Overexpression of SrDXS1 and SrKAH enhances steviol glycosides content in transgenic Stevia plants

The sweetness in your zero-calorie soda probably came from a leaf — a specific leaf from a plant called Stevia rebaudiana. Its active compounds are roughly three hundred times more intense than sugar by weight. The plant makes those compounds naturally, but the problem is it doesn't produce enough of them. For decades, there was no reliable way to delve into Stevia's genetics and change that. That bottleneck is exactly what Zheng and colleagues set out to crack. The molecules responsible for Stevia's sweetness are called steviol glycosides. They are built in the plant's leaves through a branching biochemical assembly line that starts in the chloroplast. The first significant step is the MEP pathway, short for methylerythritol phosphate, which assembles the basic five-carbon building blocks that feed into isoprenoid biosynthesis — the broad chemical family that includes everything from plant pigments to hormones. An enzyme called SrDXS1, or one-deoxy-D-xylulose-five-phosphate synthase 1, catalyzes the entry point of that pathway, generating the precursors that flow downstream. Several steps later, another enzyme, SrKAH, or kaurenoic acid hydroxylase, converts kaurenoic acid into steviol, the core scaffold onto which sugar units are attached to create the final glycosides. Those sugars are added by a family of enzymes called UDP-glycosyltransferases. One of them, SrUGT76G1, specifically converts stevioside into rebaudioside A, the glycoside with the cleanest, most pleasant taste profile. The commercial case for engineering these enzymes is clear: if you can push more carbon through the pathway, you get more sweetener per leaf. But this project hit a wall. To overexpress a gene in a plant, you first need a way to stably introduce DNA into that plant's cells and then coax those cells into becoming whole, fertile plants. For most model species, Agrobacterium-mediated transformation — using a soil bacterium to deliver DNA into plant tissue — is routine, but for Stevia, it wasn't. Several groups had tried, and none had produced a broadly reproducible protocol. As Zheng and colleagues put it, a reliable Stevia transformation technology remains to be developed. The bottleneck wasn't the gene delivery step itself; it was what came after. Leaf tissue could be coaxed into forming callus, an undifferentiated mass of dividing cells, but getting that callus to regenerate into actual shoots was inconsistent and poorly understood. So, before any metabolic engineering could happen, the team had to solve a cell biology problem. They started with second and third leaves from in vitro cultured Stevia plants and systematically tested combinations of two plant hormones — cytokinin and auxin — across six different media conditions. The baseline cytokinin was six-benzylaminopurine at one milligram per liter. The auxin varied; some conditions used naphthaleneacetic acid while others used indoleacetic acid. Callus formation was high across nearly all conditions, ranging from eighty-seven to ninety-nine percent of explants, so that wasn't the variable that mattered. Shoot regeneration was. And here the differences were stark. Conditions containing naphthaleneacetic acid produced poor shoot regeneration, while conditions using indoleacetic acid did much better. The best single result came from Condition D, which yielded sixty-five point eight percent of explants with regenerated shoots. A further optimized version, Condition F, maintained regeneration near fifty-three percent while producing healthier, more numerous shoots per callus clump. Then came the unexpected insight that made the whole system work. When the team kept explants in complete darkness during callus induction and shoot regeneration, they observed a striking improvement. The shoots grew pale and elongated — etiolated, in botanical terms, like seedlings that have been deprived of light. More importantly, the proportion of explants that successfully regenerated shoots jumped to fifty-three percent under dark conditions, compared to just twenty-nine point five percent under normal light. That's nearly a twofold difference just from turning off the lights. The darkness served a second purpose, and this one was almost elegantly practical. To identify which plants had actually incorporated the foreign DNA, the team used green fluorescent protein as a visual reporter — a gene that makes cells glow green under fluorescence microscopy. The problem with using green fluorescent protein in plant tissue is that chlorophyll also fluoresces, drowning out the signal. Etiolated shoots, grown in the dark, have not developed their chlorophyll yet, leading to a quiet background. The green glow comes through clearly. By combining the optimized hormone regime with prolonged dark incubation, the team gained both better regeneration and a clean visual screen for genuine transformants — a two-for-one that turned an unreliable process into a reproducible one. Molecular validation confirmed it worked. Across the lines they produced — thirteen overexpressing SrDXS1 and nine overexpressing SrKAH — genomic PCR detected the expected transgene fragments in every line and not in wild-type controls. Southern blotting with an nptII probe confirmed stable integration. Transcript measurements showed the transgenes were actively expressed: SrDXS1 lines showed up to a thirteen-fold increase in SrDXS1 messenger RNA, while high-expressing SrKAH lines showed roughly forty to sixty-fold higher transcript levels than wild-type. Now for the results that matter commercially. In leaves of the SrDXS1-overexpressing lines, total steviol glycoside content reached as high as five point nine percent of dry weight, compared to three point eight percent in the vector-only control — an increase of up to fifty-four percent. The gains tracked with expression level: lines with higher SrDXS1 transcripts accumulated more glycosides, demonstrating a dose-response relationship that strengthens the interpretation that the enzyme itself is the bottleneck being relieved. Overexpressing SrKAH pushed the numbers further. The highest-expressing lines reached six point zero percent total steviol glycosides by dry weight, increasing by sixty-seven to eighty-eight percent over the vector-only control. Again, the dose-response was evident: the low-expressing SrKAH line, with only about a four-fold transcript increase, showed only modest gains, while the high expressers drove the largest accumulations. The larger boost from SrKAH, relative to SrDXS1, is consistent with where it sits in the pathway. SrDXS1 acts at the very start of the MEP pathway, supplying precursors that flow into many different isoprenoid products. SrKAH acts directly upstream of steviol, specifically pulling carbon into the glycoside branch. The more targeted the intervention, the more directly the flux goes where desired. The profile of individual glycosides shifted too, especially in the SrKAH lines. Stevioside rose by fifty-seven to seventy-one percent in the high expressers. Rebaudioside A, the more pleasant-tasting compound, rose by one hundred thirty-three to two hundred percent in the highest-expressing SrKAH line — a disproportionate gain suggesting that the overexpression is pushing flux not just into total glycosides but specifically toward the more desirable downstream compounds. A standard concern with metabolic engineering is that forcing a pathway to run harder will rob resources from elsewhere, stunting the plant or shifting other metabolite pools in unpredictable ways. That didn't happen here. Both sets of transgenic plants were morphologically indistinguishable from wild-type and vector-only controls — they had the same height, same leaf size, and same internode length at two months of age. Because the MEP pathway also supplies chlorophylls and carotenoids, the team directly measured those. Chlorophyll A, chlorophyll B, and total carotenoids were unchanged in nearly all lines. Gas chromatography and mass spectrometry analysis found no significant shifts in linalool, alpha-pinene, or beta-pinene either. The glycoside gains were real, and they occurred without measurable cost to the plant. Zheng and colleagues frame the takeaway explicitly: the transformation protocol itself is a tool, independent of the specific genes tested here. The same system can be used for knockdown studies — silencing genes to investigate what happens when pathway steps are removed — and for overexpressing other enzymes, particularly the glycosyltransferases that control which glycosides accumulate. The paper specifically highlights rebaudioside D and rebaudioside M as targets. These are minor glycosides present in tiny amounts in natural Stevia, but they have superior taste profiles and are increasingly valuable. Engineering their production at scale requires exactly the kind of handle this work provides. For a plant that spent decades being used commercially but resisted genetic manipulation, this is the moment metabolic engineers finally got a lever — and the first results suggest it works. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.