Genome-Wide Identiﬁcation and Analysis of the Growth-Regulating Factor Family in Zanthoxylum armatum DC and Functional Analysis of ZaGRF6 in Leaf Size and Longevity Regulation

: Growth-regulating factors (GRFs) are plant-speciﬁc transcription factors that play an important role in plant growth and development. In this study, ﬁfteen GRF gene members containing QLQ and WRC domains were identiﬁed in Zanthoxylum armatum . Phylogenetic and collinearity analysis showed that ZaGRFs were closely related to CsGRFs and AtGRFs, and distantly related to OsGRFs. There are a large number of cis-acting elements related to hormone response and stress induction in the GRF gene promoter region of Z. armatum . Tissue-speciﬁc expression analysis showed that except for ZaGRF7 , all the ZaGRFs were highly expressed in young parts with active growth and development, including terminal buds, seeds, and young ﬂowers, suggesting their key roles in Z. armatum growth and development. Eight ZaGRFs were selected to investigate the transcriptional response to auxin, gibberellin and drought treatments. A total of six ZaGRFs in the NAA treatment, four ZaGRFs in the GA 3 treatment, and six ZaGRFs in the PEG treatment were induced and signiﬁcantly up-regulated. Overexpression of ZaGRF6 increased branching and chlorophyll content and delayed senescence of transgenic Nicotiana benthamiana . ZaGRF6 increased the expression of CRF2 and suppressed the expression of ARR4 and CKX1 , indicating that ZaGRF6 is involved in cytokinin metabolism and signal transduction. These research results lay a foundation for further analysis of the GRF gene function of Z. armatum and provide candidate genes for growth, development, and stress resistance breeding of Z. armatum . ZaGRF7 These results indicate that ZaGRF genes likely in a manner that is to hormone signal transduction and drought stress response.


Introduction
Growth-regulating factors (GRFs) are plant-specific transcription factors that play an important role in growth and response to abiotic stress [1][2][3]. The GRF protein contains two conserved domains, QLQ (Gln, Leu, Gln) and WRC (Trp, Arg, Cys) [1]. The QLQ domain interacts with GRF interaction factor (GIF) to form a transcription co-activator complex [4]. The WRC domain is the DNA-binding region of the GRF protein, which consists of a nuclear localization signal motif and a DNA-binding zinc finger structure [5]. The first member of the identified GRF gene is OsGRF1, which plays a regulatory role in gibberellin (GA 3 )induced stem elongation [1]. Studies have shown that OsGRF1 has multiple functions. The OsGRF1 gene is involved in regulating growth at the juvenile stage, and may be involved in the regulation of heading in rice as well [1,6]. There is a close relationship between plant hormones and GRFs in plant growth and development. In tobacco, twelve NtGRFs were enhanced by GA 3 treatment [7]. Exogenous GA 3 application activated AhGRF5 and showed a more positive response [8]. In apple, exogenous 1-naphthylacetic acid (NAA) and GA 3 can elevate the expression level of five and seven MdGRF genes, 2 of 18 respectively [9]. In Arabidopsis, AtGRF5 stimulates chloroplast division, leaf expansion and longevity [10]. GRFs have been reported to coordinate plant growth with cytokinin to affect plant senescence. The suppressive role of GRFs in leaf senescence may be explained by the GRF-cytokinin interaction, as 35S:AtGRF5 increases the sensitivity of leaves to cytokinins. Studies have shown that GRFs can respond positively to drought stress, suggesting that GRFs play an important role in plant response to abiotic stress. The overexpression of AtGRF7 results in upregulation of DREB2A, which confers increased tolerance to salt and drought stress [11]. MsGRF2 and MsGRF6 were significantly up-regulated under mannitol treatment in Medicago, indicating that they are most likely involved in drought stress tolerance [12]. To date, GRFs have been identified in multiple species. Among them are nine GRFs in Arabidopsis [5], thirteen in tobacco [7], twelve in rice [6], sixteen in apple [9], and thirty in wheat [13].
Zanthoxylu armatum DC, an important aromatic woody shrub in Rutaceae, has a long history of usage and cultivation as an important spice and medicinal plant in Asia [14,15]. The fruit of Zanthoxylum is important as a seasoning and for sesame oil, with good edible value [16]. Due to its high tolerance to drought and calcareous soils, Zanthoxylum is an economically valuable plant, grows well under moderately arid to semi-arid conditions, and has the potential to restore degraded land [17]. At present, most of the research on Zanthoxylum has focused on its pharmacological and antioxidant activity, while the regulatory mechanism of its growth and development and related gene functions are remain less studied. Although GRF family members of many plants have been identified and studied, studies on GRF genes in Z. armatum are not yet available. The release of the whole genome data of Z. armatum is of great significance for understanding its biological characteristics [14]. Therefore, identifying and characterizing GRF genes in Z. armatum is of great interest.
In this study, fifteen ZaGRF gene family members were identified from Z. armatum. The structural characteristics, phylogenetic relationships, chromosome localizations, synteny analysis, and expression patterns of the ZaGRF gene family were analyzed. In addition, the expression profiles of ZaGRF genes in response to hormone and drought treatment were ascertained. Furthermore, we identified a possible functional gene, ZaGRF6, involved in cytokinin metabolism and signal transduction. Genome-wide identification and expression analysis of the ZaGRF gene family provides a basis for further clarifying the important functions of GRF genes in the growth and development of Z. armatum.

Identification of the GRF Gene Family in Z. armatum
Via Hidden Markov Model searches, twenty GRF genes were identified from the Z. armatum genome. Finally, fifteen GRF genes were obtained by NCBI-CDD and SMART and the ZaGRFs were designated as ZaGRF1 to ZaGRF15 based on their positions on their distribution in the Z. armatum genome. The deduced polypeptides ranged in length from 272 (ZaGRF13) to 652 (ZaGRF8) amino acids, with molecular weights between 31.14 kD (ZaGRF5) and 70.82 kD (ZaGRF8). The isoelectric points ranged from 6.24 (ZaGRF8) to 9.90 (ZaGRF7). The instability coefficient results showed that all ZaGRFs were unstable proteins (instability coefficient > 40) ( Table 1).
The multiple sequence alignment indicated that all of the putative ZaGRF proteins have both QLQ and WRC domains in the N-terminal region ( Figure 1). The N-terminal QLQ domain was conserved with one Leu and two Gln residues in all the ZaGRF proteins except for ZaGRF2. The WRC domain was highly conserved, with one Trp, Arg, and Cys in each of the ZaGRF proteins. The zinc finger structure (CCCH) was found within the WRC domain in all ZaGRF proteins except for ZaGRF12, ZaGRF14, and ZaGRF15 ( Figure 1).
MiRNAs typically recognize their target mRNAs through partial complementarity [18]. Previous studies have found that GRF genes are the targets of miR396. MiR396 directly targets GRF transcripts, thereby negatively regulating gene expression levels [19]. Here, sRNATarget targeting software was used to predict the target sites of miR396 in ZaGRFs.
The results showed that a total of thirteen ZaGRFs, excepting ZaGRF2 and ZaGRF5, contained target sites for miR396. Therefore, we speculated that these thirteen genes might be the target genes of miR396 ( Figure 2). The multiple sequence alignment indicated that all of the putative ZaGRF prote have both QLQ and WRC domains in the N-terminal region ( Figure 1). The N-termi QLQ domain was conserved with one Leu and two Gln residues in all the ZaGRF prote except for ZaGRF2. The WRC domain was highly conserved, with one Trp, Arg, and C in each of the ZaGRF proteins. The zinc finger structure (CCCH) was found within t WRC domain in all ZaGRF proteins except for ZaGRF12, ZaGRF14, and ZaGRF15 (Figu 1).  MiRNAs typically recognize their target mRNAs through partial complementarity [18]. Previous studies have found that GRF genes are the targets of miR396. MiR396 directly targets GRF transcripts, thereby negatively regulating gene expression levels [19]. Here, sRNATarget targeting software was used to predict the target sites of miR396 in ZaGRFs. The results showed that a total of thirteen ZaGRFs, excepting ZaGRF2 and ZaGRF5, contained target sites for miR396. Therefore, we speculated that these thirteen genes might be the target genes of miR396 ( Figure 2).

The Protein Structure, Subcellular Localization Prediction, and Post-Translational Modification of ZaGRFs
The protein subcellular localization of ZaGRF was analyzed by Protcomp. The results showed that eleven members of ZaGRF were presumably localized in the nucleus, while four members (ZaGRF2, ZaGRF12, ZaGRF14, and ZaGRF15) were presumably localized in the extracellular matrix. The secondary structure prediction results showed that the ZaGRF proteins contained three types of secondary structures, namely, α-helix, extended chain, and random coil, of which random coil accounted for the largest proportion at 66.58% on average (Table S2). The results of the 3D structure showed that ZaGRFs had great similarities in the spatial conformation of protein three-dimensional structures, all containing a large number of irregular curls, which is essentially the same as the secondary structure prediction structure (Figure 3a).
The post-translational modifications of ZaGRFs were predicted in terms of phosphorylation. A total of 924 potential phosphorylation events on the serine, threonine, and ty-

The Protein Structure, Subcellular Localization Prediction, and Post-Translational Modification of ZaGRFs
The protein subcellular localization of ZaGRF was analyzed by Protcomp. The results showed that eleven members of ZaGRF were presumably localized in the nucleus, while four members (ZaGRF2, ZaGRF12, ZaGRF14, and ZaGRF15) were presumably localized in the extracellular matrix. The secondary structure prediction results showed that the ZaGRF proteins contained three types of secondary structures, namely, α-helix, extended chain, and random coil, of which random coil accounted for the largest proportion at 66.58% on average (Table S2). The results of the 3D structure showed that ZaGRFs had great similarities in the spatial conformation of protein three-dimensional structures, all containing a large number of irregular curls, which is essentially the same as the secondary structure prediction structure (Figure 3a).
The post-translational modifications of ZaGRFs were predicted in terms of phosphorylation. A total of 924 potential phosphorylation events on the serine, threonine, and tyrosine were identified within ZaGRFs (Figure 2b). The phosphorylation events were predicted to be related to serine (666), followed by threonine (191), and then by tyrosine (67). Among the ZaGRFs, most of the phosphorylation sites (99 sites) were predicted in ZaGRF8, whereas the phosphorylation events ranged from 44 to 81 sites in other ZaGRFs. rosine were identified within ZaGRFs (Figure 2b). The phosphorylation events were predicted to be related to serine (666), followed by threonine (191), and then by tyrosine (67). Among the ZaGRFs, most of the phosphorylation sites (99 sites) were predicted in ZaGRF8, whereas the phosphorylation events ranged from 44 to 81 sites in other ZaGRFs.

Gene Structure, Motif Distribution, and Conserved Domain Alignment
To understand the gene structure diversity and conserved motifs of the Z. a GRF protein, the intron-exon regions in genes and motifs in proteins were system analyzed. Gene structure diagram analysis showed that fifteen ZaGRFs had variou bers of exons, ranging from three to six (Figure 5a). Most ZaGRF genes contain exons. Among them, the number of exons of ZaGRF3 and ZaGRF8 were at most six for ZaGRF7 and ZaGRF11 there were at least two. Ten motifs were revealed in Z (Figure 5b, c). Conserved motif analysis showed that all fifteen members contain tify1 (WRC) and Motify2 (QLQ) (Figure 5b, c). The motifs of proteins that cluste gether within the phylogenetic tree presented similarities in the distribution of m an extent (Figure 4b).

Gene Structure, Motif Distribution, and Conserved Domain Alignment
To understand the gene structure diversity and conserved motifs of the Z. armatum GRF protein, the intron-exon regions in genes and motifs in proteins were systematically analyzed. Gene structure diagram analysis showed that fifteen ZaGRFs had various numbers of exons, ranging from three to six (Figure 5a). Most ZaGRF genes contained four exons. Among them, the number of exons of ZaGRF3 and ZaGRF8 were at most six, while for ZaGRF7 and ZaGRF11 there were at least two. Ten motifs were revealed in ZaGRFs (Figure 5b,c). Conserved motif analysis showed that all fifteen members contained Motify1 (WRC) and Motify2 (QLQ) (Figure 5b,c). The motifs of proteins that clustered together within the phylogenetic tree presented similarities in the distribution of motifs to an extent ( Figure 4b).

Chromosome Localization and Synteny Analysis
Eight genes (ZaGRF1-ZaGRF8) were anchored on eight chromosomes, while ZaGRF9-ZaGRF15 were not anchored on chromosome. The chromosomal location map of ZaGRF genes showed that ZaGRF1-ZaGRF8, were anchored on chromosomes 11,15,18,20,21,26,27, and 31, respectively. Chromosome 27 contained two ZaGRF genes (ZaGRF7 and Zardc38469.t1), and ZaGRF3 and ZaGRF8 were segmental duplication genes (Figure 6a). The collinear relationship between Z. armatum and C. sinensis and A. thaliana was significantly greater than that of O. sativa. Five ZaGRF genes shared homology with C. sinensis, seven ZaGRF genes shared homology with A. thaliana, and only one ZaGRF gene shared homology with O. sativa (Figure 6b). This may be related to the fact that Z. armatum, C.

Chromosome Localization and Synteny Analysis
Eight genes (ZaGRF1-ZaGRF8) were anchored on eight chromosomes, while ZaGRF9-ZaGRF15 were not anchored on chromosome. The chromosomal location map of ZaGRF genes showed that ZaGRF1-ZaGRF8, were anchored on chromosomes 11,15,18,20,21,26,27, and 31, respectively. Chromosome 27 contained two ZaGRF genes (ZaGRF7 and Zardc38469.t1), and ZaGRF3 and ZaGRF8 were segmental duplication genes (Figure 6a). The collinear relationship between Z. armatum and C. sinensis and A. thaliana was significantly greater than that of O. sativa. Five ZaGRF genes shared homology with C. sinensis, seven ZaGRF genes shared homology with A. thaliana, and only one ZaGRF gene shared homology with O. sativa (Figure 6b). This may be related to the fact that Z. armatum, C. sinensis, and A. thaliana are all dicotyledonous plants, and C. sinensis belongs to Rutaceae, which is closely related.

Promoter Analysis
Surveying cis-elements in promoter regions can help in better understanding the potential function and regulatory mechanism of the GRF gene in Z. armatum. The analyses of cis-regulatory elements in promoter regions revealed the presence of binding sites for key transcription factors related to CAAT-box and TATA-box elements, light-responsive elements, hormone-responsive elements, stress-related elements, and growth-response elements ( Figure S1). Among them, cis-elements related to the hormone response include ARE (auxin), ABRE (abscisic acid), CGTCA-motif (methyl jasmonate), and P-box (gibberellin). Elements related to the stress response include STRE (stress response), LTR (low temperature), ACE (anaerobic induction), TC-rich repeats (defense and stress), MBS (MYB binding site involved in drought), and WUN-motif (mechanical injury). The O2-site (zein metabolic regulatory element) is related to growth and development ( Figure S1).

In Silico Tissue-Specific Expression of ZaGRF Genes
To further explore the function of ZaGRFs, we analyzed the expression levels of ZaGRFs in nine tissues: young leaf, mature leaf, petiole, terminal bud, stem, young flower, prick, seed, and husk. RNA sequencing (RNA-Seq) data were downloaded from the SRA database published by Wang et al. [14]. Differential expression was noted for ZaGRF genes in various tissues of Z. armatum. In general, the ZaGRF genes were constitutively expressed in all tested tissues except for ZaGRF5 and ZaGRF11. ZaGRF5 was only expressed in seeds, young flowers, and terminal buds, while ZaGRF11 was not expressed in leaves. Except for ZaGRF7, all ZaGRF genes were expressed at higher levels in terminal buds, seeds, and young flowers than in other tissues, suggesting that these genes play an important role in Z. armatum growth and development. Among them, the expression levels of ZaGRF3 and ZaGRF8 in these three tissues were significantly higher than those of other genes, and ZaGRF1 had the highest expression in leaf and husk [5]. In addition, ZaGRF13 was highly expressed in young flowers, and ZaGRF9 was highly expressed in seeds (Figure 7). These expression results showed that ZaGRF family members may play important roles in the development of Z. armatum shoot tip meristems, seeds, and flowers.

Promoter Analysis
Surveying cis-elements in promoter regions can help in better understanding the potential function and regulatory mechanism of the GRF gene in Z. armatum. The analyses of cis-regulatory elements in promoter regions revealed the presence of binding sites for key transcription factors related to CAAT-box and TATA-box elements, light-responsive elements, hormone-responsive elements, stress-related elements, and growth-response elements ( Figure S1). Among them, cis-elements related to the hormone response include ARE (auxin), ABRE (abscisic acid), CGTCA-motif (methyl jasmonate), and P-box (gibberellin). Elements related to the stress response include STRE (stress response), LTR (low temperature), ACE (anaerobic induction), TC-rich repeats (defense and stress), MBS (MYB binding site involved in drought), and WUN-motif (mechanical injury). The O2-site (zein metabolic regulatory element) is related to growth and development ( Figure S1).

In Silico Tissue-Specific Expression of ZaGRF Genes
To further explore the function of ZaGRFs, we analyzed the expression levels of ZaGRFs in nine tissues: young leaf, mature leaf, petiole, terminal bud, stem, young flower, prick, seed, and husk. RNA sequencing (RNA-Seq) data were downloaded from the SRA database published by Wang et al. [14]. Differential expression was noted for ZaGRF genes in various tissues of Z. armatum. In general, the ZaGRF genes were constitutively expressed in all tested tissues except for ZaGRF5 and ZaGRF11. ZaGRF5 was only expressed in seeds, young flowers, and terminal buds, while ZaGRF11 was not expressed in leaves. Except for ZaGRF7, all ZaGRF genes were expressed at higher levels in terminal buds, seeds, and young flowers than in other tissues, suggesting that these genes play an important role in Z. armatum growth and development. Among them, the expression levels of ZaGRF3 and ZaGRF8 in these three tissues were significantly higher than those of other genes, and ZaGRF1 had the highest expression in leaf and husk [5]. In addition, ZaGRF13 was highly expressed in young flowers, and ZaGRF9 was highly expressed in seeds (Figure 7). These expression results showed that ZaGRF family members may play important roles in the development of Z. armatum shoot tip meristems, seeds, and flowers.

ZaGRF Gene Expression Profiles in Response to Hormone and Drought Treatment
Previous studies have demonstrated that GRFs play important roles in plant responses to hormone and abiotic stresses. Our cis-element analysis showed that there are a large number of hormone-responsive and drought-responsive elements in the ZaGRF gene promoter region. We selected eight ZaGRFs for qRT-PCR to investigate the transcriptional response to auxin, gibberellin and drought treatments. Comprehensive expression profiles of ZaGRF genes under hormone and drought treatment are shown in Figure 8. The results show that most of the ZaGRFs exhibited significantly altered transcript levels after treatment. A total of six ZaGRFs in the NAA treatment, four ZaGRFs in the GA3 treatment, and six ZaGRFs in the PEG treatment were up-regulated in transcription by twofold to 64.9-fold. The highest-fold inductions in the transcriptional responses to hormones and drought were exhibited by ZaGRF3 (64.9-fold to NAA), ZaGRF6 (13.3-fold to GA3) and ZaGRF8 (9.1-fold to PEG). Notably, ZaGRF1 and ZaGRF3 accumulated higher transcription levels in response to all the treatments. ZaGRF6 was strongly induced by GA3 and responded only slightly to PEG. The expression of ZaGRF4 and ZaGRF5 was suppressed by GA3 and PEG, and ZaGRF4 was not sensitive to NAA. ZaGRF7 was inhibited by GA3 and up-regulated by NAA treatment. These results indicate that ZaGRF genes likely function in a manner that is responsive to hormone signal transduction and drought stress response.

ZaGRF Gene Expression Profiles in Response to Hormone and Drought Treatment
Previous studies have demonstrated that GRFs play important roles in plant responses to hormone and abiotic stresses. Our cis-element analysis showed that there are a large number of hormone-responsive and drought-responsive elements in the ZaGRF gene promoter region. We selected eight ZaGRFs for qRT-PCR to investigate the transcriptional response to auxin, gibberellin and drought treatments. Comprehensive expression profiles of ZaGRF genes under hormone and drought treatments are shown in Figure 8. The results show that most of the ZaGRFs exhibited significantly altered transcript levels after treatment. A total of six ZaGRFs in the NAA treatment, four ZaGRFs in the GA 3 treatment, and six ZaGRFs in the PEG treatment were up-regulated in transcription by two-fold to 64.9-fold. The highest-fold inductions in the transcriptional responses to hormones and drought were exhibited by ZaGRF3 (64.9-fold to NAA), ZaGRF6 (13.3-fold to GA 3 ) and ZaGRF8 (9.1-fold to PEG). Notably, ZaGRF1 and ZaGRF3 accumulated higher transcription levels in response to all the treatments. ZaGRF6 was strongly induced by GA 3 and responded only slightly to PEG. The expression of ZaGRF4 and ZaGRF5 was suppressed by GA 3 and PEG, and ZaGRF4 was not sensitive to NAA. ZaGRF7 was inhibited by GA 3 and up-regulated by NAA treatment. These results indicate that ZaGRF genes likely function in a manner that is responsive to hormone signal transduction and drought stress response.   p values were determined according to one-factor ANOVA test (* p < 0.05, ** p < 0.01).

Overexpression of ZaGRF6 Alters Leaf Size and Longevity in Transgenic Nicotiana Benthamiana
To further investigate the function of ZaGRF6, the overexpression construct 35S: ZaGRF6-GFP was transformed into Nicotiana benthamiana. Thirty-two independent transgenic lines were obtained. The expression levels of ZaGRF6 in transgenic lines ZG1 and ZG2 were detected, then the two lines were used for further analysis. The overexpression transgenic lines showed various physiological and morphological abnormalities, including shorter petioles, thicker and smaller leaves, darker green leaf color, and increased branch number compared with the wild type (Figure 9a,b). A previous study showed that GRF5 could enhance chlorophyll retention after dark-induced senescence [10], indicating that leaf senescence is postponed by cytokinins, which could serve as an additional commonality between GRF5 and cytokinin functions. Leaves from the transgenic lines and WT were kept in darkness in vitro to observe their senescence responses. Compared with the WT, the transgenic lines showed delayed leaf senescence (Figure 9c). The chlorophyll (Chl) content was significantly higher in transgenic leaves than in wild-type leaves under normal greenhouse conditions (Figure 9d). The leaves of WT plants started to turn yellow after 5 d in the dark, and completely turned yellow and withered after 12 d. After 5 d in the dark, the leaves retained 75% of the total Chl that was present before incubation, and the leaves retained only 45% after 9 d. The leaves of ZG1 began to turn yellow after 9 d in the dark, and the Chl retention was 70%. However, the leaves of ZG2 showed no significant change in leaf color after 12 d, and the leaf Chl was retained at 80% (Figure 9e). Furthermore, we detected the expression levels of ZaGRF6, PORB, GLK1, ARR4, CRF2, and CKX1 in transgenic plants. The results showed the expression level of ZaGRF6 in ZG2 was 32-fold higher than that in ZG1, and the expression levels of GLK1, ARR4, and CKX1 in transgenic plants were significantly decreased compared with those in wild type plants ( Figure 10). The expression level of CRF2 was up-regulated 1.5-and 2.2-fold in ZG1 and ZG2, respectively. There was no obvious difference in the expression of PORB ( Figure 10). The leaves of ZaGRF6-overexpressing plants showed more intense greening and Chl retention, suggesting an increase in Chl levels and senescence delay, which could be caused by altered cytokinin signaling.

Discussion
GRFs are a class of transcription factors that are widely involved in leaf and seed development, pistil development, flower regulation, abiotic stress responses, and exogenous hormone responses [4,10,11,[20][21][22]. In the present study, we identified fifteen GRF genes from Zanthoxylum armatum genome. The structural features of ZaGRF proteins contained QLQ and WRC domains, which are similar to those of the GRF proteins of Arabidopsis and rice [5,6]. Although the Gln-Leu-Gln residues of the QLQ domain are absolutely conserved in ZaGRFs, this feature is absent in ZaGRF2 due to the substitution of Leu by Phe. It is well known that transcription factors act in the nucleus to regulate the expression of target genes. The results of subcellular localization prediction showed that eleven ZaGRFs were located in the nucleus, suggesting that most ZaGRFs are subject to normal transcription. Protein phosphorylation is one of the most common and important post-translational modifications, and occurs as a mechanism to regulate the biological activity of a protein [23]. We predicted phosphorylation sites in all ZaGRFs, ranging in number from 44 to 99. The phylogenetic analysis showed that ZaGRFs clustered preferentially with CsGRFs and AtGRFs, indicating that Z. armatum is closely related to C. sinensis and Arabidopsis and distantly related to the monocotyledon rice. The collinearity analysis showed that there was a strong linear homologous relationship between Z. armatum and C. sinensis, followed by Arabidopsis and rice. Collinear analysis showed that only two pairs of segmental duplication genes (ZaGRF3:ZaGRF8, ZaGRF7:Zardc38469.t1) were found in Z. armatum. We found that Zardc38469.t1 only contains the WRC domain, not the QLQ domain. This suggests that fewer gene duplication events have occurred in the ZaGRF family during evolution. Synteny analysis showed that seven pairs of duplication genes were found in Z. armatum and Arabidopsis and five pairs of duplication genes in Z.

Discussion
GRFs are a class of transcription factors that are widely involved in leaf and seed development, pistil development, flower regulation, abiotic stress responses, and exogenous hormone responses [4,10,11,[20][21][22]. In the present study, we identified fifteen GRF genes from Zanthoxylum armatum genome. The structural features of ZaGRF proteins contained QLQ and WRC domains, which are similar to those of the GRF proteins of Arabidopsis and rice [5,6]. Although the Gln-Leu-Gln residues of the QLQ domain are absolutely conserved in ZaGRFs, this feature is absent in ZaGRF2 due to the substitution of Leu by Phe. It is well known that transcription factors act in the nucleus to regulate the expression of target genes. The results of subcellular localization prediction showed that eleven ZaGRFs were located in the nucleus, suggesting that most ZaGRFs are subject to normal transcription. Protein phosphorylation is one of the most common and important post-translational modifications, and occurs as a mechanism to regulate the biological activity of a protein [23]. We predicted phosphorylation sites in all ZaGRFs, ranging in number from 44 to 99. The phylogenetic analysis showed that ZaGRFs clustered preferentially with CsGRFs and AtGRFs, indicating that Z. armatum is closely related to C. sinensis and Arabidopsis and distantly related to the monocotyledon rice. The collinearity analysis showed that there was a strong linear homologous relationship between Z. armatum and C. sinensis, followed by Arabidopsis and rice. Collinear analysis showed that only two pairs of segmental duplication genes (ZaGRF3:ZaGRF8, ZaGRF7:Zardc38469.t1) were found in Z. armatum. We found that Zardc38469.t1 only contains the WRC domain, not the QLQ domain. This suggests that fewer gene duplication events have occurred in the ZaGRF family during evolution. Synteny analysis showed that seven pairs of duplication genes were found in Z. armatum and Arabidopsis and five pairs of duplication genes in Z. armatum and C. sinensis (Table S4). However, only one pair of duplication genes was found between Z. armatum and rice (ZaGRF5: OsGRF1) (Table S4). Consistent with the phylogenetic tree, the collinearity analysis revealed that ZaGRF3, ZaGRF8, AtGRF1, AtGRF2, and CsGRF2 probably share the same ancestor. These results demonstrate that ZaGRF5 completed duplication before the separation and evolution of monocotyledonous and dicotyledonous plants. Therefore, ZaGRF5, AtGRF5, AtGRF6, and OsGRF1 might originate from the same ancient ancestor.
Previous studies have indicated that the GRF gene family is involved in hormone signal transduction and abiotic stress responses. There are several cis-acting elements related to hormone response and stress induction in the promoter region of the GRF genes of Z. armatum. Our study found that most of the ZaGRFs were induced by hormone and drought stress treatments. Under NAA treatment, the expression levels of most ZaGRFs were significantly upregulated. Among them, the expression of ZaGRF3 increased 65-fold compared to that before treatment, whereas ZaGRF4 and ZaGRF5 were downregulated. These results suggest that different genes may perform different functions in the same gene family [24]. In rice, GA 3 can suppress the expression of miR396, thereby upregulating the transcript level of OsGRF6 [25]. In our study, under GA 3 treatment, the transcript levels of five predicted miR396 target genes (ZaGRF1, ZaGRF3, ZaGRF6, ZaGRF8 and ZaGRF15) were significantly upregulated. It has been reported that GRF transcription factors play key roles in plant growth by coordinating stress responses and defense signals [11,26,27].
In several studies, drought treatment has been found to increase the expression level of GRFs [13,28]. In the present study, we found that six ZaGRF genes were upregulated under PEG treatment, especially ZaGRF1 and ZaGRF8. This shows that ZaGRF1 and ZaGRF8 may play important roles in the responses to abiotic stress in Z. armatum. The expression levels of ZaGRFs peaked 24 h to 48 h after PEG6000 treatment. We speculate that the slow response of ZaGRF genes to drought stress suggests that these genes are not upstream of the osmotic stress signaling pathway [13].
GRFs generally play positive roles in plant growth and development. Studies have shown that overexpression of GRFs leads to increase cell proliferation and leaf expansion in rice, Arabidopsis, poplar, and lettuce [10,21,[29][30][31]. However, GRF genes have been reported to play negative roles in plant growth as well. Overexpression of ZmGRF10 resulted in a reduction in leaf size and plant height [20]. AtGRF9 negatively regulates Arabidopsis leaf growth by restricting cell proliferation in leaf primordia. The grf9 mutants had larger rosette leaves and petals than WT, while plants overexpressing AtGRF9 produced smaller leaves and petals [24]. In rice, overexpression of OsGRF7 causes a semi-dwarf and compact plant architecture with increased culm wall thickness and narrowed leaf angles [32]. In this study, plants overexpressing ZaGRF6 showed smaller leaves and shorter petioles, indicating that ZaGRF6 plays a negative role in regulating N. benthamiana leaf development. However, in addition to the smaller leaves, transgenic plants showed cytokinin-overproducing phenotypes, such as increased chlorophyll content and branching and delayed leaf senescence [33,34]. It is suggested that the regulatory mechanisms of ZaGRF6 and these GRFs that have opposite effects on growth and development may be different.
Cytokinins generally function in promoting mitotic cell division [35], regulating shoot apical meristem (SAM) activity and organizing shoot architecture and leaf senescence [36]. Preventing decline in cytokinin levels by expressing the ipt gene has been shown to delay the senescence of tobacco and Zanthoxylum [33,34]. In our study, the transgenic lines ZG1 and ZG2 showed darker green leaves with higher chlorophyll contents than the wild type. Delayed leaf senescence was found in the transgenic plants overexpressing ZaGRF6. The higher ZaGRF6 expression level caused a more severe phenotype in the transgenic line.
These results suggest that there may be a relationship between ZaGRF6 and cytokinin. Currently, there is evidence that GRFs directly regulate the cytokinin degradation gene CKX1 and are involved in cytokine signal transduction. In rice, OsGRF4 regulates two cytokinin dehydrogenase precursor genes (OsCKX5 and OsCKX1), resulting in increased cytokinin levels [21]. PpnGRF5-1 binds to the PpnCKX1 promoter directly and represses its expression. Overexpression of PpnGRF5-1 repressed cytokinin degradation and significantly increased zeatin and isopentenyladenine in poplar apical buds and third leaves [30]. In Arabidopsis, AtGRFs interact with AtGIF1/AN3 to directly activate AtCRF2 and repress AtARR4 [29]. High AtGRF3 and AtGRF5 activity delayed leaf senescence. Evidence has demonstrated that GRF5 and cytokinin functions are interconnected during senescence. The AtGRF5 gene stimulates chloroplast proliferation, resulting in a higher chloroplast number per cell with a concomitant increase in chlorophyll levels in 35S:GRF5 leaves. 35S:GRF5 leaves showed enhanced sensitivity to cytokinin-driven stimulation of Chl retention after dark-induced senescence [10]. The cytokine response factor AtCRF2 was upregulated in 35S:GRF plants. The A-type ARRs ARR4, ARR5, ARR6, and ARR9 were significantly repressed in 35S:GRF5 plants [10]. In our study, the expression levels of the CKX1 and CRF2 genes in ZG1 and ZG were significantly increased, while the ARR4 gene was suppressed. Evolutionary tree analysis results show that ZaGRF6 has high homology with OsGRF4, indicating that ZaGRF6 and OsGRF4 may have similar functions. We hypothesized that ZaGRF6 may regulate the growth and development of N. benthamiana plants by participating in cytokinin metabolism and signal transduction. Further research is required to explore the relationship between the ZaGRF6 gene and cytokinin in transgenic plants.

Plant Materials and Treatments
Three-year-old plants of Dingtan pepper (Zanthoxylum armatum var. dintanensis) were selected for expression pattern experiments. In this experiment, the pepper plants were irrigated with 20% PEG for drought treatment, and the young leaves were sprayed with 300 µg/L NAA and 400 µg/L GA 3 for hormone treatments. Leaves (0.2 g) were collected at 0, 1, 2, 4, 6, 12, 24, and 48 h post-treatment. All the leaves were immediately snap-frozen in liquid nitrogen and then stored in a −80 • C ultralow temperature freezer. Nicotiana benthamiana was used as the wild-type (WT) control, and all transgenic lines were generated in the background of Nicotiana benthamiana in this study. All transgenic and WT plants were grown in a glasshouse (16 h light/8 h dark cycles; 24 • C; relative humidity 70%).

Chromosome Location and Collinearity Analysis
The chromosomal locations of the ZaGRF genes were determined according to the GFF3 file and mapped with TBtools software [39]. MCScanX v1.1 with default parameters in TBtools was used to analyze the collinearity among the ZaGRFs [40]. Then, MCScanX v1.1 and Circos 0.69.8 in TBtools were used to retrieve and map collinearity among Z. armatum, C. sinensis, A. thaliana, and O. sativa.

Cis-Acting Element and Gene Expression Pattern Analysis
The promoter sequences, 2000 bp regions upstream of the ZaGRF translational start sites were extracted from the genome sequence by SequenceToolkit in TBtools. The cis-regulatory elements of promoter sequences were analyzed by PlantCARE (http:// bioinformatics.psb.ugent.be/webtools/plantcare/html/, accessed on 16 September 2021) and visualized by TBtools. Transcriptome data for nine different tissues (young leaf, mature leaf, petiole, terminal bud, stem, young flower, prick, seed, and husk) were downloaded from the NCBI database (PRJNA721257) by using the SRA Toolkit software [14]. The SRA files were converted to fastq files by FASTQ software [41]. Then, FastQC software was used for data quality control [42]. Trimmomatic was used for data filtering and removing adapters [43]. Kallisto was used for quantitative data processing [44]. The TBtool software was used for heatmap visualization.

Plasmid Construction and Plant Transformation
The full length CDS of ZaGRF6 was cloned from Dingtan pepper and constructed into the overexpression vector pCambia1300-GFP by the 35S promoter. The 35S:ZaGRF6-GFP vector was transformed into Nicotiana benthamiana by Agrobacterium-mediated transformation following the leaf disc method. The roots of transgenic plants were used to investigate the subcellular localization of ZaGRF6. The green fluorescent protein (GFP) fluorescence signal was observed with a laser confocal microscope (Leica SP8 STED) under excitation at 488 nm.

Chl Measurements after Dark-Induced Senescence
To test the tolerance of plant leaves under dark-induced senescence conditions, leaves from WT plants and transgenic lines were detected on moist filter paper in 10-cm Petri dishes and then placed in the dark at 28 • C. The leaf chlorophyll concentration was measured by a SPAD-502PLUS meter (KONICA MINOLTA, Tokyo, Japan).

RNA Isolation and qRT-PCR Analysis
Total RNA was extracted by using Plant RNA Kit (Omega Bio-Tek, Doraville, GA, USA) according to the manufacturer's instructions. The cDNAs were generated by RT-PCR using the StarScript III RT Mix Kit (GenStar, Beijing, China). The qRT-PCR analyses were performed using a qTower3G Real-time PCR System (Analytik Jena AG, Jena city, Germany) and SYBR ® Green Fast Mixture (GenStar, Beijing, China). The ZaActin and NbActin genes were used as internal references to normalize the gene expression levels. The relative gene expression levels were calculated using the 2 −∆∆Ct method [45]. The primers used for real-time PCR are listed in Table S1.