Introduction

Stevia rebaudiana Bertoni recently became an intensively studied species due to its unique ability to produce and accumulate steviol glycosides (SvGls). These secondary metabolites are 250–400 times sweeter than sucrose and are considered to be metabolically inert. For this reason, SvGls could be safely used as a sugar substitute by people with diabetes and obesity (Cramer and Ikan 1986; Ghosh et al. 2008; Ceunen and Geuns 2013; Gupta et al. 2013, 2016; Gasmalla et al. 2014).

Biosynthesis and accumulation of SvGls in S. rebaudiana plants is also an interesting scientific subject in the context of their engagement in the processes of plant adaptation to the environmental conditions. One of the functions of plant secondary metabolites is their participation in plant response to biotic and abiotic stress factors (Bohnert 1995; Srivastava and Srivastava 2007; Selmar and Kleinwächter 2013; Ashraf et al. 2018). It is commonly known that adverse environmental stimuli lead to the changes in the plant metabolism and the induction of oxidative stress due to overproduction of reactive oxygen species (ROS) (Apel and Hirt 2004). Some secondary metabolites should be mentioned among the main antioxidants (Edreva et al. 2008) that, due to their specific chemical structure, function as electron donors or acceptors protecting against oxidative stress. It has been shown that some transcription factors associated with the oxidative stress reaction in plants participate in controlling the synthesis of secondary metabolites (Hong et al. 2013). However, so far, only limited information has been shared on the involvement of some secondary metabolites in overcoming oxidative stress in plant cells (Selmar and Kleinwächter 2013). SvGls are ent-kaurenoid diterpene glycosides synthesized through the methylerythritol phosphate pathway (MEP) from the isopentenyl diphosphate (IPP) skeleton, the fundamental unit of all isoprenoids (Totté et al. 2000). These, in turn, are known to complement the functional roles of antioxidant enzymes under recurrent combinations of excess temperature, light, and drought, reducing oxidative stress by quenching ROS/reactive nitrogen species (RNS) (Tattini et al. 2015).

The biosynthetic pathway of SvGls is a multiphase process sharing some steps with the biosynthetic pathway of gibberellins (Helliwell et al. 1999). Divergence between these two pathways occurs after the stage of hydroxylation of kaurenoic acid to produce aglycone steviol, catalyzed by kaurenoic acid 13-hydroxylase (KAH). At this step, steviol is glycosylated by a set of uridine diphosphate (UDP)-dependent glucosyltransferases (UGT) (Brandle and Telmer 2007; Ceunen and Geuns 2013). The general sequence of modifications in the steviol glucoside pathway has been already determined and particular enzymes and genes have been identified (Brandle et al. 2002; Richman et al. 2005; Lee et al. 2019). This provides a good insight into the glycosylation process and a possible template for engineering of steviol glycosides biosynthesis (Lee et al. 2019).

Due to the many beneficial properties of stevia plants that can affect human health, as well as a growing interest in the results of studies of the engagement of oxidative stress in the regulation of secondary metabolites concentration in plants, the easy access and ability to obtain large quantities of SvGls in a short period of time attract a lot of attention (Mathur et al. 2017). Propagation of stevia is usually done by stem cuttings, which root easily, but require high labor inputs (Yadav et al. 2011). An alternative way of breeding stevia plants is micropropagation using in vitro techniques initiated by Yang and Chang 1979 and since then, successfully used for obtaining a high number of stevia plants on an industrial scale. The usefulness of this method is further enhanced by the fact that stevia seeds are characterized by a very low germination strength (Carneiro et al. 1997). Moreover, it has been shown that the level of SvGls content can be manipulated by application of drought or salt stress during in vitro micropropagation (Hajihashemi and Ehsanpour 2014; Zeng et al. 2013; Fallah et al. 2017; Ghaheri et al. 2019), some combination of phytohormones (Bondarev et al. 2003a; Luwańska et al. 2015; Pazuki et al. 2019), or signaling molecules, such as glutamine (Esmaeili et al. 2018). However, using the whole plant for SvGls production introduces also some disadvantages, given that SvGls concentration changes during the plant’s growth and development, as well as SvGls content varies across stevia organs, with the highest concentration found in leaves and the lowest in roots (Bondarev et al. 2003b; Kang and Lee 1981; Bondarev et al. 2003b; Yang et al. 2015; Libik-Konieczny et al. 2018).

In vitro plant cell, tissue, and organ cultures offer a possibility of producing plant material containing a high level of stable selected secondary metabolites (Ramakrishna and Ravishankar 2011; Hussain et al. 2012). In vitro culture techniques for callus, cell suspension, shoots, or adventitious roots formation have been already developed for S. rebaudiana (Bondarev et al. 1997, 2001, 2003a, 2003b; Lee et al. 1982; Hsing et al. 1983; Swanson et al. 1992; Sivaram and Mukundan 2003; Jadeja et al. 2005; Ladygin et al. 2008; Janarthanam et al. 2010; Taware et al. 2010; Reis et al. 2017; Mathur and Shekhawat 2013; Gupta et al. 2014; Khalil et al. 2014; Gupta et al. 2015; Michalec-Warzecha et al. 2016; Pandey et al. 2016). However, the data on the content of steviol glycoside in cultured samples differ in literature, with some studies reporting contradictory results. Namely, some authors did not discover any presence of steviol glycosides in callus or cell suspension cultures, while others reported that the concentration of SvGls in the callus cultures might even be twice as high as in stevia leaves (Luwańska et al. 2015; Pandey et al. 2016). Similarly, disparate data concerning biosynthesis of SvGls by hairy roots induced after transformation of S. rebaudiana with Agrobacterium strains were published by two scientific groups (Yamazaki and Flores 1991; Pandey et al. 2016). In the study concerning the first attempt to produce hairy roots by stevia plants, SvGls were not detected in the transformed tissue (Yamazaki and Flores 1991). On the contrary, in the studies performed by Pandey et al. 2016, the presence of stevioside was noted in both HR and in the culture medium. Because of these discrepancies, the technique for production of transgenic hairy root (HR) cultures as a source of secondary metabolites from S. rebaudiana requires better development and more scientific attention, especially that HR exhibit unique features such as rapid increase in biomass, intensified biosynthesis of many molecules, genetic stability, and the ability of unlimited growth in hormone-free media (Srivastava and Srivastava 2007). It was discovered that bacterial genes: rolA, rolB, rolC, and rolD from Ri plasmid of Agrobacterium rhizogenes, apart from being responsible for the expression of hairy roots phenotype (White et al. 1983), are also potential activators of secondary metabolism in transformed cells of plants from different families (Bulgakov 2008; Pistelli et al. 2010). RolB and rolC genes were found to be most interesting for manipulating the secondary metabolite biosynthesis. It was shown that the rolB gene significantly increased the secondary metabolites production in transformed plant cells, but it inhibited cell growth. RolC, on the other hand, was less important in the promotion of secondary metabolites synthesis but it had an ability to increase cell growth (Shkryl et al. 2008). These facts make HR culture an even more promising method not only for mass production of secondary metabolites in in vitro cultures but also as a primarily source of material for scientific studies on steviol glycosides biosynthesis via genetic manipulation of genes involved in this pathway. We investigated the influence of light and osmotic active compounds present in the medium on steviol glycosides biosynthesis and accumulation in hairy roots of S. rebaudiana. We hypothesized that transformation of stevia leaf explants to produce HR and an application of stress factors during HR growth would alter the biosynthetic pathway and the accumulation of steviol glycosides, leading to changes in the secondary metabolite profile. The aim of this study was also to obtain transformed callus tissue derived from HR cultures that could be used for plant regeneration and for further investigation of genetic manipulation on steviol glycosides biosynthesis genes.

Material and methods

Experimental material

The study was conducted on hairy roots obtained after transformation of S. rebaudiana Bertoni leaf explants with the wild-type hypervirulent Agrobacterium rhizogenes strain LBA 9402 (a gift from Dr. David Tepfer, INRA Versailles, France). All details concerning the transformation protocol were previously described by Michalec-Warzecha et al. (2016). The hairy roots formed at wounded sites of infected explants were individually separated as clones and transferred onto solid MS medium and kept in the dark at 26 °C. The culture medium was supplemented with cefotaxime 100 mg l−1 to eliminate A. rhizogenes. The explants were kept on the medium until a complete removal of the bacteria from the culture. One hairy root line that showed sufficient growth was chosen for further investigation. After 3 weeks of growth, the hairy roots multiplicated from one origin line were transferred individually into 250-ml Erlenmeyer flasks containing 30 ml of liquid MS medium containing 85 mM sucrose. Cultures were kept at 25 ± 2 C on a rotary shaker with constant agitation at 80 rpm under dark conditions, and were used in the experiment as control samples.

Manipulation of in vitro culture conditions

Some of the hairy roots liquid cultures were kept in the medium composed as described above in the conditions of continuous light provided by cool fluorescent light intensity 40 μm m−2 s−1 (Pandey et al. 2016), or they were grown in the dark but in the liquid MS media supplemented with sucrose in concentration 43 mM or 175 mM, or with NaCl in concentration 45 mM (0.2% NaCl according to Gupta et al. 2014), or polyethylene glycol (PEG) 6000 in concentration 1.0 mM (6% PEG according to Hajihashemi and Ehsanpour 2014). The osmotic potential of culture media was measured using a psychrometer HR 33T (WESCOR, Inc., Logan, USA) equipped with type C52 (WESCOR) leaf chamber and Metex M 3640D digital gauge (measurer, indicator). Paper discs, 0.5 cm in diameter, soaked with solutions of estimated media served as samples. They were placed in a chamber and left for 25 min. Measurements were conducted in the “dew point” mode. For data analysis, Metex graphic program was used; therefore, it was possible to monitor the changes. The values of the osmotic potential of different media used for the hairy root cultures are presented in Table 1.

Table 1 Osmotic potential values of different media used for hairy root culture. Fresh weight (FW) and dry weight (DW) of hairy roots grown for 0, 10, 20, 30, and 40 days in media with different osmotic potential and under light. Comparison of hairy root length and root hair length depending on the culture conditions, values within a column sharing the same letter are not statistically different (P ≤ 0.05, n = 6)
Full size table

HR were grown for 40 days and their fresh as well as dry weight was recorded at 0, 10, 20, 30, and 40 days of culture. The growth potential of cultured hairy roots was monitored by accumulation of fresh biomass, according to the equation: AFB = FFW–IFW, where AFB is the accumulated fresh biomass, FFW the final fresh weight, and IFW the initial fresh weight of the hairy roots used for inoculation. To determine the dry biomass, the hairy roots cultured for 10, 20, 30, and 40 days were filtered through Whatman filter paper and then placed for 2 days in an oven at 105 °C. After that time, the dry weight was determined. Each result was based on 3 replications, and the experiment was repeated three times.

Biochemical and molecular studies were performed on HR collected at the 30th day of culture when the constant fresh and dry weight of cultured HR were achieved.

Assessment of oxidative stress level

The level of oxidative stress was expressed as an amount of endogenous hydrogen peroxide in HR cultured in different conditions. Endogenous concentration of hydrogen peroxide was measured using the Amplex Red Hydrogen Peroxide Assay Kit (Molecular Probes). The extract was prepared from freshly collected HR (0.1 g) which were ground in 0.5 mL of the reaction buffer (provided in the kit). Then, the extract was centrifuged for 5 min at 14000 g and 50 μL of the supernatant was incubated with 50 μL of the working solution containing 100 mM Amplex Red reagent and 0.2 units mL−1 of horseradish peroxidase for 30 min at room temperature under the dark conditions. The H2O2 concentration was measured using spectrofluorometer equipped with the 96-well microplate reader Synergy 2 (Bioteck) under the excitation wavelength of 530 nm and fluorescence detection at 590 nm. The standard curve in the range 0.05−1 μM was made using different concentrations of H2O2 provided in the kit.

RT-qPCR analysis of transcript level of particular genes involved in steviol glycosides biosynthetic pathway

Gene transcript level was studied by reverse transcription–quantitative polymerase chain reaction (RT-qPCR). Total RNA from hairy roots was extracted using TRIZOL according to Chomczynski 1987. Extracted RNA was quantified by a Nanodrop spectrophotometer (Thermo Scientific), and only high purity samples (A260/280 > 2.00) were used for cDNA synthesis. RNA in the quantity of 500 ng RNA was used for the preparation of cDNA, and the synthesis was obtained with the use of iScript cDNA Synthesis Kit (BioRad, USA) according to the manufacturer’s instructions. Primers for all target genes (kaurenoic acid hydroxylase, UDPglucosylotransferase-74G1, UDPglucosylotransferase-76G1, UDPglucosylotransferase-85C2), and control gene (actin) were designed using the Blast tool from NCBI and Primer 3 (v. 0.4.0) software. The list of primers is presented in Suppl. 1. The reactions of qPCR were prepared using SsoFast EvaGreen Supermix (BioRad) according to the manufacturer’s instructions, and they were run as triplicates in an C1000 Thermal Cycler (BioRad). The quantity of targeted gene transcripts was normalized to actin.

UHPLC analysis of some steviol glycosides content in hairy roots cultures

Steviol glycosides concentration after extraction to water was analyzed by UHPLC (Agilent Infinity 1260) in HILIC (Hydrophilic Interaction Liquid Chromatography) mode with UV detection at 210 nm. For analysis of steviol glycosides (SGs), about exactly weighted 10-mg samples (balance sensitivity 0.1 mg) were extracted to 1 ml of deionized water (PureLab Option R, Elga, England). After 5 min of centrifugation (21,000×g, Universal 32, Hettich, Germany), the supernatant was collected and 100 μl aliquot was diluted with acetonitrile (1:3 v/v). Diluted samples were centrifuged (5 min, 21,000×g) and finally filtered through 0.2 μm membrane (Costar X, Cornig, USA) and used for UHPLC (Agilent Infinity 1260, Agilent, Wolbrum, Germany) analysis. SvGls were separated in HILIC (hydrophilic interaction liquid chromatography) mode with UV detection at 210 nm. Separation was achieved on BlueOrchid NH2 (100 × 2 mm, 1.8 μm, Knauer, Germany) in gradient mode A) water, B) acetonitrile, 0–4 min 100–75% B, then 4–5 min 75–35% B, then back to 100% B for 1 min. To enhance separation, 0.005% of formic acid was added to both mobile phases. Results were compared with those obtained for pure standards of stevioside (St), rubusoside (Rub), dulcoside A (DulcA), steviolbioside (Stev/Biol), rebaudioside A (RebA), rebaudioside B (RebB), rebaudioside C (RebC), rebaudioside D (RebD), and rebaudioside F (RebF). All standards were of highest available purity (Chromadex, Irvine, CA, USA).

Induction of callogenesis in S. rebaudiana hairy roots culture

Some hairy roots, before transfer into the liquid medium, were used for callus induction. Hairy roots were cut into 0.5-cm fragments under sterile conditions and transferred onto MS solid medium supplemented with 1.0 mg/l TDZ (thidiazuron), 30.0 mg/l sucrose (85 mM) and 7 g/l agar, and then, they were cultured in the darkness.

Confirmation of transformation process

Transformation of hairy roots as well as transformed nature of induced callus was confirmed by PCR amplification of rol C, rolB, and virC genes. RNA extraction was performed according to Chomczyński 1987. To determine the RNA purity, the ratio of absorbance on two wavelengths (OD260 nm and OD280 nm) was measured, and it was considered pure when the value A (260)/A (280) was within 1.6–2.0. iScript cDNA Synthesis Kit (BioRad, USA) and reverse transcription reaction were used to obtain cDNA. For each PCR reaction, DreamTaq Green PCR Master Mix (ThermoScientific) was used. For each reaction, we used 25 μL of DreamTaq Green Master Mix, 0.5 μM of each forward and reverse primer, and 1 μg of template DNA, all complemented with water nuclease-free up to 50 μL. Primers used for PCR were as follows rolB 5′-GCTCTTGCAGTGCTAGATTT-3′ and 5′- GAAGGTGCAAGCTACCTCTC-3′; rolC 5′-CTCCTGACATCAAACTCGTC-3′ and 5′-TGCTTCGAGTTATGGGTACA-3′; virC 5′-ATCATTTGTAGCGACT-3′ and 5′-AGCTCAAACCTGCTTC-3′ (Królicka et al. 2001). PCR was conducted in 40 cycles in order as follows: initial denaturation at 95 °C for 3 min, denaturation at 95 °C for 30 s, primer annealing at 53 °C for 30 s, elongation at 72 °C for 1 min, and followed by final elongation for 5 min (C1000 Thermal Cycler, BioRad). Amplified fragments of cDNA were separated by electrophoresis on 1% agarose gel in TE buffer. For visualization, PCR products on gel were stained with ethidium bromide and observed under UV light.

Statistical analysis

To assess the statistical significance of the differences between the control and experimental samples, one-way analysis of variance (ANOVA) and Tukey’s test at the level P < 0.05 were performed.

Results

Morphology and growth of transgenic hairy roots

Hairy roots grown in liquid media under various cultural conditions showed typical morphological features, such as branching and rapid mass increase (Fig. 1a and Table 1).

Fig. 1
figure 1

Macroscopical studies of transformed stevia hairy roots. a Morphology of hairy roots grown for 30 days under different conditions, scale bar 1 mm. b Representative agarose gel after electrophoretic separation of PCR products of rolB and rolC gene fragments in hairy roots. VirC gene PCR product was identified to exclude the A. rhizogenes contamination of investigated material. Actin (Act) was used as a reference gene.

Full size image

On day 40 of the experiment, the growth of hairy roots was arrested under all conditions tested. The highest increase in weight was noted for HR grown in the dark and were maintained in the medium with the highest osmotic potential. The weight of these samples on the 30th day of culture was as much as four times higher than at the beginning of the experiment. Hairy roots cultured in the light grew slower than those kept in media with changed osmotic conditions, and only doubled their weight after 30 days of culture (Table 1).

The length of investigated hairy roots varied significantly depending on growing conditions (Table 1). The longest HR were found in the medium with the highest osmotic potential, while the shortest HR were described in the medium supplemented with PEG. Inspected HR produced root’s hair of various lengths depending on culture conditions. The longest root’s hair were found on HR growing under light, the shortest root’s hair were visible in HR cultured in medium with the lowest osmotic potential, while no root’s hair were detected on HR cultured in PEG or NaCl supplemented media (Table 1).

The transgenic nature of HR cultured in different media was verified at the molecular level by the PCR method using primers for rolB and rolC genes. The bacterial rolB and rolC genes were successfully amplified using cDNA obtained from RNA isolated from HR indicating the integration of the Ri T-DNA of bacterial plasmid into the plant genome (Fig. 1b).

Assessment of oxidative stress level

Spectrophotometric analysis of hydrogen peroxide content revealed that the conditions of continuous light or application of osmotic active compounds in the culture medium led to the increase in H2O2 concentration in HR, leading to an enhancement of oxidative stress (Fig. 2) The highest concentration of H2O2 was measured in HR cultured under light or in the medium with the highest osmotic potential indicating that these conditions were the most stressful.

Fig. 2
figure 2

Endogenous concentration of hydrogen peroxide. Data represent means ± SDs. Values sharing the same letter(s) are not significantly different (P < 0.05, n = 4)

Full size image

Alteration of KAH and UGTs transcript levels in response to different culture conditions

The activity of all investigated genes, KAH encoding ent-kaurenoic acid hydroxylase as well as UGTs encoding, was upregulated in HR cultured under either light or in the media supplemented with different osmotic active compounds (Fig. 3). The highest increase in KAH transcriptional activity was noted in HR cultured in light condition or in the highest osmotic potential of culture medium. The highest activity of UGT85c2 and UGT74g1 was found in HR cultured in media with the highest osmotic potential. The highest activity of UGT76g1 was noted in HR grown in the medium with the lowest osmotic potential or under PEG treatment.

Fig. 3
figure 3

Relative transcriptional activity of kaurenoic acid hydroxylase (KAH) and UDP glucosyltransferase (UGTs) genes in hairy roots cultured under light or in the media supplemented with different osmotic active compounds

Full size image

Accumulation of steviol glycosides in HR cultured in the light or in media with different osmotic potential

UHPLC analysis of SvGls concentration in cultured HR showed the presence of the following compounds: stevioside (St), rubusoside (Rub), dulcoside A (DulcA), steviolbioside (Stev/Biol), rebaudioside A (RebA), rebaudioside D (RebD), and rebaudioside F (RebF). The obtained results indicated that the culture conditions affected different levels of accumulation of these SvGls (Fig. 4). Reduction in St content was found in HR grown in culture conditions modified by light or by application of different osmotic active compounds. In contrast, Rub concentration was either higher in HR grown in the media supplemented with different osmotic active compounds, or it did not change in HR exposed to light in comparison to the control conditions. A decrease in DulcA concentration was observed only in HR grown in the medium with the highest osmotic potential, while in the samples from other conditions a similar level of this compound was found and it was the same as in HR from the control conditions. The content of RebA increased significantly in HR cultured in the medium with the lowest osmotic potential or supplemented with PEG. The other tested culture conditions did not influence the content of RebA in HR. Interestingly, Stev/Biol and Reb F were detected only in HR cultured in the medium with the lowest osmotic potential, while Reb D was detected also in HR cultured in the medium supplemented with NaCl or PEG.

Fig. 4
figure 4

Concentration of some steviol glycosides (μg·mg−1 D.W.) in hairy roots cultured under light and under different osmotic conditions. Data means ± SDs. Values sharing the same letter(s) are not statistically different (P ≤ 0.05, n = 3)

Full size image

Investigation of possibility to induce callogenesis in S. rebaudiana hairy roots

HR from the same line as used for liquid culture were applied in experiments on induction of callogenesis potential. Small segments of HR cultured in the presence of thidiazurone in the medium produced callus after 10 days of culture. Callus appeared on the cuttings’ ends of HR, and it was friable and yellowish (Suppl. 2a). Prolonged culture led to further growth of callus tissue that could be separated from the original explant and cultured separately.

The transgenic nature of achieved callus was verified at the molecular level by the PCR method using primers for rolB and rolC genes. The bacterial rolB and rolC genes were successfully amplified using cDNA obtained from RNA isolated from HR (Suppl. 2b). This confirmed that the transformation with A. rhizogenes LBA 9402 strain is permanent, and it continues within callus formation.

Discussion

The results presented in this paper revealed that HR of S. rebaudiana have an ability to synthesize and accumulate SvGls, and that an application of light or different osmotic active compounds changes the biosynthetic profile of these secondary metabolites to varying degrees depending on the culture conditions. This observation, however, does not confirm previous studies carried out by Yamazaki and Flores (1991) in which SvGls were not found in HR cultures, as well as by Pandey et al. (2016), where biosynthesis of steviosides was recorded only in HR cultured under light conditions and exhibited photosynthetic activity and the presence of chlorophyll as factors regulating the production of stevioside.

In our study, we have found the levels of stevioside concentration in HR cultured in the dark ranging between 0.3 and 1.3 μg/mg DW just after 30 days of culture (Fig. 4). It might indicate that the light and active photosynthesis are not necessary factors in SvGls biosynthesis pathway, as it was earlier suggested by Yamazaki and Flores (1991) and Bondarev et al. (2001). Although chloroplasts are involved in the steviol glycosides biosynthesis, only the initial stage of this pathway—ent-kaurene synthesis—occurs in these organelles. Subsequent steps take place in the endoplasmic reticulum, and then in the cytoplasm followed by SvGls accumulation in the vacuole (Kumari and Chandra 2015). We can, therefore, postulate that the active photosynthetic apparatus is not necessary for the occurrence of steviol glycoside biosynthesis pathway. Apparently, the presence of amyloplasts or chloroplasts precursors is sufficient for this process to take place. As it was already shown by Balmer et al. (2006), amyloplasts may also participate in partitioning of resources such as carbon between different biosynthetic pathways via the expression of ent-kaurene oxidase activity responsible for the production of kaureonic acid—an output molecule for both steviol glycosides and gibberelins biosynthesis.

It was previously shown that SvGls are very potent ROS scavengers involved in the detoxication of both hydroxyl and superoxide radicals (Woźniak et al. 2014), and may contribute to abiotic stress resistance within the plant of origin (Stoyanova et al. 2011). This finding is in agreement with the hypothesis stated by Bolouri-Moghaddam et al. (2010), suggesting that sugar-like compounds, possibly in combination with other antioxidants, form an important antioxidant system. The results of our studies showed that all cultured conditions caused an increase in the oxidative stress in the cultured HR (Fig. 2). However, its highest and comparable level was recorded in the samples grown in the light or in the medium with the highest osmotic potential. Based on the studies of HR growth and morphology, it can be concluded that they can better adapt to the changes in media osmoticum than to light-induced stress. This is in agreement with previous studies conducted by Pandey et al. (2016), which showed that green HR grew more slowly and reached smaller size compared to those cultured in the dark. Continuous light exposure might generate photooxidative stress in newly synthesized chloroplasts of green HR and some metabolic changes have to be induced to combat this effect. According to Mukherjee et al. (2014), inhibition of HR growth by light may be related to metabolic shift to the methylerythritol phosphate pathway (MEP) in order to synthesize the chlorophyll phytophilic chain in green HR. It should be remembered, however, that this pathway is also responsible for the production of steviol glycosides, starting from the induction of ent-kaurenoic acid hydroxylase (KAH) activity. Indeed, in our studies, we observed a significant increase in the transcriptional activity of the gene encoding this enzyme in HR cultured in the light, as well as in those growing in the dark, especially under the conditions of high osmotic potential (Fig. 3), indicating the redirection of metabolic pathway.

It had previously been found that both the transcript levels of UGTs genes as well enzymatic activity of glucosyltransferases involved in further steps of SvGls biosynthesis are strongly affected by different biotic and abiotic stress factors (Horvath and Chuai 1996; Vogt and Jones 2000; Mazel and Levine 2002; Bowles et al. 2005; Langlois-Meurinne et al. 2005; Sepúlveda-Jiménez et al. 2005; Vanderauwera et al. 2005; Meissner et al. 2008; Sun et al. 2013; Tognetti et al. 2010; Ahrazem et al. 2015; Yang et al. 2015; Le Roy et al. 2016; Vazquez-Hernandez et al. 2019). In our studies, the analysis of genes encoding three different glucosyltransferases—UGT85C2, UGT74G1, and UGT76G—has revealed that they were upregulated in HR under all culture conditions. We have shown that the level of analyzed UGTs upregulation was different in applied conditions and depended on the type of stress factor causing oxidative stress rather than on the level of the oxidative stress. Therefore, our results do not confirm the previous studies concerning the positive correlation between upregulation of some UGTs and the alleviation of oxidative stress (Ahrazem et al. 2015), but they are the agreement with the findings published earlier by Hajihashemi et al. (2013) and Hajihashemi and Geuns (2016), who found that the effects of treatment on the activity of genes involved in the steviol glycosides biosynthesis pathway can differ individually. In accord with the paper cited above, we have found that transcriptional activity of KAH does not correlate with the level of investigated UGTs expression. Moreover, it can be concluded that, while KAH is sensitive to oxidative stress level, the regulation of UGTs activity is more complicated and not directly dependent on hydrogen peroxide concentration (Figs. 2 and 3).

The role of steviol glycosides as osmoprotectant molecules was previously suggested by Ceunen and Geuns (2013). Follow-up research was performed and confirmed that water or ionic stress significantly influenced the transcriptional activity of some genes involved in SvGls biosynthesis as well as the content of particular steviol glycosides in in vitro and in vivo S. rebaudiana culture (Hajihashemi et al. 2013; Zeng et al. 2013; Hajihashemi and Ehsanpour 2014; Gupta et al. 2014; Gupta et al. 2015; Hajihashemi and Geuns 2016; Pandey and Chikara 2015; Cantabella et al. 2017; Fallah et al. 2017; Hussin et al. 2017; Debnath et al. 2018; Shahverdi et al. 2018; Ghaheri et al. 2019; Javed and Gürel 2019; Lucho et al. 2019). However, the direction of identified changes in SvGls biosynthesis depended significantly on the type of endogenously applied compound. The concentration of NaCl or PEG used in our study was chosen according to the studies mentioned above, and because it was indicated to be the most potent in modifying SvGls content or transcriptional activity of genes involved in SvGls biosynthesis. The growth rate of HR from NaCl or PEG treatment was higher than in HR from control conditions (Table 1), and both stress factors had a positive impact on transcriptional activity of investigated genes (Fig. 3). It has been observed in our study that NaCl or PEG application during HR culture led to a significant decrease in St concentration, an increase in Rub content, and an enhanced production of Reb A after PEG treatment, as well as the presence of detectable amounts of Reb D in HR cultured in both stress-inducing conditions (Fig. 4). Interestingly, all the samples investigated in our experiment exhibited a high ratio of rebaudioside A to stevioside, indicating a higher concentration of Reb A than St—a result very seldom described in the whole plant. A similar effect of pronounced higher concentration of Reb A than St in the samples cultured in vitro was achieved by Gupta et al. (2014) in suspension culture of stevia cells treated with NaCl stress. Both findings may be an important indication for alternative approaches to metabolic engineering of stevia to enhance SvGls that are more desirable due to sensory attributes that are more favorable than those of other major SGs (Tanaka 1997).

It is generally difficult to correlate the expression of a specific gene encoding an UGTs enzyme with the concentrations of a particular product resulting from the activity of a given enzyme, since plant UGTs are not believed to be highly specific and exhibit broad substrate specificity (Vogt and Jones 2000; Bowles et al. 2005; Mohamed et al. 2011). Previous studies concerning in vitro analysis of stevia UGTs (SrUGTs) for their reactivity against various known substrates of the pathway of steviol glycoside biosynthesis revealed that these enzymes possess a broad acceptor specificity (Humphrey et al. 2006). Therefore, it cannot be excluded that various bifurcations to main biosynthetic route of SvGls might exist. Indeed, the existence of alternative pathway of steviol glycoside biosynthesis pathway was recently demonstrated by several authors (Richman et al. 2005; Guleria and Yadav 2013; Kim et al. 2019). As an example, it was found that the enzymes encoded by UGT74G1 and UGT76G1, in addition to their catalytic role in the main biosynthetic route from steviolmonoside via steviolbioside to stevioside, and then to rebaudioside A, might be also involved in an alternate route of steviol glycoside biosynthesis by catalyzing 19-O-b-glucopyranosyl steviol and rubusoside (Guleria and Yadav 2013). In this study, it has been shown that the biosynthetic pathway of steviol glycosides in HR cultured under different conditions occurred mainly through the rubusoside production, especially under conditions of osmotic active compounds application, where the increase in this SvGl was observed. Since the presence of steviolbioside, which is one of the steps of the main SvGls biosynthetic pathway, was noted only in HR cultured in the medium with the lowest osmotic potential, it can be concluded that only in these conditions, the main biosynthetic pathway of SvGls biosynthesis was supported by an alternative route.

The present work confirms that HR of S. rebaudiana Bertoni are an easy tool for the manipulation of SvGls content through the application of external factors generating enhancement in oxidative stress level, which influence the transcriptional activity of genes involved in SvGls biosynthesis, and remodel their accumulation pattern. In this way, the resultant experimental material might exhibit elevated levels of desired SvGls that might function as an element of plant antioxidant system, as well as be of interest to food and pharmaceutical industry. Special attention should be paid to Steviolbioside, which is a minor sweet compound found in leaves of S. rebaudiana (Kohda et al. 1976) and is considered extremely important due to its specific anticancer properties (Chen et al. 2016). Although the content of two major SvGls as St and Reb A is lower in HR than in the whole plants (Libik-Konieczny et al. 2018), preferential accumulation of Reb A as well as the selective accumulation of some minor SvGls provides an interesting opportunity for metabolic engineering of stevia, and for further experiments concerning the induction of cell suspension culture or regeneration of whole plants from callus tissue already induced from S. rebaudiana HR (Suppl 2).