Genome-wide identification and characterization of the SPL gene family and its expression in the various developmental stages and stress conditions in foxtail millet (Setaria italica)

Among the major transcription factors, SPL plays a crucial role in plant growth, development, and stress response. Foxtail millet (Setaria italica), as a C4 crop, is rich in nutrients and is beneficial to human health. However, research on the foxtail millet SPL (SQUAMOSA PROMOTER BINDING-LIKE) gene family is limited. In this study, a total of 18 SPL genes were identified for the comprehensive analysis of the whole genome of foxtail millet. These SiSPL genes were divided into seven subfamilies (I, II, III, V, VI, VII, and VIII) according to the classification of the Arabidopsis thaliana SPL gene family. Structural analysis of the SiSPL genes showed that the number of introns in subfamilies I and II were much larger than others, and the promoter regions of SiSPL genes were rich in different cis-acting elements. Among the 18 SiSPL genes, nine genes had putative binding sites with foxtail millet miR156. No tandem duplication events were found between the SiSPL genes, but four pairs of segmental duplications were detected. The SiSPL genes expression were detected in different tissues, which was generally highly expressed in seeds development process, especially SiSPL6 and SiSPL16, which deserve further study. The results of the expression levels of SiSPL genes under eight types of abiotic stresses showed that many stress responsive genes, especially SiSPL9, SiSPL10, and SiSPL16, were highly expressed under multiple stresses, which deserves further attention. In this research, 18 SPL genes were identified in foxtail millet, and their phylogenetic relationships, gene structural features, duplication events, gene expression and potential roles in foxtail millet development were studied. The findings provide a new perspective for the mining of the excellent SiSPL gene and the molecular breeding of foxtail millet.


Introduction
Foxtail millet (Setaria italica) was first domesticated from the ancestor green foxtail around 16 000 Years Before Present (YBP) [1][2][3] and became the main crop in Northern China around 5000-6000 YBP [4][5][6]. As a diploid C4 crop, it has a great significance for energy utilization, with their photosynthesis especially improving nitrogen and water utilization efficiency [7,8]. In addition, foxtail millet is self-compatible and has a highly conserved genome structure, which makes forward or reverse genetics and trait location easier [9]. Foxtail millet is rich in various nutrients, including phenolic compounds (phenolic acids, flavonoids, and tannins) [10], organic acids (gallic acid, caffeic acid, and chlorogenic acid) [11], lipophilic oxidants (vitamin E and carotenoids) [12], proline-rich prolamin (prolamin) [13] and trace elements (zinc and iron) [14]. As a model plant, foxtail millet is of great value in many aspects [2,8,15], but foxtail millet does not seem to have received enough attention at present. It is worth noting that, as a nutritionally balanced crop, it plays an important role in maintaining human health. Under the Covid-19 pandemic, studies have shown that foxtail millet has the potential to become a new staple food crop, especially in hunger hotspots [16]. Therefore, the study of foxtail millet is of both biological and socio-economic significance.
SPL is a key regulator of a variety of biological processes in plants, including changes from nutrition to the reproductive phase, leaf development, tiller/branching, plastid pigment, spike/spike structures, grain ripening, fertility, and response to stress [36]. In A. thaliana, AtSPL2, AtSPL9, AtSPL10, AtSPL11, AtSPL13, and AtSPL15 play important roles in the transition from young plants to mature plants and from vegetative growth to reproductive growth. AtSPL3, AtSPL4, and AtSPL5 play important roles in promoting the transformation of floral meristems [28]. AtSPL14 can resist fumonisin B1 [37], and AtSPL8 affects gibberellic acid biosynthesis and ultimately regulates reproductive development [38,39]. There are also studies on the SPL gene in rice, such as OsSPL14, which regulates plant structure by inhibiting the number of tillers in rice, increasing grain weight and enhancing disease resistance [40][41][42]. OsSPL16 promotes grain filling and improves the quality and yield of rice [43,44]. In addition, OsSPL3 can improve the cold resistance of plants [45]. Some of the wheat SPL genes have been found to be down regulated under NaCl and PEG treatments [33], while 13 ZmSPLs in maize were found to be involved in the drought stress response [35]. Since the SPL transcription factor family was found to be targeted by miR156 [18,46], the gene function of SPL has been more widely studied. The regulatory pathway of the miR156/SPL module is highly conserved in different plant species and plays an important role in regulating plant fitness, biomass, and yield [36]. Moreover, as plants age, the level of miR156 decreases, thus alleviating the inhibition of SPL targets [47][48][49].
SPL plays a crucial role in plant growth, development, and stress response. Overall, the identification, classification, evolution, and gene function studies of SPL gene families in foxtail millet are not systematic. Therefore, this study comprehensively analyzed the sequence composition, gene structures, cis-acting elements, miR56 binding sites, chromosome positions, and gene replication events of 18 SPL gene families in the millet genome. The evolutionary relationship of SiSPL genes were also analyzed among several species, including rice, A. thaliana, sorghum, maize, tomato, and tartary buckwheat. The grouping, motif composition, collinearity, and evolutionary relationship between the SiSPL genes and other plants were analyzed. We also studied the spatial expression and tissue expression patterns of SPL genes in different tissues during foxtail millet development. As a result, the roles of specific SPL gene members in different biological processes of foxtail millet were determined. In addition, the expression of SPL genes in foxtail millet under eight types of abiotic stress were investigated. In this study, a comprehensive analysis of the SPL gene family of foxtail millet was carried out, which not only screened important SPL genes under growth, development, and stress treatment, but also provided insights for the study of the SPL gene family of other plants.

Identification of SPL proteins in foxtail millet
We used two blast methods to extract SPL genes from the whole genome of foxtail millet. After removing redundant genes, a total of 18 SPL genes were obtained. These 18 SPL genes were mapped to the nine chromosomes and renamed SiSPL1 to SiSPL18 according to their location. The characteristics of these SiSPL genes, including coding sequence length, amino acid sequence length, molecular weight of protein, isoelectric point of protein, and prediction of protein subcellular localization, were analyzed. Among the 18 SPL genes, the smallest protein had only 181 amino acids (SiSPL4), while the largest protein had 1118 amino acids (SiSPL14). The molecular weight of the protein ranged from 19.2 kDa to 122.18 kDa with an isoelectric point ranging from 5.58 to 9.91. The results of subcellular localization showed that a total of 16 genes were located in the nucleus, one gene was localized in the plastid, and the other was located in the chloroplast (Additional file Table S1).

Multiple sequence alignment, phylogenetic analysis, and classification of SiSPL genes
To explore the evolutionary relationship of SPL genes in foxtail millet, 16 A. thaliana SPL genes and 18 SiSPL genes were used to construct a phylogenetic tree using MEGAX software. According to the classification method of the AtSPL gene family, as shown in the evolutionary tree, the SPL gene family can be divided into eight subfamilies, namely groups I to VIII (Fig. 1a). However, there were no SiSPL members in group IV, and the final SiSPL gene family was divided into seven subfamilies. Among these subfamilies, group I contained the least SiSPL gene members, only one (SiSPL8), followed by group VI with two members (SiSPL5 and SiSPL6), while the other subfamilies contained three SiSPL members. According to the location of the SBP domain, we extracted the SBP domain sequence of the AtSPL and SiSPL genes (approximately 76 amino acids) and used them for multiple sequence alignment. As shown in Fig. 1b, the foxtail millet SBP domain contains highly conserved sequences, such as CQQC, SCR, and RRR. In addition, a highly conservative nuclear localization signal (NLS) was found in the foxtail millet domain. Two zinc finger-like structures, Zn-1 and Zn-2, were also found in the SBP domain. There was a certain mutation in the Zn-1 (Cys-Cys-Cys-His) sequence of subfamily I, and the fourth histidine was mutated to cysteine. Therefore, the conformation of zinc finger binding site 1 of subfamily I may be mutated, conferring a special function to subfamily I. However, this phenomenon was not observed in other subfamilies containing highly conserved Zn-1 (Cys-Cys-Cys-His) and Zn-2 (Cys-Cys-His-Cys).

Gene structure and motif composition of the SiSPL gene family
The evolution of the SPL gene family in foxtail millet was further explored by studying the intron-exon structures of all SiSPL genes. As shown in Fig. 2, while the intronexon structures of the same subfamily were similar, while the differences of different subfamilies were large. For example, subfamily II (SiSPL11, SiSPL14, and SiSPL18) had the largest number of introns, with an average intron number as high as 9.7. Second, SiSPL8 (subfamily I) contained six introns, while the other subfamilies had a small To explore the differences in the conserved motifs of each foxtail millet SPL protein, we used the MEME website to analyze the motif composition of the entire sequence, including the sequence of the SBP domain. Ten different conserved motifs were identified from these sequences, namely motifs 1 to 10 ( Fig. 2c, Additional file Table S2). Motifs 1 and 2 were distributed among all SiSPL members, motifs 9/6/5/10 were unique to subfamily II (SiSPL11, SiSPL14, and SiSPL18), and none of the other subfamilies contained these four motifs. In addition, subfamily VIII (SiSPL4 and SiSPL13) also contained a special motif, motif 8. The motif arrangement of the same subfamily was similar, indicating that these protein structures were conserved relatively, which supported the reliability of the SiSPL subfamily classification.

Cis-acting element for the SiSPL gene family
The promoter region 2-kb upstream of the SiSPL genes were further analyzed. A total of 86 cis-acting elements were found in the promoter region of these SiSPL genes (Additional file Table S3). They were divided into seven categories, including promoter-related elements, lightresponse elements, hormone response elements, environmental stress-related elements, development-related elements, binding site-related elements, and other elements. Among these elements, light reaction-related elements (20) and development-related elements (14) accounted for a larger proportion. Light reaction-related elements (G-box, Sp1, and TCCC-motif ) are present in most SiSPL genes. Development-related elements, such as root-specific expression (as-1) and meristem expression element (CAT-box), are also widely distributed in the SiSPL gene family. In addition, we also found that most SiSPL genes had cis-acting elements involved in abscisic acid (ABRE), methyl jasmonate (TGACG-motif, CGTCA-motif ), and auxin (TGA-element), while a few genes also had cis-acting elements involved in gibberellin (P-box, GAREmotif ), ethylene (ERE), and salicylic acid (TCA-element). Among the environmental stressrelated elements, hypoxia inducible (GC motif ), low temperature response (LTR), and drought inducible-related elements (MBS) were found to be widely distributed in SiSPL genes. Among the promoter-associated elements, the core elements associated with transcription  Table S2 initiation (TATA and CAAT box) were present in all SiSPL genes, which proves that our promoter analysis is reliable. Finally, 25 other cis-acting elements were found in foxtail millet SPL genes, of which STRE (AGGGG) and Unnamed_4 (CTCC) were found in all SiSPL genes; however, the functions of these elements are unknown.
In addition, Venn analysis was performed on cis-acting elements contained in more than 10 genes (Fig. 3) to further analyze the significance of these elements. For example, among the environmental stress-related elements, the three SiSPL genes (SiSPL9, SiSPL16, and SiSPL2) contained ARE (AAA CCA ), GC-motif (CCC CCG ), LTR (CCG AAA ), and MBS (CAA CTG ) cis-acting elements. Among the light response-related elements, four SiSPL genes (SiSPL10, SiSPL15, SiSPL14, and SiSPL4) contained G-Box, Sp1, and TCCC-motif elements at the same time.

MiR156 family in foxtail millet and their target site in SiSPL genes
In order to understand the post-transcriptional regulation of SiSPL mediated by miR156, we searched all SiSPL coding regions and 3'-UTR to find the target sites for miR156 to bind SiSPL. To this end, we studied the miR156 family in the whole millet genome and further analyzed their targets on these SiSPL genes. First, we downloaded the foxtail millet miR156 family from PMRD (http:// bioin forma tics. cau. edu. cn/ PMRD/) and found four members of the miR156 family: miR156a, miR156b, miR156c, and miR156d ( Fig. 4a, b). The sequences of mature miR156a, miR156b, and miR156d were identical, except for the fact that the sequence of mature miR156c differed by two bases. However, the precursor sequences of the four members differed considerably. We found a total of nine SiSPL with possible complementary binding to miR156 among the 18 genes. miR156-targeted SiSPL genes were distributed across groups V-VIII, with potential sites of binding to miR156 found in the coding regions of eight genes, namely SiSPL2, SiSPL4, SiSPL10, SiSPL12, SiSPL13, SiSPL15, SiSPL16, and SiSPL17 (Fig. 4c, Additional file Table  S4). The 3'-UTR region of SiSPL5 identified a potential site for binding to miR156. These results indicate that miR156 may specifically regulate the expression of these SiSPL genes in foxtail millet. miR156 complementary sites are relatively conserved among these SiSPL genes, and the sequence differences are mainly limited to the first, second, and eighth nucleotides of the complementary sequence. This means that these complementary sites may be under great selection pressure in the process of evolution, even those located in the 3'-UTR. These results suggest that miR156-mediated post-transcriptional regulation plays an important role in the function of the foxtail millet SPL gene. The chromosomal location of 18 SiSPL genes was determined using the foxtail millet genome, which distributed in nine chromosomes of foxtail millet. The least number of SiSPL genes were distributed in chromosome VII, chromosome VIII, and chromosome IX with only one SiSPL gene each. The remaining chromosomes contained two or three SiSPL genes. Gene duplication events are essential for the evolution of gene families and tend to play an important role in gene expansion and the generation of new functional genes. Therefore, we investigated the duplication events of the SiSPL gene in the whole foxtail millet genome and found that there was no tandem repeat event for the SiSPL gene. However, four pairs of segmental duplication events were found in the SiSPL gene family (Fig. 5, Additional file Table S5) using the BLASTP and MCScanX methods. These segmental duplication genes are distributed on chromosomes 1/2/4/5/6, with each pair of segmental duplication genes is contained in the same subfamily. These results indicate that segmental duplication events play a major role in the expansion of the SiSPL gene in the evolution of the SiSPL gene family.
To further study the phylogenetic mechanism of the foxtail millet SPL gene family, we constructed syntenic maps between six representative species with foxtail millet, which are three monocotyledons (rice, sorghum, and maize) and three dicotyledons (A. thaliana, tartary buckwheat, and tomato) (Fig. 6). Among the homologous genes of the foxtail millet SPL gene and other species, the homologous genes with maize was the most (42 pairs). This was followed by rice (28 pairs), sorghum (26 pairs), Arabidopsis (7 pairs), tomato (2 pairs), and tartary buckwheat with the least homologous genes (one pair). Interestingly, SiSPL11 had homologous genes in all six species (Additional file Table S6). SiSPL12 has homologous genes with five other species except tartary buckwheat, suggesting that SiSPL11 and SiSPL12 may have existed before the differentiation of monocotyledons and dicotyledons, and play important roles in these species after differentiation. In general, the foxtail millet SPL genes have the best homology with maize, and these highly homologous genes may have evolved from the common ancestor of different plants. To better understand the evolutionary constraints acting on the SiSPL gene family, the Ka/Ks values of the SiSPL gene pairs were calculated (Additional file Table S7). The results showed that the Ka/Ks values of the gene pairs of each subfamily and the gene pairs of all segmental repeats were all less than 1, suggesting that the SPL gene family of foxtai millet may have undergone strong purification selection pressure during evolution.

Evolutionary analysis of SiSPL proteins and the SPL genes of several other species
We constructed phylogenetic trees of SiSPL proteins with SPL proteins from monocots and dicots using Mega X. Among them, monocots included rice (19  Table S8). As shown in Fig. 7, SPL proteins can be divided into six subfamilies in the phylogenetic tree (labeled a-f ). Each of the seven species contributed to at least one SPL gene in different subfamilies, and 18 SiSPL genes were almost evenly distributed in these six subfamilies.
The conserved motifs of the SiSPL protein were analyzed using online MEME analysis. The motifs 3, 2, 1, and 6 were conserved and distributed almost alternately across the whole subfamily. This indicates that this motif composition is very important for these SPL genes and may be of great significance for the SPL protein. However, the differences between different subfamilies were also relatively large. This included subfamily a that contained conserved motifs 4/7/5/8, while the remaining subfamilies did not contain. Subfamilies b and e contained special motifs 9 and 10. Subfamily c contained special motif 9 but lacked motif 10. The situation in subfamily d was the opposite to that in subfamily c. Subfamily f showed the most differences, containing motifs 3, 2, 1, and 6, and no other conserved motifs.

Expression patterns of the SiSPL genes in different foxtail millet tissues
To better characterize SPL gene function in foxtail millet, qRT-PCR was used to determine SiSPL gene expression in different tissues (roots, stems, young leaves, mature leaves, husks, and seeds) at the mid-grain filling stage of foxtail millet. We selected at least one representative gene from each subfamily, and nine SiSPL genes were detected for their expression patterns in different tissues. As shown in Fig. 8a, different genes have differential expression patterns in different tissues. All genes   Table S2 had extremely high expression levels in seeds (about 10 folds more than controls), and the high expression levels of six genes (SiSPL3, SiSPL8, SiSPL9, SiSPL10, SiSPL11, and SiSPL17) were observed in young leaves. SiSPL10 and SiSPL17 were expressed at higher levels in roots (10 and 15 folds more than the control, respectively). Interestingly, SiSPL17 also had an extremely high expression profile in husk and mature leaves, unlike the expression profiles of other genes, as reflected by the broader expression abundance of SiSPL genes.
In addition, the correlation between gene expression patterns in different tissues was analyzed (Fig. 8b). The results showed that the expression patterns of most genes were positively correlated (P < 0.05). For example, three genes, SiSPL11, SiSPL14, and SiSPL16, showed significant positive correlations with each other, and they also showed significant correlations with SiSPL3 and SiSPL6. Interestingly, six genes were negatively correlated with SiSPL17, indicating that the expression pattern of SiSPL17 in tissues was different from that of other genes.

Expression patterns of SiSPL genes in foxtail millet during grain development
The expression of the SiSPL gene in different tissues showed that it was mainly highly expressed in seeds. Therefore, we further investigated the expression levels of SiSPL in millet grains and husks of early, middle, and late grain filling. As shown in Fig. 9a, the SiSPL gene was expressed to varying degrees during seed and husk development at these stages. In addition, SiSPL17 was highly expressed in the early stage of seed filling and decreased gradually during the filling process, while the other eight genes were highly expressed in the middle stage of seed filling. In addition, genes that had higher expression levels were also observed in the husk during the early stage of filling (SiSPL8, SiSPL10, SiSPL11, and SiSPL14) and the later stage of filling (SiSPL6, SiSPL10, SiSPL14, and SiSPL17).
We also analyzed the correlation between SiSPL gene expression patterns and found that most of these genes were positively correlated (Fig. 9b). For example, significant positive correlations were found between pairs of genes, namely SiSPL3, SiSPL6, SiSPL8, and SiSPL9, which were also significantly correlated with SiSPL11 and SiSPL16, respectively. SiSPL17 was only significantly correlated with SiSPL10 and positively correlated with SiSPL11 (P < 0.05), but negatively correlated with all other genes.

Expression patterns of SiSPL genes in response to different abiotic stresses
To further determine whether the expression of the SiSPL gene was affected by different abiotic stresses, qRT-PCR experiments were carried out on nine SiSPL members to analyze their expression patterns in roots, stems, and leaves under different treatments (acid, alkali, NaCl, PEG, flooding, dark, heat, and cold treatments). In general, some genes were significantly expressed under stress while others were inhibited (Fig. 10a). For example, SiSPL9, SiSPL10, SiSPL11, SiSPL14, SiSPL16, and SiSPL17 in roots, stems, and leaves showed different levels of induced expression under all stress conditions, especially SiSPL9, SiSPL10, and SiSPL16(more than 15 folds higher than the control), which showed extremely high levels of induced expression under stress. In contrast, some genes, such as SiSPL6, were inhibited under stress, and were down regulated to varying degrees under acid, alkali, NaCl, PEG, and flooding stress. Under stress treatment, more genes tended to reach the highest level of expression at 24 h. For example, under salt stress, the expression of most genes reached a maximum at 24 h.
In addition, a correlation between SiSPL gene expression patterns under stress was also observed, and most of the SiSPL genes were positively correlated (P < 0.05) (Fig. 10b). For example, there was a significant positive correlation between SiSPL8 and SiSPL9, which also showed positive correlations with SiSPL3, SiSPL6, SiSPL11, SiSPL14 and SiSPL16, respectively. A significant positive correlation was found between SiSPL11 and SiSPL14, as well as with SiSPL17.

Characteristics of SiSPL genes
The analysis of the physicochemical information of 18 SiSPL genes showed that the relative molecular weight of the proteins ranged from 19.2 kDa to 122.18 kDa, with an isoelectric point ranging from 5.58 to 9.91. The following were predicted for subcellular localization: nucleus (16), plastid (1), and chloroplast (1). This result is consistent with the results of SiSPL gene domain information alignments, in which a highly conserved nuclear localization signal (NLS) is present in the SBP domain. The analysis of SPL gene families in other species also showed that most SPL genes were located in the nucleus [23]. The SBP domain contains highly conserved sequences, such as CQQC, SCR, RRR, NLS, Zn-1, and Zn-2, which is the same as the SBP domain of the SPL gene family in other species [31,33]. In AtSPL7 and OsSPL9 (group IV), the His of the first Zn fingerlike structure was replaced by a Cys residue, which was consistent with SiSPL8, suggesting that the Zn fingerlike structure of subfamily I was relatively special. The SPL gene family in foxtail millet can be divided into seven subfamilies, which lack group IV and are different from Arabidopsis [18]. The AtSPL6 gene in group IV plays an important role in plant immunity [50]. In addition, PpSBP3, a member of group IV in the bryophyte Physcomitreiia patens, can inhibit reproductive development [51]. The results of conserved motif analysis of millet SPL protein showed that foxtail millet SPL genes contained the motifs 1 and 2 generally. The conserved motif composition of members of the same subfamily was similar, indicating that the protein structure was conserved in a specific subfamily. However, there are large differences in the conserved motifs among different subfamilies, such as motifs 9, 6, 5, and 10, which are unique to subfamily II (SiSPL11, SiSPL14, and SiSPL18), and none of the other subfamilies contain these four motifs. These motifs may confer special physiological functions to this subfamily, but their specific functions remain to be explored in depth. The regulation of gene expression by cis-elements in the promoter region has become the main mechanism for organisms to adapt to different environments [52]. Therefore, in addition to studying the structural characteristics of the SiSPLs, we also carried out an in-depth analysis of the upstream promoter region of these genes. A large number of cis-acting elements were found in the promoter region. Light response-related elements (G-box, Sp1, and TCCC-motif ) were present in most SiSPL genes, suggesting that these cis-acting elements may play an important role in the pathway of the SiSPLs in plant light response. We also identified root-specific expression elements (as-1) and meristem expression elements (CAT-box) in most of the SiSPLs promoter regions. In addition, a large number of SiSPL genes have environmental stress-related elements, such as hypoxia-induced (ARE and GC-motif ), low-temperature response (LTR), and drought-induced (MBS)-related elements. This also shows that the SiSPLs plays an important role in external stress; for example, miR156 in alfalfa improves drought tolerance by silencing SPL13 [53]. We also found that most SiSPLs have cis-acting elements involved in abscisic acid (ABRE), methyl jasmonate (TGACG-motif and CGTCA-motif ), and auxin (TGA-element). Methyl jasmonate, a plant hormone and signal molecule associated with injury, exists widely in plants, and its exogenous application can stimulate the expression of defense genes. Previous studies have found that SPL9 interacts with JAZ3 and attenuates the jasmonate (JA) response, making plants resistant to insects in seedlings [54]. At the organ and whole plant levels, auxin acts from seedling to grain ripening. SPLs have been found to be involved in the auxin synthesis pathway. Previous studies have shown that SPL2, SPL10, and SPL11 can inhibit root regeneration by inhibiting callus-induced auxin synthesis [55]. It was also found that SPL2, SPL10, and SPL11 in older plants can directly bind to the promoters of AP2/ERFs to inhibit their expression, thus destroying auxin accumulation in callus [55].

Evolution of SiSPL genes
In higher organisms, introns can regulate gene expression at many levels, and their main role is to produce different exon combinations through alternative splicing to translate different proteins and improve protein complexity [56,57]. Interestingly, recent studies have shown that introns have important functions independent of the genes they encode. They not only mediates cell responses to starvation [58], but also regulate cell growth rate under stress conditions to improve the adaptability of yeast [59]. The analysis of the SPL gene structure in foxtail millet showed that the number of introns in the SiSPL gene of groups I and II was much higher than that of other subfamilies. Thus, members of the foxtail millet SPL subfamilies I and II may be older. At present, most studies tend to favor the early intron hypothesis, that is, the existence of a large number of introns in relatively old ancestors, which means that the loss of a large number of introns in evolution is common [60][61][62][63]. Research showed that most of the repetitive events of millet genes are generated in whole genome duplication (WGD) events shared by gramineous plants [64]. We investigated the repeat events of the SiSPL gene in the whole genome of foxtail millet and found that there were no tandem duplication events; however, four pairs of segmental duplication events existed in the SiSPL gene family. This suggests that segmental duplication events play a major role in the expansion of the SiSPL gene over the evolutionary course of the SiSPL gene family. In addition, collinearity analysis with other six species showed that the foxtail millet SPL gene had the most homologous genes with maize, followed by rice and sorghum, while tartary buckwheat had the least. Therefore, the millet SPL gene has the highest (See figure on next page.) Fig. 9 Expression pattern and correlation of 9 S. italica SPL genes during grain development. a qRT-PCR was used to detect the expression patterns of 9 S. italica SPL genes in husk and grain before, during and after grain filling. Error bars were obtained from three measurements. homology with maize. Interestingly, SiSPL11 has homologous genes with all six species. Thus, we speculated that SiSPL11 may come from their common ancestor and may have existed before monocotyledon and dicotyledonous differentiation. However, the intron of SiSPL11 (Fig. 2) also confirmed that this gene may be relatively primitive. Finally, to better understand the evolutionary constraints of the SiSPL gene family, we analyzed the nonsynonymous and synonymous substitutions of the SiSPL gene pairs. The results showed that the ka/ks of both the gene pairs of each subfamily and the segmental duplication gene pairs was less than 1, such that the foxtail millet SPL gene family should have experienced strong purification and selection pressure in the process of evolution.
The function of miR156 in regulating plant flowering seems to be conserved, with the overexpression of miR156 in rice [65], Arabidopsis [48], maize [66], potato [67], and Arabis alpina [68] being associated with a lateflowering phenotype. Among the 17 SPL genes in A. thaliana, 11 are miR156 targets [18,28]. In rice, 11 OsSPL genes are predictive targets of OsmiR156 [65]. Among the 18 SiSPL genes of foxtail millet, nine members were complementary to miR156. Among them, eight genes were found in the coding region, and one gene (SiSPL5) found a potential site that could bind to miR156 in the 3'-UTR region. Interestingly, AtSPL3, AtSPL4, and AtSPL5 in group VI also have binding sites for miR156 in the 3'-UTR region, and miR156 can regulate the expression Fig. 10 Gene expression of 9 S. italica SPL genes in plants subjected to abiotic stresses (acid, alkali, NaCl, PEG, dark, flooding, heat, cold) at the seedling stage. a qRT-PCR was used to detect the expression patterns of 9 S. italica SPL genes in roots, stems and leaves at different times. Error bars were obtained from three measurements. Lowercase letter above the bar indicates significant difference (α = 0.05, LSD) among treatments. b Positive number: positively correlated; negative number: negatively correlated. Red numbers indicate a significant correlation at the 0.05 level of AtSPL3 through transcriptional cleavage and translation inhibition to maintain normal plant morphology and flowering time [69]. Therefore, we speculated that SiSPL5 may be very important for the growth and development of foxtail millet. In addition, SiSPL and miR156 complementary site sequence differences were mainly observed in the first, second, and eighth nucleotides, with previous studies reporting similar results [70]. It means that these complementary sites are under great selection pressure in the process of evolution, indicating that miR156-mediated regulation is very important for foxtail millet SPL gene expression.

Spatio-temporal expression patterns of the SiSPL gene and its response to abiotic stress
Members of different SPL subfamilies often have different functions, which has been proven in previous studies. Generally, SPL transcription factors are encoded by miR156, but not all SiSPL genes have binding sites. For example, SiSPL8, a member of subfamily I, does not have binding sites with miR156, which is similar to AtSPL7. However, previous studies have shown that AtSPL7 plays an important role in maintaining copper homeostasis in plants [71]. Similar to the SPL subfamily I genes, the SPL genes of subfamily II members are widely expressed in different tissues and at different stages of plant development, but lack the negative regulation of miR156 [22,31,72]. Arabidopsis thaliana has four SPL subfamily II members, namely AtSPL1, AtSPL12, AtSPL14, and AtSPL16, which are widely expressed in seedlings, stem leaves, stem apical meristem, flowers, fruits, and roots [31]. In tomato, the genes slysbp12a and slysbp12b, two group II members of the SPL subfamily, are widely expressed from seedlings to mature fruits [22]. In this study, SiSPL11 and SiSPL14, members of subfamily II, were also expressed in different tissues (root, stem, leaf, husk, and grain).
The members of rice and Arabidopsis SPL subfamily III also lack miR156 binding sites [18,65], which is consistent with SiSPL9 (subfamily III). Functional studies of AtSPL8, a subfamily III member of Arabidopsis, were earlier and clearer. Its genetic mutants greatly affected the seed setting rate, petal trichome production, and root growth, and were found to do so by positively and negatively regulating gibberellin (GA) signaling in flowers and roots, respectively [38,39]. The analysis of the SiSPL9 promoter sequence revealed that it contains the participation of a gibberellin response element P-box (CCT TTT G); thus, SiSPL9 may also be involved in the response pathway to GA signals. In addition, the SiSPL9 promoter also contains cis-acting elements for hypoxia (ARE and GC-Motif ), low temperature (LTR), and drought (MBS), which may result in high levels of SiSPL9 expression under multiple stress treatments, is supported by the experimental results.
The dicotyledonous Arabidopsis subfamily IV contains only AtSPL6, and no homologous gene of AtSPL6 is found in monocotyledons, indicating that this gene pedigree has been lost in evolution [28]. AtSPL6 and its homologous gene NbSPL6 in tobacco play a positive regulatory role in plant innate immunity mediated by the NTIR-NB-LRR receptor [73]. In A. thaliana, AtSPL2, AtSPL10, and AtSPL11 of SPL subfamily V have binding sites for miR156 [74], and the overexpression of miR156b can reduce the accumulation of AtSPL2, AtSPL10, and AtSPL11 mRNA [75]. Functional analysis also showed that AtSPL2, AtSPL10, and AtSPL11 can regulate the morphological characteristics of cauline leaves and flowers [76]. Foxtail millet SPL subfamily V contains SiSPL2, SiSPL3, and SiSPL10. SiSPL3 and SiSPL10 were expressed in different tissues and periods, and SiSPL10 was significantly induced under multiple stresses.
The most extensive function of the SPL genes are to promote plant growth transition from seedling to mature plant [77,78]. Overexpression analysis of AtSPL3, AtSPL4, and AtSPL5 genes of SPL subfamily VI showed that they are related to changes in plant morphological characteristics [79], and these functions are realized through the negative regulation of miR156 [47,48,80]. The SPL subfamily VI in foxtail millet contains SiSPL5 and SiSPL6, and the analysis of the expression pattern of SiSPL6 showed that it is highly expressed in seeds at the middle filling stage (Fig. 8a) indicating an important role in the adult vegetative growth. In addition, the functions of AtSPL3, AtSPL4, and AtSPL5 were verified under different sunshine conditions [81,82]. Under shortday exposure, miR156 negatively regulates these three genes, and SUPPRESSION OF OVEREX-PRESSION OF CONSTANS1 (SOC1) positively regulates these genes through the GA pathway. Under long-day exposure, SOC1, FLOWERING LOCUS T (FT), and FLOWER-ING LOCUS D (FD) regulate these three motifs in the leaves to cope with light signals. Both SiSPL5 and SiSPL6 have light response elements, G-Box and TCCC motifs, and Sp1, respectively (Additional file Table S3), which suggests that SiSPL5 and SiSPL6 may also be involved in light response pathways similar to the Arabidopsis SPL subfamily VI genes. Interestingly, the results showed that the expression of SiSPL6 was not inhibited under dark, cold, or heat stress, but was inhibited to varying degrees under other stress treatments. In addition, OsSPL13, a homolog of SiSPL5 in rice, can positively regulate the cell size in rice husk, thus improving the grain length and yield of rice.
The expression of SiSPL17 in the same subfamily of AtSPL13 was highly in roots, leaves, husks, and seeds, except for in the stems (Fig. 8a). SiSPL12 and SiSPL15 are found in the same subfamily, which are homologous to AtSPL13, OsSPL2, and OsSPL16 (Additional file  Table S6). AtSPL13, a single member of subfamily VII, was widely expressed in different tissues of Arabidopsis, delaying leaf growth after cotyledon emergence during seed germination [83,84]. OsSPL16 encodes a protein that positively regulates cell proliferation, and the high expression of this gene can promote cell division and grain filling, thus increasing the grain width and rice yield [44]. In addition, OsSPL16 directly binds to the GW7 promoter and inhibits its expression, thereby affecting rice yield and quality [43]. Therefore, the functions of SiSPL12 and SiSPL15 in foxtail millet are worthy of further discussion.
In this study, SiSPL16 was not only highly expressed in different tissues but was also induced under abiotic stress. Subfamily VIII of SPL in foxtail millet includes SiSPL4, SiSPL13, and SiSPL16, in which SiSPL4 and SiSPL13 are homologous to upland rice OsSPL14 and OsSPL17 (Additional file Table S6). OsSPL14, also known as IPA1, has been studied extensively and thoroughly, and OsSPL14 is regulated by miR156 in vivo [41]. OsSPL14 can inhibit the growth of tiller buds, but promotes the branches of the panicle to increase yield [40,42,85,86], as well as being able to enhance immunity [87]. OsSPL17 plays an important role in the regulation of rice panicle structure [88,89]. AtSPL9 belongs to subfamily VIII of SPL, which participates in petal trichome initiation by activating TRICHOME-LESS (TCL1) and the accumulation of anthocyanin pigments in vegetative stems [90,91]. In addition, researchers have also found that the miR156-AtSPL9-miR172 regulatory pathway plays an important role in the development of young leaves to mature leaves of A. thaliana [92]. The gene function of other members of the subfamily VIII of SiSPL warrants further study.

Plant materials, growth conditions, and abiotic stress treatments in foxtail millet
Yugu 1, a cultivar from northern China, was used as the experimental material in this study. And this material was supplied by Prof. Cheng Jianping. For the experimental materials planted in the greenhouse, we obtained the husk and seed in the early, middle, and late filling stages, as well as the roots, stems, and leaves in the middle filling stage. Samples were taken from plants grown under the same conditions (five replicates). The samples were immediately frozen in liquid nitrogen and stored at -80℃ for future use. The expression levels of nine SiSPL genes from different subfamilies were determined. In addition, foxtail millet plants at seedling stage (28 days) were treated with eight abiotic stresses (salt: 5% NaCl, acid: 0.1 mol/L HCl, alkali: 0.2 mol/L NaOH, drought: 30% PEG6000, flooding: whole plant, darkness: complete shading, heat: 40℃, and cold: 4℃). For acid, alkali, salt and drought stress, the same volume of liquid immersion root system was maintained in different repetitions under the same stress, and the treatment method was referred to Zhang et al. [93,94]. These genes were sampled for qRT-PCR analysis at 0, 2, and 24 h.

Total RNA extraction, cDNA reverse transcription, and qRT-PCR analysis
Total RNA was extracted from all samples using an RNA extraction kit (TaKaRa Bio, cat: 9769) and then reverse transcribed (HiScript II Q RT SuperMix for qPCR, cat: R223-01). The qRT-PCR primers for these genes (Additional file Table S9) were designed using Primer 5.0. Si001873 mg (Actin) was used as an internal control. Standard qRT-PCR with SYBR Premix Ex Taq II (TaKaRa Bio, cat: RR820A) was repeated three times on a CFX96 Real-Time System (Bio-Rad). The gene expression was evaluated using the 2 −ΔΔCt method.

Phylogenetic analysis and classification of the SiSPL gene family
Phylogenetic trees were constructed using protein sequences (Oryza sativa, A. thaliana, Solanum lycopersicum, Sorghum bicolor, maize, and Fagopyrum tataricum) downloaded from the UniProt Database (https:// www. UniPr ot. org) with muscle wrappers and built an ML phylogenetic tree with IQ-tree wrapper (bootstrap number set to 1000). The classification of the foxtail millet SPL gene family was based on the classification method of the Arabidopsis SPL gene family.