A cytochrome P450 CYP87A4 imparts sterol side-chain cleavage in digoxin biosynthesis

Carroll, Emily; Ravi Gopal, Baradwaj; Raghavan, Indu; Mukherjee, Minakshi; Wang, Zhen Q.

doi:10.1038/s41467-023-39719-4

Download PDF

Article
Open access
Published: 08 July 2023

A cytochrome P450 CYP87A4 imparts sterol side-chain cleavage in digoxin biosynthesis

Nature Communications volume 14, Article number: 4042 (2023) Cite this article

5377 Accesses
8 Citations
36 Altmetric
Metrics details

Subjects

Abstract

Digoxin extracted from the foxglove plant is a widely prescribed natural product for treating heart failure. It is listed as an essential medicine by the World Health Organization. However, how the foxglove plant synthesizes digoxin is mostly unknown, especially the cytochrome P450 sterol side chain cleaving enzyme (P450_scc), which catalyzes the first and rate-limiting step. Here we identify the long-speculated foxglove P450_scc through differential transcriptomic analysis. This enzyme converts cholesterol and campesterol to pregnenolone, suggesting that digoxin biosynthesis starts from both sterols, unlike previously reported. Phylogenetic analysis indicates that this enzyme arises from a duplicated cytochrome P450 CYP87A gene and is distinct from the well-characterized mammalian P450_scc. Protein structural analysis reveals two amino acids in the active site critical for the foxglove P450_scc’s sterol cleavage ability. Identifying the foxglove P450_scc is a crucial step toward completely elucidating digoxin biosynthesis and expanding the therapeutic applications of digoxin analogs in future work.

Promiscuous CYP87A enzyme activity initiates cardenolide biosynthesis in plants

Article Open access 18 September 2023

Identification of a novel enzyme from E. pacifica that acts as an eicosapentaenoic 8R-LOX and docosahexaenoic 10R-LOX

Article Open access 26 November 2020

Effective prediction of biosynthetic pathway genes involved in bioactive polyphyllins in Paris polyphylla

Article Open access 13 January 2022

Introduction

Cardiac glycosides extracted from the foxglove plant Digitalis lanata have been used for treating congestive heart failure since 1785¹. Digoxin, a widely prescribed cardiac glycoside, is listed as an essential medicine by the World Health Organization². About 400,000 patients are prescribed digoxin in the United States, making it one of the most prescribed plant natural products³. Recent research has broadened the medicinal applications of cardiac glycosides for treating viral infection, inflammation, cancer, hypertension, and neurodegenerative diseases^{4,5,6,7,8,9,10,11}.

Due to the prominence of digoxin in medicine, the study of cardiac glycoside biosynthetic pathways dates back to the 1960s. Radiolabeling studies suggested cholesterol as the precursor for digoxin¹². While this is generally accepted, controversies remain since cholesterol is a minor sterol in plants. The exact biosynthetic pathway of digoxin remains enigmatic half a century after the initial work¹³. The hypothetical cardiac glycoside biosynthetic pathway starts with cholesterol, which undergoes nine enzyme-catalyzed steps to digoxigenin, the aglycone of digoxin¹³. Currently, the only known enzymes in the pathway are 3β-hydroxysteroid dehydrogenase (3βHSD) and progesterone-5β-reductase (P5βR and P5βR2)^14,15,16. The first and rate-limiting enzyme, cytochrome P450 sterol side chain cleaving enzyme (P450_scc), along with all other enzymes, has not been identified yet^13,17. The foxglove P450_scc is thought to convert cholesterol to pregnenolone through a reaction identical to mammalian P450_scc, catalyzing the rate-limiting step in animal steroid hormone synthesis¹⁸. However, the plant P450_scc has not been isolated and characterized since its first description by Pilgrim in 1972¹⁹. Hence, the direct sterol precursor for digoxin biosynthesis remains ambiguous. Indirect evidence suggests that phytosterols, including campesterol, stigmasterol, and sitosterol, may also be precursors for digoxin^{12,19,20,21,22,23}.

In this study, we utilized a high-quality transcriptome of D. lanata to identify the foxglove P450_scc. Characterizing the foxglove P450_scc validated its sterol cleaving activity in tobacco and yeast. Investigating the foxglove P450_scc substrate preference uncovered the identity of sterol precursors for digoxin biosynthesis. Phylogenetic analysis suggests that this enzyme evolved from the CYP87A family. Protein modeling and mutagenesis revealed critical amino acids for foxglove P450_scc’s sterol-cleaving activity. The foxglove P450_scc is the first plant P450_scc identified and does not share substantial homology with the animal P450_scc.

Results

Transcriptome assembly and annotation

Total RNA from leaf and root tissues, including three biological replicates and two technical replicates from each tissue, were pooled to generate a reference transcriptome of D. lanata. We performed de novo assembly of the transcriptome from 173,448,870 Illumina raw reads with an average length of ~100 bp. The assembled transcriptome contains 317,983 transcripts with an N50 of 1712 bp (Table 1, Supplementary Fig. 1). The Benchmarking Universal Single-Copy Orthologs (BUSCO) score for the transcriptome was 94.6%, indicating that the transcriptome was near complete. A total of 190,755 transcripts at least 300-bp long were annotated using publicly available databases, including the NCBI non-redundant protein database (nr) and the UniProt database, each annotated 84.6% and 48.3% of the 190,755 transcripts, respectively^24,25. The transcripts were found to match best with genes of other Lamiales species, including Sesamum indicum, which covered 75.3% of the transcriptome (Supplementary Fig. 1). 113,221 non-redundant unigenes were categorized by gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway classification^26,27 (Supplementary Fig. 2). KEGG analysis revealed 5,683 unigenes in 412 KEGG pathways, among which were pathways for terpenoid and steroid biosynthesis (Supplementary Fig. 3). UniProt annotation identified 7517 transcription factors and regulators, 4226 protein kinases, and 22,549 simple sequence repeats (SSR) as genetic markers (Table 1, Supplementary Fig. 4, Supplementary Table 1). The annotated transcriptome presented here provides a comprehensive representation of transcripts in the root and leaf tissues of D. lanata.

Table 1 Summary of Transcriptome data

Full size table

Genes for sterol biosynthesis are differentially expressed

Since cardiac glycosides are only present in leaves but not roots (Fig. 1a), we asked if phytosterol and cholesterol biosynthetic genes are overexpressed in leaves. Indeed, genes encoding rate-limiting enzymes in phytosterol and cholesterol pathways were overexpressed in leaves (Fig. 1b). Squalene epoxidase (SQE), a rate-limiting step in sterol biosynthesis, showed higher relative transcript abundance in leaves²⁸. Sterol side-chain reductase (SSR1) is a known bottleneck enzyme²⁹ in cholesterol and phytosterol biosynthesis. Its transcript is also more abundant in D. lanata leaves. C4 sterol methyl oxidase 3 (SMO3), unique to the cholesterol pathway, is also more abundant in leaves. It catalyzes the rate-limiting step of 4-methyl elimination in the cholesterol pathway²³. Indeed, D. lanata leaves have higher cholesterol levels than roots, whereas the total sterols in these two tissues are comparable (Supplementary Fig. 5). Another gene with higher transcript abundance in leaves is the sterol C-14 reductase (C14-R)³⁰, a shared enzyme between phytosterol and cholesterol pathways.

**Fig. 1: Quantifying the relative transcript abundance of genes in the sterol and digoxin biosynthetic pathways.**

Analysis of the three known genes involved in digoxin biosynthesis shows that only 3βHSD’s transcript was more abundant in leaves. While P5βR’s relative transcript abundance is the same in both tissues, P5βR2’s transcript is more abundant in roots. Since digoxin and sterols are triterpene derivatives, we also analyzed the differential relative transcript abundance of terpenoid biosynthetic genes (Supplementary Fig. 6). The methylerythritol phosphate (MEP) pathway for terpenoid synthesis and triterpene pathways are induced in leaves, agreeing with the compartmentalization of the MEP pathway in the chloroplast³¹.

Identifying two candidate genes as D. lanata P450 _scc

The first step of the digoxin pathway is cleaving a sterol by a cytochrome P450_scc to generate pregnenolone¹³. Interpro scan identified 438 enzymes annotated as cytochrome P450s (CYPs) (Pfam: PF00067)^32,33 from the transcriptome. CYPs from Arabidopsis and CYPs in D. lanata were used to construct a phylogenetic tree for CYP subfamily classification (Supplementary Fig. 7). Quantifying relative transcript abundance identified 104 CYP transcripts overexpressed in the leaves (Supplementary Fig. 8). Among these CYPs, only those members of subfamilies relevant to sterol/brassinosteroid biosynthesis were included for future analysis. Thirteen full-length CYP transcripts were identified as potential P450_scc (Fig. 2a). We focused on DlCYP87A4 and DlCYP90A1 because DlCYP87A4 was highly induced in leaves, and CYP90A1 is known to oxidize the 22(S)-hydroxycampesterol³⁴. qRT-PCR confirmed that DlCYP87A4 was expressed much higher in leaves compared to DlCYP90A1 (Fig. 2b). Therefore, these two transcripts were identified as P450_scc candidates and cloned from cDNA for functional validation by tobacco transient expression assay.

**Fig. 2: Identifying and characterizing candidate *D. lanata* P450_sccs in tobacco.**

Tobacco expression identified D. lanata P450_scc as CYP87A4

To test the two candidates, we employed the tobacco transient expression experiment. Tobacco does not produce digoxin but has sterol substrates for the P450_scc³⁵. Therefore, it is an ideal system for functionally characterizing the P450_scc enzyme. The two candidates, DlCYP87A4 and DlCYP90A1, and the two known pathway enzymes, 3βHSD and P5βR, were expressed in tobacco leaves (Fig. 2c, set 1). Following their expression, products of these enzymes, including progesterone (compound 2), 5β-pregnane-3,20-dione (compound 3), and 3β-hydroxy-5β-pregnane-20-one (compound 4), were detected (Fig. 2c, set 1, Supplementary Fig. 9). The direct product of P450_scc, pregnenolone (compound 1), was not seen in set 1 potentially due to its quick turnover through 3βHSD. Note that the minor peak is not pregnenolone due to the different retention time compared to the pregnenolone standard. In fact, D. lanata leaves do not produce detectable amounts of pregnenolone but produce the downstream pathway intermediates (Supplementary Fig. 10). Omitting the DlCYP87A4 abolished the reactions (set 2), whereas taking out the DlCYP90A1 (set 3) had no effect. Expressing DlCYP87A4 alone resulted in the production of pregnenolone (set 4), whereas expressing DlCYP90A1 alone (set 5) did not produce pregnenolone. These data strongly support the hypothesis that DlCYP87A4 is the P450_scc of the digoxin pathway.

Determining the sterol substrates of CYP87A4

The tobacco expression system cannot determine the sterol substrate of D. lanata P450_scc because tobacco contains a mixture of cholesterol and phytosterols. Thus, we turned to the in vivo yeast expression since yeast does not produce cholesterol or phytosterols. However, feeding yeast with different sterols is challenging due to their hydrophobicity. Therefore, we used previously engineered yeast strains that produce various sterols, including cholesterol, campesterol, 7-dehydrocholesterol, and desmosterol (Fig. 3a, Supplementary Fig. 11)³⁵. We also included a wildtype yeast that produces ergosterol. When expressing the D. lanata CYP87A4 and Arabidopsis thaliana cytochrome P450 reductase 2 (ATR2) as a redox partner, the yeast strains containing campesterol and cholesterol, respectively, produced pregnenolone (Fig. 3b, Supplementary Fig. 12). Pregnenolone is toxic to yeast; thus, yeast acetylates pregnenolone to detoxify it, generating pregnenolone acetate (compound 5) as a byproduct³⁶. We also expressed the human P450_scc (CYP11A1) with its redox partners in these yeast strains. Only the cholesterol-producing yeast generated pregnenolone, as expected (Fig. 3b)¹⁸. These data indicate that both campesterol and cholesterol are substrates of D. lanata CYP87A4, which identifies as the DlP450_scc.

**Fig. 3: Characterizing the *D. lanata* CYP87A4 in *S. cerevisiae* strains producing various sterols.**

Neofunctionalization of CYP87A4 unique to Digitalis

To understand the evolutionary history of the D. lanata CYP87A4, we constructed a phylogenetic tree with transcripts homologous to the DlCYP87A4 from the 1000 transcriptome project (Fig. 4)³⁷. Four transcripts in the D. lanata transcriptome fall into the CYP87A subfamily. DlCYP87A1 and DlCYP87A2 are 72.2 and 74.2% identical to the characterized DlCYP87A4 (Supplementary Table 2). DlCYP87A3 is 97.4% identical at the protein level to the characterized DlCYP87A4 but does not cleave the side chain of campesterol when expressed in yeast (Supplementary Fig. 14). DlCYP87A1 and DlCYP87A2 likely represent the canonical enzymes of the CYP87A subfamily. Indeed, DlCYP87A1 and DlCYP87A2 are expressed almost constitutively in leaves and roots (Fig. 2a). DlCYP87A4 may be duplicated from a canonical CYP87A and neofunctionalized to gain its sterol cleaving activity. The DlCYP87A4 is distinct from the human P450_scc as these two proteins only share 29.8% identical amino acids (Supplementary Fig. 13). None of the other Lamiales in the 1000 transcriptome project had duplicates in the CYP87A subfamily (Fig. 4). Species in the Oenothera genus also have multiple copies of CYP87A, but their function is unclear since these species are not known to produce cardiac glycosides.

**Fig. 4: Maximum-likelihood phylogenetic tree of the CYP87A family members from eudicot plants.**

We included CYP87A transcripts from other plants that produce cardiac glycosides since they would have an enzyme with a similar sterol cleaving function^10,13. We searched the publicly available transcriptomes of Digitalis purpurea, Calotropis gigantea, and Asclepias syriaca for transcripts that were a close match to the DlCYP87A4. A. syriaca did not have any transcripts that matched over 55% at the protein level with DlCYP87A4. C. gigantea had one transcript, which matched 69% to DlCYP87A4 (Fig. 4, Supplementary Table 2). The C. gigantea CYP87A is likely the canonical CYP87A enzyme since there is only one copy. The D. purpurea transcriptome had one transcript that matched 97% to the DlCYP87A4. Thus, this transcript likely has the same sterol cleaving ability. Our analysis indicates that the expansion and neofunctionalization of DlCYP87A as P450_scc is probably unique to the Digitalis species.

Identify amino acids critical for D. lanata P450_scc’s function

To gain mechanistic insights into DlCYP87A4’s sterol cleaving ability, a protein model was created using AlphaFold2. Campesterol and cholesterol were docked to the active site of the protein model (Fig. 5a, b). Aligning canonical CYP87As and DlCYP87A4’s protein sequences identified three unique amino acids in the active site of DlCYP87A4: S123, A355, and L357. These amino acids are conserved between the DlCYP87A4 and the putative D. purpurea P450_scc (DpCYP87A) but differ from the canonical CYP87As (Fig. 5c), suggesting that they are important for the sterol cleaving activity. Reverting A355 to leucine or L357 to alanine, as in the canonical DlCYP87A1, abolished the campesterol side chain-cleaving activity, whereas the S123A mutation had no effect (Fig. 5d). A355 and L357 likely stabilize the steroid by forming hydrophobic interactions with the four steroid rings (Fig. 5a, b). However, these two amino acids are insufficient to impart the sterol side-chain-cleaving activity as the canonical DlCYP87A1 mutated with these two amino acids was unable to cleave campesterol (Supplementary Fig. 14). The wildtype canonical DlCYP87A1 could not cleave campesterol, suggesting it is not involved in digoxin biosynthesis (Fig. 5e).

**Fig. 5: Protein modeling identified critical amino acids for DlCYP87A4’s sterol cleaving activity.**

Discussion

We identified and characterized the first and rate-limiting enzyme in the plant cardiac glycoside biosynthetic pathway, P450_scc, which has long been speculated but not found before. We used differential transcriptomic analysis to identify that a CYP87A family protein acts as the P450_scc in Digitalis. This protein is distinct in sequence from its mammalian counterpart, CYP11A1. The similarities and differences between the mammalian and plant P450_scc indicate that the “cholesterol side-chain-cleaving” activity evolved independently and serves distinct functions. While mammalian P450_sccs for steroid hormone biosynthesis are essential for the normal development of animals, plant P450_sccs evolved for synthesizing specialized metabolites, such as cardiac glycosides, which are unique to very specific plant families.

The DlP450_scc is a crucial “gatekeeping” enzyme that connects plant primary and secondary metabolisms. It channels sterols essential for maintaining cell membrane homeostasis to produce cardiac glycosides, secondary metabolites important for defense²³. Such an enzyme acting as the “gatekeeper” for specialized metabolism is not surprising. The rate-limiting nature of DlP450_scc was evident because feeding cholesterol to Digitalis produced a trace amount of pregnenolone, whereas administering progesterone increased various pregnane intermediates¹⁷. Unlike P450_scc in animals, DlP450_scc is promiscuous as it catalyzes the side-chain-cleaving reaction for cholesterol and campesterol (Fig. 3). This promiscuity is somewhat expected since campesterol is one of the major sterols in plants. In order to test if the other major plant sterols, β-sitosterol and stigmasterol, may also serve as substrates for DlP450_scc, we performed a docking simulation (Fig. 6). While the C20 and C22 of campesterol, cholesterol, and β-sitosterol are within 4.6–5.6 Å to the heme center, docking with stigmasterol put these two carbons over 7 Å away from the heme. The 22:23 double bond of stigmasterol prevents bond rotation resulting in the bulky 22-ethyl group pointing towards the heme center, preventing C20 and C22 from accessing the heme. Thus, we surmise that β-sitosterol could also be DlP450scc’s substrate, along with cholesterol and campesterol, but not stigmasterol. The discovery of DlP450_scc as a promiscuous protein will likely end the half-a-century controversy over the sterol precursor for digoxin.

**Fig. 6: Docking cholesterol and phytosterols to the active center of the DlP450_scc.**

Future work is necessary to understand if the DlP450_scc acts by the exact catalytic mechanism as the mammalian P450_scc, which catalyzes three-step sequential oxidations through 22-hydroxylation, 20-hydroxylation, and cleavage between C20 and C22¹⁸. Previous in vitro assay using 20- or 22-hydroxycholesterol as substrates support this mechanism¹⁷. We show that A355 and L357 are essential for DlP450_scc’s activity (Fig. 5d). These two amino acids are unique to DlP450_scc and distinct from canonical CYP87As (Fig. 5c). They likely form a conformationally optimized “floor” ideal for binding sterols in the enzyme’s active site (Fig. 5a, b). However, since the activity assay used cell lysate instead of purified proteins, which are difficult to isolate despite repetitive attempts, we do not rule out the possibility that these two mutations may also affect protein folding or stability. Comparing the substrate recognition sites of DlP450_scc and the human P450_scc revealed that the amino acids are distinct, although both sites are comprised mainly of non-polar amino acids, indicating hydrophobic interactions are the main driving force for substrate binding (Fig. 7). It remains unclear, however, if the stereochemistry of the 24-methyl group of campesterol affects the catalytic activity of DlP450_scc. Many plant species contain an epimeric mixture of 24(R)- and 24(S)- campesterol^38,39,40; the latter is called dihydrobrassicasterol. It is unclear if D. lanata contains both epimers or only the 24(S) stereoisomer.

**Fig. 7: Substrate binding sites of the DlP450_scc and the human P450_scc.**

The identification of DlP450_scc will enable the study of cardiac glycoside biosynthesis in other plant species, such as milkweed (Asclepias, Calotropis), wallflower (Erysimum), and oleander (Nerium oleander), to name a few⁴¹. Phylogenetic analysis showed that Calotropis gigantea might not have a duplicated CYP87A gene (Fig. 4), assuming the publicly available Calotropis transcriptome is complete. Interestingly, the CYP87A is in the same phylogenetic clade as CYP90B1 that catalyzes the 22(S)-hydroxylation of campesterol, which is one of the three steps in the sterol side-chain cleaving reaction (Supplementary Fig. 7)¹⁸. It is likely that cytochrome P450s within this clan, including CYP708A, CYP88A, CYP702A, CYP85A, CYP90, CYP720A, and CYP724A, have the potential to evolve the sterol side-chain-cleaving activity. It remains unclear what is the function of the canonical CYP87A. It may oxidize a sterol or a triterpenoid since CYP87D16 from Maesa lanceolata oxidizes the C16 of β-amyrin⁴².

In conclusion, this work identified the rate-limiting and long-speculated P450_scc in Digitalis for the biosynthesis of digoxin. It is an essential step toward ultimately elucidating the digoxin biosynthetic pathway. This work will also open the door for biomanufacturing novel digoxin analogs with expanded medicinal value in microbial or plant systems.

Methods

Plant material, RNA isolation, and sequencing

Digitalis lanata Ehrh seeds were procured from Strictly Medicinal (Williams, Oregon, USA). Seeds were germinated on the soil mix (57 g triple superphosphate, 85 g calcium hydroxide, 57 g bone meal, 369 g Osmocote (14-14-14), 99 g calcium carbonate, 25 L perlite, 50 L loosened peat and 25 L coarse vermiculite) and maintained in a growth chamber (Invitrogen, Clayton, Missouri, USA) under a light period of 16-h at 25 °C and a relative humidity of 60–80%.

Leaf and root tissues from three different seedlings were used to prepare the Illumina sequence library. Each seedling represents one biological replicate, and the total RNA from each replicate is split into two technical replicates. Total RNA was isolated using the RNeasy Plant Mini Kit (Qiagen, Germantown, MD, USA). The sequencing library was prepared from total RNA using the TrueSeq Ribo-Zero Plant RNA library prep kit (Illumina, San Diego, CA, USA) that removes ribosomal RNA. A quality check of the library was carried out with an Agilent 2100 bioanalyzer. The library was sequenced using Illumina HiSeq 2500 to generate 100 bp paired-end raw reads.

Gene isolation and cloning

The detailed cloning method is included in the Supplementary Methods. Primers used are listed in Supplementary Table 3, and plasmid constructs are listed in Supplementary Table 4, respectively.

Tobacco transient expression

Agrobacterium transformation

pEAQ plasmids carrying genes of interest were transformed into the Agrobacterium tumefaciens strain AGL1 individually by the freeze-thaw method⁴³. The resulting strains were prepared for infiltration using a modified protocol as in Saxena et al.⁴⁴. Briefly, A single Agrobacterium colony containing one of the pEAQ plasmids was inoculated into 5 ml yeast extract broth (YEB) [5 g/L tryptone, 1 g/L yeast extract, 2.5 g/L Luria broth (Fisher Scientific, Waltham, MA, USA), 5 g/L sucrose,0.49 g/L MgSO₄·7H₂O] with 50 mg/L kanamycin for pEAQ plasmid selection and 25 mg/L rifampicin for A. tumefaciens strain AGL1 selection. The bacterial cultures were grown 24 h at 28 °C with shaking at 220 rpm. Afterward, 0.5 ml of the seed culture was used to inoculate 25 ml of YEB with kanamycin (50 mg/L) and rifampicin (25 mg/L). The flasks were grown overnight at 28 °C, 220 rpm. The cultures were pelleted at 3000 g for 15 min, washed once with 10 mL sterile double-distilled water (ddH₂O), and resuspended in MMA [10 mM MES (2-N-morpholinoethanesulfonic acid), pH 5.6, 10 mM MgCl₂, 100 μM acetosyringone]. The individually transformed strains were pooled together so that the final volume was 10 ml and each A. tumefaciens strain had a final OD₆₀₀ of 0.4. Then cultures were incubated for 2 to 4 h at 28 °C before infiltrating tobacco leaves.

Tobacco infiltration

The pooled A. tumefaciens was infiltrated into the underside of four- to six-week-old Nicotiana benthamiana new leaves using a needleless plastic syringe. The tobacco plants were grown in 16-h light and 8-h dark periods at 21 °C with a relative humidity of 60–80% and photon intensity of 120–150 μmol/m². Three leaves were infiltrated for each experimental set, and each set was completed on a single plant. As a negative control, A. tumefaciens transformed with pEAQ_GFP was infiltrated into a separate plant. Plants were maintained in dark for 12 h to increase the agrobacterial infection and then shifted to light. The plants were maintained in normal conditions for four to six days. Once the fluorescence from GFP was intense when exposed to UV light, all infiltrated leaves were detached from the petiole, snap-frozen in liquid nitrogen, and ground into a fine powder. Metabolites were extracted with 1 ml 100% methanol (Fisher Scientific, Waltham, MA, USA) and heating at 65 °C for 10 min. They were centrifuged at 17,000 g for 10 min, filtered through a 0.45 μm filter (VWR, Randor, PA, USA), and stored at −20 °C before LC/MS analysis.

Yeast in vivo expression assay

Sterol-producing yeasts were kindly provided from the Riezman lab (Supplementary Table 5)³⁵. Competent cells of these strains were prepared using the Frozen EZ Yeast Transformation II Kit^TM (Zymo Research, Irvine, CA, USA). Starter cultures were grown in the SD-Leu medium at 30 °C overnight and then used to inoculate 25 mL SD-leu medium in triplicates in a shaking flask with an initial OD₆₀₀ of 0.2–0.4. Samples were harvested at 18 h and pelleted 3000 g for 5 min. Yeast cells were resuspended in 200 μL of TES buffer (50 mM Tris-HCl pH = 7, 600 mM sorbitol, 10 g/L bovine serum albumin, 1.5 mM β-mercaptoethanol) and homogenized with an equal volume of 0.5 mm glass beads in a BBX24 Bullet Blender® homogenizer (Next Advance, Troy, NY, USA) at setting 8 at 4 °C for 4 min. A total of 300 μL of TES buffer was added to the lysed cells, and 400–500 μL of the yeast lysate was transferred into a capped glass tube, followed by adding 1 mL chloroform immediately. The sample was vortexed for 1 min, and the organic phase was transferred into a new glass test tube and dried under a stream of air. The sample was resuspended in 100 μL methanol, centrifuged at 17,000 g for 10 min, and the supernatant was transferred into a LC/MS vial and stored at −20 °C until use.

LC/MS analysis for pregnane intermediates in the digoxin pathway

Samples were analyzed using a LC/MS² instrument, a Thermo Scientific Q-Exactive Focus^TM, a hybrid quadrupole and orbitrap mass analyzer (Fisher Scientific, Waltham, MA, USA) and Thermo Scientific UltiMate 3000 UHPLC^TM (Fisher Scientific, Waltham, MA, USA). A Waters XSelect CSH^TM C18 HPLC column (SKU: 186005257, Waters, Milford, MA, USA) with a particle size of 3.5 μm, an internal diameter of 2.1 mm, and a length of 150 mm was used for separation. The column was set to 25 °C with the back pressure in the range of 130–150 psi. Digitalis lanata and tobacco extract samples in biological triplicates were analyzed as previously described^45,46. For analyzing pregnenolone and pregnenolone acetate from yeast samples, the following protocol was developed. The mobile phase A was water with 0.1% formic acid and mobile phase B was acetonitrile with 0.1% formic acid, with a flow rate of 200 μL min⁻¹. Gradient started with 40% mobile phase B for 2 min followed by a linear gradient of 40% to 95% B from 2 to 11 min, held for 5 min, and brought back to initial conditions of 40% mobile phase B in 6 min. The sample injection volume was 20 μL, and the injection cycle time set to automatic with no sample splitting. The eluents were ionized by electron spray ionization (ESI) and analyzed in the positive ion mode. The full scan range is 100–1200 m/z at resolution of 70,000, inclusion error of ±5 ppm and automatic gain control (AGC) of 1 million. The resolution for ddMS² is 17,500 with collision energies at 10, 30, and 60 eV, isolation window of 3.0 m/z, and AGC of 8000. The scan rate was set at automatic. The mass spectrometer was regularly calibrated to ensure mass accuracy. Qualitative analysis was performed using the XCalibur™ (v. 4.4.16.14) software. Raw files were converted to.mzML files using MSConvert (v. 3.0.21040), and chromatograms and spectra were generated in R (v. 4.2.0) using the XCMS (v. 3.18.0) and Spectra (v. 1.6.0) packages^47,48.

Phylogenetic analysis

Transcriptome sequences used for the CYP87A tree were retrieved by BLASTing the DlCPY87A4 transcript against the 1000 Plant Transcriptome (1KP) database using tBLASTx^37,49 under default parameters. Sequences were filtered, and only those within 400–600 amino acids long and had a start codon were retained. A. thaliana CYP87A2 was used as a reference, and CYP708 was used as an outgroup.

All trees were constructed by aligning protein sequences using MAFFT⁵⁰. Aligned sequences were trimmed using trimAl⁵¹. The phylogenetic tree was constructed using RAxML-NG (v. 1.0.1) with the all-in-one Maximum likelihood (ML) tree search and slow bootstrapping with 1000 replicates⁵².

Protein modeling and docking

A protein model for the D. lanata P450_scc (DlCYP87A4) was generated using Alphafold2 through the ColabFold platform v1.4^53,54. The MSA mode used was MMseqs2, and all other parameters were the default⁵⁵. Five models were generated, and model 3 was used for further analysis based on Predicted Aligned Error (PAE) and predicted local distance difference test (pLDDT) scores (Supplementary Fig. 15). Docking of sterols was performed using Chimera version 1.16 and Autodock Vina version 1.1.2^56,57.

Statistics

Total lanatoside quantification

The data are normalized by dry weight and represent the average ± SD of three biological replicates.

Transcriptome heatmaps

Three biological replicates with two technical replicates each from roots and leaves were used to generate heatmaps. EdgeR was used to conduct the differential expression analysis to obtain log2FPKM (transcript per million mapped reads). The significance cutoff for overexpression in leave is P < 0.05.

qRT-PCR

Three biological replicates and two technical replicates were included for each sample. The mean of the two technical replicates’ Ct values was normalized against that of the polyubiquitin 10 gene (UBQ10) to calculate the ΔCt. ΔCt value was then normalized against the mean ΔCt of roots to derive the ΔΔCt value.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Data generated and analyzed are included in the published article and its supporting information files. D. lanata raw RNA-seq reads and the assembled transcriptome are deposited into the Gene Expression Omnibus database (Accession: GSE224014). LC/MS² and GC/MS data are available in the Metabolites database (Accession: MTBLS7993). D. lanata CYP87A1-4 sequences are available in Supplementary Data 1, 2 and GenBank database (Accession: OR134561, OR134562, OR134563, OR134564). Source data are provided with this paper.

References

Withering, W. An account of the foxglove, and some of its medical uses. (Swinney, 1785).
World Health Organization. WHO model list of essential medicines - 22nd list, 2021. Tech. Doc. (2021).
Medical Expenditure Panel Survey. Digoxin drug usage statistics, United States, 2013–2020. ClinCalc DrugStats Database (2022).
Mekhail, T. et al. Phase 1 trial of Anvirzel^TM in patients with refractory solid tumors. Invest. N. Drugs 24, 423–427 (2006).
Article CAS Google Scholar
Srivastava, M. et al. Digitoxin mimics gene therapy with CFTR and suppresses hypersecretion of IL-8 from fibrosis lung epithelial cells. Proc. Natl Acad. Sci. USA 101, 7693–7698 (2004).
Article CAS PubMed PubMed Central Google Scholar
Su, C.-T. et al. Anti-HSV activity of digitoxin and its possible mechanisms. Antivir. Res. 79, 62–70 (2008).
Article CAS PubMed Google Scholar
Prassas, I. & Diamandis, E. P. Novel therapeutic applications of cardiac glycosides. Nat. Rev. Drug Discov. 7, 926–935 (2008).
Article CAS PubMed Google Scholar
Piccioni, F., Roman, B. R., Fischbeck, K. H. & Taylor, J. P. A screen for drugs that protect against the cytotoxicity of polyglutamine-expanded androgen receptor. Hum. Mol. Genet. 13, 437–446 (2004).
Article CAS PubMed Google Scholar
Kim, N. et al. Cardiac glycosides display selective efficacy for STK11 mutant lung cancer. Sci. Rep. 6, 1–11 (2016).
CAS Google Scholar
Botelho, A. F. M., Pierezan, F., Soto-Blanco, B. & Melo, M. M. A review of cardiac glycosides: structure, toxicokinetics, clinical signs, diagnosis and antineoplastic potential. Toxicon 158, 63–68 (2019).
Article CAS PubMed Google Scholar
Ziff, O. J. & Kotecha, D. Digoxin: the good and the bad. Trends Cardiovasc. Med. 26, 585–595 (2016).
Article CAS PubMed Google Scholar
Wickramasinghe, J. A. F., Hirsch, P. C., Munavalli, S. M. & Caspi, E. Biosynthesis of plant sterols. VII. The possible operation of several routes in the biosynthesis of cardenolides from cholesterol. Biochemistry 7, 3248–3253 (1968).
Article CAS PubMed Google Scholar
Kreis, W. The foxgloves (Digitalis) revisited. Planta Med. 83, 962–976 (2017).
Article CAS PubMed Google Scholar
Herl, V., Fischer, G., Müller-Uri, F. & Kreis, W. Molecular cloning and heterologous expression of progesterone 5β-reductase from Digitalis lanata Ehrh. Phytochemistry 67, 225–231 (2006).
Article CAS PubMed Google Scholar
Finsterbusch, A., Lindemann, P., Grimm, R., Eckerskorn, C. & Luckner, M. Δ5-3β-hydroxysteroid dehydrogenase from Digitalis lanata Ehrh. - A multifunctional enzyme in steroid metabolism? Planta 209, 478–486 (1999).
Article CAS PubMed Google Scholar
Pérez-Bermúdez, P., Moya García, A. A., Tuñón, I. & Gavidia, I. Digitalis purpurea P5βR2, encoding steroid 5β-reductase, is a novel defense-related gene involved in cardenolide biosynthesis. N. Phytol. 185, 687–700 (2010).
Article Google Scholar
Lindemann, P. & Luckner, M. Biosynthesis of pregnane derivatives in somatic embryos of Digitalis lanata. Phytochemistry 46, 507–513 (1997).
Article CAS Google Scholar
Strushkevich, N. et al. Structural basis for pregnenolone biosynthesis by the mitochondrial monooxygenase system. Proc. Natl Acad. Sci. USA 108, 10139–10143 (2011).
Article CAS PubMed PubMed Central Google Scholar
Pilgrim, H. Cholesterol side-chain cleaving enzyme Aktivitat in keimlingen und in vitro kultivierten geweben von Digitalis purpurea. Phytochemistry 11, 1725–1728 (1972).
Kreis, W., Hensel, A. & Stuhlemmer, U. Cardenolide biosynthesis in foxglove. Planta Med. 64, 491–499 (1998).
Article CAS Google Scholar
Aberhart, D. J., Lloyd-Jones, J. G. & Caspi, E. Biosynthesis of cardenolides in Digitalis lanata. Phytochemistry 9, 1539–1543 (1970).
Google Scholar
Milek, F., Reinhard, E. & Kreis, W. Influence of precursors and inhibitors of the sterol pathway on sterol and cardenolide metabolism in Digitalis lanata ehrh. Plant Physiol. Biochem. 35, 111–121 (1997).
CAS Google Scholar
Raghavan, I., Ravi Gopal, B., Carroll, E. & Wang, Z. Q. Cardenolide Increase in foxglove after 2,1,3-benzothiadiazole treatment reveals a potential link between cardenolide and phytosterol biosynthesis. Plant Cell Physiol. (2022). pcac144.
Sayers, E. W. et al. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 48, D9–D16 (2020).
Article CAS PubMed Google Scholar
Bateman, A. et al. UniProt: the universal protein knowledgebase in 2021. Nucleic Acids Res. 49, D480–D489 (2021).
Article Google Scholar
Ashburner, M. et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29 (2000).
Article CAS PubMed PubMed Central Google Scholar
Kanehisa, M., Sato, Y., Kawashima, M., Furumichi, M. & Tanabe, M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 44, D457–D462 (2016).
Article CAS PubMed Google Scholar
Yoshioka, H. et al. A key mammalian cholesterol synthesis enzyme, squalene monooxygenase, is allosterically stabilized by its substrate. Proc. Natl Acad. Sci. USA 117, 7150–7158 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lange, I., Poirier, B. C., Herron, B. K. & Lange, B. M. Comprehensive assessment of transcriptional regulation facilitates metabolic engineering of isoprenoid accumulation in Arabidopsis. Plant Physiol. 169, 1595–1606 (2015).
CAS PubMed PubMed Central Google Scholar
Schrick, K. et al. FACKEL is a sterol C-14 reductase required for organized cell division and expansion in Arabidopsis embryogenesis. Genes Dev. 14, 1471–1484 (2000).
Article CAS PubMed PubMed Central Google Scholar
Banerjee, A. & Sharkey, T. D. Methylerythritol 4-phosphate (MEP) pathway metabolic regulation. Nat. Prod. Rep. 31, 1043–1055 (2014).
Article CAS PubMed Google Scholar
Quevillon, E. et al. InterProScan: protein domains identifier. Nucleic Acids Res. 33, 116–120 (2005).
Article Google Scholar
Finn, R. D. et al. Pfam: the protein families database. Nucleic Acids Res. 42, 222–230 (2014).
Article Google Scholar
Ohnishi, T. et al. CYP90A1/CPD, a brassinosteroid biosynthetic cytochrome P450 of Arabidopsis, catalyzes C-3 oxidation. J. Biol. Chem. 287, 31551–31560 (2012).
Article CAS PubMed PubMed Central Google Scholar
Souza, C. M. et al. A stable yeast strain efficiently producing cholesterol instead of ergosterol is functional for tryptophan uptake, but not weak organic acid resistance. Metab. Eng. 13, 555–569 (2011).
Article CAS PubMed Google Scholar
Cauet, G., Degryse, E., Ledoux, C., Spagnoli, R. & Achstetter, T. Pregnenolone esterification in Saccharomyces cerevisiae. A potential detoxification mechanism. Eur. J. Biochem. 261, 317–324 (1999).
Article CAS PubMed Google Scholar
Leebens-Mack, J. H. et al. One thousand plant transcriptomes and the phylogenomics of green plants. Nature 574, 679–685 (2019).
Article Google Scholar
Tsukagoshi, Y. et al. Ajuga Δ24-sterol reductase catalyzes the direct reductive conversion of 24-methylenecholesterol to campesterol. J. Biol. Chem. 291, 8189–8198 (2016).
Article CAS PubMed PubMed Central Google Scholar
Yamada, J. et al. 24-Methyl- and 24-ethyl-Δ24(25)-cholesterols as immediate biosynthetic precursors of 24-alkylsterols in higher plants. Tetrahedron 53, 877–884 (1997).
Article CAS Google Scholar
Mulheirn, L. J. Identification of C-24 alkylated steranes by P.M.R. spectroscopy. Tetrahedron Lett. 14, 3175–3178 (1973).
Article Google Scholar
Oerther, S. E. Plant poisonings: common plants that contain cardiac glycosides. J. Emerg. Nurs. 37, 102–103 (2011).
Article PubMed Google Scholar
Moses, T. et al. Unraveling the triterpenoid saponin biosynthesis of the african shrub maesa lanceolata. Mol. Plant 8, 122–135 (2015).
Article CAS PubMed Google Scholar
Wise, A. A., Liu, Z. & Binns, A. N. Three methods for the introduction of foreign DNA into Agrobacterium. Methods Mol. Biol. (Clifton, N. J.) 343, 43–53 (2006).
CAS PubMed Google Scholar
Saxena, P., Thuenemann, E. C., Sainsbury, F. & Lomonossoff, G. P. Virus-derived vectors for the expression of multiple proteins in plants. Methods Mol. Biol. 1385, 39–54 (2016).
Article CAS PubMed Google Scholar
Ravi, B. G., Guardian, M. G. E., Dickman, R. & Wang, Z. Q. High-resolution tandem mass spectrometry dataset reveals fragmentation patterns of cardiac glycosides in leaves of the foxglove plants. Data Br. 30, 1–8 (2020).
Google Scholar
Ravi, B.G., Grace, M., Dickman, R. & Wang, Z. Q. Profiling and structural analysis of cardenolides in two species of Digitalis using liquid chromatography coupled with high-resolution mass spectrometry. J. Chromatogr. A 1681, 460903 (2020).
Smith, C. A., Want, E. J., O’Maille, G., Abagyan, R. & Siuzdak, G. XCMS: Processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal. Chem. 78, 779–787 (2006).
Article CAS PubMed Google Scholar
Rainer, J. et al. A modular and expandable ecosystem for metabolomics data annotation in R. Metabolites 12, 173 (2022).
Article CAS PubMed PubMed Central Google Scholar
Carpenter, E. J. et al. Access to RNA-sequencing data from 1,173 plant species: The 1000 Plant transcriptomes initiative (1KP). Gigascience 8, 1–7 (2019).
Article CAS Google Scholar
Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability article fast track. Mol. Biol. Evol. 30, 772–780 (2013).
Article CAS PubMed PubMed Central Google Scholar
Capella-Gutiérrez, S., Silla-Martínez, J. M. & Gabaldón, T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25, 1972–1973 (2009).
Article PubMed PubMed Central Google Scholar
Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).
Article CAS PubMed PubMed Central Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article CAS PubMed PubMed Central Google Scholar
Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).
Article CAS PubMed PubMed Central Google Scholar
Steinegger, M. & Söding, J. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat. Biotechnol. 35, 1026–1028 (2017).
Article CAS PubMed Google Scholar
Tettersen, E. et al. UCSF Chimera–a visualization system for exploratory research and analysis. J. Comput Chem. 13, 1605–1612 (2004).
Article Google Scholar
Trott, O. & Olson, A. J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Comput. Chem. 31, 455–461 (2010).
CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank Dr. Howard Riezman at the University of Geneva for providing the sterol-producing yeast strains, Rian Hammond for providing plasmids encoding the human P450_scc, Dr. George Lomonossoff at the John Innes Center and Leaf Systems for supplying the pEAQ vector, Dr. Valerie Freichs for assistance with chromatography work, and Dr. Donald Yergeau for RNA-seq at University at Buffalo. This project was supported by the Research Foundation for the State University of New York [71272] to Z. Q. Wang and the National Science Foundation [CHE-1919594] to the University at Buffalo Chemistry Instrument Center.

Author information

These authors contributed equally: Emily Carroll, Baradwaj Ravi Gopal.

Authors and Affiliations

Department of Biological Sciences, University at Buffalo, the State University of New York, Buffalo, NY, USA
Emily Carroll, Baradwaj Ravi Gopal, Indu Raghavan, Minakshi Mukherjee & Zhen Q. Wang

Authors

Emily Carroll
View author publications
You can also search for this author in PubMed Google Scholar
Baradwaj Ravi Gopal
View author publications
You can also search for this author in PubMed Google Scholar
Indu Raghavan
View author publications
You can also search for this author in PubMed Google Scholar
Minakshi Mukherjee
View author publications
You can also search for this author in PubMed Google Scholar
Zhen Q. Wang
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

E.C., B.R.G., and Z.Q.W. designed research; E.C., B.R.G., I.R. and M.M. carried out experiments; E.C., B.R.G, I.R., M.M. and Z.Q.W. analyzed data; E.C., B.R.G, I.R. and Z.Q.W. wrote the paper.

Corresponding author

Correspondence to Zhen Q. Wang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Reuben Peters, Yong Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Supplementary Data 5

Supplementary Data 6

Supplementary Data 7

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Carroll, E., Ravi Gopal, B., Raghavan, I. et al. A cytochrome P450 CYP87A4 imparts sterol side-chain cleavage in digoxin biosynthesis. Nat Commun 14, 4042 (2023). https://doi.org/10.1038/s41467-023-39719-4

Download citation

Received: 25 January 2023
Accepted: 20 June 2023
Published: 08 July 2023
DOI: https://doi.org/10.1038/s41467-023-39719-4

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.