Seed tubers from the table potato cultivars Bintje, Juliette, King Edward, and Princess were obtained from certified producers. Desiree and Magnum Bonum seed tubers were generated from in vitro clones obtained from Prof. Erland Liljeroth, Swedish University of Agricultural Sciences, Alnarp, Sweden. During years 1 and 2 of the study, tubers were planted in 15-L black plastic pots filled with fertilized peat and grown in an outdoor cultivation facility in Uppsala, Sweden. Tubers were planted at the end of May and harvested after 20 weeks. Watering and fertilization were according to common practice at the facility. All cultivars were grown, harvested, stored, and treated in parallel. Tuber storage after harvest was in a dark room kept at +3°C ( ± 1°C) and at ambient air humidity. Wounding or light exposure treatments of tubers were as described (Petersson et al., 2013a). For all treatments, samples consisted of four tubers of equal size and weight, which were either pooled for metabolic and real-time quantitative polymerase chain reaction (QPCR) analyses or analyzed separately for transcriptomics (n = 3). Samples were frozen in liquid nitrogen and stored at −70°C.

The complementary-DNA fragments (cDNAs) encoding 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase 1 (HMGR1) and transaminase isoform 2 (TAM2) were reverse transctiptase-PCR (RT-PCR)-amplified from tuber RNA from cv. Magnum Bonum, verified by DNA sequencing, and cloned in sense orientation behind the Cauliflower Mosaic Virus (CaMV) 35S promoter in plasmid pK7WG2.0 (Karimi et al., 2002). As a control, TAM2 cDNA was also amplified from the EST-clone STMCB25 (Arizona Genomics Initiative, and cloned in sense orientation behind the 35S promoter in plasmid PCV702 (Koncz and Schell, 1986). Plasmids 35S:HMGR1(pK7WG2) and 35S:TAM2(pK7WG2) were used to transform cv. Bintje, whereas 35S:TAM2(PCV702) was used to transform cv. Desiree, as described (Beaujean et al., 1998). Potato transformants were grown in a certified greenhouse with natural daylight supplemented with mercury lamps giving a photon flux of ca. 70 µmol m⁻² s⁻¹, and a 16-h/8-h photoperiod.

Metabolome fingerprinting was performed using an ultra-high pressure liquid chromatography high-resolution mass spectrometry (UHPLC-HRMS/MS) system consisting of an Acquity UHPLC coupled with Synapt G2 mass spectrometer (Waters, USA). The analytical column was Acquity BEH HILIC (100 mm × 2.1 mm i.d., 1.7 µm; Waters, USA) with column temperature at 40° C. Mobile phases were as follows: A) 5 mM aqueous solution of ammonium formate with 0.1% formic acid (v/v) and B) 5 mM solution of ammonium formate in 90% acetonitrile with 0.1% formic acid volume per volume (v/v). Gradient elution was as follows: 0-8 min linear gradient elution from 95% to 5% of solvent B, 8% to 12 min isocratic elution at 5% of solvent B, 12–15 min isocratic elution at 95% of solvent B with the flow rate 0.3 mL min⁻¹. The High-resolution tandem mass spectrometry (HR-MS/MS) detection used electrospray ionization (ESI) operated in positive and negative ionization modes in the scan range from 100 to 1,200 mass-to-charge ratio (m/z). For nebulization nitrogen gas was used. Optimized ESI conditions were as follows: capillary voltage, 1.0 kV in both positive and negative modes; sampling cone voltage, 25 V; extraction cone voltage, 4.0 V; ESI source temperature, 110°C; desolvation temperature, 450°C; cone gas flow, 30 L h⁻¹; desolvation gas flow, 800 L h⁻¹; collision gas flow 0.5 mL min⁻¹; and collision energy ramp in range of 15–45 V. Identification of markers was done on the basis of retention time, exact m/z values, elemental composition generated by the MassLynx software (version 4.1), and mass spectra were obtained in full scan HR-MS mode. The compounds were further studied using HR-MS/MS in MS^e mode to confirm the identification of compounds according to their fragment ions and MS/MS spectra comparison with databases “Metlin” and “MZcloud.”

Glycoalkaloids were from Extrasynthèse (Genay, France). Calystegines A₃, B₂, and B₄ were purified from potato sprouts, and the purity and identity were established by LC-MS and NMR (Schulzova; unpublished work). For all types of alkaloid, the same preparations were used for identification and quantification purposes.

For determination of glycoalkaloids and calystegines, 1 g of freeze-dried sample was shaken with 30 mL of methanol:water (1:1, v/v) for 30 min and centrifuged (5 min, 10, 000 revolutions per minute (rpm)). The supernatant was filtered, and the filtrate was quantitatively made up to 50 mL with extractant solution, filtered through a polytetrafluoroethylene (PTFE) microfilter, and diluted 20× before analysis. UHPLC-MS/MS analyses were performed using an UHPLC Dionex UltiMate 3000 RSLC coupled to a Q-Exactive high-resolution tandem mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Chromatographic separation for 2 μL of injected sample extract was carried out using an analytical column Atlantis HILIC Silica (100 mm × 3 mm i.d., 1.7 μm) (Waters, USA) at 30° C column temperature. The mobile phases were A) 20 mM aqueous solution of ammonium acetate and B) acetonitrile, with gradient elution as 0–5 min isocratic elution at 90% of solvent B, 5–8 min of linear gradient elution from 90% to 60% of solvent B, 8–10 min 90% of solvent B with the flow rate 0.45 mL min⁻¹.

Mass spectrometric detection was performed using a Q-Exactive high-resolution tandem mass spectrometer equipped with heated ESI, operated in positive ionization modes. The mass spectrometer was operated at a spray voltage of 3.3 kV, heater temperature of 220°C, capillary temperature of 300°C, and S-lens radiofrequency (RF) level of 60. Full MS mode parameters were with a resolution of 70,000 full width at half maximum (FWHM) and scan range of 70−1,000 m/z.

Alkaloid identification was performed by comparing the analyte with authentic standards on the basis of retention time and exact mass m/z. Quantification of analytes was performed by comparison of the peak area to a standard curve made from serial dilutions of the corresponding reference substance. Analytes and their limit of quantification (LOQ) are given in Table S2 .

RNA was extracted from leaves and tubers as described (Nahar et al., 2017). Complementary DNA was synthesized using Superscript III reverse transcriptase (Invitrogen Corp., Carlsbad, CA, USA). QPCR was performed with ABsoluteTm QPCR SYBR® Green Fluorescein mix (Thermo Fisher Scientifics Inc., Surrey, UK), analyzed on an iQ5 Real-time PCR detection system (Bio-Rad Laboratories Inc., Hercules, CA, USA), and evaluated using the QBASE software (Hellemans et al., 2007). QPCR was made in triplicates, normalized against two reference genes (60S rRNA protein and β-tubulin), and included an internal reference RNA to standardise for between-run variation. Reference genes were selected from QPCR analyses of six potato genes showing minimal expression changes in a microarray study of wound- and light-inducible genes (Nahar et al., 2017). Primer DNA sequences are listed in Table S3 .

RNA for transcriptomic analyses was purified from single tubers in biological triplicates. Sequencing libraries, cluster generation, and pair-end sequencing were made by the SNP&SEQ Technology Platform at Uppsala University, following an accredited method using NovaSeq and v1 sequencing chemistry (Illumina Inc.).

Quality control of reads was performed using FastQC algorithm (Andrews, 2010). Reads were aligned against potato genome v.4.03, and transcript counts were estimated using featureCount algorithm in the Subread aligner package (Liao et al., 2013). Transcript coordinates were based on PGSC_DM_V403_representative_genes.gff obtained3from the Solanaceae genome website (http://solanaceae.plantbiology.msu.edu/pgsc_download.shtml). DESeq2 R package was used for differential expression analyses, and a gene was considered as a differentially expressed gene (DEG) when having a fold change (FC) >4.0 and an adjusted p-value <0.01. Transcripts with average values below 20 reads were not included. For better estimates of logfold changes, apeglm shrinkage was performed using lfcShrink function (Zhu et al., 2019).

To characterize metabolic responses in potato tubers during SGA-inducing conditions, an untargeted metabolomic analysis was undertaken in tubers from six table potato cultivars that had been subjected to either wounding or light exposure treatment. To also investigate the effect of tuber storage, tubers were analyzed both at harvest and after a cold storage in darkness during 6 months.

The metabolome fingerprint analysis using UHPLC-HRMS/MS acquired a total of 43 features in the negative ionization mode and 73 features using positive ionization. Data acquired in the positive ionization mode were, hence, used for further evaluations. An orthogonal partial least squares discriminant analysis (OPLS-DA) showed cultivar separation on the scores plot ( Figure 1A ). A set of parallel quality control samples grouped near the center of the loading plot, confirming reliable data acquisition ( Figure 1B ). Loading and variable influence on projection (VIP) analyses showed that a number of features, e.g., with the retention time and m/z of 2.47_852.5 (α-chaconine), 2.70_868.5 (α-solanine), and 3.58_258.11 (glycerophosphocholine), were important to discriminate between the cultivars ( Figure 1C ). Analyses of the wound and light treatments separately showed that the tuber responses to both treatments were progressive processes ( Figures S2 and S3 ). On the basis of VIP values, a number of features that were important for the discrimination of cultivars in the different OPLS-DA models (cultivar, wound, and light) could be identified, including α-solanine, α-chaconine, and 2.18_868.504 (solamargine) and 2.53_884.499 (solasonine). A summary of response marker metabolites and the corresponding statistical model characteristics is given in Table S4 .

The results show that glycoalkaloids such as α-chaconine and α-solanine are important parts of metabolic differences between table potato cultivars. This was observed under all conditions investigated, i.e., at harvest, after storage, and after wounding or light exposure.

To characterize the tuber alkaloid responses in greater detail, levels of different SGA forms were analyzed by targeted LC-MS analysis of the same materials that had been used for metabolomics. The levels of CAs were analyzed for comparison.

To investigate the genetic basis for the cultivar differences in SGA production, we performed an RNAseq analysis of the cultivars Bintje and Magnum Bonum, i.e., two cultivars that showed significant SGA response differences in tubers subjected to wounding or light exposure. The analyzed tuber samples corresponded to those used for SGA analyses during year 1 of the study. Because the expression of genes in SGA metabolism precedes the SGA accumulation (Nahar et al., 2017), the RNAseq analysis was performed before the end-point SGA measurements, i.e., at 24 h after wounding or at 48 h after light exposure.

Principal component analysis (PCA) of gene expression patterns showed that samples clustered within the treatment ( Figure S6 ). A certain separation of the cultivars was evident for the light exposure treatment, in line with the observed SGA differences. Depending on the cultivar and treatment, there were in total between 2,500 and 3,500 DEGs ( Figure S7A ). Venn display of DEGs revealed 287 genes whose expression in the two cultivars were significantly upregulated during both wound and light treatments and 179 DEGs that were downregulated ( Figures S7B, C ; Tables S5 and S6 ).

Genes accounting for significant response differences were determined from analyses of gene × cultivar interactions in a multivariate linear regression model, and their different expression patterns were illustrated by Volcano and Heatmap displays ( Figure S8 ). Analyses of the 100 most significant gene × cultivar interactions showed a cultivar separation for both wounding and light treatments ( Figure S8 ; Tables S7 and S8 ), but none of the genes could be attributed to a role in sterol or alkaloid metabolism. Rather, they seemed to be part of cultivar differences related to other types of wound- or light-regulated processes. Gene ontology (GO) analysis further illustrated this and showed that genes associated with the photosynthetic metabolism were overrepresented in Bintje during both treatments ( Figure S9 ). In Magnum Bonum, the regulation of metabolic and oxidative processes was overrepresented in wounded and light-exposed tubers, respectively.

To more directly pinpoint the genes that might underlie the observed cultivar differences in SGA accumulation, we specifically monitored the expression of 58 genes encoding proteins with a known or plausible role in sterol- and SGA biosynthesis. Heatmap display of this gene set indicated a signficant shift in the expression profile of a number of genes after both wounding and light exposure ( Figure 3 ; Table S9 ). Cluster analysis of gene expression profiles identified three clusters (#1, #2, and #3) that were positively correlated with increased SGA levels ( Figure S10 ). These clusters included not only genes encoding sterol-biosynthetic enzymes such as HMGR1, SMO1-L, SMO2-L, and DWF7-L, as well as proteins acting in the post-cholesterol SGA metabolism, e.g., CYP72A188, 2-ODD (GAME11), TAMisoform2 (TAM2 and GAME12), SGT1, and SGT3, but also the SGA-related transcription factor GAME9. These results correlated well with the wounding and light induction of HMGR1, SMO1-L, CYP72A188, GAME11, TAM2, SGT1, and SGT3, as observed in a potato microarray study (Nahar et al., 2017). Interestingly, the transcription factor GAME9 grouped in the same cluster (#3) as some of its supposed target genes in potato, e.g., DWF1-L, DWF5-L, and GAME11 (Cárdenas et al., 2016), indicating that the gene in itself was inducible by both wounding and light exposure. This was confirmed by QPCR analyses, showing not only significant wound and light inductions of GAME9 expression in Bintje and Magnum Bonum during both years of the study, but also that the degree of induction did not differ between the cultivars ( Figure S11 ).

In light of recent work in tomato showing two allelic GAME9 variants with different affinities for their target genes GAME17 and GAME7, i.e., GAME9^135V (low) and GAME9^135A (high) (Yu et al., 2020), we analyzed the full-length GAME9 cDNA sequence in Bintje and Magnum Bonum. This showed the GAME9^135A variant in both cultivars and that the deduced GAME9 amino acid sequences were identical over the entire protein ( Figure S12 ). The GAME9^135A form was present also in two other potato cultivars for which corresponding DNA sequences were available, as well as the wild potato species S. Phureja.

To further illustrate the most important gene expression differences between the two cultivars, the set of 58 sterol- and SGA-related genes was filtered both for a significant gene × cultivar interaction (P_adj < 0.05) and an over two-fold treatment difference between cultivars. This identified 10 genes after wounding and 21 genes after light exposure ( Figure 4 ). The span of between-genotype FC ratios was smaller after wounding than after light exposure, in line with the observed smaller SGA differences between genotypes after wounding than after light treatment. The largest FC difference between the cultivars after wounding was a five-fold higher TAM2 (GAME12) induction in Bintje than in Magnum Bonum ( Figure 4A ). In the light experiment, the span of FC-ratios was greater than after wounding. In particular, three genes —HMGR1, SMO1-L, and TAM2 —showed much higher FC values in Magnum Bonum than in Bintje ( Figure 4B ). In addition, downregulation of four genes —SMT1, SQE, HMGR3, and CAS— was significantly stronger in Magnum Bonum after light exposure.

The significantly stronger light-induced expression of certain SGA genes, such as HMGR1, SMO1-L, and TAM2 in Magnum Bonum, might explain the cultivar’s higher SGA levels after light exposure. To investigate a role of these genes in SGA production, we generated transgenic Bintje plants overexpressing the Magnum Bonum StHMGR1_MB and StTAM2_MB cDNAs from the 35S promoter. As a control, transgenic Desiree plants were created expressing an independent StTAM2_Ken cDNA clone from cv. Kennebec.

Compared with control plants, Bintje 35S:StHMGR1_MB lines accumulated up to about 100% more SGA in leaves, whereas 35S:StTAM2_MB transformants accumulated up to 900% more, the increases being statistically significant for both types of transformant ( Figure 6 ). There was a variation of SGA levels within the transgenic populations (not shown), presumably related to the degree of transgene expression. QPCR analyses showed that both the StHMGR1_MB and StTAM2_MB transgene expressions were correlated to SGA levels ( Figure S14 ). In line with SGA increases in 35S:StTAM2_MB transformants, overexpression in Desiree of the StTAM2_Ken cDNA also led to significant leaf SGA increases, but, in this case, up to 100% more ( Figure S15 ). The different relative SGA increases in Bintje vs. Desiree StTAM2 transformants may be due to the use of different cDNAs, vectors, or cultivars. Further analyses of SGA levels in the corresponding Bintje transgenic tubers revealed only insignificant SGA alterations compared with controls (not shown), possibly due to the tissue-specific expression differences between the 35S promoter activity in the tuber flesh (Kim et al., 2009) and the localization of the general SGA synthesis machinery in tuber phelloderm (Krits et al., 2007).

The results show that an increased expression of StHMGR1or StTAM2 alone is sufficient to increase SGA levels. Enzymatic activity of the encoded proteins can thus be seen as being limiting for SGA biosynthesis.

Metabolomic profilings showed that α-solanine and α-chaconine were important parts of the general metabolic differences that existed between the six potato cultivars that were investigated. This was shown for tubers analyzed directly at harvest and after wounding or light exposure ( Figure 1 ). In addition, CA levels varied between the cultivars but were not significantly altered by the SGA-inducing treatments ( Figure S5 ). These findings agree well with previous studies (Petersson et al., 2013a; Shepherd et al., 2016) and demonstrate that increased SGA levels do not occur at the expense of CA biosynthesis, not even during the 10-fold SGA increases that were here observed in light-induced Magnum Bonum tubers. The SGA and CA biosynthesis pathways can thus be seen as two metabolically independent routes for alkaloid production in the potato. In a food quality perspective, analyses of both SGA and CA are thus needed to evaluate the tuber quality at harvest, but additional analyses mainly of SGA might be sufficient to monitor later post-harvest changes. The negligible levels of additional SGAs in all samples shows that α-solanine and α-chaconine are the most important SGAs to monitor in table potato tubers, also when stresses and post-harvest conditions are to be considered.

We have previously reported that the relative degree of SGA increases in tubers after wounding or light exposure differs between cultivars (Petersson et al., 2013a). We here confirm and extend these findings and show that cultivar differences in tuber SGA accumulation after both wounding and light exposure are stable for a time period of at least 2 and 7 years, respectively. This strongly suggests that the between-cultivar differences in SGA accumulation are genetically determined. Results also suggest that the relative degree of wound or light responsiveness are separable genetic traits.

Cold storage of tubers in the dark for 6 months did not significantly alter SGA levels in control tubers, although a certain reduction in both wound and light responsiveness was observed. A similar decrease in light responsiveness after cold storage has been shown also for the cultivars Kennebec and Russet Burbank (Mondy and Seetharaman, 1990). These results may reflect a reduction of tuber viability during storage but without affecting the original SGA level attained at harvest. However, also stronger light inductions have been reported after tuber storage (Percival et al., 1993; Griffiths et al., 1998). The reason for these contrasting results is not known, and the effect of storage on tuber SGA responses is seemingly a subject that merits more investigation.

The strongest SGA increases in the present study were detected after light exposure of cv. Magnum Bonum, where 8-day light treatments during 3 separate cultivation years increased the SGA levels about 10-fold compared with the harvest level ( Table 2 ). These increases were significantly higher than those observed for any of the other five cultivars investigated in parallel. Investigations using comparable experimental conditions have shown between two-fold and five-fold SGA inductions for the over 30 potato cultivars analyzed, the average being 2.4-fold (Percival, 1999; Chuda et al., 2004; Petersson et al., 2013a). This inidicates that the degree of light induction of SGA in Magnum Bonum is unusually high among table potato cultivars. By contrast, the relative SGA induction in Magnum Bonum after wounding was not stronger than for any of the five additional cultivars that we investigated here or among the 21 cultivars investigated by Petersson et al. (2013a). In fact, the wounding response of Magnum Bonum was in the present study significantly lower than that of Bintje ( Figure 2 ). Together, the results show that Magnum Bonum compared with most table potatoes has a relatively normal tuber SGA level both at harvest and after a wounding treatment, but that its SGA level can become extremely high as a response to light exposure.

Magnum Bonum was withdrawn from commerce in Sweden after having demonstrated high SGA levels after a cold and wet autumn in year 1986 (Hellenäs et al., 1995). SGA levels reported from analyses of tubers from over 300 commercial lots ranged between 61 and 665 mg kg⁻¹, the mean level being 251 ± 106 mg kg⁻¹. The range of SGA levels for light-exposed Magnum Bonum tubers observed in our present study, between 60 and 680 mg kg⁻¹, is thus remarkably similar to that described by Hellenäs et al. (1995). As analyzed in a questionnaire to the Magnum Bonum farmers in 1986, both frost and mechanical vine killing showed significant correlations to high SGA levels (Diedrichs, 1988), but a potentially causal relationship was not further investigated.

We suggest that the sudden increase of high SGA levels in Magnum Bonum tubers at least, in part, can be explained by the cultivar’s unusual strong light responsiveness. The SGA induction in potato tubers subjected to light exposure is triggered already at low light intensities, e.g., 7–14 µmol m⁻² s⁻¹ (Grunenfelder et al., 2006; Olsen et al., 2018), and natural daylight has a spectral composition that is active in the SGA induction (Sinden et al., 1984; Percival et al., 1996). We speculate that an increased soil erosion around tubers caused by the heavy raining or mechanical vine killing, possibly in combination with inadequate hilling, may have led to an exposure of tubers to low levels of daylight that was sufficient to trigger a high SGA synthesis in this light-sensitive cultivar. In support of this view is that potatoes that are planted with a small soil cover and not are hilled during the growing season form higher SGA levels than those that are planted deeply and kept well covered (Hutchinson and Hilton, 1955). Similarly, covering the soil with black film during the growth season lowered SGA levels in tubers at harvest with over 50%, depending on the cultivar (Nie et al., 2021). In addition, in line with our hypothesis is that the range of Magnum Bonum SGA levels detected in year 1986 compared well to those we here obtained by light exposure but were clearly different to those obtained by wounding ( Figure 2 ). The greening response of Magnum Bonum tubers after light exposure was weak ( Figure S4 ), the reason why a low light exposure of tubers might have been undetected by visual observation at harvest but still could have been able to increase SGA production.

However, other explanations, such as low-temperature stress (van Swaaij, 1992) and/or flooding/anoxia responses (Hellenäs et al., 1995), are also plausible. Interestingly, trancriptome analysis of tubers submerged in water (i.e., “flooded”) did not reveal any significant upregulations of SGA-associated genes (Peivastegan et al., 2019). Instead, the main SGA-related effect was a downregulation of SGT1 and SGT3 genes. Because the downregulation of SGT1 and SGT3 decreases SGA levels in transgenic plants (McCue et al., 2005; McCue et al., 2007), the results in the work of Peivastegan et al. (2019) indicate lowered SGA levels during flooding, rather than increased. The influence of low temperature or frost remains to be investigated.

Whatever the exact cause of high SGA levels in Magnum Bonum was in year 1986, our results demonstrate that Magnum Bonum has the strongest light responsiveness with respect to the SGA accumulation among more than 30 table potato cultivars that have investigated by us and by others. To what extent a strong light responsiveness also applies for other high-SGA cultivars, such as Lenape, is not known. Our results should stimulate further analyses of the influence of light and/or other conditions on SGA levels in table potato cultivars, as well as breeding lines, to identify the most sensitive genotypes for each condition.

Transcriptomic analyses identified a small number of SGA-associated key genes that were more strongly and significantly induced by light in Magnum Bonum than in Bintje. The most significant examples were HMGR1, SMO1-L, and TAM2, but 14 other genes were also shown to be important ( Figure 4 ). In addition, four genes were identified whose expression was negatively correlated to SGA levels, SMT1 being the most significant gene in this case. At present, we do not know what regulatory processes that might underly these cultivar differences in gene expression. The spectral response for the light induction of SGA levels in cv. King Edward potato tubers is consistent with a role for both cryptochrome (blue light) and phytochrome (red light) photoreceptors, similar to the parallel light induction of chlorophyll biosynthesis genes (Okamoto et al., 2020). In addition, in line with a role for the light-regulated gene expression machinery is that downregulation in tomato plants of the light signal transduction factors ELONGATED HYPOCOTYL 5 (SlHY5) and PHYTOCHROME-INTERACTING FACTOR 3 (SlPIF3) modulated expression of SGA biosynthetic genes and reduced SGA levels (Wang et al., 2018). Furthermore, a comparison of the SGA-associated genes in Figure 4 with those that have been shown to be correlated to the expression of the SGA transcription factor GAME9 in GAME 9-overexpressing potato transformants (Cárdenas et al., 2016) showed both similarities and differences. For instance, expression GAME11 (ODD) and GAME12 (TAM2) were elevated in potato GAME9-overexpressing transformants, but HMGR1 or SMO1-L was not reported as upregulated. It should, however, be noted that different tissues (leafs vs. tubers) were used in these two studies. We show that the GAME9 expression in tubers in itself is upregulated by wounding as well as light exposure, both in Bintje and Magnum Bonum ( Figure S11 ). However, the degree of wound- or light-induced GAME9 expression did not differ significantly between the cultivars, and their GAME9 amino acid composition was identical ( Figure S12 ), together indicating that GAME9-regulated expression or activity was not main parts of the between-cultivar SGA differences in our investigation.

The increased SGA levels in leafs from transgenic potato plants overexpressing StHMGR1 and StTAM2 indicate that induced expression of one single key SGA gene can be sufficient to increase SGA synthesis. The increased leaf SGA levels in our HMGR1-overexpessing transformants are well in agreement with other reports on HMGR1-overexpressing potatoes (Ginzberg et al., 2012; Moehninsi et al., 2020; Villano et al., 2020) and suggest a limiting function of the enzyme. The effect of TAM2 overexpression on SGA levels in table potatoes has not been investigated directly, although Nakayazu et al. (2021) showed that overexpression of the potato PGA4 cDNA (GAME12 and TAM2) could restore SGA production in hairy roots obtained from Solanum stipuloideum, a wild potato species lacking SGA due to a natural mutation in the PGA4 gene. Conversely, we have shown that overexpression of a soybean SMT1 cDNA in transgenic potato plants reduces SGA levels due to a decreased production of the SGA precursor cholesterol and increased levels of the non-precursors campesterol, sitosterol, and stigmasterol (Arnqvist et al., 2003). The downregulation of SMT1 expression by light exposure in all cultivars here investigated might thus be anticipated as having the opposite effect, i.e., increasing cholesterol synthesis. Well in line with this is that Arabidopsis SMT1 mutants have increased cholesterol levels (Diener et al., 2000) and that cholesterol levels increase during both wound- and light-induced SGA accumulation in potato tubers (Nahar et al., 2017). The role in SGA synthesis for the other three genes (SQE, HMGR3, and CAS/LAS) that were negatively correlated to SGA after light exposure is not clear, but also the expression profiles of these genes can be interpreted in terms of increasing cholesterol production through a reduced synthesis of competing steroidal substances.

These results are consistent with a situation where differences in SGA levels between potato cultivars depend on their individual expression or activity of genes/enzymes that act at key steps in both the pre- and post-cholesterol SGA metabolism, as well as on their activity of genes/enzymes regulating competing steroidal reactions. A model for the main SGA-biosynthetic differences between light-exposed Bintje and Magnum Bonum tubers is depicted in Figure S16 .