Firstly, we obtained the PYL full-length protein sequence of rice from the NCBI (http://www.ncbi.nlm.nih.gov/nucleotide/) (Yadav et al., 2020). The rice PYLs gene sequences were used as query sequences in the Phytozome database (https://phytozome-next.jgi.doe.gov/) to find PYLs in the finger millet genome. This was done by using BLASTp searches (http://blast.ncbi.nlm.nih.gov/) against the finger millet genome E. coracana v1.1 (Goodstein et al., 2012; Hittalmani et al., 2017; Hatakeyama et al., 2018). The appeared duplicate and redundant sequences were eliminated by venny2.1 (https://bioinfogp.cnb.csic.es/tools/venny/) (Oliveros, 2016). Then, the NCBI open reading frames (ORF) Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) was used to determine ORFs. On the other hand, we used Pfam (https://www.ebi.ac.uk/interpro/entry/pfam/) to search for the polyketide cyclase 2/START domain (PF10604), which is a characteristic of the PYR/PYL protein sequence. Further, multiple sequence alignment by MEGA11 was used to identify conserved amino acid residues among all putative PYL sequences of finger millet (Tamura et al., 2021). Genes that lack specific conserved domains were sorted and removed. Further the domain position, physicochemical attributes and subcellular localization were predicted by HMMSCAN (http://www.ebi.ac.uk/Tools/hmmer/search/hmmscan, ExPaSy Protparam (https://web.expasy.org/protparam/) and Wolf-Psort online tool (https://www.genscript.com/tools/wolf-psort) respectively (Gasteiger et al., 2005; Horton et al., 2007).

The PYL gene family in finger millet has been designated as EcPYL1-1A, which indicates for the first PYL genes of E. coracana and is found on chromosome 1A. Similarly, other finding was also annotated and used for evolutionary study. An evolutionary analysis was made between putative PYLs genes of finger millet and the PYLs genes of rice (Yadav et al., 2020), sorghum (Dalal and Inupakutika, 2014), and foxtail millet (Duarte et al., 2019), which were retrieved from the Phyozome and NCBI database. All of the retrieved sequences of the PYL genes family were aligned through ClustalW, and the phylogenetic tree was constructed through MEGA11 using the Maximum Likelihood method at 1,000 bootstrap levels (Larkin et al., 2007; Tamura et al., 2021). MapGene2 Chromosome v2.1 (http://mg2c.iask.in/mg2c_v2.1/) was used for physical mapping of PYLs gene of finger millet (Devos et al., 2023), and TB tool was used to display the exon-intron structure (Chen et al., 2020; Chao et al., 2021). The MEME Suite Vs. 5.5.3 (https://meme-suite.org/meme/tools/meme) predicted the conserved motif within each finger millet protein sequence. The maximum number of motifs was set to 6, keeping others at default parameters (Bailey et al., 2015).

A ratio of non-synonymous (Ka) versus synonymous (Ks) substitution rates of nucleotides was calculated among finger millet PYLs gene family by using Ka/Ks calculation tool (http://services.cbu.uib.no/tools/kaks) and the gene duplication event was calculated by using T = Ks/2λ), where (λ)= (6.5 × 10⁻⁹ (Gaut et al., 1996; Zhang et al., 2006). It measures the selective pressures placed on genes and can be used to identify pairs of genes where encoded proteins may have changed their function. When the ratio of Ka/Ks > 1, it indicates that there is a positive selective pressure or Darwinian selection. If the ratio is 1, it indicates neutral evolution and if the ratio is less than 1, then it indicates stabilizing selection, i.e., the protein sequence is pressured to be conserved.

The 2,000 bp upstream sequences from the translations tart site of all of the EcPYLs genes were obtained from finger millet genome at Phytozome database (Goodstein et al., 2012) and the putative cis regulatory elements were predicted by using the PlantCARE web server (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) to know the putative function associated with the genes (Lescot et al., 2002). According to the cis-regulatory elements, PYL genes were categorized into three groups based on their response to phytohormones, stress, and growth and development.

For the prediction of miRNA target sites in the EcPYL genes, the miRNA dataset of finger millet was retrieved from PmiREN server (https://www.pmiren.com/) (Guo et al., 2020) and cDNA sequence of EcPYLs genes were examined against miRNA dataset using the pSRNATarget server (https://www.zhaolab.org/psRNATarget/) (Dai and Zhao, 2011). To determine the interrelationships among the 14 putative EcPYLs genes, the STRING database (https://string-db.org/) (Szklarczyk et al., 2021) was used at confidence level (0.07) against rice (Yu et al., 2002), Arabidopsis (Cheng et al., 2017), sorghum (McCormick et al., 2018) and foxtail millet (Zhang et al., 2012) genome one by one. Using the combination of Cytoscape v. 3.10.3 with MCODE (Molecular Complex Detection), the most accurate prediction of the interaction network was displayed. Additionally, the integration of Cytoscape v. 3.10.3 with BinGO program were used for Gene Ontology (GO), functional annotation and gene enrichment profiling of EcPYL genes of finger millet (Paul et al., 1971; Bader and Hogue, 2003).

After characterization of EcPYL genes, machine learning algorithm was used to predict the expression pattern of these genes. For the prediction of probability of expression of EcPYL genes under drought and salinity stresses we used ASRPro webserver (http://iasri-sg.icar.gov.in/asrpro/result.php) (Meher et al., 2024). This web server utilized machine learning algorithms to assess the likelihood of gene expression under the given stress conditions, and generated a probability score for each gene. Genes with a probability score of ≥0.70 were considered highly likely to be expressed in response to drought and salinity stress. On the basis of the probability of gene expression, we constructed a heatmap and scatter plot to predict the candidate genes for qRT-PCR expression profiling.

Five PYL genes (EcPYL4-2A, EcPYL7-2B, EcYL11-5A, EcPYL12-5A, and EcPYL14-5B) with probability score of 0.70 were selected for qRT-PCR analysis. Primers for each gene were designed using Primer3 input software (https://bioinfo.ut.ee/primer3-0.4.0/), and the primers specificity was done with genome sequences of finger millet to ensure that primers only bind to the intended target sequence. The primer sequences, annealing temperature, and product size of each primer were listed in Table 1. Healthy and viable seeds of finger millet genotype (IC87469) was selected for this experiment, and grown under drought (15% PEG-induced drought stress) and salinity (200 mM NaCl) stress conditions according to previous protocols (Rakkammal et al., 2023b; Rakkammal et al., 2023a). Both shoot and root tissues were harvested after 28 days of growth, and they were used for gene expression analysis. Total RNA was isolated from shoot and root tissues using the Trizol method (Rakkammal et al., 2023a; Rakkammal et al., 2023b), and cDNA was synthesized using the QuantiTect Reverse transcription kit (Qiagen, Germany). EF-1α was used as a reference gene to normalize the expression pattern of each PYL gene. Cycle threshold values of each PYL genes were analyzed using the formula 2^–∆∆Ct. A total of three biological and three technical replicates were used for each treatment. Whereas, statistical analysis were performed through origin 2021 software (https://www.originlab.com/origin) using ANOVA, followed by Tukey test for multiple comparisons for significantly difference genes (p ≤ 0.05).

The SWISS MODEL predicted the in silico structure of EcPYLs proteins (Waterhouse et al., 2018). PDBsum (http://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) and the SAVES v6.0 server (https://saves.mbi.ucla.edu/) were used to validate, analyze, and estimate the quality of the build model in two dimensions (Laskowski et al., 2018). Whereas, ResProx (http://www.resprox.ca/) (Berjanskii et al., 2012) was used to estimate the resolution of the model and SuperPose v. 1 (http://superpose.wishartlab.com/) and Biovia Discovery Studio 2019 was used for superposition and visualization of model, respectively.

A total of 27 PYL genes were retrieved from the finger millet genome (Supplementary Table S1). Moreover, based on conserved domain analysis, the redundant sequences were removed, and 14 PYL genes were considered for further in silico studies. Table 2 shows the gene position, ORF length, domain position, and chromosome location of these 14 PYL genes. These genes have the polyketide cyclase 2/START domain and have the Pfam accession number PF10604. These 14 putative EcPYLs genes were unevenly distributed among four out of nine sets of chromosomes (Figure 1). The chromosomes 2A and 2B possess three EcPYL genes, chromosomes 5A and 5B has two EcPYL genes while chromosomes 1A and 1B and 3A and 3B possess only one EcPYL gene. EcPYL genes were not observed on chromosomes 4, 6, 7, 8, and 9. In general, both A and B sub-genomes have seven EcPYL genes, showing that there is not much difference in the number of PYL genes at the sub-genome level. For a better understanding, we named the possible genes EcPYL (E. coracana PYL) followed by the number of chromosomes located, for example, EcPYL_1A. We identified 14 EcPYL genes in finger millet and sequentially named them as EcPYL1 to EcPYL14, followed by their respective chromosomal numbers. The physicochemical attributes of putative PYLs of finger millet (Table 3) revealed that EcPYLs ranged from 181 to 222 aa. Accordingly, the molecular weight ranged from 19.5 to 23.7 kDa, and PI ranged from 5.5 to 9.3. The ORF lengths predicted by the ORF finder for EcPYLs ranged from 546 to 669 bp (Table 2). On the other hand, WolfsPort’s predicted subcellular localization of EcPYLs genes revealed that the majority of the putative PYLs genes of finger millet were located in the cytoplasm and chloroplast (Table 3). Further, all the putative PYLs of finger millet were hydrophilic in nature as they possess negative value of Grand Average of Hydropathicity.

It was demonstrated that EcPYL can be classified into two distinct clades based on the intron/exon structure analyzed using TB tool’s (Figure 2). In subfamilies I and II, no introns were detected, whereas all members of subfamily III except EcPYL14-5B contain two introns (Figure 2). As shown in Figure 3a, motifs 1, 2 and 3 were conserved across all EcPYL genes. Among the 14 EcPYLs studied, motif 1 contains the conserved characteristic Gate–Latch domain (Figure 4). However, motif 4 was found to be conserved across all receptors except EcPYL3/5/8. Motif 5 was found to be present only in the EcPYL3/6/9/10 genes, whereas EcPYL11 and EcPYL13 of the 5A and 5B chromosomes contained Motif 6. The members of the subfamily share conserved motifs and possess unique motifs (Figure 3a). The unique motifs could indicate that each member performs a unique or specialized biological function, while similar motifs reveal functional similarities. Figure 3b displays the logos of these identified motif groups, indicating the frequency of amino acid occurrence at that site.

In terms of sequence similarity, EcPYL gene can be grouped in three major subfamilies, namely, subfamily-I, subfamily-II, and subfamily-III (Figure 5). Among the 14 EcPYL genes, EcPYL1/2/11/13/9/10 (six PYLs distributed among chromosomes 1A,1B, 5A, 5B and 3A, 3B) belong to subfamily I; EcPYL3/6/5/8 belong to subfamily II (four PYLs distributed among chromosomes 2A and 2B), while EcPYL4/7/12/14 belong to subfamily III (four PYLs distributed among chromosomes 2A, 2B and 5A,5B). A comparative evolutionary tree (Figure 5) shows that the PYL genes of rice (13 OsPYL), sorghum (8 SbPYL), finger millet (9 SiPYL) and foxtail millet (14 EcPYL) are grouped together into subfamilies I, II, and III. The genes EcPYL11/13 in subfamily I are closely related to one another and distantly related to OsPYL5, suggesting that these genes might be involved in signaling pathways that help plants resist the rice blast fungus (Yadav et al., 2020). Along with this, EcPYL10 may also play a similar role because it is closely related to OsPYL6 and has the same inhibitory phosphatase activity (Kim et al., 2014). EcPYL9 is closely related to SiPYL7/4 and also has some evolutionary similarities with EcPYL10-B from finger millet and OsPYL6 from rice. These evolutionary similarities are made up of conserved amino acid residues. EcPYL3/6 belong to subfamily II of the evolutionary tree and are closely related to foxtail millet SiPYL2 as well as to sorghum SbPYL2. EcPYL4/7 are distantly related to rice OsPYL10, suggesting that EcPYL4/7 could inhibit phosphatase activity without ABA (He et al., 2014). In addition, the OsPYL7/8/9/13 appear in different subclades (Figure 5) as a result of mutations in conserved amino acid residues of the latch loop (HRL), which are replaced by HML amino acid residues (Figure 4) in the same evolutionary tree.

To determine the degree of selection pressure during the divergence of PYL genes, we evaluated the (Ka) versus (Ks) analysis for the paralogous gene pairs of finger miller PYL receptor protein. Based on the Ka versus Ks calculation tool, seven pairs of paralogous genes with their position on the chromosome were identified for finger millet PYLs proteins, as shown in Table 4 and Figure 6. The selective pressure on genes due to which encoded protein shows functional alteration can be predicted by nucleotide substitution rate, i.e., the ratio of Ka (non-synonymous) versus Ks (synonymous) (Table 4). All gene pairs with Ka/Ks < 1 indicate the possibility of synonymous substitution and stabilizing selection during evolution. Moreover, the evolutionary time study of these paralogous genes revealed that they evolved approximately 3.1 million years ago.

CREs in the promoter region are known to influence transcriptional regulation greatly. Across fourteen EcPYL genes, 37 putative CREs were identified and mapped on the promoter region of EcPYL genes (Figure 7a; Supplementary Table S2). Several cis-elements were found to have different functions in finger millet. These included role in hormonal signaling (ABRE, TGACG/CGTCA-motif, P-box, GARE-motif, AuxRR-core TGA-element, and TCA-element), developmental processes (RY-element and circadian), light-responsiveness (G-box, C-box, Box 4, ACE, GATA motif MRE and Sp1) and various stresses (TC-rich repeats, LTR, MBS) in finger millet (Figure 7b). In addition to transcription-related CRE, a lot of the ABREs (abscisic acid-responsive element) and MeJA (methyl jasmonate/TGACG-motif/CGTCA-motif), which are linked to plant defense systems, were found. Furthermore, the fact that several EcPYL genes have GARE (gibberellic acid-responsive elements) and ARE (auxin-responsive elements) in their promoter region suggests that hormones may interact at the level of ABA receptor expression. Among stress-responsive factors, MBS (MYB binding site involved in drought-inducibility) and LTR (low-temperature responsive elements) are commonly observed in finger millet. Further, one additional CRE has been observed to be ubiquitously present in all the PYL genes of finger millet, called CCAAT. This five nucleotide base motif is associated with controlling flowering in plants (Wenkel et al., 2006) and contains a binding region for Nuclear Factor-Y transcription factor (NF-Y), which has been reported to regulate several biotic and abiotic stress factors (Rani et al., 2023; Rani et al., 2024a; Rani et al., 2024c). Hence, the presence of different CREs (8% stress responsive, 25% light responsive, 29% growth and development related and 38% hormone responsive) in the promoter region of finger millet provides an insight into its putative role in plant development and stress tolerance (Figure 7c).

The PPI results reveal a total of 44 nodes and 289 edges, and the MCODE of Cytoscope classifies them into three clusters. There are four EcPYL genes in cluster 3, and they interact with nine Arabidopsis PYL genes and 35 other genes performing the similar biological functions (Figure 8). In most cases, EcPYLs interact with PP2C, SnRK2 (the core ABA signaling complex), and the MYB transcription factor, which are vital components of the ABA-auxin signaling complex (Figure 8). Based on functional annotation, a group of EcPYLs proteins was discovered to play a role in different biological functions (BF), molecular functions (MF), cellular component (CC) (Figure 9).

Gene ontology and enrichment analysis reveals that EcPYL genes are enriched in plasma membrane (GO: 0005886), nucleus (GO: 0005634), protein phosphatase inhibitor complex (GO: 0062049) in terms of cellular components. Molecular function is enriched in abscisic acid binding (GO: 0010427), protein phosphatase inhibitor activity (GO: 0004864) and signaling receptor activity (GO: 0004864). Biological processes revealed that EcPYL genes are involved in regulating protein serine/threonine phosphatase activity (GO: 0080163) and the abscisic acid-activated signaling pathway (GO: 0009738). Besides these functions EcPYL genes also involved in two KEEG pathways viz. mitogen-activated protein kinase (MAPK) signaling pathway (ath04016) and plant hormone signal transduction pathway (ath04075).

In order to determine the miRNA targets and their regulation post-transcriptionally, EcPYL-miRNA interaction analysis was conducted, which revealed only three unique miRNAs targeting specific members of a particular subfamily of PYL genes in finger millet (Figure 10). Eco-miRN5585 targets EcPYL9-3A and EcPYL10-3B genes of subfamily I, while Eco-miRN34a targets EcPYL3-2A and EcPYL6-2B of subfamily-II and Eco-miRN529a targets only one gene EcPYL4-2A of subfamily III (Supplementary Table S3) through cleavage activity. Inhibition through cleavage activity appears to be the predominant mechanism by which most of the miRNAs predicted to target EcPYL members and regulating their expression post-transcriptionally either by positive or negative mechanism. However, miR529a were found to be involved in modulating panicle architecture in plants; however, miR5585 and miR34a were reported to play a crucial role as tumour suppressor cells (Farooqi et al., 2017). Thus, controlling EcPYL genes through sequence-specific interaction mediated by miRNAs could be very important for plants’ ability to respond to growth and pathogen infection.

By combining scatter plot visualization (Figure 11b), hierarchical clustering (Figure 11a), and numerical probability assessments (Table 5), the expression probabilities of these genes were analyzed under both the stress conditions. Scatter plot analysis (Figure 11b) identified distinct clusters of EcPYL genes under salt and drought conditions, each exhibiting varying degrees of expression. EcPYL14-5B, EcPYL12-5A, EcPYL11-5A and EcPYL7-2B were most likely to be expressed during drought stress, which suggests they may be involved in mechanisms that help plants deals with drought. Conversely, EcPYL4-2A and EcPYL7-2B displayed the highest expression probabilities under salt stress (≥0.72), indicating their potential role in salinity adaptation. The hierarchical clustering heatmap (Figure 11b), which grouped genes with similar expression patterns, further supported these observations. Drought stress strongly linked the groups of EcPYL14-5B, EcPYL12-5A, and EcPYL11-5A. On the other hand, EcPYL4-2A and EcPYL7-2B grouped separately, indicating their role in salinity stress. Further, EcPYL9-3A and EcPYL10-3B showed a strong preference for drought stress, while EcPYL6-2B exhibited a lower expression probability under both conditions. Table 5 shows that these trends were supported by the probability analysis. EcPYL14-5B had the highest expression probability (0.839) during drought stress, followed by EcPYL12-5A (0.818) and EcPYL11-5A (0.811). The EcPYL4-2A (0.749) and EcPYL7-2B (0.721) were also among the most highly expressed genes under salt stress. Although some genes displayed moderate expression probabilities under both stress conditions, they did not meet the threshold for high likelihood (≥0.70). For example, EcPYL5-2A showed a fairly balanced expression pattern (0.604 when exposed to salt and 0.618 when exposed to drought), which suggests it might play a small but important role in adapting to abiotic stress in general. However, genes such as EcPYL9-3A (0.13 under salt, 0.781 under drought) and EcPYL10-3B (0.164 under salt, 0.819 under drought) showed a distinct preference for drought-related expressions. Overall, our results show that EcPYL14-5B, EcPYL12-5A, EcPYL11-5A and EcPYL7-2B are important candidate genes and these genes are likely to be involved in regulating drought stress. While EcPYL4-2A and EcPYL7-2B are likely to regulate salinity stress.

Through ML based algorithm, a total of five PYL genes (EcPYL4-2A, EcPYL7-2B, EcPYL11-5A, EcPYL12-5A, and EcPYL14-5B) that respond to drought and salinity stresses were selected (Figures 11a,b), and their expression pattern was analyzed in shoot and root tissues of finger millet under drought and salinity stress (Figures 12a,b). Notably, all genes exhibited expression in both tissues under both the stress conditions (Figures 12a,b,d,f); however, their expression patterns varied between tissues and conditions (Figures 12c,e). For instance, compared to EcPYL4-2A, only EcPYL12-5A showed a 9.69-fold increase in shoot tissue under drought stress conditions (Figure 12c). When we compared the expression pattern of EcPYL12-5A in shoot tissue between salinity and drought, it was observed that drought stress increased its expression level up to 2.1-fold compared to salinity stress (Fig. d–f). The expression level of EcPYL11-5A was higher in the shoot after EcPYL12-5A under drought. Under drought stress, EcPYL7-2B increased more than 2-fold in shoot tissue compared to EcPYL4-2A and EcPYL14-5B. Three genes (EcPYL4-2A, EcPYL11-5A, and EcPYL14-5B) were comparatively less active in the shoot tissue of finger millet under salt stress (Figure 12e). However, of these, EcPYL4-2A increased its expression level by 0.76-fold higher in shoot tissue under salinity stress than drought (Figure 12f). Two genes (EcPYL7-2B and EcPYL11-5A) showed their expression level higher (>3.5-fold) in shoot tissues under salinity stress than the other three genes. In conclusion, three genes (EcPYL7-2B, EcPYL11-5A, and EcPYL12-5A) and two genes (EcPYL7-2B and EcPYL12-5A) were found to be highly expressed in shoot tissue of finger millet under drought and salinity stresses, respectively.

In the root, EcPYL2-7B exhibited highest expression among the five PYLs under both stresses. Notably, expression level of EcPYL12-7B increased by more than 15-fold compared to EcPYL11-5A, EcPYL12-5A, and EcPYL14-5B under salinity stress. Similarly, the same gene demonstrated 1.96-fold higher expression in the root under salinity stress compared to EcPYL4-2A. However, EcPYL4-2A expression was up to 9.3 times higher than EcPYL11-5A, 13.5 times higher than EcPYL12-5A, and 17.6 times higher than EcPYL14-5B in the root under salinity stress. Interestingly, the expression levels of the same three genes—EcPYL11-5A, EcPYL12-5A, and EcPYL14-5B was enhanced by 6.15, 12.26, and 4.6 times in root tissues under drought stress. Moreover, EcPYL7-2B alone had highest expression in the root under drought stress among the five genes. For instance, the expression level of this gene was almost 1.5 times higher than that of the other genes. Overall, EcPYL4-2A and EcPYL7-2B showed higher expression in the root tissue, under both the stress conditions suggesting that EcPYL7-2B is the most responsive gene under drought and salinity stress, particularly in root tissues, while EcPYL12-5A and EcPYL11-5A play significant roles in shoot tissues under drought stress. This differential expression highlights the potential role of these genes in stress adaptation and provides insights for future functional studies on drought and salinity tolerance mechanisms in finger millet.

Protein sequences, EcPYL13-5B, EcPYL6-2B, EcPYL8-2B, and EcPYL14-5B representing each clade of the phylogenetic tree, were used for homology modeling for structural prediction (Figure 5). The widely preferred method for modeling protein structures is to compare a sequence of our interest, i.e., the target sequence (unknown structure), with a template (known structure) and look for structural similarities. The SWISS-MODEL server (https://swissmodel.expasy.org/) was used in the present study for homology modeling of EcPYL6-2B, EcPYL8-2B, EcPYL13-5B, and EcPYL14-5B (Figure 13) against best-matched template 4oic, 5ujv and 5gwo, respectively, based on the high percent of identity, as shown in Table 5. The model generated by these templates was further validated by the SAVES server by analyzing the number of amino acid residues in allowed or disallowed areas through PROCHECK, accuracy of non-bonded contact by ERRAT (Table 6). All the models had ERRAT scores greater than the acceptable value of 80%. After the accuracy measurement, the modeled structure was subjected to the degree of similarity search with the crystal structure of (PDB ID 4oic, 5ujv and 5gwo) by using SuperPose V.1. The RMSD (Root Mean Square Deviation) and NRES (Number of Overlapped Residue) were analyzed for the superposed structure as shown in Table 4. By evaluating the result of the superposition, it can be concluded that the quality of the model is good, with the least deviation of residues from the template. Further, validation of the predicted model was performed with ResProx resolution criteria (Good: 0–1.5, Middle: 1.5–2.5, Bad: >2.5), indicating the models to be of good resolution. Table 6 summarizes the detailed analysis of homology modeling using SWISS-MODEL, structural assessment, SAVES server validation, and quality and resolution estimation using SuperPOSE v.1 and ResProx. The 2-D structural analysis performed by the PDBSum server revealed that finger millet EcPYLs genes (EcPYL6-2B, EcPYL8-2B, EcPYL13-5B, and EcPYL14-5B) contains same number of strand but having different number of helices as 5:4:5:4 respectively (Table 6). Figure 13 depicts each model’s conserved residues present in CL2 (GATE loop) and CL3 (LATCH loop), specific for ABA binding.