Cardenolide and glucosinolate accumulation in shoot cultures of Erysimum crepidifolium Rchb.

Erysimum crepidifolium Rchb. is one of the few Brassicaceae species accumulating glucosinolates as well as cardenolides. This is possibly providing a selective advantage in evolution as both compounds are part of a chemical defense system. In order to study the biosynthesis of these compounds, a regeneration protocol for E. crepidifolium using in vitro shoot cultures derived from seeds has been developed. Murashige and Skoog (MS) culture medium supplemented with various combinations of cytokinins and auxins was used. MS medium containing NAA (naphthaleneacetic acid, 0.04 mg mL−1) and BAP (6-benzylaminopurine, 0.2·10−2 mg mL−1) proved to be optimal for root formation. Plantlets developed well on modified MS medium without the use of phytohormones. About 80% of the plantlets rooted in vitro developed into intact plants after transfer to the greenhouse. Cardenolides (1.75 mg g−1 dry weight (DW)) were detected in cultured shoots on solid DDV media while glucosinolates mainly accumulated in roots where 0.025 mg g−1 FW were detected in shoots cultured on the same medium (DDV). The expression of two progesterone 5β-reductase and three Δ5-3β-hydroxysteroid dehydrogenase genes were measured in shoot cultures since the encoded enzymes are supposed to be involved in cardenolide biosynthesis. E. crepidifolium shoot cultures propagated on solid media meet the necessary requirements, i.e., clonal homogeneity, product accumulation, and gene expression, for a suitable model to study cardenolide but not glucosinolate biosynthesis.


Introduction
Arabidopsis thaliana (L.) Heynh., a member of the Brassicaceae family, is widely used as a model organism in genetic studies. The genus Erysimum L. belongs to the same family and comprises more than 200 species (Bailey et al. 2006;Polatschek 2013a, b). The production of cardenolides by species in the genus Erysimum is one of the best-studied examples of an evolutionarily recent gain of a novel chemical defense (e.g., Jaretzky and Wilcke 1932;Makarevich et al. 1994;Züst et al. 2020). Most Brassicaceae species accumulate glucosinolates which are activated by myrosinases to produce repellent mustard oils ( Fig. 1; Fahey et al. 2001). However, several herbivores have evolved to overcome this defense mechanism (Winde and Wittstock 2011). Cardenolide biosynthesis may have evolved further since the 'mustard oil bomb' (Matile 1980) to provide a new type of repellent. In fact, several glucosinolate-adapted beetles refused to feed on dietary cardenolides at levels found in Erysimum species (Nielsen 1978a, b).
Cardenolides and cardenolide-containing plant extracts including Erysimum species (Zhu et al. 1989;Burger and Wachter 1998) have been used for centuries in the therapy of heart insufficiency in humans. The biosynthesis of cardenolides has mainly been studied in Digitalis (Kreis andMüller-Uri 2010, 2013;Kreis 2017). The likely existence of a very similar cardenolide pathway in Erysimum has not been demonstrated. Munkert et al. (2011Munkert et al. ( , 2014 proposed Erysimum species as model plants for the study of cardenolide biosynthesis in Brassicaceae when investigating the enzymes involved and their respective genes. This idea was further elaborated by Züst et al. (2018).
In previous efforts, greenhouse plants originating from seeds were used (Munkert et al. 2014); however, the generation of reliable data is often hindered by the limited availability of homogeneous plant material (Schöner et al. 1986;Kreis et al. 2015). Therefore, the main purpose of this work was to develop a rapid micropropagation protocol for genetically homogenous E. crepidifolium shoot cultures and plants. Plant growth, cardenolide content, and glucosinolate content were quantified during plant regeneration and until permanent shoot cultures were established (Fig. 1, 2).

Materials and methods
Chemicals Myrosinase was prepared from Sinapis alba (mustard) seeds as described by Meitinger and Kreis (2018). All cardenolides analyzed (erysimoside, helveticoside, kstrophantoside, β-methyldigitoxin; Fig. 1 (Murashige and Skoog 1962) containing 30 g glucose and 2.25 mg 6benzylaminopurine (BAP) per liter (solid DDV medium). After 6 to 8 wk, developing shoots were transferred to liquid MS medium containing 30 g glucose and 2.25 mg BAP per liter (liquid DDV medium). After 6 wk of adaption to the liquid media, the shoots were transferred to a liquid MS medium containing 30 g glucose and varying concentrations of naphthaleneacetic acid (NAA; 0 to 80 mg) and BAP (0 to 2.25 mg) per liter to initiate rooting. The optimal composition was established at 40 mg NAA and 0.02 mg BAP per liter (liquid EDDV medium) (Fig. 2c). After 40 d in rooting medium, plantlets with well-developed roots were transferred to solid OH medium that is based on DDV medium with increased amounts of KH 2 PO 4 (340 mg L −1 ), glycine (4 mg L −1 ), and glucose (33 g L −1 ) but without additional phytohormones. After 2 mo, the plantlets were transferred to autoclaved soil and allowed to acclimatize. They were kept in seeding boxes for about 3 wk and were sprayed with water twice a day during the first week to guarantee a humid atmosphere. In order to reduce the humidity during the acclimatization period, spraying with water was reduced gradually (Arikat et al. 2004). Following this procedure the plants were cultivated at standard greenhouse conditions (Fig. 2d).

Permanent shoot cultures
Shoots developed different phenotypes depending on whether they were kept in liquid DDV or on solid DDV medium (Fig. 2a, b). This resulted in two types of permanent shoot cultures, and the cardenolide and glucosinolate contents (see below) have been determined for both. All stock cultures were kept under permanent white light (fluorescent lightning, approximately 40 μmol photons m −2 s −1 ) and at 21 to 24°C. Permanent cultures cultivated in liquid media were kept on MULTITRON gyratory shakers (Infors GmbH, Sulzemoos, Germany) at 55 rpm. Stress treatment with either 2-cyclohexene-1-one or methylvinyl ketone (MVK) was applied as described by Munkert et al. 2014.
RNA extraction and cDNA synthesis RNA was isolated from the individual tissue samples. These samples were frozen with liquid nitrogen and then ground to a fine powder using mortar and pestle. Total RNA was extracted using the illustra™ RNAspinMini RNA Isolation Kit (GE Healthcare, Munich, Germany). cDNA was synthesized employing the SuperScript™ III First-Strand Synthesis for reverse transcription PCR (RT-PCR) Kit (Invitrogen, Karlsruhe, Germany).
Quantitative real-time analysis (qPCR) Real-time PCR was carried out in the StepOnePlus Real-Time PCR System (Life Technologies, Erlangen, Germany) using the SYBR Green Mastermix Kit (Applied Biosystems, Erlangen, Germany) according to the manual provided. The relative gene expression levels were calculated using the 2 −ΔΔCt -method (Livak and Schmittgen 2001) with actin as the reference gene. Primers for actin are based on A. thaliana actin (At3g18780). Primer pairs used for qPCR are listed in Table 1  Cardenolide extraction and determination Cardenolides were extracted from plant and shoot samples as described by Wiegrebe and Wichtl (1993). The extracts were analyzed by HPLC according to Wichtl et al. (1982) with slight modifications (Schebitz et al. 2010). For HPLC analysis, a Waters 1525 Binary HPLC Pump system with a Waters 2487 Dual   (100%B), and 6 min (20%B). Erysimoside and helveticoside were detected and quantified at 220 nm using a UV detector. Erysimoside and helveticoside standards were purchased from Merck KGaA (Darmstadt, Germany) and β-methyldigitoxin served as the internal standard for quantification. In several experiments, cardenolides were hydrolyzed prior to quantification ("total cardenolides") following the protocol of Mannich and Siewert (1942) with slight modifications only.
Glucosinolates Glucosinolates were extracted as described by Meitinger and Kreis (2018). For extraction, 1 mg ascorbic acid as well as 250 μL of a sinigrin solution was added to 0.35 g of fresh plant material until a final concentration of 1.8 mM was reached. In order to ensure complete hydrolysis of endogenous glucosinolates, myrosinase prepared from mustard seeds (Meitinger and Kreis 2018) was added.
After 2 h of incubation at 23°C and 10 min of centrifugation at 10,000g, 3 μg μL −1 benzylisothiocyanate was added as an internal standard and the supernatant extracted with dichloromethane. After separation, the organic phase was analyzed by GC-MS using the following conditions: DB-5 ms (Agilent Technologies Deutschland GmbH, Waldbronn, Germany) column (30 m × 0.25 mm, film 0.25 μm); column oven temperature: 60°C; injector temperature: 230°C; injection volume: 1 μL; split ratio: 1:100; ion source temperature: 200°C; interface temperature: 210°C; and MS in scan mode: 35-200 m/z. The temperature gradient with helium as the carrier gas was implemented at a flow rate of 46.3 cm s −1 . The temperature was kept constant for 3 min at 60°C before it was increased to 210°C at a rate of 10°C per min.

Results and discussion
Micropropagation protocol of Erysimum crepidifolium Rchb. Surface-sterilized single seeds were placed on a modified MS solid medium containing 2 mg L −1 BAP (DDV). In previous studies, BAP proved to be the most suitable phytohormone to initiate shoot multiplication without root formation (Zimmerman and Scorza 1992).
Alternatively, rooting was induced on solid media supplemented with NAA. Murashige and Skoog (1962) previously reported that NAA is a more potent auxin than natural occurring indoleacetic acid (IAA). We here tested 20 different combinations of BAP and NAA in order to initiate root formation. Optimal results were achieved when liquid MS medium containing 2 mg L −1 BAP and 40 mg L −1 NAA was used and root formation began after 2 wk. George and Sherrington (1984) also reported that rooting can be induced with a nutrient medium containing high auxin and low cytokinin concentration. Even higher concentrations of BAP and NAA stimulated callus formation as had been reported before for other plants (Hussey and Stacey 1981;Iriondo 1990aIriondo , 1990b. Plantlets were transferred onto solid, hormone-free MS medium after 6 wk on rooting medium. After an additional 8 wk, the plantlets were ready to be potted (Fig. 2a-d). Plantlets were transferred to the greenhouse and kept initially on a sterilized mixture of sand and soil to avoid microbial damage. Shoot cultures grown in vitro fail to form sufficient layers of epicuticular wax (Grout and Aston 1977;Sutter and Langhans 1979). In order to still guarantee a humid atmosphere, the plantlets were kept in seeding boxes for 3 more wk. Two months after their transfer to the greenhouse, 77.5% of the plants had survived and became well developed (Fig. 2d). This protocol provides genetically homogenous plants that can be used in further studies; however, vegetative propagation by scions and runners is also feasible (unpublished observations).
Glucosinolates and cardenolides in permanent cultures and during regeneration Developing shoots from stock cultures on solid medium were transferred into liquid MS medium for further proliferation. Cultivation in liquid medium facilitates the delivery of biosynthetic precursors or chemical stressors and makes it easier to monitor an increase in biomass (Fig. 2b). Permanent shoot cultures are suitable for studying the regulation of cardenolide biosynthesis or for detailed analyses of the influence of light and chloroplast differentiation which has been demonstrated in somatic embryos of D. lanata (Scheibner et al. 1989). After transfer into liquid medium, E. crepidifolium shoots developed an abnormal, clinched phenotype (Fig. 2a). The chlorophyll content was 0.3 mg to 0.9 mg g −1 fresh weight (FW) and the biomass only doubled within 3 wk. A 30 mg mL −1 glucose medium was provided after 30 d of culture, and the partially submerged shoot cultures accumulated neither glucosinolates nor cardenolides (see below). On the solid DDV medium, shoot tufts developed successfully (Fig. 2c) and could be propagated easily. Shoot cultures kept at these conditions contained cardenolides but no glucosinolates (see below).
Glucosinolates The glucosinolate content was calculated from iberin released from glucoiberin by the action of myrosinase (Fig. 1). Myrosinase from mustard seeds was added to ensure the complete conversion of glucosinolates to their respective isothiocyanates since the endogenous myrosinase activity was very low from shoots cultured in liquid medium (data not shown). Hydrolysis of glucosinolates usually occurs when plant tissue is damaged and glucosinolates come in contact with endogenous myrosinases (Rask et al. 2000;Vaughn and Berhow 2005;Loebers et al. 2014). After prolonged cultivation on a solid hormone-free medium, E. crepidifolium shoot cultures contained 0.025 ± 0.01 mg g −1 FW (n = 3) glucosinolates. Glucosinolates did not accumulate in shoots cultured in a liquid medium (Table 2).
Only low levels of glucosinolates were detected in leaves of greenhouse plants ( Table 2). The glucosinolate content of roots was considerably higher and reached 4.55 mg iberin g −1 FW. Concentrations as high as 5% of the dry seed mass were reported for individual glucosinolates in flowers and seeds (Bennett et al. 2004).
Cardenolides E. crepidifolium accumulates approximately 1% cardenolides in the dry mass of its above ground organs, most of them bearing k-strophanthidin (str) as their aglycone. Digitoxose (dox) and glucose (glc) are the sugar components of the main glycosides erysimoside (str-dox-glc) and helveticoside (str-dox), representing about 90% of the cardenolide mixture ( Fig. 1; Gmelin and Bredenberg 1966). Since only traces of the minor cardenolides such as erycordin and glucostrophalloside (Makarevich et al. 1974) were detected, we only quantified erysimoside and helveticoside either individually or as their degradation product, k-strophanthidin, released by chemical hydrolysis of the cardenolide extract (Fig. 1). Approximately 80 μg g −1 FW cardenolides were detected in shoots cultivated on the DDV medium even after prolonged subculture for more than 2 y. The concentrations decreased dramatically when stock shoot cultures were transferred from the solid into the liquid DDV medium where they developed an abnormal phenotype (Fig. 2a). While minor amounts of cardenolides were still present after 3 wk in liquid medium (about 10 μg g −1 FW), they had vanished completely after two mo. Cardenolide formation also ceased when shoots were kept on the solid DDV medium but cultivated in permanent darkness. After 1 wk, the cardenolide content dropped down to 10% of that seen in control shoots grown at standard conditions. This was paralleled by the degradation of chlorophyll which dropped from 0.3 to 0.03 mg g −1 FW within 12 d.
This corroborates the findings of others (Luckner and Diettrich 1985;Eisenbeiß et al. 1999) who demonstrated that cardenolide formation depends on morphological differentiation and light. This is also consistent with observations concerning other natural compounds. For example, Coste et al. (2011) reported that hypericin biosynthesis is impaired after transfer from a solid into a liquid culture medium. Hugentobler and Renwick (1995) discussed possible reasons for differential formation of cardenolides and glucosinolates in Erysimum as a result of nutrient deficiency. Shoots transferred into rooting medium also ceased to produce cardenolides which can be explained by the altered medium composition (Table 2; 0.04 mg mL −1 NAA) as was suggested by Patil et al. (2013). Plantlets grown on the hormone-free medium prior to their transfer into soil produced only minor quantities of cardenolides. However, cardenolides accumulated in a similar range in young leaves (0.38 mg g −1 dry weight (DW)) of Erysimum crepidifolium plants as in liquid shoot cultures (0.47 mg g −1 DW), but did not reach cardenolide levels of shoot cultures that were cultured on solid media ( Table 2).
Expression of two EcP5βR genes and three Ec3β-HSD genes in E. crepidfolium shoot cultures Other than the cardenolide content, gene expression levels of EcP5βRs and Ec3β-HSDs were measured since the encoded enzymes, namely, progesterone 5β-reductase and Δ 5 -3β hydroxysteroid dehydrogenases, are reported to be involved in cardenolide biosynthesis in Digitalis (Kreis andMüller-Uri 2010, 2013). Two EcP5βR (EcP5βR1; EcP5βR2) genes have been reported to be expressed in E. crepidifolium greenhouse plants by Munkert et al. (2011Munkert et al. ( , 2015 who also characterized recombinant forms of the encoded progesterone 5β-reductases. In permanent shoot cultures (solid medium), the expression of EcP5βR1 was about six times higher than that of EcP5βR2 (Fig. 3a). The expression of EcP5βR2, but not EcP5βR1, was stimulated by abiotic stress. When treated with 2-cyclohexen-1-one or methyl vinyl ketone (MVK), expression increased 6.8-fold and 3.4-fold, respectively, after 24 h, after 7 d, basic expression levels, similar to levels prior to abiotic stress treatment, were reached (Fig. 3b). Heterologous P5βR1 of D. purpurea was also expressed constitutively in plants, whereas the expression P5βR2 dramatically increased in stress conditions (Pérez-Bermúdez et al. 2010). Munkert et al. (2014) identified three genes (Ec3β-HSD1; Ec3β-HSD2; Ec3β-HSD3) in E. crepidifolium. Ec3β-HSD1 did not respond to various forms of stress, whereas Ec3β-HSD2 and Ec3β-HSD2 were upregulated when plants were treated with methyl jasmonate or methyl vinyl ketone. We observed that during rooting, the expression levels of Ec3β- HSD1 remained almost constant, whereas the expression level of Ec3β-HSD2 and Ec3β-HSD3 increased by almost eightfold immediately after transfer to the rooting medium which was interpreted as a stress response (data not shown). Exposing 4mo-old plants to various stress conditions, Munkert et al. (2014) found that cardenolide content paralleled gene expression of Ec3β-HSD2 and Ec3β-HSD3 in several but not all tested stress conditions. In our studies, shoots grown in liquid medium developed an abnormal phenotype and did not accumulate cardenolides although Ec3β-HSD2 and Ec3β-HSD3 were upregulated. This indicates that morphological differentiation plays a pivotal role in triggering cardenolide formation and that Ec3β-HSDs may also be involved in other physiological processes.
After rooting and transfer to solid medium, the relative expression of Ec3β-HSD1 steadily increased from 0.51 to 4.19 and thus paralleled cardenolide accumulation, whereas the expression levels of Ec3β-HSD2 and Ec3β-HSD3 remained nearly unaffected. Ec3β-HSD1 expression seems to be related to plant development and differentiation, including cardenolide formation, whereas Ec3β-HSD2 and Ec3β-HSD3 are stress-sensitive (Fig. 4a, b).
All three 3β-HSD isoforms investigated here were expressed in each phase of the regeneration protocol and in permanent shoot cultures in stages with or without cardenolide accumulation. Similar results were obtained for 3β-HSD and P5βR1 of Digitalis lanata (Ernst et al. 2010); the latter is considered to be a key enzyme in cardenolide biosynthesis (Herl et al. 2008). Like the 3β-HSD1 of E. crepidifolium,  the P5βR1 of D. purpurea was expressed constitutively in plants. The expression P5βR2 was mainly induced by stress (Pérez-Bermúdez et al. 2010).
Like the glucosinolates, cardenolides were only detected after shoot differentiation and during plant regeneration. Cardenolide formation returned after approximately 2 mo on solid medium (Fig. 4a) but did not correlate with the expression of the 3β-HSD genes. Cardenolide content and pattern can change during plant growth and development as has been demonstrated widely for Digitalis species (Weiler and Zenk 1976;Wichtl and Freier 1978;Braga et al. 1997).

Conclusion
We conclude that E. crepidifolium shoot cultures propagated on solid and liquid media meet the essential requirements (clonal homogeneity, cardenolide production) of a suitable model to study cardenolide but not glucosinolate biosynthesis. The development of this micropropagation protocol additionally offers access to genetically identical mature plants for future studies.