Vitamin D is a lipid-soluble sterol that plays an essential role in human health. Deficiency of this vitamin increases the risk of osteoporosis, hypertension, autoimmune diseases, infectious disease, diabetes and cancer. Vitamin D exists in two major forms: vitamin D₃ (cholecalciferol), mainly found in animal food source, and vitamin D₂ (ergocalciferol), typically present in sundried and ultraviolet-B (UV-B) exposed fungi and yeast (Jäpelt et al., 2013). Vitamin D₃ is produced in human skin upon sunlight exposure, where pro-vitamin D₃ (7-dehydrocholesterol; 7-DHC) is converted to vitamin D₃ by UV-B light (290–315 nm). Unfortunately, vitamin D₃ deficiency is common in both children and adults worldwide. Endogenous synthesis of vitamin D₃ in human skin is inhibited by several factors such as melanin presence, sunlight intensity, pollution and geographic location. Therefore, dietary sources are essential for maintaining consistent vitamin D₃ levels. Unfortunately, few dietary sources and supplements naturally contain vitamin D₃ and most of these are animal-based foods (e.g. meat and eggs), which raises concerns about vitamin D₃ levels among those populations that consume low amounts of animal products (Black et al., 2017).

Plants harbour an enormous reservoir of diverse steroidal molecules and, in principle, could be a source of vitamin D₃. However, although vitamin D₃ has been identified in some plants and algae, the levels are much lower compared to animal-based sources. The precursor of vitamin D₃, 7-DHC, is also the immediate precursor for cholesterol biosynthesis in plants (Figure 1a) (Sonawane et al., 2017). Since most plants produce cholesterol in very low amounts, 7-DHC levels are low as well. Notably, Solanaceae family members (e.g. tomato and Nicotiana benthamiana) accumulate naturally high levels of cholesterol. In tomato and other Solanum food crops such as potato and eggplant, cholesterol serves as a starting precursor for biosynthesis of defensive steroidal glycoalkaloids (SGAs) (Sonawane et al., 2017). Using the recently elucidated cholesterol pathway in plants along with gene editing strategies, it is now possible to engineer high levels of 7-DHC and therefore, vitamin D₃ in plants. Here, we report metabolic engineering approaches to enhance vitamin D₃ production in tomato (Solanum lycopersicum) and N. benthamiana plants.

The sterol 7-dehydrocholesterol (7-DHC) is an intermediate in the cholesterol biosynthetic pathway and is converted to vitamin D₃ upon sunlight exposure. Thus, overproduction of vitamin D₃ in plants requires a significant accumulation of 7-DHC. In Solanaceae plants, for example tomato, where high levels of cholesterol are produced, two related sterol-Δ⁷ reductase enzymes (7-dehydrocholesterol reductases; 7-DR1 and 7-DR2, 80% amino acid identity) were identified. Moreover, 7-DR1/DWARF5 catalyses the reduction of pathway intermediates (e.g. Δ^5,7-episterol) in phytosterols biosynthesis, while 7-DR2, which evolved from 7-DR1 through duplication and divergence, produces cholesterol from 7-DHC (Figure 1a) (Sonawane et al., 2017). Thus, the 7-DR2 gene is an ideal target for altering cholesterol metabolism and 7-DHC accumulation can be achieved by genome editing of 7-DR2 in tomato. Recently, Li et al. (2022) reported the production of vitamin D₃ in tomato (S. lycopersicum cv. Money Maker) by knocking out the 7-dr2 gene (7-dr2ko), achieving vitamin D₃ yields of 200 μg/g dry weight (DW) in leaves, 0.3 μg/g DW in green fruit and 0.2 μg/g DW in red fruit. We also generated the loss-of-function 7-dr2 mutants in two tomato varieties (cv. Money Maker and cv. Micro Tom) using CRISPR-Cas9 genome editing (Figures S1a and S2a). Loss of function of 7-dr2 had no visible effect on the growth, development and fruit yield of the mutant lines as compared to wild type (WT) plants (Figure S2b). Leaves and fruits from homozygous 7-dr2 mutant lines were analysed for 7-DHC and cholesterol metabolites using gas chromatography–mass spectrometry (GC–MS) (Figure 1B; Figures S1b,c,e,f and S2c,d). In both cultivars, the 7-dr2 mutant leaves and green fruits showed accumulation of 7-DHC, the expected product. Treatment of these tissues with UV-B light for 30 min with an intensity of 22 W/m² resulted in vitamin D₃ yields in Money Maker of 5.24 ± 1.2 μg/g of DW in leaves and 0.15 ± 0.01 μg/g in green fruit and no quantifiable levels in ripe fruit (Figure 1c; Figure S1d,g). In the previous study, Li et al. reported higher levels of vitamin D₃ in Money Maker mutant plant tissues, particularly in the leaves, despite targeting the same gene knockout. We hypothesize that this discrepancy could be attributable to differences in the UV-B treatment, which was performed for 1 h at an intensity of 30 W/m²—both longer and at a higher intensity compared to our treatment (30 min. at 22 W/m²). Yields in Micro Tom were 6.7 ± 1.73 μg/g DW in leaves, 0.38 ± 0.06 μg/g DW in green fruit and 0.27 ± 0.02 μg/g dry weight in ripe fruits (Figure 1C; Figure S2e–g). Therefore, while deletion of 7-dr2 leads to production of vitamin D₃, we concluded that alternative approaches to improve the levels of vitamin D₃ in plants could be explored.

We noted that the levels of cholesterol were significantly higher in leaves of 7-dr2 mutant lines compared to wild type, consistent with earlier reports (Figures S1b and S2c). It is likely 7-DR1, a close homologue of 7-DR2 that is responsible for production of brassinosteroids, may be compensating for the loss of 7-DR2. Therefore, we hypothesized that an enzyme that could convert cholesterol back to 7-DHC, for example a cholesterol 7-desaturase enzyme, could mitigate this effect. A cholesterol 7-desaturase enzyme has never been reported in any land species, but these enzymes are common in insects that convert dietary cholesterol to ecdysones (e.g. 20-hydroxyecdysone), steroid hormones crucial for insect development. In fact, the first step in ecdysone biosynthesis is conversion of cholesterol to 7-DHC. This irreversible step is catalysed by Neverland (NVD), an evolutionarily conserved oxygenase-like protein (Yoshiyama et al., 2006, 2011). Therefore, we decided to take an advantage of this unique enzyme for vitamin D₃ engineering by overexpressing a codon-optimized version of NVD (synthetic gene) in the 7-dr2 tomato mutant lines via stable genetic transformation.

We selected the silkworm Bombyx mori Neverland (BmNVD) enzyme that has been reported to catalyse the conversion of cholesterol to 7-DHC (Yoshiyama et al., 2006, 2011). Since expression of properly folded, active insect enzymes is challenging in heterologous plant hosts, we first tested the capacity of codon-optimized version of BmNVD enzyme in leaves of N. benthamiana, a substantial producer of cholesterol, by Agrobacterium tumefaciens-mediated transient expression. Metabolic profiling of the leaf extracts by GC–MS showed that transient expression of BmNVD clearly led to the accumulation of 7-DHC (Figure 1d; Figure S3). Moreover, treatment of transiently infiltrated leaves (after 3 days) with UV-B light indeed resulted in the production of vitamin D₃ (Figure 1d). Confocal microcopy analysis of the BmNVD:RFP infiltrated N. benthamiana disks showed that BmNVD is localized to the endoplasmic reticulum (ER) (Figure 1e). In plants, cholesterol is synthesized in the ER; therefore, co-localization of BmNVD enzyme in the same compartment likely allows the direct access to the cholesterol substrate, facilitating its conversion to 7-DHC. Altogether, these results demonstrate that the codon optimized BmNVD is active, and able to function normally in heterologous plant host system.

Inspired by these results, we next overexpressed BmNVD (codon optimized) in the 7-dr2ko mutant tomato (cv. Money Maker) plants (Figure S4a,b). The commercial variety Money Maker was selected for the appealing flavour of the fruit, which is in contrast to the Micro Tom, whose fruits are more bitter and therefore less suitable for consumption. Homozygous 7-dr2ko mutant lines overexpressing BmNVD (BmNVDOx + 7-dr2ko) were further analysed for altered steroidal metabolite profiles (Figure 1F; Figure S4c,d,f,g). Though high levels of 7-DHC were observed in leaves of BmNVDOx + 7-dr2ko transgenic lines compared to WT, we still noticed the significant accumulation of cholesterol in these lines (Figure S4c,f), suggesting compensatory activity from 7-DR1, involved in brassinosteroid biosynthesis. Subsequent UV-B treatment of leaves and green fruits of BmNVDOx + 7-dr2ko genotype resulted in ~3 and ~ 5-fold increase in vitamin D₃ levels, respectively as compared to the ones produced by 7-dr2ko mutant alone (18 ± 2.1 Vs 5.1 ± 1.2 μg/g DW in leaves and 0.76 ± 0.14 Vs 0.15 ± 0.01 μg/g DW in green fruit) (Figure 1C; Figure S4e,h). As observed earlier with the 7-dr2ko mutant lines, treatment of red fruits from BmNVDOx + 7-dr2ko plants did not yield any quantifiable amount of vitamin D₃. The levels of vitamin D₃ produced in tomato waste material, for example the leaves, offers a promising source for plant-derived vitamin D₃ supplements and can easily cover the recommended daily intake of vitamin D₃, which ranges between 10 and 20 μg depending mainly on age. In summary, our findings contribute to the exploration of innovative methods for vitamin D₃ biofortification, addressing global deficiencies and improving accessibility via a sustainable, plant-based and cost-effective platform.

维生素D是一种脂溶性甾醇，对人类健康起着重要作用。缺乏这种维生素会增加患骨质疏松症、高血压、自身免疫性疾病、传染病、糖尿病和癌症的风险。维生素 D 以两种主要形式存在：维生素 D ₃ （胆钙化醇），主要存在于动物食品来源中；维生素 D ₂ （麦角钙化醇），通常存在于晒干且暴露于紫外线 B (UV-B) 的真菌和酵母中（Jäpelt等）等， 2013 ）。维生素 D ₃是在阳光照射下人体皮肤中产生的，其中维生素 D ₃原（7-脱氢胆固醇；7-DHC）通过 UV-B 光（290-315 nm）转化为维生素 D ₃ 。不幸的是，维生素 D ₃缺乏症在全世界儿童和成人中都很常见。人体皮肤中维生素 D ₃的内源合成受到多种因素的抑制，例如黑色素的存在、阳光强度、污染和地理位置。因此，膳食来源对于维持稳定的维生素 D ₃水平至关重要。不幸的是，很少有膳食来源和补充剂天然含有维生素 D ₃ ，其中大多数是动物性食品（例如肉和蛋），这引起了对那些消耗少量动物产品的人群中维生素 D ₃水平的担忧（Black等人） ., 2017 ).

植物蕴藏着大量不同的类固醇分子，原则上可能是维生素 D ₃的来源。然而，尽管在一些植物和藻类中发现了维生素 D ₃ ，但与动物来源相比，其含量要低得多。维生素 D ₃的前体 7-DHC 也是植物中胆固醇生物合成的直接前体（图 1a）（Sonawane等人， 2017 ）。由于大多数植物产生的胆固醇含量非常低，因此 7-DHC 水平也很低。值得注意的是，茄科成员（例如番茄和本塞姆氏烟草）自然积累高水平的胆固醇。在番茄和其他茄属粮食作物（如马铃薯和茄子）中，胆固醇是防御性甾体糖生物碱（SGA）生物合成的起始前体（Sonawane等， 2017 ）。利用最近阐明的植物中胆固醇途径以及基因编辑策略，现在可以在植物中设计高水平的 7-DHC 和维生素 D ₃ 。在此，我们报告了提高番茄 ( Solanum lycopersicum ) 和N. Benthamiana植物中维生素 D ₃产量的代谢工程方法。

甾醇 7-脱氢胆固醇 (7-DHC) 是胆固醇生物合成途径中的中间体，在阳光照射下会转化为维生素 D ₃ 。因此，植物中维生素 D ₃的过量产生需要大量积累 7-DHC。在产生高水平胆固醇的茄科植物（例如番茄）中，鉴定出两种相关的甾醇-Δ ⁷还原酶（7-脱氢胆固醇还原酶；7-DR1 和 7-DR2，80% 氨基酸同一性）。此外，7-DR1/DWARF5 催化植物甾醇生物合成中途径中间体（例如 Δ ^5,7 -episterol）的减少，而 7-DR2 从 7-DR1 通过复制和分化进化而来，从 7-DHC 产生胆固醇（图1a)（Sonawane等人， 2017 ）。因此， 7-DR2基因是改变胆固醇代谢的理想靶标，并且可以通过番茄中7-DR2的基因组编辑来实现7-DHC积累。最近，李等人。 ( 2022 ) 报道了通过敲除7-dr2基因 ( 7-dr2ko ) 在番茄 ( S. lycopersicum cv . Money Maker) 中生产维生素 D ₃ ，实现了 200 μg/g 干重 (DW) 的维生素 D ₃产量叶子中，绿色果实中 0.3 μg/g DW，红色果实中 0.2 μg/g DW。我们还使用 CRISPR-Cas9 基因组编辑在两个番茄品种（ cv . Money Maker 和cv . Micro Tom）中生成了功能丧失的7-dr2突变体（图 S1a 和 S2a）。 与野生型 (WT) 植物相比， 7-dr2功能丧失对突变株系的生长、发育和果实产量没有明显影响（图 S2b）。使用气相色谱-质谱法 (GC-MS) 分析纯合 7 -dr2突变株系的叶子和果实中的 7-DHC 和胆固醇代谢物（图 1B；图 S1b、c、e、f 和 S2c、d）。在这两个品种中， 7-dr2突变体的叶子和绿色果实都显示出 7-DHC（预期产物）的积累。用强度为 22 W/m ²的 UV-B 光处理这些组织 30 分钟，Money Maker 中维生素 D ₃的产量为 5.24 ± 1.2 μg/g DW（叶子）和 0.15 ± 0.01 μg/g（绿色）果实中没有可量化的水平（图 1c；图 S1d，g）。在之前的研究中，Li等人。据报道，尽管针对相同的基因敲除，但 Money Maker 突变植物组织中的维生素 D ₃水平较高，特别是在叶子中。我们假设这种差异可能是由于 UV-B 处理的差异造成的，该处理以 30 W/m ²的强度进行 1 小时，与我们的处理（22 小时 30 分钟）相比，时间更长且强度更高。瓦/米² )。 Micro Tom 中叶片的产量为 6.7 ± 1.73 μg/g DW，绿色果实的产量为 0.38 ± 0.06 μg/g DW，成熟果实的干重为 0.27 ± 0.02 μg/g（图 1C；图 S2e-g）。因此，虽然删除7-dr2会导致维生素 D ₃的产生，但我们得出结论，可以探索提高植物中维生素 D ₃水平的替代方法。

我们注意到，与野生型相比， 7-dr2突变株系的叶子中胆固醇水平显着升高，这与之前的报道一致（图 S1b 和 S2c）。 7-DR1（7-DR2 的密切同源物，负责产生油菜素类固醇）可能会补偿 7-DR2 的损失。因此，我们假设一种可以将胆固醇转化回 7-DHC 的酶，例如胆固醇 7-去饱和酶，可以减轻这种影响。胆固醇7-去饱和酶从未在任何陆地物种中被报道过，但这些酶在昆虫中很常见，它们将膳食胆固醇转化为蜕皮激素（例如20-羟基蜕皮激素），这是对昆虫发育至关重要的类固醇激素。事实上，蜕皮激素生物合成的第一步是将胆固醇转化为 7-DHC。这一不可逆的步骤由梦幻岛（NVD）催化，这是一种进化上保守的加氧酶样蛋白（Yoshiyama等， 2006，2011 ）。因此，我们决定利用这种独特的酶进行维生素 D ₃工程，通过稳定的遗传转化在7-dr2番茄突变株系中过表达密码子优化版本的 NVD（合成基因）。

我们选择了家蚕Bombyx mori Neverland (BmNVD) 酶，据报道该酶可催化胆固醇转化为 7-DHC (Yoshiyama et al ., 2006 , 2011 )。由于正确折叠的活性昆虫酶在异源植物宿主中的表达具有挑战性，因此我们首先通过根癌农杆菌介导的瞬时表达测试了本塞姆氏烟草（胆固醇的大量生产者）叶子中密码子优化版本的 BmNVD 酶的能力。通过 GC-MS 对叶子提取物的代谢分析表明，BmNVD 的瞬时表达明显导致 7-DHC 的积累（图 1d；图 S3）。此外，用 UV-B 光处理短暂渗透的叶子（3 天后）确实导致了维生素 D ₃的产生（图 1d）。 BmNVD:RFP 浸润的本塞姆氏烟草盘的共聚焦显微分析表明， BmNVD定位于内质网 (ER)（图 1e）。在植物中，胆固醇是在内质网中合成的；因此，BmNVD 酶在同一区室中的共定位可能允许直接接触胆固醇底物，促进其转化为 7-DHC。总而言之，这些结果表明密码子优化的 BmNVD 具有活性，并且能够在异源植物宿主系统中正常发挥作用。

受这些结果的启发，我们接下来在7-dr2ko突变番茄（ cv . Money Maker）植物中过表达BmNVD （密码子优化）（图 S4a、b）。选择商业品种 Money Maker 是因为其水果具有诱人的风味，这与 Micro Tom 形成鲜明对比，Micro Tom 的水果更苦，因此不太适合食用。进一步分析过度表达BmNVD ( BmNVDOx + 7-dr2ko ) 的纯合7-dr2ko突变株系的类固醇代谢物谱变化 (图 1F;图 S4c,d,f,g)。尽管与 WT 相比，在BmNVDOx + 7-dr2ko转基因品系的叶子中观察到高水平的 7-DHC，但我们仍然注意到这些品系中胆固醇的显着积累（图 S4c，f），表明 7-DR1 的补偿活性涉及在油菜素类固醇生物合成中。随后对BmNVDOx + 7-dr2ko基因型的叶子和绿色果实进行 UV-B 处理，与单独使用7-dr2ko突变体产生的水平相比，维生素 D ₃水平分别增加了约 3 倍和约 5 倍 (18 ± 2.1叶子中的 DW为5.1 ± 1.2 μg/g，绿色水果中的 DW 为 0.76 ± 0.14 Vs 0.15 ± 0.01 μg/g DW）（图 1C；图 S4e，h）。正如之前对7-dr2ko突变株系所观察到的，对来自BmNVDOx + 7-dr2ko植物的红色果实的处理没有产生任何可定量的维生素 D ₃ 。 番茄废料​​（例如叶子）中产生的维生素 D ₃水平为植物源性维生素 D ₃补充剂提供了有前景的来源，并且可以轻松满足维生素 D ₃的每日推荐摄入量（根据具体情况而定，其范围在 10 至 20 μg 之间）主要看年龄。总之，我们的研究结果有助于探索维生素 D ₃生物强化的创新方法，解决全球性的缺陷，并通过可持续、植物性且具有成本效益的平台提高可及性。