To study the transcriptome profiles, three biological replicates were collected using three separate S. guaranitica plants. Each replicate included a pooled sample of both young and mature leaves. The plants utilized in this study were two years old. Moreover, three biological replicates were gathered from mature leaves, tender leaves, flowering parts, flower buds, stems, and roots for the qRT-PCR assays. The samples were swiftly stored in liquid nitrogen and thereafter held onto -80°C till needed (Mehmood et al., 2021). TRIzol™ Reagent (Invitrogen, CA, US) was employed to obtain the total RNA from different samples as per the guidelines provided by the manufacturer. After treating the extracted RNA from various samples with DnaseI (Takara, China), its overall quality was assessed by subjecting it to electrophoresis on a 1.25% agarose-formaldehyde gel, followed by visualization with ethidium bromide. Additionally, the NanoDrop™ 2000/2000c Spectrophotometers (MA, USA) was utilized to calculate the RNA quality and concentration from different samples. Ten µg of RNA was used for cDNA synthesis via the reverse transcription kit (M-MLV, China) (Huang et al., 2012; Rastogi et al., 2014; Zhu et al., 2019; Santa Cruz et al., 2021). High-quality RNAs extracted from different samples were utilized to construct cDNA libraries (Ali et al., 2017, 2018). Sequencing was achieved on the high-quality libraries using an Illumina HiSeq 2500 platform, generating paired-end reads. Clean reads were acquired by filtering out adapters, poly-N sequences, and low-quality reads. Subsequently, the assembly was performed by the Trinity platform https://github.com/trinityrnaseq/trinityrnaseq/wiki with the parameters “min_kmer_cov set to 2”. Subsequently, the values of Q20, Q30, GC content, and sequence duplication level were estimated (Yang et al., 2017).

To examine the putative transcription levels of SgGPI, SgT6PS, and SgSUS across various tissues, the A. thaliana eFPbrowsers (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) was employed. Moreover, the predicted subcellular localizations of their orthologs from A. thaliana were retrieved. Subsequently, the image envisioned their cellular localizations was built, and then the putative domains were estimated through the InterPro database (https://www.ebi.ac.uk/interpro/) (El-ramah et al., 2022; Elsherbeny et al., 2022; Makhadmeh et al., 2022a, 2022b; Abdelhameed et al., 2024b).

Towards examining the activity levels of glycolysis/gluconeogenesis, starch, and sucrose biosynthesis genes in S. guaranitica at different tissues, twenty candidate genes were chosen. The expression profiles for these selected genes were compared within various tissue samples to disclose their ‘transcriptional control’, offering insights into the epistatic relationship regarding mRNA copies, and the products and the end-products. The expression profiles of our chosen candidates: SgGPI, SgT6PS, SgSUS, SgPFK9, SgALDH, SgALDO, SgPYK, SgFBP, SgACS, SgPCKA, SgGlGA, SgGlGC, SgBMY, SgGBE1, SgAGL, SgBGL, SgHK, SgPYG, SgUGDH and SgINV, across different tissues were investigated.

The SgGPI, SgT6PS, and SgSUS full-length cDNAs were amplified using gene-specific primers designed from the Illuimina sequencing data of S. guaranitica leaves ( Supplementary Table S1 ). The initial PCR was conducted with the short primers, KOD-Plus-DNA polymerase (Toyobo, Japan), and leaf cDNA using the program: 96°C for 4 min, 98°C for 12 s, 60°C for 4 s, 68°C for 2.5 min, and 34 cycles followed by 68°C for 15 min. Regarding the following PCR, the products from the initial PCR were used as templates with the long primers under the same PCR conditions. The products were then cleaned and then transferred to the Gateway entry vector (pDONR221), then subsequently sub-cloned into the destination vector (pB2GW7). The vectors were incorporated into A. thaliana flowers via Agrobacterium tumefaciens strain GV3101 by electroporation. The cloning steps were verified using Sanger sequencing (Rehman et al., 2018; Ali et al., 2022a, 2022c, 2022b).

The genes SgGPI, SgT6PS and SgSUS were chosen to be characterized and expressed in A. thaliana utilizing the Agrobacterium-mediated floral dip technique. The transformation was performed using A. tumefaciens GV3101 harboring pB2GW7-SgGPI, pB2GW7-SgT6PS, and pB2GW7-SgSUS plasmids driven by the 35S promoters. Using the procedure outlined by (Ali et al., 2018, 2022a, 2022c, 2022b; Ali, 2023). Briefly, the A. thaliana seeds were germinated in advance and the plants were made ready for transformation after two months. The secondary inflorescences were immersed in a solution containing Agrobacterium carrying the pB2GW7-vector, specifically targeting the gynoecium of the flower. The plants were cultivated until the siliques reached a brown and dried state. Subsequently, the seeds were collected, cultivated again, and subjected to BASTA treatment – a herbicide containing glufosinate-ammonium – to select the desired transgenic seedlings carrying the resistance gene against BASTA. Moreover, the presence of target genes in positive transgenic lines was verified through semi-quantitative RT-PCR (semi-qRT-PCR). The physiological and biochemical parameters of different transgenic lines were evaluated. A total of twelve 45-day-old plants, including putative transgenic and wild-type plants, were chosen for the purpose of harvesting mature leaves. These leaves were then subjected to semi-qRT-PCR for measuring the activity levels of the Salvia-derived genes (SgGPI, SgT6PS and SgSUS) into transgenic A. thaliana plants.

Soluble sugars, including sucrose, glucose, and fructose, were analyzed as described by (Sonnewald et al., 1991; Ortiz-Marchena et al., 2014). Briefly, the quantification involved extraction from both wild-type and genetically modified A. thaliana plant leaves using 80% ethanol in 10 mM HEPES-KOH (pH 7.7) at 80°C for 2 hours. The supernatant was utilized to measure glucose, fructose, and sucrose concentrations through the sequential addition of specific enzymes—5 units each of glucose-6-phosphate dehydrogenase and hexokinase, 2 units of glucose-6-phosphate isomerase, and 20 units of invertase—followed by the monitoring of NAD⁺ reduction at 340 nm absorbance at various intervals.

Each parameter was tested with three biological replicates. Furthermore, the quantities of chlorophyll a, b, and total (a+b), were measured following (El-Mahdy et al., 2024). Concisely, about 20-30 mg of fresh leaf samples were weighted from each treatment, then each sample was transferred to a centrifuge tube with 4 mL dimethylformamide (DMF) and maintained away from light to preserve chlorophyll integrity. The level of chlorophyll a, b, and total (a+b) in the extracts were accomplished following (El-Mahdy et al., 2024). The absorbance readings of the chlorophyll samples were taken at 664 and 647 nm through JENWAY 6505 UV/Vis spectrophotometer.

The results were analyzed using SPSS (IBM Corp., 2020), incorporating three biological replicates. Significance levels were indicated as (*) for P-values less than 0.05, (**) for P < 0.01, (***) for P < 0.001, and (****) for P < 0.0001, demonstrating the highest degree of significance.

Lately, the Illumina sequencing technology has emerged as a robust technique for genome analysis and discovery in non-model plants. In this research, transcriptome sequences were obtained from pooled leaves of Salvia guaranitica using the Illumina HiSeq 2500 platform. This process yielded approximately 8.2 Gb of raw data from the S. guaranitica leaves. Post-filtering and removal of adapter sequences, 38,521,658 reads (38.52 million) were obtained, containing 210,521,170 high-quality nucleotide bases. The quality metrics indicated that 94.95% of the bases had a quality score of Q20, 90.54% had a quality score of Q30, and the GC content was 48.58%. Our findings were consistent with previously obtained results of many other studies which utilized transcriptome tools to detect and identify key genes associated with various biomolecules in several species, including Vicia sativa, Dendrobium nobile, S. officinalis, Ocimum sanctum, Ocimum basilicum, and Cunninghamia lanceolata (Rastogi et al., 2014; Ali et al., 2017; Zhu et al., 2019; Huang et al., 2021; Zhang et al., 2023).

With respect to de novo assembly and transcriptome study, high-quality and pure reads were assembled using Trinity program (Ali et al., 2017, 2018; Mehmood et al., 2021). The assembly results yielded 200,298 RNA variants, the N50 length was 1,850 bp, the N90 length was 520 bp, and the mean length was 1,125 bp. Additionally, 75,100 unigenes were identified, with N50 equals to 1,524 bp, N90 is 320 bp, and a mean length of 965 bp. The assembled lengths were 200 to approximately 2,000 bp. The majority of transcripts (83,387 transcripts, 41.652%) were between 200 and 500 bp, then 48,252 transcripts (24.102%) between 1,000 and 2,000 bp, and 41,197 transcripts (20.578%) between 500 and 1,000 bp. Conversely, the fewest transcripts (27,462 transcripts, 13.731%) were longer than 2,000 bp. Similarly, the lengths of the unigene assemblies ranged from 200 to over 2,000 bp, with the majority (39,145 unigenes, 52.125%) between 200 and 500 bp, followed by 16,866 unigenes (22.458%) between 500 and 1,000 bp, and 13,837 unigenes (18.425%) between 1,000 and 2,000 bp. The fewest unigenes (5,251 unigenes, 6.992%) were longer than 2,000 bp. The length profile of the transcripts and unigenes is presented in Supplementary Table S2 . Our findings correspond with results observed in other species, such as, Boehmeria nivea, Curcuma longa, M. sativa, S. officinalis, Centella asiatica, and Apium graveolens, where transcript and unigene lengths predominantly fell within 75 to 500 bp (Srividya et al., 2015; Zhu et al., 2019; Xiao et al., 2024).

Nearly 75,100 unigenes served as search queries across NR (http://www.ncbi.nlm.nih.gov/), NT, KO, Swiss-Prot (http://www.ebi.ac.uk/uniprot/), PFAM, GO (http://www.geneontology.org/), and KEGG (https://www.kegg.jp/kegg/kegg2.html) Supplementary Table S3 . The BLAST2GO program facilitated the sorting and ranking of the functions of all annotated unigenes, with 29,695 unigenes (39.54% of the assembled unigenes) assigned with at least one GO term. Inferred from the homology results, the unigenes were sorted into 56 functional groups across three main categories: 63,008 assigned to biological processes (BP), 48,517 to cellular components (CC), and 20,546 to molecular functions (MF). In the CC section, the most enriched GO terms were “cell part” (9,357) and “cell” (9,281). In the MF section, “binding” (14,258) and “catalytic activity” (12,754) were predominant. Within the BP section, “metabolic process” (15,326) and “cellular process” (14,820) were highly enriched ( Figure 1 ). These findings are consistent with previous studies on the RNA profiles of S. miltiorrhiza, S. officinalis, O. sanctum, and O. basilicum, that also reported high percentages of these GO terms (Ali et al., 2017, 2018; Mehmood et al., 2021).

Most transcripts belonged to the Metabolism section (5,081), subsequently, Genetic Information Processing (2,187), Organismal Systems (1,868), Cellular Processes (1,222), and Environmental Information Processing (976). From the data analysis, 1,250 transcripts were related to carbohydrate metabolism, with 410 putatively linked to starch, sucrose, and glycolysis metabolism. This included 175 genes involved in glycolysis/gluconeogenesis biosynthesis and 235 genes related to starch and sucrose biosynthesis. The levels of gene expressions were estimated through the UniProt database for annotation against the transcriptome libraries. Normalization and calculation were performed using the DESeq package (1.10.1), represented as fragments per kilobase of transcripts per million mapped fragments (FPKM) as shown in Figure 3 and presented in Supplementary Tables S4 and S5 .

To identify the physiological roles of the SgGPI, SgT6PS, and SgSUS genes, we investigated their accumulation patterns across forty-seven tissues. The analysis was facilitated by the high similarity between SgGPI, SgT6PS, SgSUS, and the AT4G24620, AT1G06410, and AT5G37180 from A. thaliana, respectively. The relative expressions of SgGPI, SgT6PS, and SgSUS were observed in various tested tissues ( Figures 4A–C ). These results align with results reported (Ali et al., 2017, 2018, 2022a) who used similar tools in the BAR database to predict the putative expression patterns for several genes, such as, SoNEOD, SoHUMS, SoFLDH, SoLINS2, GmTPS21, SgTPSV, SgGERIS, and SgFARD, from S. officinalis, Glycine max, and S. guaranitica, respectively, showing heightened manifestation in leaves, roots, and seeds. Additionally, the localizations of SgGPI, SgT6PS, and SgSUS revealed that they are present in various cell organelles. For example, the SgGPI is predominantly located in the plastid, cytosol, extracellular space, and mitochondria. However, SgT6PS is mainly found in the mitochondrion, cytosol, plasma membrane, Golgi apparatus, nucleus, peroxisome, and vacuole. Meanwhile, SgSUS is primarily present in the cytosol, mitochondrion, plastid, nucleus, and plasma membrane ( Figures 4D–F ).

To elucidate the differential gene expression of glycolysis/gluconeogenesis, starch and sucrose biosynthesis genes ( Supplementary Table S6 ; Figure 5 ) across different treatments, we employed the Bio-Rad Nucleic Acid Amplification and Detection systems (CFX384) with SYBR Green fluorescence and ROX as a passive reference dye (Newbio Industry, China), following protocols outlined in previous studies (Rastogi et al., 2014; Zhu et al., 2019; Huang et al., 2021; Santa Cruz et al., 2021). Primers were designed using the IDTdna tool (https://eu.idtdna.com/scitools/Applications/RealTimePCR/), as listed in Supplementary Tables S6 and S2 . The cycle threshold (CT) of the target genes was calculated using SgB-ACTIN as a reference gene to normalize gene expression levels. Relative gene expression levels were then determined through delta-delta C_t method ( Supplementary Tables S6 ; Figure 5 ). The relative expression levels of several genes, including SgGPI, SgT6PS, SgSUS, SgPFK9, SgALDH, SgALDO, SgPYK, SgFBP, SgACS, SgPCKA, SgGlGA, SgGlGC, SgBMY, SgGBE1, SgAGL, SgBGL, SgHK, SgPYG, SgUGDH, and SgINV, were detected. For instance, SgBMY, SgGlGA, SgSUS, SgALDO, SgGlGC, SgPFK9, SgGPI, and SgAGL genes exhibited the top expression levels in mature leaves. In contrast, SgUGDH and SgACS genes showed the peak expression levels in immature leaves. Additionally, the SgT6PS gene demonstrated the maximum expression levels in flowers, whereas the SgALDH gene had the highest expression levels in flower buds. Moreover, SgBGL, SgINV, SgHK, SgPYK, and SgFBP genes demonstrated the peak levels in stems ( Figure 5 ). Finally, SgPCKA, SgPYG, and SgGlGC genes were found with the greatest levels in roots. Interestingly, the qPCR analyses of these genes were in line with their expressions as measured by Illumina HiSeq 2500 (Hussain et al., 2017; Darwish et al., 2022; Ali et al., 2023b).

To further investigate the biological function of SgGPI, SgT6PS, and SgSUS, we transferred the overexpression vector pB2GW7-SgGPI, pB2GW7-SgT6PS and pB2GW7-SgSUS directed by 35S promoter into A. thaliana. We obtained twelve homozygous transgenic lines from each transformed gene and the successful transformed lines was confirmed by PCR and semi-qRT-PCR analysis ( Figures 6A, B ). Compared to the WT, a significant difference in leaf development and plant growth were observed in all lines overexpressing these genes ( Figure 6A ). Interestingly, the flowering process of transgenic A. thaliana was accelerated earlier than that of W.T plants ( Figure 6A ).

Sugars serve as the primary carbon source for synthesizing various pathways involved in both primary and secondary metabolites. The levels of soluble sugars were assessed in transgenic A. thaliana plants overexpressing SgGPI, SgT6PS, and SgSUS, revealing significant increases compared to the wild-type (WT). Furthermore, starch, total sugar, glucose, and fructose contents were quantified in these transgenic plants. Results indicated higher levels of these components in all transgenic lines compared to WT ( Figure 7 ). Specifically, the average starch content increased by approximately 5.033-fold (10.033/5.0) for A. thaliana plants overexpressing SgGPI, 6.566-fold (11.866/5.3) for A. thaliana plants overexpressing SgT6PS, and 7.133-fold (11.700/4.6) for A. thaliana plants overexpressing SgSUS compared to WT. Similarly, average sugar contents visually increased by 2.2-fold (5.2/3) for A. thaliana plants overexpressing SgGPI, 3.766-fold (6.566/2.8) for A. thaliana plants overexpressing SgT6PS, and 6.233-fold (8.733/2.5) for A. thaliana plants overexpressing SgSUS. Glucose levels showed respective increases of 0.87-fold (1.97/1.1), 1.2-fold (2.4/1.2), and 1.04-fold (2.25/1.21), while fructose levels increased by 1.0-fold (2.2/1.2), 0.94-fold (2.24/1.3), and 1.52-fold (2.733/1.25) for A. thaliana plants overexpressing SgGPI, A. thaliana plants overexpressing SgT6PS, and A. thaliana plants overexpressing SgSUS compared to WT.

Additionally, the total chlorophyll, chlorophyll a, and chlorophyll b contents were evaluated in these transgenic plants. Results indicated higher levels of these chlorophyll components in all transgenic lines compared to WT ( Figure 7 ). For instance, the average total chlorophyll content increased by 0.017-fold (0.0508/0.033) for A. thaliana plants overexpressing SgGPI, 0.030-fold (0.0640/0.034) for A. thaliana plants overexpressing SgT6PS, and 0.033-fold (0.067/0.034) for A. thaliana plants overexpressing SgSUS compared to WT. Similarly, average chlorophyll a level increased by 0.009-fold (0.031/0.022) for A. thaliana plants overexpressing SgGPI, 0.0131-fold (0.0351/0.022) for A. thaliana plants overexpressing SgT6PS, and 0.016-fold (0.039/0.023) for A. thaliana plants overexpressing SgSUS, while chlorophyll b levels increased by 0.068-fold (0.0795/0.011) for A. thaliana plants overexpressing SgGPI, 0.0747-fold (0.0867/0.012) for A. thaliana plants overexpressing SgT6PS, and 0.071-fold (0.082/0.011) for A. thaliana plants overexpressing SgSUS compared to WT.