants ( Figure S4A-C). When grown on +Cu medium, neither the single mutants nor the double mutant showed any obvious phenotypic differences compared with wild type plants. In contrast, when exposed to −Cu medium, all mutants displayed Cu-deficient symptoms, and the double mutant had the severest phenotypes ( Figure S4D). Similarly, when grown in hydroponic solution, the growth of citf1-1 spl7-1 was severely retarded ( Figure S4E). We then determined the expression of their common target genes, COPT2, FRO4, and FRO5. In consistence with the phenotypes, the citf1-1 spl7-1 mutant had the lowest levels of COPT2, FRO4, and FRO5 compared with citf1-1 and spl7-1 ( Figure 4). Taken together, our data suggest that CITF1 functions downstream of SPL7, and it plays an additive role in the regulation of Cu uptake genes.
x FOR PEER REVIEW 6 of 15

CITF1 and SPL7 Coregulate Cu-Deficiency Response
We have confirmed that CITF1 plays a key role in the maintenance of Cu homeostasis. Given that SPL7 and CITF1 are two important transcription factors, we wondered how they control the Cu-deficiency response network. To dissect the roles of CITF1 and SPL7 in controlling transcription during Cu depletion, we performed transcriptional profiling of the wild type, citf1-1, and spl7-1 in response to Cu-deficiency. We identified 220 genes responding to Cu deficiency with an at least twofold upward change in transcript levels, and 637 genes with an at least twofold downward change, and we designated them Cudeficiency upregulated (−Cu_Up) and downregulated (−Cu_Down) genes, respectively ( Figure 5A; Table S1). Among those -Cu_Up genes, the transcript abundance of 74 genes is at least twofold lower in the spl7-1 than in the wild type under -Cu conditions, and they are named SPL7 dependent −Cu_Up (−Cu_Up_SPL7) genes. Of the −Cu_Down genes, 157 genes, whose transcript abundance is at least two-fold higher in the spl7-1 than in the wild type under − Cu conditions, are named SPL7 dependent − Cu_Down (−Cu_Down_SPL7) genes. By a similar filtration, we identified 61 CITF1 dependent -Cu_Up (−Cu_Up_CITF1) genes and 138 CITF1 dependent − Cu_Down (−Cu_Down_CITF1) genes. We further analyzed and compared the SPL7 and CITF1 regulation networks ( Figure 5B; Table S1). Fifty-four genes including the well-known − Cu_Up genes are induced by Cu deficiency in the wild type and citf1-1, but not in the spl7-1, suggesting that their upregulation by −Cu is under the control of SPL7, but not of the CITF1. Correspondingly, ninety-six genes are regulated negatively only by SPL7. Similarly, we identified 41 and 77 genes which are upregulated and downregulated respectively in a CITF1 dependent fashion. Meanwhile, we found that 81 genes are responsive to Cu deficiency under the control of both SPL7 and CITF1 (Table S1).
From the transcriptome data, we found that the expression of Cu-economy genes is controlled specifically by SPL7. It is well-known that Cu-miRNAs (miR397, miR398, miR408 and miR857) negatively regulate the expression of Cu-economy genes CSDs, CCS, LACs, and ARPN [12,[15][16][17]. In support of this, our transcriptome data also show the opposite expression patterns between Cu-miRNAs and Cu-economy genes. Consistent with the previous reports [12,21], the expression of FSD1 is activated significantly by Cu deficiency, further supporting that FSD1 compensates for the lack of CSDs when Cu Expression of Cu uptake genes in citf1-1 spl7-1. Seedlings were grown on +Cu or −Cu medium for 7 days. Roots were used for RNA extraction. Data represent arithmetic means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).

CITF1 and SPL7 Coregulate Cu-Deficiency Response
We have confirmed that CITF1 plays a key role in the maintenance of Cu homeostasis. Given that SPL7 and CITF1 are two important transcription factors, we wondered how they control the Cu-deficiency response network. To dissect the roles of CITF1 and SPL7 in controlling transcription during Cu depletion, we performed transcriptional profiling of the wild type, citf1-1, and spl7-1 in response to Cu-deficiency. We identified 220 genes responding to Cu deficiency with an at least twofold upward change in transcript levels, and 637 genes with an at least twofold downward change, and we designated them Cu-deficiency upregulated (−Cu_Up) and downregulated (−Cu_Down) genes, respectively ( Figure 5A; Table S1). Among those −Cu_Up genes, the transcript abundance of 74 genes is at least twofold lower in the spl7-1 than in the wild type under −Cu conditions, and they are named SPL7 dependent −Cu_Up (−Cu_Up_SPL7) genes. Of the −Cu_Down genes, 157 genes, whose transcript abundance is at least two-fold higher in the spl7-1 than in the wild type under −Cu conditions, are named SPL7 dependent −Cu_Down (−Cu_Down_SPL7) genes. By a similar filtration, we identified 61 CITF1 dependent −Cu_Up (−Cu_Up_CITF1) genes and 138 CITF1 dependent −Cu_Down (−Cu_Down_CITF1) genes. We further analyzed and compared the SPL7 and CITF1 regulation networks ( Figure 5B; Table S1). Fifty-four genes including the wellknown −Cu_Up genes are induced by Cu deficiency in the wild type and citf1-1, but not in the spl7-1, suggesting that their upregulation by −Cu is under the control of SPL7, but not of the CITF1. Correspondingly, ninety-six genes are regulated negatively only by SPL7. Similarly, we identified 41 and 77 genes which are upregulated and downregulated respectively in a CITF1 dependent fashion. Meanwhile, we found that 81 genes are responsive to Cu deficiency under the control of both SPL7 and CITF1 (Table S1).
From the transcriptome data, we found that the expression of Cu-economy genes is controlled specifically by SPL7. It is well-known that Cu-miRNAs (miR397, miR398, miR408 and miR857) negatively regulate the expression of Cu-economy genes CSDs, CCS, LACs, and ARPN [12,[15][16][17]. In support of this, our transcriptome data also show the opposite expression patterns between Cu-miRNAs and Cu-economy genes. Consistent with the previous reports [12,21], the expression of FSD1 is activated significantly by Cu deficiency, further supporting that FSD1 compensates for the lack of CSDs when Cu availability is limiting in cells. SPL7 also controls the expression of many Cu translocation genes which are not affected by CITF1, such as YSL2, YSL3, CCH and COPT1 (Table S2). Thus, SPL7 may control the Cu distribution between tissues or organs. deficiency in roots, dependently on SPL7 (Table S2; Figure 5C). Similarly, CLE2 was also induced by Cu deficiency in roots, which is involved in expansion of plant root systems in a nitrogen-dependent manner [26]. Therefore, it is likely that NRT2.7 and CLE2 are involved in the crosstalk between Cu signaling and nitrogen signaling. Interestingly, we found that SPL7 and CITF1 coregulate Cu-deficiency-responsive genes including COPT2, FRO4, FRO5, and MT1C (Metallothionein 1C) (Table S2; Figure 5C). Taken together, our data suggest that SPL7 and CITF1 not only coregulate Cu-deficiency signaling pathway, but also have functions independent of each other in the Cu-deficiency response. Cu is required for ethylene perception since Cu acts as a cofactor of ethylene receptors [22]. Among those genes regulated by CITF1, EIL2 (ETHYLENE-INSENSITIVE3-like 2) and ERF042 (ETHYLENE-RESPONSIVE TRANSCRIPTION FACTOR 042) encodes transcription factors involved in ethylene signaling transduction (Table S2). Therefore, EIL2 and ERF042 may function as the nodes in the crosstalk between Cu signaling and ethylene signaling. It was reported that CITF1 transcription is responsive to jasmonic acids (JA) treatment and involved in regulation of JA-synthesis genes in anthers [19,23]. The expression levels of JA-synthesis genes were retrieved (Table S3). We observed that the expression of LOX3, LOX4, AOS, AOC1, AOC2, and AOC3 was suppressed by Cu deficiency, and their expression levels were lower in both spl7-1 and citf1-1 than in the wild type. Therefore, SPL7 and CITF1 mediate the interaction between Cu signaling and JA synthesis.
Recently, it was reported that SPL7 mediates the copper-deficiency response in the presence of high nitrogen in Arabidopsis thaliana [24]. Our data suggest that the nitrogen signaling is also affected by SPL7 under Cu-deficiency conditions. The transcript levels of NRT2.7, encoding a tonoplast nitrate transporter [25], were upregulated under Cu deficiency in roots, dependently on SPL7 (Table S2; Figure 5C). Similarly, CLE2 was also induced by Cu deficiency in roots, which is involved in expansion of plant root systems in a nitrogen-dependent manner [26]. Therefore, it is likely that NRT2.7 and CLE2 are involved in the crosstalk between Cu signaling and nitrogen signaling. Interestingly, we found that SPL7 and CITF1 coregulate Cu-deficiency-responsive genes including COPT2, FRO4, FRO5, and MT1C (Metallothionein 1C) (Table S2; Figure 5C). Taken together, our data suggest that SPL7 and CITF1 not only coregulate Cu-deficiency signaling pathway, but also have functions independent of each other in the Cu-deficiency response.

CITF1 and SPL7 Directly Bind to the Promoters of Cu Uptake Genes
Given that the transcript abundance of COPT2, FRO4, and FRO5 is coregulated by CITF1 and SPL7, we proposed that COPT2, FRO4 and FRO5 are the direct targets of SPL7 and CITF1. Generally, bHLH transcription factors recognize an E-box [27] and SPL7 recognizes a GTAC box. Sequence analysis indicated that there are several E-boxes and GTAC boxes in the promoters of COPT2, FRO4, and FRO5 ( Figure 6A), implying their interactions with CITF1 and SPL7. To investigate this hypothesis, we conducted EMSAs. The 6×His-tagged CITF1 was expressed in and purified from Escherichia coli. When His-CITF1 was incubated with the biotin-labeled pCOPT2-E probe with an E-box, His-CITF1 could bind to it, and this binding could be competed by the unlabeled pCOPT2-E, but not the mutation version pCOPT2-mE. In contrast, His-CITF1 could not bind to the promoter of COPT1 ( Figure S5). This suggests that CITF1 directly binds to the promoter of COPT2 ( Figure 6B). Similarly, we also confirmed that CITF1 binds to the promoters of FRO4 and FRO5 ( Figure 6B). Meanwhile, we performed EMSAs with His-SPL7 and the probes of COPT2, FRO4, and FRO5 with GTAC boxes, finding that SPL7 also directly binds to the promoters of COPT2, FRO4, and FRO5, but not the promoter of COPT1 (Figures 6B and S5). Cu medium for 7 days. Roots were used for RNA extraction. Data represent means ± SD (n = 3). Different letters above each bar indicate statistically significant differences (ANOVA, p < 0.01).

CITF1 and SPL7 Directly Bind to the Promoters of Cu Uptake Genes
Given that the transcript abundance of COPT2, FRO4, and FRO5 is coregulated by CITF1 and SPL7, we proposed that COPT2, FRO4 and FRO5 are the direct targets of SPL7 and CITF1. Generally, bHLH transcription factors recognize an E-box [27] and SPL7 recognizes a GTAC box. Sequence analysis indicated that there are several E-boxes and GTAC boxes in the promoters of COPT2, FRO4, and FRO5 ( Figure 6A), implying their interactions with CITF1 and SPL7. To investigate this hypothesis, we conducted EMSAs. The 6×His-tagged CITF1 was expressed in and purified from Escherichia coli. When His-CITF1 was incubated with the biotin-labeled pCOPT2-E probe with an E-box, His-CITF1 could bind to it, and this binding could be competed by the unlabeled pCOPT2-E, but not the mutation version pCOPT2-mE. In contrast, His-CITF1 could not bind to the promoter of COPT1 ( Figure S5). This suggests that CITF1 directly binds to the promoter of COPT2 ( Figure 6B). Similarly, we also confirmed that CITF1 binds to the promoters of FRO4 and FRO5 ( Figure 6B). Meanwhile, we performed EMSAs with His-SPL7 and the probes of COPT2, FRO4, and FRO5 with GTAC boxes, finding that SPL7 also directly binds to the promoters of COPT2, FRO4, and FRO5, but not the promoter of COPT1 (Figures 6B and  S5). To further investigate the regulation of COPT2, FRO4, and FRO5 by SPL7 and CITF1, we used the reporter-effector transient expression system mentioned above. The promoters of COPT2 and FRO5 were respectively fused with the NLS-GFP reporter gene as the reporters (Figure 7). Pro 35S :bHLH60, Pro 35S :SPL9, Pro 35S :CITF1 and Pro 35S :SPL7 were used as the effectors. When Pro COPT2 :NLS-GFP was used as the reporter, both Pro 35S :CITF1 and Pro 35S :SPL7 effectors considerably promoted the expression of GFP compared with the control effectors Pro 35S :bHLH60 and Pro 35S :SPL9. Similar results were also obtained when Pro FRO5 :NLS-GFP was used as the reporter. Taken together, our data suggest that both SPL7 and CITF1 directly bind to and activate the promoters of COPT2, FRO4, and FRO5.
To further investigate the regulation of COPT2, FRO4, and FRO5 by SPL7 and CITF1, we used the reporter-effector transient expression system mentioned above. The promoters of COPT2 and FRO5 were respectively fused with the NLS-GFP reporter gene as the reporters (Figure 7). Pro35S:bHLH60, Pro35S:SPL9, Pro35S:CITF1 and Pro35S:SPL7 were used as the effectors. When ProCOPT2:NLS-GFP was used as the reporter, both Pro35S:CITF1 and Pro35S:SPL7 effectors considerably promoted the expression of GFP compared with the control effectors Pro35S:bHLH60 and Pro35S:SPL9. Similar results were also obtained when ProFRO5:NLS-GFP was used as the reporter. Taken together, our data suggest that both SPL7 and CITF1 directly bind to and activate the promoters of COPT2, FRO4, and FRO5. Figure 7. Activation of COPT2 and FRO5 promoters by SPL7 and CITF1. The COPT2 and FRO5 promoters were used to drive the NLS-GFP as the reporters. bHLH60, CITF1, SPL7, and SPL9 respectively driven by the CaMV 35S promoter were used as the effectors. The bHLH60 and SPL9 were used as negative control effectors. The GFP/NPT II ratio represents the GFP transcript abundance relative to the internal control NPT II transcript. Data represent means ± SD (n = 3). Different letters indicate statistically significant differences of the mean values (p < 0.01).

Discussion
Plants have evolved sophisticated mechanisms to maintain Cu homeostasis. It has been documented that plants respond to Cu deficiency through upregulation of a number of genes involved in the acquisition of Cu from the extracellular environment, its transport into the cells, and its metabolism. It is well known that SPL7 is a master regulator of Cudeficiency response in Arabidopsis. In this study, we report that SPL7 regulates the expression of CITF1 by directly associating with its promoter, and both of them are required for the activation of the Cu uptake system under Cu deficiency conditions. bHLH proteins are a superfamily of transcription factors and important regulatory components in transcriptional networks in plants. There are 162 bHLH members in Arabidopsis [28]. Many bHLH transcription factors have been characterized to control the Figure 7. Activation of COPT2 and FRO5 promoters by SPL7 and CITF1. The COPT2 and FRO5 promoters were used to drive the NLS-GFP as the reporters. bHLH60, CITF1, SPL7, and SPL9 respectively driven by the CaMV 35S promoter were used as the effectors. The bHLH60 and SPL9 were used as negative control effectors. The GFP/NPT II ratio represents the GFP transcript abundance relative to the internal control NPT II transcript. Data represent means ± SD (n = 3). Different letters indicate statistically significant differences of the mean values (p < 0.01).

Discussion
Plants have evolved sophisticated mechanisms to maintain Cu homeostasis. It has been documented that plants respond to Cu deficiency through upregulation of a number of genes involved in the acquisition of Cu from the extracellular environment, its transport into the cells, and its metabolism. It is well known that SPL7 is a master regulator of Cu-deficiency response in Arabidopsis. In this study, we report that SPL7 regulates the expression of CITF1 by directly associating with its promoter, and both of them are required for the activation of the Cu uptake system under Cu deficiency conditions. bHLH proteins are a superfamily of transcription factors and important regulatory components in transcriptional networks in plants. There are 162 bHLH members in Arabidopsis [28]. Many bHLH transcription factors have been characterized to control the expression of genes involved in metal homeostasis [29][30][31][32][33][34][35][36][37][38][39][40][41][42]. CITF1 was previously identified as a Cu-deficiency-responsive transcription factor regulating Cu homeostasis [19]. Here, we further established the functions of CITF1 in the Cu-homeostasis-signaling pathway. Cu deficiency leads to a significant increase in the CITF1 transcript abundance [12,15,19]. When the function of CITF1 is lost, the expression of its target genes (COPT2, FRO4, and FRO5) is reduced ( Figure S2) [19]. Lower expression levels of COPT2, FRO4, and FRO5 can lead to decreased solubility and transport of Cu into root epidermal cells [12,43]. The overexpression of CITF1 results in the increased expression of Cu uptake genes and elevated Cu concentration ( Figure 1C,D). We also provide evidence that CITF1 directly regulates the expression of Cu uptake genes (COPT2, FRO4, and FRO5) in response to Cu deficiency (Figures 6 and 7). Thus, the maintenance of Cu homeostasis requires the appropriate expression levels of CITF1.
We confirmed that there is a direct linkage between SPL7 and CITF1 because SPL7 directly binds to and activates the promoter of CITF1 (Figure 2). However, it remains unknown how the transcription of CITF1 is suppressed under Cu-sufficiency conditions. We assume that under Cu-sufficiency conditions, either a negative regulator which targets the CITF1 promoter is activated, or its positive regulator SPL7 is inactivated, or both. From our transcriptome data, we identified many transcription factors which are downregulated under Cu-deficiency conditions, such as MYB87, WRKY46, ZAT7, and ERF021 (Table S1). Further investigation is required to clarify whether these transcription factors negatively regulate the expression of CITF1 under Cu-deficiency conditions. Given that CITF1 is transcriptionally activated in response to Cu deficiency, it is expected that SPL7 is more active and (or) more stable under Cu-deficiency conditions than under Cu-sufficiency conditions. Therefore, further investigation on the SPL7 activity in response to Cu deficiency will help us understand the regulation mechanism of Cu homeostasis.
Plants can sense the internal Cu status and reprogram the transcription of Cu-homeostasisassociated genes which are responsible for Cu uptake and allocation. Although the Arabidopsis Cu sensor is not identified yet, SPL7 is a major regulator upstream of the Cu-deficiencyresponsive signaling network. The interaction between the SPL7 protein and the promoter of CITF1 also provides evidence that the Cu-homeostasis regulation mediated by SPL7 is partly dependent on CITF1. Our genetic evidence suggests that CITF1 acts downstream of SPL7 to take part in the regulation of Cu homeostasis (Figure 3). Although CITF1 is not induced by Cu deprivation in the spl7-1, the citf1-1 spl7-1 double mutants displayed severer Cu-deficiency symptoms than the spl7-1 ( Figure S4D,E) [19]. It implies that the basal CITF1 expression is required for the maintenance of Cu homeostasis. In support of this, we found that some genes are only impacted by CITF1, not by SPL7, under −Cu conditions (Table S2), which indicates the functions of CITF1 are independent of SPL7 in Cu homeostasis. The lower levels of Cu uptake genes in the citf1-1 spl7-1 than in the citf1-1 and spl7-1 suggest that SPL7 and CITF1 have additive roles in regulating COPT2, FRO4, and FRO5 ( Figure 4). Our data also provide evidence that both SPL7 and CITF1 directly activate their expression (Figures 6 and 7). Additionally, Yan et al., (2017) reported that the citf1-1 spl7-1 plants are unable to grow in soil under Cu-sufficient conditions, but we did not observe this phenomenon. It is possible that the difference of soil components accounts for the different phenotypes in soil.
The regulatory network mediated by SPL7 was well established in Arabidopsis [12,15,44]. Our transcriptome data show that, among the genes responsive to Cu deficiency, 27% and 23% are affected by SPL7 and CITF1, respectively ( Figure 5B). Bernal et al. [12] identified 206 SPL7-dependent Cu-deficiency-responsive genes and we identified 231. Most of them are shared in both studies. Therefore, we feel confident that this study captured a large number of transcriptional responses controlled by SPL7 and CITF1. By transcriptome analysis, we identified CITF1-dependent and -independent Cu-deficiency-responsive genes. The expression of several ethylene signaling genes is affected only by CITF1 (Table S2). Our transcriptome data show that COPT2, FRO4 and FRO5 are coregulated by CITF1 and SPL7, and we show evidence that CITF1 and SPL7 target different cis elements in the promoters of COPT2, FRO4, and FRO5. It was reported that CITF1 cooperates with SPL7 to regulate JA synthesis in flowers by affecting the expression of LOX3, LOX4, AOS, AOC1, AOC2, AOC3, and so on [19]. A recent study also uncovered that jasmonate induces alternative splicing of CITF1 [23]. In our root transcriptome, the expression of LOX3, LOX4, AOS, AOC1, AOC2, and AOC3 in the spl7-1 and citf1-1 is lower than in the wild type under Cu sufficient conditions (Table S3), implying that SPL7 and CITF1 participate in the regulation of JA synthesis in roots. Further investigation is required to establish if there is a direct connection between CITF1 and the JA-synthesis-associated genes. Unexpectedly, we found that some Fe-deficiency-responsive genes are positively regulated by both SPL7 and CITF1 (Table S1), inconsistent with the previous reports [12,44]. In our experiments, EDTA was used to chelate iron in the 1/2 MS media for plant growth. It was reported that the medium containing EDTA becomes iron deficient because iron becomes unchelated and forms an unavailable precipitate as EDTA is oxidized by light [45]. It is likely that the Fe-deficient environment caused by EDTA oxidation resulted in the unexpected upregulation of several Fe-deficiency-responsive genes.
CITF1 and SPL7 are two key transcriptional regulators that control the expression of genes involved in a range of biological activities important for responding to Cu-deficient conditions. Our work sheds light on a new regulatory mechanism driving the root high affinity Cu uptake system in plants, and unravels the essential role of the CITF1 transcription factor directly stimulating Cu uptake gene expression during the Cu-deficiency responses. Further elucidating the interplay among genes in the CITF1-SPL7 network should provide much-needed insights into the mechanism by which plants adapt to Cu deficiency.
Seeds were surface-sterilized with 75% ethanol for 15 min and then washed three times with double-distilled water. After plated on 1/2 MS media, plates stayed for 2 d at 4 • C before shifting in greenhouse. Cu-sufficient (+Cu) medium is 1/2 MS medium with 1% sucrose, 0.7% agar and 1 µM CuSO 4 at pH 5.8, and Cu-deficient (−Cu) medium is the same without CuSO 4 . In all experiments on medium, contaminant metals from agar were removed by washing 3 times with 10 mM EDTA for 12 h, then 6 times with ultrapure water. Plates were placed in a culture room at 22 • C under a 16 h light/8 h dark photoperiod. For phenotype analysis, three biological replicates with each containing fifteen plants were conducted.

Plasmid Construction and Plant Transformation
The seamless cloning kit (Beyotime, Shanghai, China) was used for plasmid construction. One microgram genomic DNA from Arabidopsis was used as the template for amplification of the upstream regulatory promoter sequences. For complementation of the citf1-1 mutant, a 1.4 k genomic fragment upstream of CITF1 translation start site was PCR-amplified from Col-0 and used to drive the HA-tagged CITF1. For the generation of CITF1-OE or spl7-1/CITF1-OE plants, the CaMV 35S promoter was used to drive the HA-tagged CITF1 in the pOCA30 binary vector. For the generation of Pro CITF1 :GUS plants, a 1.4 k genomic fragment upstream of CITF1 translation start site was fused with the GUS reporter in the pOCA28 binary vector. For the construction of reporters, the corresponding promoters of CITF1, COPT2, and FRO5 were respectively fused with the NLS-GFP sequence in the pOCA28 vector [35]. For the construction of effectors, CITF1, bHLH60, SPL7, and SPL9 were inserted respectively between the CaMV 35S promoter and poly(A) in the pOCA30 vector. Primers used for these constructs are listed in Table S4. All constructs were introduced into Agrobacterium tumefaciens strain EHA105.
Arabidopsis transformation was conducted by the floral dip method [46]. Pro 35S :HA-CITF1 was transformed into the wild type and spl7-1 plants. Pro CITF1 :HA-CITF1 was transformed into the citf1-1 plants. Pro CITF1 :GUS was transformed into the wild type plants. Transgenic plants were selected with the use of 50 mg/L kanamycin.

Transient Expression Assays
The plasmids (Pro 35S :bHLH60, Pro 35S :SPL9, Pro 35S :CITF1, Pro 35S :SPL7, Pro CITF1 :NLS-GFP, Pro COPT2 :NLS-GFP, and Pro FRO5 :NLS-GFP) were transformed into Agrobacterium tumefaciens strain EHA105. Agrobacterial cells were infiltrated into leaves of Nicotiana benthamiana by the infiltration buffer (0.2 mM acetosyringone, 10 mM MgCl 2 , and 10 mM MES, pH 5.6). In the transient expression assays, the final optical density at 600 nm value was 0.5 for reporters and effectors. After infiltration, plants were kept in dark for two days. The infiltrated leaves were harvested and used for RNA extraction. The transcript abundance of GFP was normalized to NPT II in the reporter vector.

EMSA
We used the Chemiluminescent EMSA Kit (Beyotime, Shanghai, China) following the manufacturer's protocol to do EMSA Assays. The SBP domain of SPL7 (from amino acid position 122 to 228) and the full-length CITF1 were respectively fused in frame with the 6×His tag in the expression vector pET-28a. Primers used for these constructs are listed in Table S4. The resulting plasmids were introduced into Escherichia coli BL21 (DE3) for recombinant His-SPL7 and His-CITF1 protein expression. Cultures were grown overnight at 28 • C with 0.5 mM isopropyl β-D-1-thiogalactopyranoside. Proteins were extracted and purified by the His-tag Protein Purification Kit (Beyotime, Shanghai, China) following the manufacturer's protocol. The DNA fragments used for probes were synthesized and biotin was labeled to the 5 terminal of DNA. Biotin-unlabeled fragments of the same sequences or mutated sequences were used as competitors, and the negative control was His protein alone.

Cu Measurement
Plants were grown in soil for 4 weeks and leaves were harvested for the measurement of Cu concentration. Leaves were rinsed with deionized water and dried at 65 • C for 3 d. About 100 mg dry weight was wet-ashed with 5 mL of 11 M HNO 3 and 1 mL of 12 M HClO 4 for 20 min at 220 • C. Each sample was diluted to 16 mL with 18 MΩ water and Cu concentration was analyzed on a Thermo SCIENTIFIC ICP-MS (iCAP6300), Waltham, MA, USA. Reference standards were used for quantification. Each biological sample was measured three times and a mean value was calculated.

Gene Expression Analysis
Roots were collected from plants grown as described above and were frozen in liquid nitrogen. Roots were ground in liquid nitrogen to a fine powder. Total RNA was extracted by the Trizol reagent (Invitrogen, Carlsbad, CA, USA). One microgram of total RNA was used for oligo (dT)18 primed cDNA synthesis by the use of HiScript II Q RT. SuperMix for qPCR (+gDNA wiper) (Vazyme, Nanjing, China) according to the manufacturer's protocol in a 20 µL volume. Prior to RT-qPCR analysis, primer and cDNA concentrations were optimized to reach the gene amplification efficiency of 100% ± 10%. Two microliters of 15-fold diluted cDNA were used in a 20 µL qPCR reaction on a Light-Cycler 408 realtime PCR machine (Roche, Basel, Switzerland). Ten microliters of AceQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) was used for a 20 µL qPCR reaction. The PCR program consisted of one cycle (95 • C, 4 min), 40 cycles (95 • C, 10 s; 60 • C, 20 s; 72 • C, 15 s), and one cycle (72 • C, 5 min), followed by a melting curve program (55 to 90 • C in increasing steps of 3 • C). Real-time PCR experiments were conducted using three independent experiments with each containing three technical replicates. The relative expression of genes was normalized to ACT2 and TUB2. All primers used for qRT-PCR are listed in Table S4.

High-Throughput Sequencing of mRNA, and Differential Gene Expression Analysis
For transcriptome analysis, roots of wild-type, citf1-1, and spl7-1 plants grown on +Cu or −Cu 1/2 MS medium for 7 days were used for RNA extraction. Three replicates of the whole roots analysis were performed. Total RNA of each sample was extracted according to the instruction manual of the TRlzol Reagent (Invitrogen, Carlsbad, CA, USA). RNA integrity and concentration were checked using an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). The mRNA was isolated by NEBNext Poly (A) mRNA Magnetic Isolation Module (NEB, E7490). The cDNA library was constructed following the manufacturer's instructions of NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, E7530) and NEBNext Multiplex Oligos for Illumina (NEB, E7500). In briefly, the enriched mRNA was fragmented into approximately 200nt RNA inserts, which were used to synthesize the first-strand cDNA and the second cDNA. The double-stranded cDNA was performed end-repair/dA-tail and adaptor ligation. The suitable fragments were isolated by Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA), and enriched by PCR amplification. Finally, the constructed cDNA libraries were sequenced on an Illumina sequencing platform.
Transcript abundance was concluded to increase/decrease under Cu deficiency for a gene when arithmetic means of transcript abundance differed by a factor of at least 2. Changes in transcript levels were concluded to be dependent on SPL7 if log 2 FC (wild type −Cu versus spl7-1 −Cu) > 1 for log 2 FC (wild type −Cu versus wild type +Cu) > 1, and log 2 FC (wild type −Cu versus spl7-1 −Cu) < 1 for log 2 FC (wild type −Cu versus wild type +Cu) < −1. The CITF1 dependent transcripts were obtained by a similar filtration.