Identification, heterologous expression, and characterization of the tolypodiol biosynthetic gene cluster through an integrated approach

Abstract

Cyanobacteria are tremendous producers of biologically active natural products including the potent anti-inflammatory compound tolypodiol. However, linking biosynthetic gene clusters with compound production in cyanobacteria has lagged behind other bacterial genera. Tolypodiol is a meroterpenoid originally isolated from the cyanobacterium HT-58–2. Here we describe the identification of the tolypodiol biosynthetic gene cluster through heterologous expression in Anabaena, and in vitro protein assays of a methyltransferase found in the tolypodiol biosynthetic gene cluster. We have also identified similar biosynthetic gene clusters in cyanobacterial and actinobacterial genomes suggesting that meroterpenoids with structural similarity to the tolypodiols may be synthesized by other microbes. We also report the identification of two new analogs of tolypodiol that we have identified in both the original and heterologous producer. This work further illustrates the usefulness of Anabaena as a heterologous expression host for cyanobacterial compounds and how integrated approaches can help link natural product compounds with their producing biosynthetic gene clusters.

Graphical Abstract

Introduction

Cyanobacteria, which are Gram-negative photosynthetic bacteria, are rich sources of specialized secondary metabolites (aka natural products) with a wide range of biological activities including anti-viral, anti-bacterial, and anti-cancer.^1–4 Peptides (linear and cyclic), depsipeptides, polyketides, polyketide-(depsi)peptide hybrids, meroterpenoids, and terpenes have all been isolated from cyanobacteria.^5–10 To date, few cyanobacterial compounds have been assessed in pre-clinical trials, but two structural classes have entered clinical trials (cryptophycins and dolastatins), and brentuximab vedotin (trade name: Adcetris^™), an antibody drug conjugate (ADC) containing monomethyl auristatin E (dolastatin analog), was approved by the FDA in 2011 for the treatment of relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma.^11–13 More recently, other ADCs containing the monomethyl auristatin E warhead have been approved for the treatment of other cancers (e.g., Tisotumab vedotin approved in 2021).

The cyanobacterial strain HT-58–2, originally classified as a Tolypothrix sp., but subsequently shown to align more closely with Brasilonema sp., is known to produce two unique classes of natural products, the tetrapyrrole macrocyclic tolyporphins (e.g., tolyporphin A, 1)^14–15 and the meroterpenoid tolypodiol (2).¹⁶ Tolypodiol was reported to have potent anti-inflammatory activity in a mouse ear edema assay, with the reported activity similar to hydrocortisone.¹⁶ More recently, we reported the isolation, structure determination, and biological activity of 6-deoxy-tolypodiol (3) and 11-hydroxytolypodiol (4) from the extract of Brasilonema sp. HT-58–2.¹⁷

Here, we report a biosynthetic gene cluster (BGC) for the production of tolypodiol, successful heterologous expression of the tolypodiol (tyl) BGC in Anabaena sp. UTEX 2576, and propose a biosynthetic pathway. We also identified similar BGCs in both cyanobacterial and actinobacterial genomes suggesting that these molecules play an undefined role in the life of the producing organisms. Furthermore, we demonstrate that the methyltransferase, TylH, is responsible for converting a carboxylic acid to a methyl ester, most likely as one of the final steps in the tolypodiol biosynthetic pathway. Lastly, we report two new analogs of tolypodiol, 6-oxo-tolypodiol (5) and 1-oxo-tolypodiol (6) that were isolated from biomass of HT-58–2 after extraction of its lyophilized material as previously reported.¹⁷ The structures of 5 and 6 were determined largely by spectroscopic and spectrometric experiments such as LC-MS and 1D and 2D NMR.

MATERIALS AND METHODS

General.

Optical rotation measurements were taken on a JASCO DIP-370 Digital Polarimeter at the sodium line (589 nm) at 22°C. Electronic Circular Dichroism (ECD) and Ultraviolet-Visible (UV-Vis) measurements for 5 and 6 were taken on a Chirascan Circular Dichroism Spectrometer with the sample dissolved in methanol and placed in a 1 cm quartz cuvette with a solvent subtraction for baseline correction. Infrared (IR) spectra were recorded on a CaF₂ salt plate using a Shimadzu IRAffinity-1 Fourier-Transform Infrared Spectrophotometer. All Nuclear Magnetic Resonance (NMR) spectra for initial structural characterization of tolypodiol analogs (5 and 6) were acquired on an Agilent Technologies 600 MHz DD2 spectrometer operating at 600 MHz for ¹H and 150 MHz for ¹³C NMR or on a Varian Unity Inova 500 MHz spectrometer operating at 500 MHz for ¹H or 125 MHz for ¹³C. The appropriate residual solvent signal was used as an internal reference. All NMR samples were placed in 3 mm Shigemi tubes for analysis. NMR spectra for tolypodiol and its hydrolysis products were recorded on a Bruker 700 MHz spectrometer with a 5 mm carbon cryo-probe with z-axis gradient. ¹H-NMR and ¹³C-NMR chemical shifts (δ) are expressed in ppm values. CDCl₃ was used as the NMR solvent (δ_H = 7.26 ppm, δ_C = 77 ppm). High-pressure liquid chromatography (HPLC) high-resolution electrospray ionization mass spectrometry (HRESIMS) was conducted on one of two instruments. System 1, Agilent 6545 (Quadrupole-Time of Flight (Q-ToF) mass spectrometer with electrospray ionization (ESI) source was used in the positive mode to acquire high-resolution mass spectrometry (HRMS) data with a C18 Agilent Eclipse Plus (2.1 × 50 mm, 1.8 μm) column (Method A); System 2, an Agilent 6545 Q-ToF equipped with a Jet Stream ion source downstream of an Agilent 1260 Infinity HPLC system consisting of a degasser, quaternary pump, autosampler (held at 12°C), and column compartment (held at 40 °C). Separation was achieved with a Kinetex C18 column (2.1 × 50 mm, 2.6 μm, Phenomenex, Torrance, CA) (Method B). The Q-ToF machines were operated using MassHunter software and data was processed offline using MassHunter Qualitative software. HPLC separations were performed using a Shimadzu system consisting of LC-20AB binary high-pressure gradient solvent delivery unit, SPD-M20A Photodiode Array Detector, and a CBM-20A System Controller. Enzymatic assays were analyzed by HPLC using an Agilent 1100 HPLC instrument consisting of a vacuum degasser, quaternary pump, autosampler (cooled to 4 °C), column thermostat (maintained at 30 °C), and diode array detector.

All chemicals and solvents were purchased either from Sigma Aldrich, Alfa Aesar or Tokyo Chemical Industry Co., Ltd. (TCI). S-adenosyl-L-methionine (SAM) was obtained as S-adenosyl-L-methionine disulfate tosylate (MilliporeSigma, Burlington, MA) and used without purification. NdeI, XhoI, T4 DNA ligase, Phusion DNA polymerase, Escherichia coli NEB5α, and Escherichia coli BL21(DE3)-T1R were purchased from New England Biolabs (Ipswich, MA) and used per the manufacturer’s recommendations. Lysogeny broth, Miller (LB) was purchased from MilliporeSigma. GenCatch Plasmid DNA Mini-Prep Kit and EconoSpin^® All-In-One Mini Spin Columns were purchased from Epoch Life Sciences (Sugar Land, TX). PrimeSTAR GXL DNA polymerase was purchased from Takara Bio (Mountain View, CA) and oligonucleotides were purchased from Integrated DNA technologies with standard desalting (Coralville, IA). The UV absorptions were measured on an Eppendorf BioSpectrometer Kinetic and the cell used was Eppendorf μCuvette G1.0 (Eppendorf, New Haven, CT). All microbial procedures were performed using good microbial practices and aseptic techniques. Kanamycin (50 μg/mL) and spectinomycin (100 μg/mL) for selection of plasmids in E. coli, while spectinomycin and streptomycin (2.5 μg/mL each) were used for the selection of Anabaena sp. UTEX 2576.

Strain source.

Anabaena sp. UTEX 2576 was acquired from the UTEX Culture Collection of Algae, University of Texas at Austin (Austin, TX, 78705). Brasilonema sp. HT-58–2 cultures were revived from cryostorage maintained by the University of Hawaii Cyanobacterial Culture Collection. Escherichia coli strains NEB10β and BL21(DE3) were obtained from New England Biolabs.

Cultivation of Anabaena sp. UTEX 2576.

Anabaena sp. UTEX 2576 was cultivated in a Hoffman incubation chamber (Hoffman Manufacturing, Inc, Corvallis, OR, USA) at 28 °C under 24 h constant light illumination. The light was supplied by two GE Lighting 49893 F40/PL/AQ Plant and Aquarium Tube bulbs with a light intensity of 25 microeinsteins/m2/s, and the atmosphere contained 1% CO₂.

Cultivation and extraction of the cyanobacterium HT-58–2.

Cultures were grown in BG-11(Nitrogen+) media. Over time, they were scaled up to 20 L Pyrex carboys with aeration at a 4–5 L/min flow rate and under continuous fluorescent illumination. The BG-11 media with sodium nitrate as the nitrogen source (BG-11(Nit+)) formulation was composed of the following: NaNO₃ (1.5 g/L), K₂HPO₄*3H₂O (0.04 g/L), MgSO₄*7H₂O (0.075 g/L), Na₂CO₃ (0.02 g/L), CaCl₂*2H₂O (0.036 g/L), Na₂EDTA (0.001 g/L), ferric ammonium citrate (0.006 g/L) and citric acid (0.006 g/L) with trace metals H₃BO₃ (2.86 mg/L), MnCl₂*4H₂O (1.81 mg/L), ZnSO₄*7H₂O (0.22 mg/L), Na₂MoO₄*2H₂O (0.39 mg/L), CuSO₄*5H₂O (0.08 mg/L), Co(NO₃)₂*6H₂O (0.05 mg/L). The phylogenetic characterization and genomic sequencing of this strain has been reported previously.¹⁸ Cell material was harvested by filtration through the equivalent of a cheese cloth after 45 days of growth in 20 L carboys and lyophilized prior to extraction. The freeze-dried HT-58–2 (43.35 g) cell mass was exhaustively extracted in 1:1 CH₂Cl₂:i-PrOH to produce the crude extract (2.33 g) after solvent evaporation.

The HT-58–2 strain was originally described in 1992,¹⁴ but was later subjected to 16S rRNA analysis revealing that it was more aligned with strains of Brasilonema,¹⁸ a genus formally described in 2007,¹⁹ rather than Tolypothrix as originally identified from morphological features. The strain has been maintained in cryostorage at University of Hawaii since the original report,¹⁶ periodically removing cryovials to grow the strain.

Isolation of 6-oxo-tolypodiol (5) and 1-oxo-tolypodiol (6).

A modified Kupchan partition was used to separate crude into hexanes, dichloromethane, and aqueous methanolic fractions. The dichloromethane fraction yielded the new compounds after C8 silica gel fractionation using a methanol in water gradient (25%, 50%, 75%, 100% MeOH). Along with tolypodiol, the 75% MeOH fraction yielded both 6-oxo-tolypodiol (5) (0.8 mg) and 1-oxo-tolypodiol (6) (0.5 mg) after preparative HPLC (Phenomenex Gemini 5u C18 110A Axia Pac, 250 × 21.20mm, Torrance, CA) purification using acetonitrile and water with 0.1% formic acid in a linear gradient from 70–100% acetonitrile over 30 min followed by 100% acetonitrile for 15 min. The fractions collected from time 12–14 mins were combined and further purified on a semi-preparative scale (Phenomenex Luna PFP(2) 100A, 250 × 10 mm) with an 80%−100% methanol:water (0.1% formic acid) gradient over 30 min (5, t_R 18.5 min; 6, t_R 24.0 min) with a purity of 93% and 90%, respectively as estimated by ¹H NMR.

Assembly and cloning of the tyl biosynthetic gene cluster.

The meroterpenoid BGC sequence originating from Brasilonema sp. HT-58–2 was codon optimized using the Build OptimizatiOn Software Tools (BOOST)²⁰ and a Anabaena sp. PCC 7120 codon frequency table to remove all synthesis constraints (refactored sequences are listed in Table S1 as Batch202D_p002). Overlapping synthetic DNA fragments were obtained from Twist Bioscience and assembled into the pPJAV550 destination vector²¹ (Batch202D_p009 and Batch202D_p010) using yeast Transformation-Associated Recombination (TAR) cloning.²² Vector sequences are available in Table S2.

All yeast transformation was performed with an auxotrophic strain CEN.PK2 (Euroscarf, Germany) using the YEAST1 yeast transformation kit (MilliporeSigma) and the manufacturer’s protocol. Yeast colonies were then collected from the single-dropout selective plates and plasmid DNA were harvested and transformed into E. coli DH10B cells. The resulting colonies were sequence verified by the PacBio sequencing platform.

During the refactoring of the BGCs, all internal SrfI restriction sites were removed and a unique SrfI site was reintroduced at the 5’ ends of each BGCs. These SrfI sites were then used to clone two promoters by using the NEBuilder HiFi DNA Assembly kit (New England Biolabs) to generate a total of 12 BGC constructs (Batch202D_p007 and p018 in Table S1). Again, all construct sequences were verified by PacBio sequencing.

Heterologous expression of tolypodiol in Anabaena sp. UTEX 2576.

The plasmids pMero_HT583 and pMero_HT583_PpetE were introduced into Anabaena sp. UTEX 2576 through triparental mating with E. coli TOP10 cells containing either pMero_HT583 and pMero_HT583_PpetE and E. coli JCM113 according to our previously published protocol.^{21, 23} The exoconjugates were selected on BG11(N+) media supplemented with spectinomycin and streptomycin (2.5 μg/mL each) for 4 weeks at 28°C with 24-hour illumination in a 1% CO₂ atmosphere. Single colonies were restreaked onto BG11(N+) media supplemented with spectinomycin and streptomycin. The colonies were then cultivated for an additional 5 weeks followed by collection of the cells by scraping them from the plate. The cells were added to a tared 20 mL Scintillation vial, lyophilized, and then extracted with 20 mL ethyl acetate overnight with stirring provided by a magnetic stir bar. The cellular debris were removed via filtration through a Buchner funnel and the extract was concentrated to near dryness in vacuo using a 30 °C water bath. The extract was transferred to a 11 mL glass vial and concentrated to dryness in a miVac Quattro Concentrator (ATS Life Sciences Scientific Products, Warminster, PA). The extract was redissolved in HPLC grade methanol at a concentration of 1 mg/mL and then diluted to a concentration of 0.2 mg/mL for LCMS analysis.

LCMS analysis of tolypodiol.

Method A.

An Agilent 6545 LC-MS Q-Tof with ESI ionization was used in the positive mode to acquire high-resolution mass spectrometry (HRMS) data with a C18 Agilent Eclipse Plus (2.1 × 50 mm, 1.8 μm) column. A flow rate of 0.5 mL/min with a water (A)-acetonitrile (B) mobile phase system containing 0.1% (v/v) formic acid was used. The column was pre-equilibrated with 95% A/5% B. Upon injection the mobile phase composition was maintained for 0.75 minutes followed by changing the mobile phase to 0% A/100% B over 9.25 minutes using a linear gradient and 100% B was maintained for an additional 7 min.

The following parameters were used with positive polarity; Gas Temp, 350 °C; Drying gas, 10 L/min; Nebulizer, 55 psi; Sheath gas temp, 400 °C; Sheath gas flow, 10 L/min; VCap, 3500 V; Nozzle voltage (Expt), 1000 V; Fragmentor, 175 V; Skimmer, 65 V; Oct 1 RF Vpp, 750 V; Mass range, 100–3000 m/z; Acquisition rate, 10 spectra/s; Time, 100 ms/spectrum. The AutoMS/MS settings were as follows: Mass range, 100–3000 m/z; 1 spectra/s; 1000 ms/spectrum; Transients/spectrum, 5889.

Method B.

High-resolution mass spectrometry was obtained using an Agilent 1260 HPLCupstream of an Agilent 6545 Q-ToF downstream of an Agilent 1260 Infinity HPLC system. Separation was achieved using Kinetex C18 column (50 × 2.1 mm, 2.6 μm, Phenomenex, Torrance, CA) at a flow rate of 0.4 ml/min and the following gradient. Line A was water with 0.1% (v/v) formic acid and line B was acetonitrile with 0.1% (v/v) formic acid. The column was pre-equilibrated with 90% A/10% B. Upon injection the mobile phase composition was maintained for 0.75 minutes followed by changing the mobile phase to 10% A/90% B over 7.5 minutes using a linear gradient and then switching the mobile phase to 0% A/ 100%B over 0.1 minutes. The mobile phase is then maintained at 0%A / 100%B for 3.65 minutes.

The Agilent Q-ToF mass spectrometer was equipped with an Agilent JetSpray source operated with the following parameters: Auto MS/MS mode, Positive polarity; Gas Temp, 325 ᵒC; Drying gas, 7 L/min; Nebulizer, 20 psi; Sheath gas temp, 270 °C; Sheath gas flow, 12 L/min; VCap, 4000 V; Nozzle voltage (Expt), 600 V; Fragmentor, 175 V; Skimmer, 65 V; Oct 1 RF Vpp, 750 V; Mass range, 100–3000 m/z; Acquisition rate, 10 spectra/s; Time, 100 ms/spectrum. The AutoMS/MS settings were as follows: Mass range, 50–3000 m/z; 10 spectra/s; 100 ms/spectrum; Transients/spectrum, 429; Isolation Width, Medium (~4 m/z); The Precursor selection parameters were as follows: 5 Max Precursor Per Cycle; Abs. Threshold 1000 counts, Rel. Threshold (%) 0.01%, Excluded after 4 spectra; Released after 0.01 min.

Computational docking and MD simulations of TylF.

The AlphaFold predicted structure of TylF was downloaded from the AlphaFold Protein Structure Database (Locus BZZ01_00205).²⁴ Next, the protein’s orientation in the membrane was calculated using the PPM 2.0 web server.²⁵ Ligands were docked into the protein using AutoDock Vina²⁶ in UCSF Chimera.²⁷ Ligand binding was restricted to within the inside of the protein near the hypothetical active site, and conformations were picked which placed the conserved residues in proximity to the predicted site of initialization of cyclization in the ligand.

Molecular dynamics (MD) simulations of TylF embedded in a lipid bilayer were carried out with the GROMACS simulation package²⁸ v2022 using the CHARMM36 force-field (FF) for lipids^29–30 and the CHARMM36m FF for proteins.^31–33 Water was simulated using CHARMM TIP3 parameters. The glutamate residue in the predicted active site (E51) was modified to be protonated, and ligands with a carboxylic acid were specified to be deprotonated. The membranes were composed of POPG, DGDG, MGDG, SQDG to roughly match the experimental values measured for cyanobacteria.³⁴ The upper and lower membrane leaflets were symmetrical and contained 4 POPG lipids, 10 DGDG lipids, 18 MGDG lipids, and 10 SQDG lipids. Potassium and chloride ions were added to a concentration of 150 mM KCl. The combined protein-membrane system was set up using the CHARMM-GUI webserver.^35–36 Two different simulations were setup with the 3-geranylgeranyl-4-hydroxybenzoic acid and compound 10 bound in the active site. FF parameters for the two substrates were generated using the CHARMM-GUI interface and CGenFF.^37–39 Systems were simulated for 100 ns after initial energy minimization and equilibration using position restraints. Equations of motion were integrated with a leap-frog algorithm using a 2-fs time step. Van der Waals interactions were computed using a force-switched Lennard-Jones potential between 1.0 and 1.2 nm. Electrostatic interactions were computed using the particle-mesh Ewald method with a real space cutoff 1.2 nm and a Fourier grid spacing of 0.12 nm. Temperature was held constant using a velocity-rescaling algorithm⁴⁰ with a time constant of 1 ps, and pressure was held constant with a stochastic cell-rescaling algorithm⁴¹ using a time constant of 5 ps. Particle positions were saved in 5 ps intervals for trajectory analysis.

Synthesis of desmethyltolypodiol (12).

In order to hydrolyze the methyl ester of tolypodiol, a protocol using lithium hydroxide in aqueous methanol was adapted.⁴² In brief, approximately 1 mg of tolypodiol was dissolved in LCMS grade MeOH (0.7 mL) and H₂O (0.2 mL) in a 5 mL round bottom flask to which a few drops of 0.1 N LiOH dissolved in LCMS grade water was added. The reaction was placed in a 40 °C water bath and stirred overnight while connected to a reflux condenser. The reaction was quenched by acidification with a few drops of 0.1 N HCl. This reaction was done twice to secure sufficient material. The reaction mixtures were combined, dried down under a gentle stream of air, reconstituted in a minimal amount of LCMS methanol/water (8:2), and then purified by HPLC (Phenomenex Luna 5u C18(2) 150 × 4.6 mm; A: H₂O, B: MeOH (0.1% formic acid in both); gradient of 70% B (0–2 min), 70–100% B (2–17 min), 100% B (17–22 min); flow rate = 0.6 mL/min) to yield 0.7 mg of the hydrolyzed product and 0.6 mg of recovered tolypodiol. For NMR data, see Figure S1–2 and Table S3. HRESIMS m/z 443.2791 [M+H]⁺ (calc. for C₂₇H₃₉O₅⁺, 443.2792, 0.2 ppm error)

Creation of expression vector for overproduction of TylH.

The DNA fragment corresponding to TylH (protein accession number ARV57258) was amplified from Brasilomena sp. strain HT-58–2 genomic DNA by PCR using PrimeSTAR GXL DNA polymerase and the oligonucleotides TylH_Fwd-NdeI (5′-ATCTCTAAGCATATGTCTAAATGGAATGCACAAGATTACC-3′) and TylH_Rev-XhoI (5′-ATATAGATCTCGAGAGACACCGACAGTATGAGTGAAG-3′) with an annealing temperature of 58 °C using an extension time of 2:00 min and 35 cycles. The restriction sites of NdeI and XhoI are indicated by the underlined bases while the italicized bases indicate random bases added to increase restriction enzyme efficiency. The PCR amplicon was purified using the EconoSpin^® All-In-One Mini Spin Column followed by digestion with NdeI and XhoI. The restricted fragment was purified using the EconoSpin^® All-In-One Mini Spin Column followed by ligation into pET28a, which had been previously digested with NdeI and XhoI and purified using the EconoSpin^® All-In-One Mini Spin Column using T4 DNA ligase. The ligation mixture was then introduced in chemically competent E. coli NEB5α cells. Plasmid containing cells were then selected for on LB agar plates supplemented with kanamycin after overnight incubation at 37 °C. Random colonies were selected and the plasmids isolated with the GenCatch Plasmid DNA Mini-Prep Kit. The DNA inserts were sequenced with Sanger sequencing at the Center for Qualitative Life Sciences at Oregon State University to ensure insert sequence fidelity. The resulting sequence verified plasmid was named pET28-TylH-1.

Overexpression, and purification of TylH.

Chemically competent E. coli strain BL21(DE3) cells were transformed with pET28-TylH-1 and the transformation mixture was plated on LB agar plates supplemented with kanamycin. A single colony of E. coli BL21(DE3) cells containing pET28-TylH-1 was picked and grown overnight at 37 °C in LB media containing kanamycin (25 mL) and cultured in a shaker at 200 rpm at 37 °C overnight. A portion of the seed culture (5 mL) was transferred into LB medium (2 × 500 mL) in 2 × 2 L Erlenmeyer flasks containing kanamycin and grown at 37 °C until the OD₆₀₀ reached 0.6 (approximately 3.5 hr). Then, the temperature was reduced to 15 °C and after 1 h adaptation, Isopropyl β-D-1-thiogalactopyranoside (IPTG, 0.1 mM final concentration) was added to each flask. After further growth overnight at 200 rpm (approximately 16 hours), the cells were harvested by centrifugation (3,200 x g, 20 min, 4 °C), the media was decanted, and the cell pellet was stored at −20 °C until use.

The frozen cells from 1 L of culture were thawed on ice and then suspended in 30 mL lysis buffer [Tris (50 mM, pH 8), NaCl (300 mM), imidazole (10 mM)]. The suspension was sonicated using a Qsonica sonicator (9 mm probe, 30% amplitude, 45 sec, five times) on ice, then cell debris were removed by centrifugation at (13,751 x g, 4 °C for 30 min). The supernatant was applied to a Ni-NTA resin (750 μL, BioRad, Hercules, CA) by batch binding at 4 °C for 1 h followed by pelleting the resin by centrifugation at 3,200 x g for 30 sec. The supernatant was discarded, and the resin was washed with lysis buffer (2.5 mL) followed by washing two times with 1 mL Wash buffer [Tris (50 mM, pH 8), NaCl (300 mM), imidazole (25 mM), β-mercaptoethanol (5 mM)]. The resin was then washed twice with 1.5 mL Elution buffer [Tris-HCl (50 mM, pH 8), NaCl (300 mM), imidazole (250 mM), β-mercaptoethanol (5 mM)] followed by another wash with elution buffer (0.5 mL). The fractions from the washing buffer contained the highest concentration of TylH (determined by Bradford stain and SDS-PAGE, Figure S3). The protein solution was desalted using an Econo-Pac 10DG Desalting column (BioRad) equilibrated with desalting buffer [Tris (50 mM, pH 8), glycerol (10% w/v), NaCl (50 mM), β-mercaptoethanol (5 mM)]. The protein concentration of the fractions collected from the washing buffer was determined using an Eppendorf BioSpectrometer at (280 nm, 1 mm cell) using a calculated extinction coefficient of 29,910 cm⁻¹ M⁻¹.

In vitro enzymatic activity assays.

Reaction mixtures containing 100 mM Tris buffer (pH 7.5), 5 mM SAM dissolved in DI water (stock: 10 mg/ml), 1 mM substrate dissolved in MeOH (stock: 10 mg/mL), 10 μM TylH (stock: 0.7 mg/mL in desalting buffer), and 1mM β-mercaptoethanol in a final volume of 100 μL were mixed gently by pipetting and incubated at 28 °C overnight. The protein was removed by the addition of acetonitrile (50 μL) followed by centrifugation (15,871 x g) for 3 min. All negative control reactions were performed in the same manner by replacing the omitted reagent with the respective solvent. A portion of the quenched reaction (15 μL) was then analyzed by HPLC as described below.

HPLC analysis of in vitro TylH reaction mixtures.

All enzymatic reactions were analyzed on an Agilent 1100 HPLC instrument consisting of a vacuum degasser, quaternary pump, autosampler (cooled to 4 °C), column thermostat (maintained at 30 °C), and diode array detector using a Luna C18 column (15 cm × 4.6 mm, 5 μm, Phenomenex, Torrance, CA) where line C was water + 0.1 % (v/v) formic acid and line D was acetonitrile with 0.1 % (v/v) formic acid at a flow rate of 1 mL/min. The column was pre-equilibrated with 90% C / 10% D and held at this composition for 1 min. The composition was changed to 100% D over the next 20 min using a linear gradient. The composition was then held at 100% D and this composition was held for 5 min. The mobile phase was then changed to 90% C/10% D over 8 min and the composition was held for 8 min prior to the next injection.

Results and Discussion:

Identification of the tolypodiol biosynthetic gene cluster (BGC).

A retrobiosynthetic analysis of tolypodiol predicted it to be biosynthesized from p-hydroxybenzoic acid and geranylgeranyl diphosphate as key building blocks. Given the structural similarities between merosterol A and tolypodiol, we first attempted to search the published genome of HT-58–2 by BLAST analysis using MstE as the probe, given its role as a terpene cyclase in the biosynthesis of the merosterol A.^43–44 While a similar protein was identified, BZZ01_14670, it did not appear to be in proximity to other encoded proteins that were predicted to be involved in the biosynthesis of tolypodiol. We then decided to use chorismate lyase as a BLAST probe as this was expected to be unique to tolypodiol biosynthesis, compared to the isoprenoid biosynthetic enzymes which could be found in multiple genetic loci. In bacteria, p-hydroxybenzoic acid is known to be synthesized by UbiC,⁴⁵ therefore, we used All0938 (UbiC from Anabaena sp. PCC 7120) as the BLAST probe sequence and identified two proteins with significant similarity in the genome of HT-58–2, BZZ01_27380 (87% identity/92% similarity) and BZZ01_00185 (73% identity/81% similarity). Examination of the genomic region near BZZ01_00185 showed the presence of encoded proteins that have the predicted activities necessary for the biosynthesis of tolypodiol (Figure 1, Table S4). This same genomic region was recently postulated to be involved in the production of the tolypodiols by the Lindsey group, although the involvement of TylF (BZZ01_00205) in terpene cyclization was not investigated.^{18, 46} We subsequently renamed BZZ01_00185 as tylB and it is predicted to encode a chorismate lyase, which is responsible for converting chorismate to p-hydroxybenzoic acid. TylC, TylD, TylI, TylK, and TylM have similarity to enzymes involved in the formation of the isoprenoid precursors dimethylallyl diphosphate (DMAP) and isopentenyl diphosphate (IPP). TylE belongs to the polyprenyl synthase family and is presumably involved in the biosynthesis of geranylgeranyl diphosphate (GGPP). TylA is predicted to be 4-hydroxybenzoate solanesyltransferase, which is likely responsible for the transfer of the geranylgeranyl unit to the 3-position of 4-hydroxybenzoate. TylG has similarity to aromatic ring hydoxylases and is likely responsible for introduction of one of the hydroxy groups while TylJ (annotated as a cytochrome P₄₅₀) is likely responsible for the installation of the second hydroxy group. TylH is predicted to be a SAM dependent methyltransferase, which is envisioned to methylate a carboxylic acid precursor to form the methyl ester seen in the tolypodiols. TylL is annotated as a member of the AAA ATPase protein family⁴⁷ and could be a regulator of the tolypodiol BGC expression or responsible for the transport of tolypodiol through membranes. No other transport proteins were identified in the vicinity of the tyl BGC. TylF is annotated as a hypothetical protein but is presumed to be the terpene cyclase (see subsequent sections).

Figure 1.

A) Biosynthetic gene cluster for the biosynthesis of tolypodiols identified in the genome of Brasilonema sp. HT-58–2. B) Predicted biosynthetic pathway for the formation of tolypodiol in HT-58–2.

Heterologous expression of the tolypodiol (tyl) BGC in Anabaena sp.

As genetic disruption of the tolypodiol producing Brasilonema sp. HT-58–2 has not been reported and the culture is unialgal and not axenic, we elected to use heterologous expression to prove that tylA-tylM was responsible for the biosynthesis of tolypodiol. We and others have previously used Anabaena sp. to successfully heterologously express cyanobacterial BGCs,^{21, 23, 48–49} and elected to express the tyl BGC in this cyanobacterial host. The tyl BGC was cloned into pPJAV550 after codon optimization for expression in Anabaena. We cloned two versions of the tyl BGC, one with the native promoter and one under the control of the P_petE promoter (see section for full description) to yield pHT583_Mero-1 and pHT583_Mero_PpetE-1 respectively.⁵⁰

Introduction of pHT583_Mero-1 and pHT583_Mero_PpetE-1 to Anabaena sp. UTEX 2576 (=Anabaena sp. PCC 7120, herein Anabaena) through triparental mating with Escherichia coli JCM113 and NEB10β containing the plasmid of interest,²³ resulted in exoconjugates after selection on BG-11(Nit+)_SpSm agar plates (BG-11 media containing sodium nitrate as the nitrogen source, supplemented with spectinomycin and streptomycin (2.5 μg/mL each)). Colonies were selected, restreaked, and cultivated for an additional 5 weeks on BG-11(Nit+)_SpSm agar plates. The cells were then collected, lyophilized, extracted with ethyl acetate, and analyzed by LC-HRMS. We were able to detect tolypodiol in the extract of pHT583_Mero_PpetE-1/Anabaena but not pHT583_Mero-1/Anabaena (Figure 2 and Figure S12). The produced compound had a protonated molecular ion consistent with the molecular formula C₂₈H₄₁O₅⁺ (M+H obs. 457.2940, calc. 457.2949, 2.0 ppm error) as well as an identical retention time (8.1 min), and MS/MS fragmentation pattern as the tolypodiol standard (Figure S13–14). We used molecular networking^51–52 to identify other analogs present in the extract (Figure 2B, Figure S15) and were able to observe 6-deoxytolypodiol (3), 11-hydroxytolypodiol (4), 6-oxo-tolypodiol (5), 1-oxo-tolypodiol (6), and a compound with the same mass and MS/MS fragmentation pattern as tolypodiol with a different chromatographic retention time, likely a conformer or stereoisomer of tolypodiol, (Figure S12–14) in the extracts of the Anabaena strains containing pHT583_Mero_PpetE-1, thereby linking these metabolites to the tolypodiol BGC. The structural elucidation of compounds 5 and 6 is described in the section “New tolypodiol structures.”. The presumed isomer of 2 was produced in quantities that precluded structural and stereochemical determination.

Figure 2. Production of tolypodiol and related compounds by expression of the tyl BGC in Anabaena.

(A) Extracted ion chromatogram for tolypodiol (457.2940–457.2960 m/z) in Anabaena containing pPJAV361 (empty vector), pMero_HT583_PpetE (tyl BGC), and purified tolypodiol. (B) Molecular network of metabolites found in Anabaena containing pMero_HT583_PpetE. The full GNPS network can be seen in Figure S14. Additional nodes with the m/z values of 457.29 are likely stereoisomers of tolypodiol. These peaks were produced in insufficient quantities for NMR structural determination.

TylF is the founding member of a new family of bacterial terpene cyclases with similarity to fungal terpene cyclases.

Using TylF as a query sequence for a BLAST analysis against the Swiss-Prot/UniProt database retrieves terpene cyclase proteins from the BGCs involved in the biosynthesis of a variety of fungal natural products (Table S5). The top hit was DpmpB, which is involved in the biosynthesis of decalin-containing diterpenoid pyrones in Macrophomina phaseolina MS6⁵³ (28% amino acid similarity/50% amino acid identity). Just outside of the top 100 sequences returned from the BLAST query was Trt1 (25% amino acid similarity/47% amino acid identity), which has been characterized as a terpene cyclase in the terretonin biosynthetic pathway. Trt1 was previously shown to cyclize 5-farnesyl-3,5-dimethylorsellinic acid methyl ester, but not the corresponding acid 5-farnesyl-3,5-dimethylorsellinic acid.⁵⁴

TylF does not contain an aspartate rich motif found in Class I (DDxxD) or Class II (DxDD) terpene cyclases.⁵⁵ Multiple sequence alignment of the proteins identified by BLAST identified only one conserved motif Lxx(D/N)xxWExx(Y/F) found in the aligned proteins (Figure S4), and we propose this motif to be the involved in cyclization. When Trt1 is included in the multiple sequence alignment (MSA) the conserved motif was identified as (L/I)xx(D/N)xxWExx(Y/F) (Figure S4). Downloading the predicted protein structure of TylF from the AlphaFold Protein Structure Database (entry A0A1Y0RDB4)²⁴ allowed the prediction of the location of the conserved motif (Figure S5). The L⁴⁴SANIAWEFLF⁵⁴ sequence was identified in a region that had very high prediction confidence (pLDDT > 90, Figure S6, green residues) and the conserved Trp and Glu residues are located inside a cavity that is presumed to be the active site (see below). The AlphaFill database entry (AF-A0A1Y0RDB4-F1-model_v4, accessed Jan 27, 2023) did not have any ligands, metal ions or substrates predicted to be bound to the protein. We have also been unable to produce soluble protein in E. coli, most likely due to the fact that TylF has seven predicted transmembrane sections (Figure S7), and therefore we have been unable to directly test the role of TylF as a terpene cyclase and the role of each individual residue in the L⁴⁴SANIAWEFLF⁵⁴ motif.

While this manuscript was under revision, a paper describing key residues in fungal meroterpenoid cyclase AdrI.⁵⁶ In AdrI, the consereved glutamate residue (Glu63 in AdrI/Glu51 in TylF) was not invovled in cyclization but a neighboring aspartate (Asp59 in AdrI) was found to be the catalytic acidic residue. In the same study IndA7 was also studied, and IndA7 contained a asparagine at position 59. Substitution of Glu63 resulted in a loss of cyclization activity in IndA7. TylF and all Tyl-like proteins identified in bacterial genomes, contain an asparagine like IndA7 (Asn47), which provides support for our assignment of the conserved Glu residue in the TylF motif (L⁴⁴SANIAWEFLF⁵⁴) as the catalytic acidic residue that initiates meroterpenoid cyclization.

We turned to MD simulations to interrogate substrate binding to TylF as Trt1 was previously shown to accept methyl 5-farnesyl-3,5-dimethylorsellinate⁵⁴ while MstE was shown to cyclize 3-geranylgeranyl-4,5-dihydroxybenzoate.^43–44 Therefore the identity of the cyclization substrate was unclear as TylF is a cyanobacterial protein (as is MstE) but it is membrane bound (as is Trt1). We examined the predicted binding of 3-geranylgeranyl-4-hydroxybenzoic acid to determine if it is a substrate for cyclization. Docking of 3-geranylgeranyl-4-hydroxybenzoic acid followed by 100 ns of MD simulations resulted in a model with stable rmsd values (Figure S 8). The 3-geranylgeranyl-4-hydroxybenzoate fit within a cavity that was bound at the bottom by three Trp residues (Trp20, Trp50, Trp72) and the geranylgeranyl side chain was located in proximity to the L⁴⁴SANIAWEFLF⁵⁴ motif (Figure 3A). The top of the cavity was bordered by Arg62 and Arg126. The guanidino groups of the Arg residues fluctuated between 2.95 and 11.87 Å (Arg62) and between 3.17 and 11.11 Å (Arg126) from the carboxylate of 9 during the course of the MD simulation. This suggests that a hydrogen bond can be formed between the substate carboxylate and Arg residues and that 9 is the correct substrate. After cyclization, Arg62 and Arg126 are no longer closely associated with the carboxylic acid, and the top of the protein opens, providing a path for the cyclized product 10 to exit the putative active site cavity (Figure 3B).

Figure 3. Protein models of TylF bound to substrates.

(A) Binding of 3-geranylgeranyl-4-hydroxy benzoic acid (9) to TylF. The carboxylate of 9 is coordinated by Arg62 and Arg126 (shown in blue), while the geranylgeranyl side chain is contained in a cavity bound at the bottom by Trp20, Trp50, Trp72 (shown in orange) and Glu51 (modeled as protonated, shown in blue). (B) Cyclized compound 10 bound to TylF. The carboxylate of 10 is coordinated by Arg62/126 (shown in blue), while the cyclized B-E rings are contained within a pocket bordered at the bottom Trp20, Trp50, Trp72 (shown in orange) and Glu51 (modeled as deprotonated, shown in blue).

MstE utilizes a similar but distinct structural motif for binding its substrate, geranylgeranyl-3,4-dihydroxybenzoate.⁴⁴ Four aromatic side chains interact with the geranylgeranyl side chain (Phe49, Y100, Trp59, and Trp210) while Asp109 catalyzes the cyclization, similar to the situation envisioned for TylF. In contrast the carboxylate of geranylgeranyl-3,4-dihydroxybenzoate is anchored through a hydrogen bond with Tyr157, while the 3’-OH group is bound by Glu339. While a structure of Trt1 is not available, the MSA shows that Trt1 contains an Ala and a Leu at the equivalent positions to Arg62 and Arg126 in TylF, which is consistent with binding a non-charged methyl ester.

Meroterpenoid BGCs encoding TylF-like proteins are widespread in bacterial genomes.

We next investigated the prevalence of BGCs encoding TylF-like proteins in sequenced bacterial genomes. Using TylF as a BLAST probe for an EFI-EST^57–58 search (e-value cutoff, 5; Maximum number of retrieved sequences, 1,000) and creation of the clustered network with a cutoff score of 50 resulted in the identification of 30 clusters and 33 singletons (Figure S 10). Proteins from cyanobacteria were identified in clusters 1 and 5 (hexagons, Figure S 10). The TylF-like proteins in cluster 1 included cyanobacterial proteins from Nostoc sp. (cyanobiont of the lichen Peltigera membranacea), but the majority of these proteins were from other bacterial genera including Lysinibacillus (most prevalent), Rufibacter, etc. Examination of the genes upstream and downstream of the TylF-like protein encoding gene revealed that they were typically in small or non-existent BGCs and are likely not responsible for the biosynthesis of natural products. Cluster 8 (Lavender, Figure S 10) contained fungal sequences including the known terpene cyclases NvfL from Aspergillus novofumigatus (strain IBT 16806), AtmB from Byssochlamys spectabilis, and DpmpB from Macrophomina phaseolina (strain MS6). Cluster 5 contained proteins encoded in the genomes of cyanobacteria (Brasilonema, Tolypothrix, Scytonema, and Nostoc but not the genome of Anabaena) and Actinobacteria (Streptomyces, Actinoplanes, Saccharopolyspora, and Frankia). Phylogenetic analysis of the TylF-like proteins encoded in bacterial genomes are distinct from those found in fungal genomes (Figure S31) and have between 28.9–100% identity to each other (Table S6).

Closer inspection of these genome neighborhoods revealed that these loci encoded Rieske 2Fe-2S proteins, polyprenyl synthase, terpene precursor biosynthetic proteins, prenyltransferases, and cytochrome P₄₅₀ proteins, and transporter proteins (e.g., ABC transporters) (Figure 4 and Figure S11). This suggests that these BGCs have the capacity to biosynthesize meroterpenoids and that TylF-like proteins are likely terpene cyclases. These data suggest that TylF is the first described member of a class/family of previously unrecognized bacterial terpene cyclases that are distributed across bacterial genera. The current products that these BGCs produce are under investigation in our labs.

Figure 4.

BGCs identified using TylF as the BLAST probe for an EFI-EST search in cyanobacterial genomes.

Characterizing the role of TylH in tolypodiol biosynthesis.

To better understand tolypodiol biosynthesis and gather further evidence to link the tolypodiol BGC to compound production, we characterized the enzymatic activity of TylH. The encoding gene was expressed in E. coli and TylH was purified using Ni-NTA chromatography. Enzymatic assays were then performed by incubating the purified TylH with desmethyltolypodiol (12), which was obtained from hydrolysis of tolypodiol (2), SAM, and β-mercaptoethanol at room temperature overnight. Analysis of the enzymatic assay, by HPLC with standards, revealed almost complete conversion of 12into tolypodiol (2) (Figure 5 and Figure S 16). This conversion was not observed in any of the control reactions lacking TylH, SAM, or 12 (Figure S 16).

Figure 5. Conversion of Desmethyltolypodiol (12) to tolypodiol (2) by TylH.

A). Reaction diagram of desmethyltopdiol (12) with TylH to produce tolypodiol(2). B).HPLC chromatograms observing at 270 nm of (i) Full reaction containing 12, SAM, and TylH; (ii) Reaction containing 12 and SAM; (iii) standard of tolypodiol; (iv) standard of desmethyltolypodiol. Full chromatograms are shown in Figure S13.

TylH substrate specificity.

The activity of TylH was then probed to determine if TylH has stringent specificity towards desmethyltolypodiol (12) or if it can be used for the O-methylation of a variety of carboxylic acids. To test this, an enzymatic assay was performed the same as with 12 but using chroman-6-carboxylic acid (13) as the substrate. Chroman-6-carboxylic acid (13) was chosen as the substrate as it is structurally identical to a portion of hydrolyzed tolypodiol containing the carboxylic acid that undergoes O-methylation. Analysis of the enzymatic assay by HPLC revealed that the substrate did not undergo conversion (Figure S 17) into the corresponding methyl ester. In complimentary experiments, TylH did not convert 4-hydroxybenzoic acid (8) to the corresponding methyl ester (data not shown) thus showing that TylH has high specificity towards substrates with a pentacyclic structure or that the C, D, or E rings play a role in substrate binding, although this needs to be confirmed with future experiments. The failure to convert chroman-6-carboxylic acid (13) and 4-hydroxybenzoic acid (8) into the corresponding methyl ester also suggests that TylH mediated O-methylation occurs after cyclization of 9.. We examined unpurified and fractionated extract of Brasilonema sp. HT-58–2 but could not observe a signal with an m/z value consistent with 12. This suggests that the conversion of the carboxylic acid to the methyl ester is not the final step in the biosynthetic pathway suggesting the pathway seen in Figure 1, however due to the low ionization efficiency of the tolypodiols this statement is made noting that caveat. In depth in vitro reconstitution experiments are required to determine the exact order of biosynthetic steps.

New tolypodiol structures.

Simultaneously with the heterologous expression, we re-analyzed the extract of the original tolypodiol producer, Brasilonema sp. HT-58–2, and identified additional tolypodiol analogs which were subsequently identified and characterized to be 6-oxo-tolypodiol (5) and 1-oxo-tolypodiol (6). The molecular formula of 5 was determined by HRESIMS to be C₂₈H₃₈O₅ based on the detection of a protonated molecule at m/z 455.2790 [M+H]⁺ (calc for C₂₈H₃₉O₅⁺, 455.2792; −0.4 ppm). It became clear early in the structure elucidation that 5 bore many similarities with tolypodiol, which is a well-characterized compound isolated from HT-58–2.¹⁶ Specifically, the ¹H NMR spectrum revealed the 1,2,4-substituted benzene ring, methoxy functional group and five methyl signals. After negotiating through the remainder of the ¹H and ¹³C NMR spectra (Figure S18–19), it became apparent that the major difference was an oxidation at the C-6 position to a carbonyl (δ_C = 210.7) with regard to tolypodiol’s hydroxy (δ_C = 68.8),¹⁶ and a concomitant loss of the methine signal at 4.5 ppm, which would account for the observed mass difference of 2 amu. In addition, the diastereotopic protons at C-7 are further deshielded (δ_H 5: 2.32, 2.20; tolypodiol: 2.04, 1.23) and appear as doublets, while H-5 exhibits a change in multiplicity from a doublet in tolypodiol to a singlet in 5. The remainder of the planar structure was confirmed with a set of 2D NMR experiments (Figure S20–22). 2D-NOESY (Figure S23–24) along with relevant coupling constants were utilized to decipher the relative stereochemistry, which was determined to be the same as tolypodiol, except, of course, at the C-6 position where a carbonyl replaced the secondary alcohol.

HRESIMS was utilized to determine the molecular formula of 6 to be C₂₈H₃₈O₅ based on the identification of a protonated molecule at m/z 455.2788 [M+H]⁺ (calc for C₂₈H₃₉O₅⁺, 455.2792; −0.9 ppm). As in the case of 5, the molecular formula was 2 amu less than tolypodiol and the planar structure of 6 was recognized to be very similar to that of tolypodiol based on the 1D and 2D NMR spectra (Figure S25–29). It became evident that the major difference was an oxidation at the C-1 position from the hydroxy of tolypodiol to a carbonyl based on the lack of a proton resonance around 3.5 ppm that corresponded to the carbinol proton at C-1 of tolypodiol. An HMBC correlation indicated the presence of a carbonyl (δ_C = 214.7) at C-1 and the multiplicity of H-2a/b indicated coupling only to their diastereomeric partner and the adjacent diastereomeric protons, H-3a/b. Additionally, H2a/b were further deshielded (δ_H 5: 3.15, 2.12) than those of tolypodiol (δ_H 2: 1.68, 1.65) providing further evidence C-1 was oxidized. Analysis of the J_H-H coupling (H-5: 2.4 Hz; H-6: 2.4, 6.0 Hz) revealed the relative stereochemistry at C-6 was the same as tolypodiol.

For the determination of absolute configuration, ECD spectra of 5 and 6 were collected and compared to that of tolypodiol (Figure S30). The negative Cotton effect at around 265 nm suggests 5 and 6 have the same absolute configuration as tolypodiol, which is supported by the fact that the terpene cyclization is predicted to happen early in the biosynthetic pathway.

Conclusion

Cyanobacteria are a source of unique chemical structures, bioactivities, and enzymology in which the tolypodiols provide all three. Here we describe the identification and successful heterologous expression of tolypodiol (2) and related compounds 3, 4, 5, and 6. Located in the biosynthetic gene cluster is a gene encoding a previously undescribed putative terpene cyclase, tylF, that evidence and our data suggests belongs to a new family of membrane-bound bacterial terpene cyclases with similarity to fungal meroterpenoid cyclases. We were also able to identify similar terpene cyclases in other bacterial genomes including cyanobacteria and actinobacteria. We identified a SAM-dependent methyltransferase, TylH, encoded in the biosynthetic gene cluster and characterization by in vitro assays demonstrated that it was capable of methylating 12 to form 2, but was not able to methylate chroman-6-carboxylic acid (13) or 4-hydroxybenzoic acid (8). While we were unable to detect desmethyltolypodiol (12) in either Anabaena sp. UTEX 2576 containing the tyl BGC or in the native producer Brasilonema sp. HT-58–2, the substrate specificity of TylH towards the cyclized substrate suggests that methylation occurs after cyclization of 3-geranylgeranyl-4-hydroxybenzoic acid (9). The identification of similar BGCs in other bacterial genomes suggests that other meroterpenoid compounds can be identified or linked to cryptic BGCs. Our successful heterologous expression opens the potential to investigate the role of the tolypodiols and having a stable production host allows the in-depth investigation of the biosynthetic pathway.