Identification and exploration of the GRF and GIF families in maize and foxtail millet

Abstract

Plant growth and development are dependent on complex regulatory networks to adapt various environments. The growth regulatory factor (GRF) and GRF-interacting factor (GIF) families have been shown to control growth in various plant species. There are growing evidences that GRFs and GIFs can improve crop genetic transformation efficiency. In this study, we identified and classified 17 ZmGRFs, 10 SiGRFs, 4 ZmGIFs and 3 SiGIFs in maize (Zea mays L.) and foxtail millet (Setaria italica L.) using updated genome data. Many ABREs (Abscisic Acid-responsive elements) were present in the promoter regions of GRFs by analysis, and the expression levels of ZmGRF4, 9, 12, 14 and ZmGIF2 were associated with the Abscisic Acid (ABA) response. Furthermore, ZmGRF9 showed collinearity with AtGRF5 between Arabidopsis and maize. ZmGRF9 conservatively interacts with ZmGIF 2, 3, and 4. As a result, we systematically identified GRF and GIF family members, analyzed the regulatory network, and found that exogenous ABA inhibited the expression of GRFs, regulating responses to stress in the environment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12298-022-01234-z.

Introduction

Plant morphogenesis is controlled by a complex regulatory network based on the shoot apical meristem (SAM). The SAM is divided into three parts, including the central zone (CZ) at the top, the rib zone (RZ) below the central zone, and the peripheral zone (PZ) on both sides. The central zone consists of the upper stem cell zone and the organizing center (OC) near the lower part of the rib zone. CLAVATA (CLV)-WUSCHEL (WUS) maintains the balance of the apical meristem (Lopes et al. 2021). The leaf primordia first becomes distinguished from the SAM and develops into leaf blades after flattening under the effects of the adaxially expressed AtMP, abaxially-enriched auxin, and middle expression WOX (Guan et al. 2017). After leaf initiation, cell proliferation and differentiation occur throughout the leaf region. Marginal and intercalary leaf growth is regulated by miR396-GRF-GIF and miR319-TCP modules. GRF-GIF maintains cell proliferation at the leaf proximal region, and miR396 inhibits GRF-mediated proliferation at the distal region; TCPII promotes the switch from cell proliferation to cell differentiation at the leaf distal region; and miR319 inhibits TCPII at the proximal region (Kim et al. 2003; Rodriguez et al. 2010; Du et al. 2018).

GRF are a class of plant-specific transcription factor containing conserved QLQ and WRC domains. The WRC domain consisting of a Trp-Ar-Cys motif and a zinc-finger-like spacing of Cys and His residues (CX9CX10CX2H), forms a nuclear localization signal as well as a DNA-binding region. QLQ domains exist in the N-terminus, consisting of the typical glutamine-leucine-glutamine (QX3LX2Q), and mediate protein–protein interactions that can regulate plant development (Kim et al. 2003; Choi et al. 2004). SSXT is a typical domain widely present in GIF (GRF-interacting factors), interacting directly with the QLQ domain of GRF proteins (Kim and Kende 2004; Zhang et al. 2008). GRF is post-transcriptionally repressed by microRNA (miR396).

Recent studies suggest that the GRF-GIF complex improves tissue culture regeneration efficiency in some crops. For example, mTaGRF4-TaGIF1 can generate a chimera that confers a two fold to nine fold boost in regeneration in wheat and is linked to the ability of GRF-GIF to regulate the transition between stem and transit-amplifying cells, as well as the ability to promote cell proliferation (Debernardi et al. 2020; Qiu et al. 2022). AtGRF5 and GRF5 orthologs in canola (Brassica napus L.), soybean (Glycine max L.), sunflower (Helianthus annuus L.), and maize (Zea mays L.) significantly boost transformation efficiency (Debernardi et al. 2020; Kong et al. 2020). It is commonly recognized that maize and foxtail millet are important for the study of plant genetics, as well as for economic and food security. However, due to their low genetic transformation efficiency, they cannot be investigated as extensively as rice and Arabidopsis. To promote maize and foxtail millet molecular biology research, the relationship between GRF and GIF of maize, foxtail millet, Arabidopsis and rice needs to be analyzed.

In rice, GRF influences on plant architecture through phytohormones. For example, OsGRF4 suppressed leaf angle by BR (Brassinosteroid), OsGRF6 regulated plant height through GA (Gibberellin) (Tang et al. 2018), and OsGRF1 was suppressed under ABA treatment and various stresses (Lu et al. 2020). OsGRF7 suppressed leaf angle through auxin and GA metabolism (Chen et al. 2020b). In maize, the Zmgif1 mutant has narrow leaves, short internodes, and inflorescences, suggesting that GIF1 participates in shoot architecture and meristem determinacy. GRF and GIF genes have been identified from Arabidopsis (Arabidopsis thaliana L.) and rice (Oryza sativa L.) and both were found to be small transcription factor families. The GRF gene family consists of 9 members in Arabidopsis, 12 members in rice (Oryza sativa L.), 14 members in maize (Zea mays L.), and 3 members of GIF genes in each family, but is unknown in foxtail millet (Setaria italica L.) (Kim et al. 2003; Choi et al. 2004; Kim and Kende 2004; Zhang et al. 2008).

Bioinformatics techniques have been used to identify GRF/GIF families in various plants, however, maize and foxtail millet have received much less attention. We can now thoroughly study maize and foxtail millet because their high-quality genomes have recently been sequenced and updated. Maize and foxtail millet are important crop species and model plants for grasses. It is necessary to update gene families to obtain a more accurate information considering the advancements in genome sequencing technology and the completion of high-quality genome assembly. In particular, the GRF and GIF gene families have recently become increasingly important, as they play a crucial role in plant regeneration (Debernardi et al. 2020; Kong et al. 2020; Qiu et al. 2022). In this work, we performed a genome-wide investigation to identify all genes encoding GRFs and GIFs in maize and foxtail millet using the updated genome. In total, 17 GRFs and 4 GIFs were identified in the maize B73 genome and 10 GRFs and 3 GIFs were found in the foxtail millet Yugu1 genome. The hidden Markov model (HMM) and blastp were used to predict genome-wide GRFs/GIFs. A phylogenetic tree was created by MEGA-X software. The neighbor-joining and 1000 bootstrap NJ (neighbor-joining) methods were used to construct the phylogenetic tree. Motif analysis was used to analyze protein structure and cis-element analysis was used to predict the promoters of GRFs/GIFs that may indicate the response to the environment. ABA treatment, expression pattern analysis, and interaction analysis were used to predict the possible functions of GRFs/GIFs. A collinearity analysis was used to predict the relatedness among maize, foxtail millet, Arabidopsis and rice. Furthermore, interaction validation, subcellular localization and upstream microRNA analysis were performed to study the gene ZmGRF9. Based on the above analysis, the foundational information of the GRF and GIF gene families was revealed in maize and foxtail millet, which is expected to provide useful information for more in-depth studies in the future.

Materials and methods

Identification of the GRF and GIF families in maize, foxtail millet, rice, and Arabidopsis thaliana

HMM included conserved QLQ (PF08880) and WRC (PF08879) domain in GRF. The GIF conserved SSXT (PF050300) domain was downloaded from the Pfam database (http://pfam.xfam.org/). Protein sequences of maize V4, foxtail millet V2.2, Arabidopsis Araport11, and rice v7.0 were downloaded from the Phytozome database (https://phytozome-next.jgi.doe.gov/) as reference sequences. To identify all possible members of the GRF and GIF gene families, the Simple HMM Search module of TBtools was used to search for proteins containing HMM models from the reference sequences of maize, foxtail millet, Arabidopsis, and rice (Chen et al. 2020a). The presence of conserved sequences was then further verified in the candidate genes using the Pfam and NCBI databases (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi). A total of 17 ZmGRF, 10 SiGRF, 12 OsGRF, 9 AtGRF, 4 ZmGIF, 3 SiGIF, 3 OsGIF, and 3 AtGIF non-redundant genes were finally identified. Basic data for GRF and GIF genes of the four species were obtained from the Phytozome database, including amino acid numbers and chromosome locations. Physical and chemical parameters of each gene, including molecular weight (kDa) and isoelectric point (pI), were calculated using ExPASy's calculation pI/Mw (http://www.expasy.org/tools/). Gene ID conversion of different versions of maize and rice was performed by plant-regulomics (http://bioinfo.cemps.ac.cn/plant-regulomics/index.php) (Ran et al. 2020). Tissue analysis of gene expression was performed at the single-cell level through the PlantscRNA database (http://ibi.zju.edu.cn/plantscrnadb/index.php). The exon and intron structures of GRF and GIF genes were identified by genomic DNA sequences and GFF files by TBtools' Gene Structure View in maize, foxtail millet, rice, and Arabidopsis. Conserved domain proteins encoded by GRF and GIF genes were identified via NCBI databases.

Phylogenetic analysis of GRF and GIF proteins in multiple species

All GRF and GIF protein sequences were obtained from the Phytozome database and default arguments and multiple sequence alignments by MUSCLE of MEGA-X with orthologs from maize, foxtail millet, rice, and Arabidopsis thaliana. A phylogenetic tree was created by MEGA-X using the neighbor-joining method with 1000 bootstrap replicates. The phylogenetic tree was visualized on ITOL (http://itol.embl.de/).

Chromosomal location of maize GRF and GIF gene families

All the chromosomal location gained and drawn by Gene Location Visualize from GFF in TBtools and phytozome (Chen et al. 2020a).

The cis-acting regulatory elements in the promoter of GRF and GIF genes

The cis-elements in the promoter 2000 bp upstream of ATG in GRFs and GIFs were analyzed by PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) (Lescot et al. 2002), common elements such as TATA-box and CAAT-box were removed, and the remaining elements were mapped by TBtools.

qRT‒PCR

B73 seedlings of maize grown in growth chambers under a 12-h-light (30 °C)/12-h-dark (22 °C) photoperiod and a photon flux density of approximately 500 μmol m^− 2 s^− 1. To verify the ZmGRFs/ZmGIFs response to ABA, the third leaf of maize seedlings was sprayed with 100 μM ABA (Sigma-Aldrich NV/SA Bornem, Belgium) or kept under drought conditions at 15 DAP (day after planting). The leaves were collected at 0 h, 0.5 h, 1 h and 4 h after ABA treatment. Seedlings were collected after drought conditions for five days. Total RNA was extracted using TRIzol (Thermofisher, 15596018, USA). First-strand cDNA was synthesized from 1 μg total RNA using Trans® Script One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). qRT‒PCR was performed with TransStart® Top Green qPCR SuperMix (TransGen Biotech, Beijing, China). Real-time PCRs were performed using the Applied Biosystems. qRT‒PCR reaction is carried out by incubation at 94 °C for 30 S followed by 40 cycles of 5 S and then at 53 °C for 30 S. The maize ubiquitin2 (ubi2) gene (Zm00001d053838) was used as normalization control, and comparative expression levels were calculated by the 2^−ΔΔCT method (Livak and Schmittgen 2001; Wang et al. 2022). Three technical replicates on each of three biological replicates were conducted. All qRT‒PCR primers are listed in Supplementary Table 8.

Structures of GRF and GIF proteins

The structure containing conserved domains of GRF and GIF proteins was confirmed by NCBI (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi). Twenty conserved motifs of these proteins were queried by the MEME tool (http://memesuite.org).

Expression profile analysis of GRF and GIF genes

The expression profile of ZmGRFs and ZmGIFs were identified by public data in maize (Liu et al. 2020). SiGRF and SiGIF genes in foxtail millet were identified by seed, flower (panicle 3), shoot (stem), root and leaf (leaf 3) (Yang et al. 2020). OsGRF and OsGIF genes in rice were identified by seed, flower (postflowering panicle), shoot, root and leaf (Sakai et al. 2011). AtGRF and AtGIF genes in Arabidopsis were identified by seed (embryo stage 10), flower (flower stage 10), shoot (adult, stem internode, second internode), root (adult, root) and leaf (adult, rosette leaf, rosette leaf 11) (Mergner et al. 2020). GRF and GIF expression patterns in 23 different tissues of maize were intensively studied (Walley et al. 2016). Gene expression patterns analyzed by hierarchical cluster analysis in different tissues and the heatmap was drawn by TBtools.

Construction of protein interaction networks and identification of upstream miRNAs

The protein interaction networks were constructed using the STRING database (https://string-db.org/). The interaction of the combined score between the ZmGRF and ZmGIF proteins was indicated by the thickness of the line and drawn with Cytoscape software (Franz et al. 2016). Upstream miRNAs were identified by the PmiREN database (https://www.pmiren.com/), and visualized with Cytoscape software (Guo et al. 2022).

Cloning ZmGRF9 and ZmGIFs and interaction analysis by the yeast two-hybrid system (Y2H)

We extracted total RNA from the roots and stems of B73 seedlings in maize by TRIzol, obtained cDNA by reverse transcription and amplified the full-length CDS of ZmGRF9 and ZmGIF1-ZmGIF4. ZmGRF9 was inserted into the PGBKT7 vector via EcoRI and BamHI, and ZmGIF1-ZmGIF4 was inserted into the PGADT7 vector via EcoRI and BamHI. ZmGRF9 and ZmGIF1-ZmGIF4 were transformed into the yeast Y2H GOLD strain, and cultured with SD/-Trp/-Leu and SD/-Trp/-Leu/-His/-Ade + x-α-gal media. The primers were listed in Supplementary Table 8. Y2H experiments were performed according to instructions for the Matchmaker Gold Yeast Two-Hybrid System (Clontech).

Firefly luciferase (LUC) complementation imaging (LCI) assay

The CDS of ZmGRF9 was inserted into the 1300-nLUC vector via BamHI and SalI, and the CDSs of ZmGIF2, ZmGIF3 and ZmGIF4 were inserted into the 1300-cLUC vector via KpnI and BamHI. A firefly luciferase complementation assay was conducted using young N. benthamiana leaves according to a previous report (Chen et al. 2008). The primers are listed in Supplementary Table 8.

Subcellular location of ZmGRF9 protein

The full-length ZmGRF9 CDS with termination codon removed was inserted into the 1300-35S-GFP vector via SbfI to obtain 1300-35S-GRF9-GFP. Then, it was transferred to GV3101 of Agrobacterium and injected into tobacco. Transformation of maize protoplast was performed as described previously (Tu et al. 2020). The subcellular location of ZmGRF9 proteins was determined by using laser scanning confocal microscopy (Leica LASX TCS SP8). The primers are listed in Supplementary Table 8.

Results and analysis

Identification of GRF and GIF genes

To identify all genes encoding GRFs and GIFs in maize, foxtail millet, rice and Arabidopsis, multiple searches were performed by TBtools software through HMM. The HMM numbers were PF08880 (QLQ) and PF08879 (WRC) for GRFs, and PF050300 (SSXT) for GIFs. To verify validity, we compared these proteins using the NCBI-CD Search website. GRFs contain both conserved QLQ and WRC domains, and GIFs contain the SSXT domain. Consistent with previous research, 12 OsGRFs, 9 AtGRFs, 3 OsGIFs, and 3 AtGIFs were identified in rice and Arabidopsis. However, the numbers of ZmGRFs and ZmGIFs were different from previously published data. Seventeen ZmGRFs and four ZmGIFs were found in maize. ZmGRF15-17 and ZmGIF4 are new to this study (Iwasaki et al. 1995; Zhang et al. 2008) (Fig. 2A). All the newly identified members were amplified by PCR and then sequenced. The results were consistent with the sequences published online except for a small number of mutations at the 3’ end (Fig. S6), indicating that the gene assembly sequences were credible. Subsequently, the protein sequences of GRFs and GIFs were extracted from the B73 reference V4 protein sequence for multispecies comparisons and phylogenetic tree construction. With little available information on foxtail millet, 10 SiGRFs and 3 SiGIFs of foxtail millet were named based on rice homologous proteins (Choi et al. 2004). Gene information was searched using the phytozome and plant-regulomics websites. As shown in Table S1, we analyzed the protein length, molecular weight, and isoelectric point of GRFs and GIFs using Expasy (http://web.expasy.org/compute_pi). The details of each protein are listed in Table S1. Most GRFs were basic proteins (39/48) in the four species, and all GIFs were acidic proteins (Table S1). In addition, the specific tissue localization at the single cell level was analyzed using PlantscRNAdb (http://ibi.zju.edu.cn/plantscrnadb/index.php). AtGRF1-4 and OsGRF3 were specifically expressed in roots, whereas ZmGRF1-3, 6, 12–16 were specifically expressed in shoot apical meristems. This result indicated that GRFs participated in different biological processes in different tissues, such as the SAM and roots.

Fig. 2.

Chromosomal location of GRFs and GIFs. The chromosomal locations of GRFs and GIFs are displayed for maize (A) and foxtail millet (B), The chromosome number is marked above each chromosome, and the approximate locations of the GRFs and GIFs on the chromosome are marked in the figure. The newly identified GRF/GIF genes are indicated in orange font. The scale is on the left

Phylogenetic analysis of GRF and GIF proteins

To further analyze the phylogenetic relationship among the GRF and GIF proteins in multiple species, a phylogenetic tree was constructed with MEGAX by the neighbor-joining (NJ) method and 1000 bootstrap replicates. Then, the phylogenetic tree was adjusted and visually improved by iTOL (https://itol.embl.de/). According to prior studies and the distance of phylogenetic tree branches, GRFs were divided into three subfamilies (A, B, C) (Fig. 1, Table S1) (Zhang et al. 2008). The GRF A subfamily was further divided into classes A1 and A2. The GRF C subfamily was divided into classes C1 and C2. AtGRF9 and ZmGRF10, which belong to the GRF A family, negatively regulated leaf size. OsGRF10 and OsGRF11 affected the development of flowers. Thus, the homologous genes named SiGRF7/10 and the newly discovered ZmGRF15 may participate in similar biological processes (Wang et al. 2020). The GRF B subfamily included 14 members, and the newly discovered ZmGRF16, ZmGRF17 and SiGRF6 were homologous to OsGRF6, indicating that they may be involved in regulating inflorescence structure and plant height in maize and foxtail millet (Tang et al. 2018). The GRF C family included the largest number of 21 members and includes two important genes AtGRF5 and OsGRF4 that affect plant regeneration. In addition, 9 ZmGRFs and 5 SiGRFs can potentially improve genetic regeneration in maize and foxtail millet (Fig. 1A, Table S1).  GIFs were divided into three subfamilies according to their branching distances in the phylogenetic tree. The GIF A subfamily comprises four members, including the first discovered ZmGIF4, GIF B includes 3 members and GIF C includes 6 members (Fig. 1B). As previously reported, 6 pairs of SiGRF genes experienced replication events (Chen and Ge 2021). Ten ZmGRF pairs, including 14 gene members, were identified by intraspecific collinearity analysis in maize, and ZmGRFs were subjected to purification options (ka/ks < 1) (Table. S2). To summarize, the phylogenetic analysis confirmed the results of the identified conserved domains of GRFs and GIFs. The proteins derived from monocots clustered separately from those of dicots, and proteins from maize and foxtail millet shared a closer phylogenetic relationship compared to maize and rice. The new genes ZmGRF15-16 and ZmGRF17 may be involved in flower development and plant height respectively. GRF C family members may also participate in plant regeneration efficiency. To ensure that ZmGRF15, ZmGRF16 and ZmGRF17 were not due to a gene assembly problem, we cloned the gDNA sequences and they did indeed contain existing ZmGRF15, ZmGRF16 and ZmGRF17 (Fig. S6).

Fig. 1.

Phylogenetic analysis of the GRF and GIF families from maize, foxtail millet, rice, and Arabidopsis thaliana. A GRF family. GRFs were divided into three subfamilies (A, B, C), and the GRF A subfamily was further divided into classes A1 and A2. The GRF C subfamily was divided into classes C1 and C2. Golden yellow represents GRF A1, dark yellow represents GRF A2, green represents GRF B, red represents GRF C1, and blue represents GRF C2. B GIF family. GIFs were also divided into three subfamilies (A, B, C), yellow represents the GIF A subfamily, green represents GIF B and red represents GRF C. A phylogenetic tree was generated using the MEGA-X and the neighbor-joining method, and 1000 bootstrap replicates were performed

Chromosomal location of GRFs and GIFs

The distribution of GRF and GIF genes in the whole genome was investigated in multiple species. Most GRFs were distributed in pairs in maize, foxtail millet, and rice but not in Arabidopsis. For example, ZmGRF13 and ZmGRF16 were distributed adjacent to each other at the base of Chr 1, and at the top of Chr 5, ZmGRF8 had a tandem duplicated gene ZmGRF17 (Fig. 2, S1). This result suggested that the GRF and GIF gene families were evolutionarily distinct in dicotyledons and monocotyledons.

The gene structure of GRFs and GIFs

All GRFs contain a QLQ and WRC domain, indicating that the GRFs including the newly identified members are reliable (Fig. S2). The amino acid lengths of GRFs fluctuate between 200–600 amino acids, and GIFs comprise approximately 200 amino acids. The gene lengths of GRFs and GIFs were mostly below 4000 bp and contained 3–4 exons including newly discovered genes named ZmGRF15 and ZmGRF16 (Fig. 3), but ZmGRF17 and ZmGIF3 were longer than 10,000 bp. These results suggest that there is minor variation at the genetic level in the GRF and GIF families. A total of 20 motifs were predicted for all GRF and GIF proteins through the MEME website (http://meme-suite.org/tools/meme). Notably, all GRFs contain motif 1 (WRC) and motif 2 (QLQ) (Fig. 3A, Fig. S3). Near QLQ, motif 5 has been present in the majority of GRFs and performed largely conserved GRF activities. Motif3 is present in most GRF C subfamily members as an important feature. Motif13 exists in AtGRF5, SiGRF3 and ZmGRF9/3, which may be a functionally conserved structure. ZmGRF15 contains fewer motifs, including three very conserved motifs 1, 2, and 5 and one GRF A subfamily specific motif 15. ZmGRF16 and ZmGRF17 contains 12 and 11 motifs, respectively. GIFs all contain motif 1 (SSXT), ZmGIF3 contains a very large intron, and the newly identified ZmGIF4 is highly similar to ZmGIF3, but it contains motif 19 similar to SiGIF3.

Fig. 3.

GRF and GIF structures and GRF and GIF protein motifs. Gene structure and protein motif analysis of GRFs (A) and GIFs (B). Exons, introns, and UTR regions are indicated by yellow shapes, black lines, and green shapes, respectively, on the right. The length of each GRF and GIF was estimated as shown by the scale at the bottom. A Twenty GRF motifs were identified by the MEME suite. Each motif is represented by a different color and number, motif1 (WRC), motif2 (QLQ) and others in Supplementary Fig. 3. B Twenty GIF motifs were identified by MEME suite. motif1 (SSXT) and others in Supplementary Fig. 4

The cis-acting regulatory elements in the promoter of GRF and GIF genes

Cis-elements in promoters play essential roles in gene expression, response to external environmental changes and transcriptional regulation. We predicted cis-acting elements of the putative promoter region (upstream 2000 bp) of each GRF in maize, foxtail millet, rice and Arabidopsis through the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) (Lescot et al. 2002). After filtering common elements such as the TATA-box, CAAT-box, TATA-box, and light-responsive elements, many drought stress-related cis-elements (e.g., MYB and MYC) were considered to be distributed in the promoters of GRFs and GIFs (Iwasaki et al. 1995; Agarwal et al. 2006; Onishi et al. 2006). In addition, numerous ABREs, ARE, as-1, TGACG-motif, and CGTCA-motif were also found in the promoters of GRFs and GIFs, which were related to ABA, SA, anaerobic induction or MeJA-responsiveness (Garretón et al. 2002; Narusaka et al. 2003; Li and Chen 2015). For instance, fourteen ZmGRF members contain ABREs, most of which contain more than two in the promoters. Among them, ZmGRF9 contains the most ABRE, while ZmGRF15, ZmGRF16 and ZmGRF17 have 2, 5 and 3 elements, respectively. All GRFs contain ABREs in foxtail millet, and SiGRF9 contains 7 ABREs. The GRF/GIF promoter contains a large number of drought related cis-elements. For example, ZmGRF1 contained 15 drought related cis-elements (MBS/MYB/MYC), and ZmGRF10 contained 14 cis-elements. Moreover, twenty GRF promoters contained auxin-responsive-related TGA-elements. For example, ZmGRF3 and the new members ZmGRF15 and ZmGRF17 contained 3 TGA-elements (Fig. 4, Table S3). Gibberellin-responsive-related P-box elements were found in 13 GRF promoters (Fig. 4, Table S3). Taken together, GRF and GIF promoters contained many stress-related elements, especially ABREs suggesting that these TFs regulated plant responses to environmental changes. ABA regulates various physiological processes in plants, such as acting as a signal to respond drought and heat stress, inducing stomatal closure and regulating the expression of stress-related genes (Huang et al. 2016; Kuromori et al. 2018). To verify whether GRFs and GIFs respond to stress, the third leaves of maize seedlings were sprayed with 100 μM abscisic acid (ABA) and treated for 0 h, 0.5 h, 1 h, and 4 h. Then, we completed qRT–PCR with RNA extracted from the third leaves (Fig. 5A). The expression levels of ZmGRF4, 12, 14 and ZmGIF2 were significantly decreased 4 h after ABA treatment. The expression levels of ZmGRF9 and ZmGRF17 were significantly increased compared to control after ABA treatment. Subsequently, many drought stress-related cis-elements (e.g., MYB and MYC) were considered to be distributed in the promoters of GRFs and GIFs. When maize seedlings were subjected to drought for five days, qRT‒PCR was used to examine ZmGRFs/ZmGIFs changed under drought conditions. The results showed that almost all ZmGRF/ZmGIF expression levels decreased. In particular, the expression levels of ZmGRF1, 2, 6, and 10 decreased more than three-fold under drought conditions (Fig. 5B). It was demonstrated that ABA regulated the expression of these GRF genes to suppress plant growth (Fig. 5).

Fig. 4.

Distribution of major stress-related and hormone-related cis-elements in the promoter sequences of the GRF and GIF genes from maize, foxtail millet, rice, and Arabidopsis thaliana. Twenty cis-elements were identified in the promoter 2000 bp upstream of ATG in GRFs (A) GIFs (B) using the PlantCARE database. ABRE cis-acting element involved in the abscisic acid responsiveness, ARE cis-acting regulatory element essential for the anaerobic induction, TGA-element cis-acting regulatory element essential for auxin-responsive element, P-box was about gibberellin-responsive element and others shown in Supplementary Table 3

Fig. 5.

ZmGRFs/ZmGIFs expression levels in ABA treatments by qRT‒PCR. A qRT‒PCR analysis of ZmGRFs/ZmGIFs expression levels by ABA treatments. The third leaf of maize seedlings was sprayed with 100 μM ABA 15 DAP (days after planting). ZmUBI2 was used as an internal control. Error bars indicate standard deviations (SDs) based on three biological replicates. B qRT‒PCR analysis of ZmGRFs/ZmGIFs expression levels under drought conditions for 5 days

Multiple collinearity relationship of GRF and GIF genes in multiple species

Collinearity analysis can account for the evolutionary relationship of genes among different species. Multiple collinearity analyses of GRFs and GIFs were completed between Arabidopsis, maize, foxtail millet, and rice using TBtools software. The results showed many ortholog pairs between the maize, foxtail millet, and rice genomes but fewer ortholog pairs with Arabidopsis. Eighteen GRF/GIF genes in maize corresponded to 12 GRF/GIF genes in foxtail millet, 14 GRF/GIF genes in rice corresponded to 13 GRF/GIF genes in foxtail millet, and a GRF9 in maize corresponded to GRF5 in Arabidopsis. In addition, certain GRF/GIF genes had one-to-two or more ortholog among different species. For example, ZmGRF15-SiGRF10-OsGRF10/12 are orthologs corresponding to each other, ZmGRF1/5/6-SiGRF3/4-OsGRF3/4 are also orthologs corresponding to each other, as were ZmGRF16/17-SiGRF6-OsGRF6 and ZmGRF4/10-SiGRF7-OsGRF11. However, only one orthologue relationship was found, AtGRF5-ZmGRF9-SiGRF1/2-OsGRF1/2 between Arabidopsis and other species (Fig. 6A, B, Table S4). According to previous studies, two AtGRF5 homologous genes named ZmGRF3 and ZmGRF9 significantly improved transformation efficiency (Kong et al. 2020). To determine the connection between ZmGRFs and TaGRF4, a crucial gene regulating callus regeneration, we also performed a collinearity study between wheat and maize. Unfortunately, despite finding 40 GRF and GIF pairs showing collinearity between wheat and maize, we were unable to identify the collinear gene of TaGRF4 in maize. Following a BLAST comparison, we discovered that TaGRF4 shared the highest similarity with ZmGRF1, 5 and 6 (Luo and Palmgren 2021; Qiu et al. 2022). However, few studies have reported the function of ZmGRF9 and SiGRF1/2. These results could provide a reference for future studies on the potential functions of ZmGRF9.

Fig. 6.

Collinearity relationship of GRFs and GIFs between four species. The collinearity between Arabidopsis and maize, maize and foxtail millet, and foxtail millet and rice, was compared separately using TBtools software, and the collinearity relationship of a conserved GRFs or GIFs is marked with a colored line. The AtGRF5-ZmGRF9-SiGRF1/2-OsGRF1/2 collinearity relationship is labeled, and the others are shown in Supplementary Table 4

Expression profiles of GRFs and GIFs

To analyze the expression patterns of GRFs and GIFs, published transcriptome data were downloaded. Most GRFs and GIFs are highly expressed in seeds, such as the seed size regulating gene OsGRF4 (Li et al. 2016). Furthermore, ZmGRF1, 2, 5–7, 10–13, 15–17, ZmGIF1, 4 and AtGRF1-6, 9 were also highly expressed in seeds, suggesting that these genes potentially affect seed development. SiGRF7 and 10 were found to be highly expressed in flowers. SiGRF1, 4, and 6 were highly expressed in the shoots. SiGRF2, 3, 9, and AtGIF3 were highly expressed in the shoots and seeds. ZmGRF3, 9, OsGRF6, and SiGIF2 were highly expressed in roots, in which OsGRF6 positively regulated root development (Tang et al. 2018). ZmGRF4, 8, 14, ZmGIF2, OsGRF11, and SiGIF8 were expressed in leaves (Fig. 7, Table S5). We also investigated the tissue expression of GRF and GIF according to recent information on single-cell technology by the PlantscRNA database (http://ibi.zju.edu.cn/plantscrnadb/index.php). The results revealed that AtGRF1-4, AtGIF1, 2, OsGRF3, and OsGIF3 were specifically expressed in roots; ZmGRF1-3, 6, and 12–16 were specifically expressed in shoot apical meristems (Chen et al. 2021) (Table S1). The expression levels of GRFs in different organs provide evidence to distinguish the gene functions of GRFs.

Fig. 7.

The different tissue expression profiles and interaction networks for GRF and GIF genes in four species. A Different tissue GRF and GIF gene expression levels. Seed (embryo stage 10), flower (flower stage 10), shoot (adult, stem internode, second internode), root (adult, root) and leaf (adult, rosette leaf, rosette leaf 11) in Arabidopsis (Mergner et al. 2020); seed (embryo), flower (tassel tissues), shoot, root and leaf (leaf-middle) in maize (Liu et al. 2020); seed, flower (panicle 3), shoot (stem), root and leaf (leaf 3) in foxtail millet (Yang et al. 2020); seed, flower (post flowering panicle), shoot, root and leaf in rice (Sakai et al. 2011). The different tissues are noted on the bottom of each lane, and cluster dendrograms are shown on the left and top, respectively. The relative color scale value is displayed on the right. B The interaction network for GRF and GIF. In the interaction network by string database (https://cn.string-db.org/), the red sphere indicates GIFs, the green sphere indicates GRFs, and the blue line indicates the strength of the interactions (combined score)

Protein interaction networks between ZmGRFs and ZmGIFs

Previous studies have shown that GRFs and GIFs regulate plant growth and development synergistically through protein interactions (Kim 2019). From this, all the potential interactions between GRFs and GIFs were predicted using the STRING database (www.string-db.org/) (Kim et al. 2003; Choi et al. 2004; Zhang et al. 2008; Szklarczyk et al. 2021). The predictions indicated that all AtGIFs interact with all AtGRFs, and the interactions were then evaluated by yeast two-hybrid assays (Liang et al. 2014). In foxtail millet, SiGIF1/2/3 interact with SiGRF1, 2, 6–9 but not SiGRF3-5, 10. OsGIF1 interacts with OsGRF1, 2, 6–9, and 11, and OsGIF2/3 interacts with OsGRF1, 2, and 11. This may explain why OsGIF1 has more vital functions in organ development, such as regulating plant height, number of grains per panicle, internode length and leaf length/width (He et al. 2017). ZmGIF1/2/3 interact with 8 members, ZmGRF2, 8, 10, 11, 13, and 15–17, while no interactions were discovered for ZmGIF4. Interestingly, ZmGRF8 interacts with 9 ZmGRFs, including ZmGRF1, 2, 5, 6, 10–12, 14, and 15. Therefore, ZmGRF8 may function as a bridge to participate in the interaction between ZmGIFs and ZmGRFs in maize. Based on GWAS SNPs in the maize GDP database (https://www.maizegdb.org/gbrowse), ZmGRF8 was identified as a candidate gene influencing plant height (Woodhouse et al. 2021). Taken together, there was a wide range of interactions between GRFs and GIFs. Moreover, some GRFs function as bridge to indirectly achieve interactions between GRFs and GIFs and thereby function together (Fig. 7, Table S6).

ZmGRF9 interactions with ZmGIFs

Although we have identified many potential interactions by updating protein data (RefGen_V4) in maize, the STRING database is still using old maize protein data (RefGen_V3) to complete predictions, resulting in no interactions being found for the conserved ZmGRF9 or ZmGIF4. The same result can be detected in foxtail millet. Exploring members interacting proteins of GRF will help to further understand their functions. The potential interacting members ZmGIF1/2/3/4 and 8 ZmGRFs (ZmGRF15 excepted) were highly expressed in the vegetative meristem (Walley et al. 2016).To investigate the function of ZmGRF9, we verified the interaction between ZmGRF9 and ZmGIFs by yeast two-hybrid assays. When ZmGRF9 was cotransferred with ZmGIF2, 3, and 4, the yeast was able to survive in QDO + x-α-gal medium (QDO: SD-Try/Leu/His/Ade). This result suggests that ZmGRF9 interacts with ZmGIF2-4, but not ZmGIF1 (Fig. 8A). Moreover, we verified the authenticity of the interaction between ZmGRF9 and ZmGIF2, 3, 4 through the LCI system using young N. benthamiana leaves (Fig. 8A). The results showed that ZmGRF9 interacted strongly with ZmGIF3, ZmGIF4, but weakly with ZmGIF2 (Fig. 8A).

Fig. 8.

Interaction and localization of ZmGRF9 protein. A Interaction test of ZmGRF9 with ZmGIFs. DDO represents SD-Trp/Leu, and QDO represents SD-Trp/Leu/His/Ade. A firefly luciferase complementation assay was conducted using young N. benthamiana leaves (right). B GFP-ZmGRF9 subcellular localization in tobacco (left) and maize protoplast (right). C ZmGRF9 structure analyzed by the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/) (Jumper et al. 2021). The purple ellipses in 15–50 indicate the QLQ domain, and the yellow ellipses in 89–133 indicate the WRC domain. D miRNA396 target sites on ZmGRF9 sequences

Subcellular location of ZmGRF9

Acting as a transcription factor family, GRFs are localized in the nucleus. Thus, we performed a transient transformation assay in tobacco leaves and maize protoplasts to verify the subcellular location of ZmGRF9. The laser confocal microscope observation revealed that ZmGRF9 was localized in the nucleus of cells, which was consistent with the prediction (Fig. 8B). We also predicted the protein structure of ZmGRF9 using the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/). The QLQ structural domain was located on the 15–50 region containing an alpha helix and the WRC was located at 89–133 containing large number of corners to form a binding structure (Fig. 8C). These results indicate that ZmGRF9 containing a helical QLQ and a folded WRC is localized in the nucleus and functions as a transcription factor.

The miR396–GRF/GIF module

MicroRNAs (miRNAs) are 21-nucleotide RNAs and play important roles in plant growth and development (Rhoades and Bartel 2004; Chorostecki et al. 2012; Bologna and Voinnet 2014). Numerous studies have demonstrated that miR396 can inhibit the expression of the GRF family by binding to the terminus of the WRC domain (Debernardi et al. 2012; Liebsch and Palatnik 2020). All the downstream target genes of Zm-miRNA396 were predicted by the latest database PmiREN2.0 (Plant miRNA Encyclopedia, https://www.pmiren.com/) (Fig. 9, Table S7). Zm-miRNA396h matches exactly with the end sequence of WRC in ZmGRF9, and miRNA396c/d differs only in the second to last base (Fig. 8D). In this network, a total of 8 Zm-miRNA396a-h members were found in maize, among which Zm-miRNA396c/d/h redundantly coregulated many downstream genes including several ZmGRFs except for ZmGRF4, and ZmGRF10 (Fig. 9). In addition, Zm-miRNA396e regulated a small amount of ZmGRFs including ZmGRF2 and ZmGRF17, and miRNA396f regulated ZmGRF16. Unlike maize, only two miRNA396 members were found in Arabidopsis thaliana. At-miRNA396a participated in regulating AtGRF1, 2, 7, and At-miRNA396b regulated the expression of AtGRF3, 4, 7, 8, indicating that At-miRNA396a/b had different roles in regulating AtGRFs. In foxtail millet, five Si-miRNA396 members such as Si-miRNA396b/e regulated most SiGRFs except SiGRF2, 6, and 7, indicating that redundant functions exist in coregulating pathways. In rice, seven Os-miRNA396 members were found, and Os-miRNA396d/e/f/g could redundantly coregulate most OsGRFs except OsGRF11, while Os-miRNA396a/b could regulate a small amount of OsGRFs containing OsGRF6-8. In conclusion, multiple miRNA396 members have high functional redundancy in the regulation of GRFs in maize, rice and foxtail millet, but minimal redundancy in Arabidopsis (Fig. 9, Fig. S5).

Fig. 9.

The miR396–GRF/GIF module in maize and foxtail millet. All the downstream target genes of Zm-miRNA396 by PmiREN (Plant miRNA Encyclopedia, https://www.pmiren.com/) (Guo et al. 2022). The red sphere represents miRNA396 of each species, the blue sphere represents the target gene, and the line with an arrow represents the regulation direction and details of the genes in Supplementary Table 7

Discussion

New members of the GRF and GIF gene families in maize and foxtail millet

Maize and foxtail millet are significant C4 crops in the food industry (Lata et al. 2013; Wang et al. 2014; Peng and Zhang 2021). The GRF/GIF gene families play important roles in regulating genetic transformation efficiency and plant development. The GRF/GIF module affects the development of the whole plant, including leaf, stem, root, seed and flower nitrogen utilization (Kim 2019; Kong et al. 2020; Wang et al. 2020; Luo and Palmgren 2021; Qiu et al. 2022). The comprehensive identification of members of the GRF/GIF family contributes to the improvement of plant breeding and biotechnology applications. A high quality reference genome is necessary to improve the accuracy of gene family analysis. The oldest genome sequencing of maize and foxtail millet was completed in 2009 and 2012, and the latest published genomes may fill a large number of gaps due to technical limitations that cannot accurately capture many details (Schnable et al. 2009; Zhang et al. 2012; Jiao et al. 2017).

Although GRFs/GIFs have been studied in some plants, with continuous updating of genome assembly and improvement in assembly quality, there is an urgent need to study the most accurate members of the gene family (Schnable et al. 2009; Wei and Chen 2018; Zhao et al. 2019; Chen and Ge 2021). In this study, we performed a genome-wide investigation of the GRF/GIF gene families in maize and foxtail millet using updated genome data. The protein structure was parsed using motif analysis and multiple sequence alignment. Meanwhile, neighbor-joining techniques were used to create the phylogenetic tree. The GRF and GIF gene families consisted of 17 ZmGRFs, 10 SiGRFs, 4 ZmGIFs and 3 SiGIFs. Based on previous research, GRFs were divided into three subfamilies (A, B, C) (Fig. 1, Table S1). Three new ZmGRF members and one new ZmGIF member were discovered compared to past studies. These new members were verified by sequencing (Fig. S6). ZmGRF15 belongs to the GRF A subfamily and is homologous to OsGRF10, which may be involved in flower development. ZmGRF16 and ZmGRF17 belong to the GRF B subfamily and are similar to OsGRF6, which may be related to flower and plant height (Gao et al. 2015; Tang et al. 2018; Wang et al. 2020). Due to high similarities in sequence and expression pattern, ZmGIF4 and ZmGIF3 may have similar functions. In addition, ZmGIF4 is highly expressed in seeds (Li et al. 2016), which is similar to the expression pattern of OsGRF4, and may play an important role in regulating the size of seeds (Fig. 6). The identification of these new members greatly extends the understanding of the functions of the GRF/GIF families, and may play important roles in improving crop yields. The specific functional mechanism of these members needs to be further studied by genetic transformation.

Exploring the functions of ZmGRFs/ZmGIFs in the ABA response

Understanding ABA signaling is crucial for improving plant performance in the future as it is the key endogenous messenger of the plant response to stress and can cope with ongoing threats in plant production, such as biotic and abiotic stresses such as drought (Raghavendra et al. 2010). The regulation of gene expression in plants occurs primarily through the interaction between multiple cis-elements and trans-factors at the transcriptional level. In response to ABA signaling, AREB/ABF transcription factors may be linked to genes whose promoters contain ABRE cis-elements (Iwasaki et al. 1995; Narusaka et al. 2003; Yoshida et al. 2014; Li and Chen 2015). The GRF family is reportedly involved in ABA signaling. For example, the expression of AtGRF7 is inhibited to activate osmotic stress response genes. AtGRF7 functions as a repressor of stress response genes to lessen the negative impact of these genes on plant growth. AtGRF7 binding to the AtDREB2a promoter inhibits ABA signaling (Kim et al. 2012; Kim and Tsukaya 2015). In this work, we predicted the cis-elements in the 2000 bp promoters of all GRF/GIF members and discovered that the GRF/GIF promoters contain various cis-elements, most notably ABREs (Fig. 4). It is likely that the relationship between the response of the GRFs/GIFs and ABA is much stronger than what was uncovered. To verify whether GRFs and GIFs responded to ABA, we performed qRT‒PCR to detect the expression patterns of GRFs/GIFs in the leaves of maize seedlings treated with ABA (Fig. 5). The expression levels of ZmGRF4, 12, 14 and ZmGIF2 were significantly decreased at 4 h after ABA treatment. This may be related to inhibition of the expression of these genes involved in growth and development under stress conditions, thus weakening the growth rate of plants to better withstand challenging environments. The expression levels of ZmGRF9 were significantly increased compared with control after ABA treatment. This could be attributed to the high expression pattern of ZmGRF9 in roots, which can increase the ability of plants to absorb water and nutrients and withstand environmental stresses. The data demonstrated that GRF/GIF genes might be involved in ABA signaling in maize.

The essential roles of ZmGRF9 in regeneration

With the progress of CRISPR and other molecular technologies, there is an urgent push to realize efficient transgenic technology. However, only a portion of inbred lines or varieties have the ability to regenerate, such as the maize inbred lines KN5585 and B104. Implementation of transgenic technology is limited by the generally low rate of regeneration of plant materials, which has been dependent on species, genotype, and tissue type (Qiu et al. 2022). When the expression of GRFs changes, plant regeneration can be initiated in a variety of plant cell and tissue types that were previously difficult to regenerate, according to studies in multiple plant species (Yadava et al. 2016; Kausch et al. 2021). Increased genetic transformation of explants can be achieved by introducing AtGRF5 and AtGRF5 homologous genes into different plant materials. In addition, in maize, two potential AtGRF5 homologs can significantly improve transformation efficiency (Kong et al. 2020). The production of the fusion protein from wheat TaGRF4 and TaGIF1 significantly increased the regeneration efficiency and speed of wheat, triticale and rice, as well as the diversity of transformable wheat genotypes (Debernardi et al. 2020). In this work, we discovered that the only collinear relationship was AtGRF5-ZmGRF9 between Arabidopsis and maize, and the similarity of protein sequences between ZmGRF9 and AtGRF5 was 82.5% (Fig. 7). In recent research, it has been reported that ZmGRF9 can promote regeneration in maize (Kong et al. 2020), revealing that these two genes have some conserved functions. However, AtGRF5 was highly expressed in seeds and flowers, whereas ZmGRF9 was highly expressed in roots, stems and seeds. This discrepancy in the expression pattern revealed some parallels and differences between AtGRF5 and ZmGRF9 (Figs. 6, 7). Further interaction analysis showed that ZmGRF9 could interact with ZmGIF2-4, but the interaction between ZmGRF9 and ZmGIF2 was weaker (Figs. 6, 8), which may be because ZmGIF3 and ZmGIF4 had similar expression patterns to ZmGRF9 in seeds and roots, but greater differences in ZmGIF2 expression patterns (Fig. 6). These members may form a GRF-GIF complex regulating regeneration, which would be helpful to promote effective genetic transformation techniques in the future. In addition to ZmGRF9, other members may also have regenerative effects. For example, ZmGRF3, which is highly homologous to ZmGRF9, may be redundant and improve regeneration efficiency together with ZmGRF9. ZmGRF1, which is highly homologous to OsGRF4 or TaGRF4, deserves further study. Due to microRNA analysis, there might be another regulatory route based on a specific upstream miR396. The expression of ZmGRF9, ZmGRF3, and ZmGRF1 can be increased by blocking the expression of miR396 to significantly improve the efficiency of regeneration in maize. Despite these insights, how ZmGRF9 and ZmGIF2-4 regulate regeneration in maize is still unknown. The functions of these ZmGRF9/ZmGIF family members in maize need to be confirmed through a series of experiments in the future.

Conclusions

In summary, we identified the GRF and GIF families in maize and foxtail millet by updating sequences, resulting in the identification of three new ZmGRFs and one new ZmGIF4. GRF promoters contain many ABA-responsive elements, and qRT‒PCR experiments indicated that ZmGRF4, 9, 12, 14 and ZmGIF2 are associated with the ABA response in maize. All GRF and GIF members were analyzed by expression and interaction, and it was found that GRFs and GIFs are extensively involved in diverse processes of plant growth. Regeneration efficiency-related ZmGRF9 was found to be conserved with AtGRF5 by collinearity analysis of the four species and ZmGRF9 could interact with ZmGIF2-4. Moreover, ZmGRF9 has been localized in the nucleus and is regulated by miRNA396c/d/h. This work provides a considerable expansion of understanding related to the miRNA396-GRF-GIF module for regulating plant growth and regeneration efficiency.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Conflict of interest

The authors declare no competing interests.