Complete biosynthesis of QS-21 in engineered yeast

Chemicals

Numbers, trivial names and International Union of Pure and Applied Chemistry (IUPAC) names, as well as chemical structures of pathway metabolites, are listed in Supplementary Table 3. All chemical standards used in this study are analytical grade and are listed in Supplementary Table 4.

Plasmid construction

All plasmids were constructed by Gibson assembly (New England Biolabs, HiFi DNA Assembly Master Mix), followed by heat shock transformation into Escherichia coli DH5α competent cells, which were plated on Luria–Bertani (LB) agar containing 100 μg ml⁻¹ carbenicillin or kanamycin and grown at 37 °C overnight. E. coli transformants were grown in 5 ml LB medium containing 100 μg ml⁻¹ carbenicillin or kanamycin at 37 °C overnight, followed by miniprep plasmid extraction (Qiagen), and were validated by Sanger sequencing. All biosynthetic genes^9,10 with the exception of LovF-TE were codon-optimized for yeast expression and synthesized by Integrated DNA Technologies. The QS-21 biosynthetic pathway genes were directly inserted into the plasmid backbone for subcellular localization studies in Nicotiana benthamiana. All genes were assembled as expression cassettes in pESC plasmids or the plant binary expression vector pCaBGi for yeast and plant expression, respectively. All enzymes used in this study are listed in Supplementary Table 1.

Strain construction

DNA integrating sequences were constructed using a previously described method³². Manufacturer protocols and standard recombinant DNA procedures were followed for DNA purification (Qiagen), DNA amplification (New England Biolabs, Q5 HighFidelity 2X Master Mix). All primers were designed using CASdesigner. In brief, DNA fragments to be integrated were PCR-amplified then co-transformed with a Cas9-based plasmid facilitating integration at the targeted locus. Alternatively, selection markers were integrated using homologous recombination. For transformations, a fresh overnight culture of parent yeast was inoculated into 25 ml 2×YPD in a 250-ml shake flask at an optical density at 600 nm (OD_600 nm) of 0.2, and was incubated at 30 °C and 200 rpm until the OD_600 nm reached 1.0. Then, 5 OD of cells were collected by centrifuging for 2 min at 3,000g, and were washed with a half volume of H₂O. The pellet was then resuspended with DNA fragments for integration (2 µg) and pCUT plasmid (0.25 µg), which was then mixed with transformation mix (260 µl of 50% PEG3350, 36 µl of 1 M LiOAc and 10 µl of ssDNA)³³. The mixture was incubated at 42 °C for 30 min and the pellet was collected by centrifuging for 2 min at 3,000g. The pellets were then resuspended with 100 µl H₂O and this was plated onto selective agar plates. The integration was validated by colony PCR and sequencing; the correct colonies were used for further engineering after pCUT plasmid curing. Oligonucleotides and codon-optimized gBlock gene fragments were obtained from Integrated DNA Technologies. Yeast culture media were purchased from BD, and all agar plates were obtained from Teknova. All strains constructed in this study are listed in Supplementary Table 2.

In vivo production, extraction and analysis of QS-21 and its precursors

Strains were grown in 2 ml of yeast extract peptone dextrose (YPD, 4% D) medium for 48 h to reach OD_600 nm = 10–15, before being resuspended in 2 ml yeast extract peptone galactose (YPG, 4% G). All strains were incubated for 72 h in 24-deep-well plates at 30 °C and 200 rpm. YL-43 to YL-51 were supplemented with fresh YPG every 24 h. The medium was supplemented with 50–500 mg l⁻¹ (S)-2-methylbutyric acid when culturing YL-42 to YL-47.

β-amyrin production and GC–MS analysis

A single method was used to extract and quantify squalene and β-amyrin. Five hundred microlitres of culture medium in a microfuge tube was first treated with Zymolyase 100T (Arthrobacter luteus, AMSBIO) for 2 h at 37 °C before it was extracted with 500 µl ethyl acetate with bead-beating (3,800 rpm, 1 min × 2). Cholesterol was used as an internal standard. Organic and inorganic layers were separated by centrifugation at 12,000g for 1 min, and samples were extracted twice using cholesterol as an internal standard. Two hundred microlitres of the combined organic layer is derived by treatment with 200 µl of pyridine and 200 µl of BSTFA (Sigma-Aldrich) at 55 °C for 1 h. The derived sample was diluted in ethyl acetate before it was subjected to GC–MS (GC model 6890, MS model 5973 inert, Agilent). An aliquot of the sample (1 µl) was injected into a DB-WAX column (Agilent) operating at a helium flow rate of 1 ml min⁻¹. The oven temperature was held at 80 °C for 4 min after injection and was then ramped to 280 °C at 20 °C min⁻¹, held at 280 °C for 25 min, ramped to 300 °C at 20 °C min⁻¹ and finally held at 300 °C for 5 min (total method of 45 min). The MS ion source was held at 300 °C throughout, with the quadrupole at 200 °C and the GC–MS transfer line at 280 °C. Full mass spectra were generated for metabolite identification by scanning within the m/z range of 40–440. Standard curves for target molecules were routinely run at the start and end of each batch of samples.

Triterpenoid production and LC–MS analysis

The extraction and detection of erythrydiol (2) follow the procedure described for β-amyrin. For the rest of the triterpenoids, 200 µl of culture was collected in a microfuge tube before it was directly extracted with 800 µl methanol with bead-beating (3,800 rpm, 1 min × 2). The mixture was centrifuged at 12,000g for 1 min to separate the pellet. Two hundred microlitres of the supernatant was transferred into an Eppendorf tube, which was then evaporated in a vacuum concentrator at room temperature and the remainders were resuspended in 200 µl methanol. Finally, samples were filtered with Amicon Ultra 0.5-ml 3-kDa filter tubes or centrifuged at 15,000g for 5 min. Products were analysed using LC–MS (1260 Infinity II LC-MSD iQ, Agilent) equipped with a reverse phase C18 column (Kinetex 2.6 µm, 250 × 4.6 mm, XB-C18, Phenomenex). A 50-min isocratic method was performed with 10:90 of water (solvent A) and acetonitrile (solvent B) using a flow rate of 0.3 ml min⁻¹. Full mass spectra were generated for metabolite identification by scanning within the m/z range of 300–600 in negative-ion mode. Data acquisition and analysis were performed using OpenLab CDS version 2.4 (Agilent).

Production of glycosylated QA and LC–MS analysis

A similar extraction procedure was followed, by collecting 500 µl of culture and mixing with 500 µl methanol with bead-beating (3,800 rpm, 1 min × 2). Two hundred microlitres of the supernatant was evaporated and was resuspended in 200 µl of methanol before C28 glycosylation; otherwise, 800 µl of the supernatant was resuspended in 160 µl of methanol. Detection of glycosylated triterpenoids was performed using an LC-MSD iQ equipped with a Kinetex column 2.6 μm XB-C18 100 Å, 50 × 2.1 mm (Phenomenex) using the following parameters⁹: MS (ESI ionization, desolvation line temperature = 250 °C, nebulizing gas flow = 15 l min⁻¹, heat block temperature = 400 °C, spray voltage positive 4.5 kV, negative −3.5 kV). Method: solvent A: (H₂O + 0.1% formic acid); solvent B: (acetonitrile (CH₃CN) + 0.1% formic acid). Injection volume: 10 µl. Gradient: 15% B from 0 to 1.5 min, 15% to 60% B from 1.5 to 26 min, 60% to 100% B from 26 to 26.5 min, 100% B from 26.5 to 28.5 min, 100% to 15% B from 28.5 to 29 min, 35% B from 29 to 30 min. The method was performed using a flow rate of 0.3 ml min⁻¹. Full mass spectra were generated for metabolite identification by scanning within the m/z range of 400–1,350 in negative-ion mode. Data acquisition and analysis were performed using OpenLab CDS v.2.4 (Agilent). The production of target molecules was confirmed by co-elution with the purified standards previously reported⁹.

Production of acylated molecules and QS-21, and LC–QTOF–MS analysis

A similar extraction procedure was followed, by collecting 500 µl of culture and mixing with 500 µl methanol with bead-beating (3,800 rpm, 1 min × 2). Eight hundred microlitres of the supernatant was evaporated and was resuspended in 40 µl methanol, which was then filtered with Amicon Ultra 0.5-ml 3-kDa filter tubes or centrifuged at 15,000g for 5 min. Detection of the acylated molecules and QS-21 was performed by LC–MS (Agilent 6545 for quadrupole time-of-flight (QTOF), Agilent) using the following parameters¹⁰: MS (ESI ionization, desolvation line temperature = 250 °C, nebulizing gas flow = 15 l min⁻¹, heat block temperature = 400 °C, spray voltage positive 4.5 kV, negative −3.5 kV). Method: solvent A: (H₂O + 0.1% formic acid); solvent B: (acetonitrile (CH₃CN) + 0.1% formic acid). Injection volume: 10 µl. Gradient: 15% B from 0 to 0.75 min, 15% to 60% B from 0.75 to 13 min, 60% to 100% B from 13 to 13.25 min, 100% to 15% B from 13.25 to 14.5 min, 15% B from 14.5 to 16.5 min. The method was performed using a flow rate of 0.6 ml min⁻¹ and a Kinetex column 2.6 μm XB-C18 100 Å, 50 × 2.1 mm (Phenomenex). Full mass spectra were generated for metabolite identification by scanning within the m/z range of 400–2,500 in negative-ion mode¹⁹. Analysis was performed using MassHunter Qualitative Analysis v.B.06.00 (Agilent). Note that the standard used to spike in the QS-21 sample was 18-Xyl, which was generated in vitro using 18 with a terminal apiose on the C28 sugar chain. Because the molecules with a C28 terminal apiose or xylose co-elute, 18-Xyl (C28 terminal apiose) was used to determine the elution time of 17-Xyl (C28-terminal-xylose).

Extraction of CoA from engineered yeast and LC–MS analysis

The extraction procedure was adapted from previous reports^28,34. Specifically, 5 OD of cells were pelleted by centrifugation for 2 min at 4 °C at 3,000g and the supernatant was discarded. Cells were quenched and extracted by 100 μl of methanol: acetonitrile: 0.1% glacial acetic acid at a 45:45:10 ratio prechilled at −20 °C. The resuspended extracts were incubated on ice with intermittent vortexing for 15 min, followed by a 3-min centrifugation at 12,000g and 4 °C. The supernatant (10 µl) was injected for LC–MS analysis. Detection of CoA was performed using an LC-MSD iQ equipped with a Hypercarb column 5 μm, 250 Å, 150 × 1 mm (Thermo Fisher Scientific) using the following parameters: MS (ESI ionization, desolvation line temperature = 350 °C, nebulizing gas flow = 13 l min⁻¹, spray voltage positive 4.5 kV, negative −6.0 kV). Method: solvent A: (H₂O + 0.1% formic acid); solvent B: (acetonitrile (CH₃CN) + 0.1% formic acid). Injection volume: 10 µl. Gradient: 2% B from 0 to 15 min, 2% to 90% B from 15 to 17 min, 90% to 20% B from 17 to 18 min, 2% B from 18 to 35 min. The method was performed using a flow rate of 0.1 ml min⁻¹. Full mass spectra were generated for metabolite identification by scanning within the m/z range of 300–1,300 in negative-ion mode. Data acquisition and analysis were performed using OpenLab CDS v.2.4 (Agilent). The 2MB-CoA standard was synthesized according to a reported procedure²⁹.

Transient expression of fluorescent fusion proteins in tobacco plants

Leaves of four-week-old N. benthamiana plants were infiltrated following a procedure adapted from a previous study³⁵. In brief, constructs assembled into binary vectors were transformed into the Agrobacterium tumefaciens strain GV3101. Transformed Agrobacterium strains were grown in LB with appropriate antibiotics at 30 °C, shaking at 200 rpm, to an OD_600 nm of 0.8–1.2. Agrobacterium cells were collected by centrifugation at 4,000g for 10 min at room temperature and resuspended in infiltration buffer (10 mM MES, 10 mM MgCl₂ and 500 µM acetosyringone) to final OD_600 nm = 0.5. Cells were incubated in the infiltration buffer for 1 h with gentle shaking. N. benthamiana leaves were infiltrated with a 1-ml syringe with no needle attached by gently pressing the syringe to the abaxial side of the leaf while applying gentle pressure to the adaxial side. N. benthamiana plants were grown and maintained in a plant growth room at 25 °C in 16-h–8-h light–dark cycles with 50% humidity. Leaves were collected three days after infiltration.

Reporting summary

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