Substrate recognition by the 4‐hydroxytryptamine kinase PsiK in psilocybin biosynthesis

Escherichia coli BL21 (DE3) was transformed with pJF44, an expression vector derived from pET‐28a (Merck KGaA, Darmstadt, Germany) that encodes full‐length P. cubensis PsiK (GenBank entry ASU62237.1↗) preceded by a hexahistidine tag and a recognition site for TEV protease. Cultures were grown at 37 °C in LB medium supplemented with 50 μg·mL⁻¹ kanamycin and 1% glucose. At an OD_600nm of 0.5, isopropyl β‐d‐1‐thiogalactopyranoside (IPTG) was added to a final concentration of 1 mm, followed by protein expression for 5 h at 16 °C. Upon centrifugation (60 min at 10 000 g, 4 °C), cells from 2 L of culture were resuspended in 30 mL buffer A (20 mm Tris/HCl pH 8.0, 400 mm NaCl) and disrupted by sonication on ice using a Sonopuls HD 2070 (Bandelin Electronic GmbH, Berlin, Germany). Insoluble material was removed by centrifugation for 1 h at 40 000 g and 4 °C. The lysate was then passed through a 0.45 μm filter (Whatman GmbH, Dassel, Germany) and applied to a 5 mL metal affinity chromatography Ni²⁺‐cartridge (GE Healthcare GmbH, Munich, Germany). After extensive washing, PsiK was eluted using a 0–500 mm imidazole gradient in buffer A. Peak fractions were pooled and dialysed for 12 h against 2 L buffer A at 4 °C to remove imidazol. This was followed by the addition of 0.5 mL TEV protease (1 mg·mL⁻¹) and ADP to a final concentration of 0.25 mm. After a further 48 h at 4 °C, the solution was passed through the Ni²⁺‐cartridge again and PsiK eluted by means of a 0–100 mm imidazole gradient in buffer A. Selected fractions were pooled and ADP was added to a final concentration of 1 mm. The protein was further purified by gel filtration (Superdex 200, GE Healthcare GmbH) in a buffer containing 20 mm Tris/HCl pH 8.0, 400 mm NaCl and 1 mm DTT. Peak fractions were pooled and concentrated to 9.4 mg·mL⁻¹ (0.20 mm) using a Vivaspin 20 ultrafiltration unit (10 kDa MWCO, Sartorius Lab Instruments GmbH, Göttingen, Germany). Protein concentration was determined spectroscopically, making use of a sequence‐based estimate of the molar absorption coefficient at 280 nm (47 650 m⁻¹·cm⁻¹). Aliquots of the purified protein were flash‐frozen in liquid nitrogen and stored at −80 °C until further use.

PsiK was crystallised at 4 °C using the sitting drop method. Drops of 0.4 μL were produced by mixing equal volumes of the protein solution and the reservoir buffer. The latter contained 100 mm HEPES pH 7.5, 2 m ammonium sulphate and 2% (w:v) PEG 550 MME. After several weeks, monoclinic crystals belonging to space group C2 began to appear.

Crystals were flash‐cooled in liquid nitrogen without the addition of further cryoprotectants. For X‐ray diffraction experiments, a temperature of 100 K was used. Data were collected at ESRF ID23‐2 [25] and processed using xds [26, 27]. Structures were solved by molecular replacement using phaser [28], in combination with the ccp4 suite [29]. A monomeric structure generated by alphafold [30, 31] was used as the search model. coot [32] and refmac [33, 34] were used for iterative rounds of model building and refinement. Data collection and refinement statistics can be found in Table 1.

All molecular graphics illustrations were produced with pymol 0.99rc6 (DeLano Scientific LLC, Palo Alto, CA, USA). Structural comparisons were performed using the SSM superposition method implemented in coot [32].

Mutagenesis of the psiK gene involved PCR amplification of the pET‐28a‐derived expression plasmid pJF23 [10]. Each gene variant was amplified in two fragments that overlapped in the area to be mutated, using the primers listed in Table S1. Thermal cycling conditions were: 98 °C for 30 s at the start; 35 cycles of 98 °C for 8 s, 61 °C for 20 s and 72 °C for 50 s; 72 °C for 5 min at the end. Phusion DNA polymerase (New England Biolabs GmbH, Frankfurt, Germany) was used according to the manufacturer's instructions. The dNTPs were added to 200 μm and oligonucleotide primers to 0.5 μm final concentration. Oligonucleotide oKR102 was used as the forward primer to amplify the upstream fragments, while oKR103 served as the reverse primer for the downstream fragments.

The 5′‐end of all upstream fragments and the 3′‐end of all downstream fragments overlapped with the multiple cloning site of vector pET‐28a, which was linearised by BamHI and HindIII restriction. Plasmids were assembled with the NEBuilder HiFi DNA Assembly Kit (New England Biolabs GmbH). The resulting expression plasmids, pKR20‐pKR23 and pTW04‐pTW07, correspond to PsiK variants W319A, W226A, D249A, E251A, F319A, Y264A, L252A and L184A, respectively.

To produce hexahistidine‐tagged PsiK wild type and variants, E. coli KRX (Promega GmbH, Walldorf, Germany) was transformed with plasmid pJF23 (for PsiK wild type) or the expression plasmids mentioned above. For each transformant, a 5 mL culture was grown overnight at 37 °C in LB medium supplemented with 50 μg·mL⁻¹ kanamycin and then used to inoculate 500 mL of 2 × YT medium with 50 μg·mL⁻¹ kanamycin. This culture was incubated at 37 °C to an OD_600nm of 0.6. After induction with l‐rhamnose to a final concentration of 0.1%, the cultures were shaken for another 20 h at 16 °C. The biomass was collected by centrifugation at 4000 g and 4 °C, resuspended in 7.5 mL lysis buffer (50 mm phosphate buffer, 500 mm NaCl, 20 mm imidazole, pH 8) and lysed with a Bandelin Sonopuls GM 3200 sonicator. To remove cell debris, the lysate was centrifuged (13 751 g, 20 min, 4 °C). The cleared supernatant was passed through a 0.45 μm filter (Whatman) and purified using a 1 mL HisTrap HP column (Cytiva Europe GmbH, Munich, Germany). Following an extensive wash, PsiK was eluted using a 0–250 mm imidazole gradient in lysis buffer. Purification was verified by SDS/PAGE (Fig. S1), and the protein was quantified photometrically at a wavelength of 280 nm. For use in subsequent activity assays, the enzyme was rebuffered on an Amicon column (Merck KGaA) and eluted with 50 mm phosphate buffer (pH 7.5) for subsequent activity assays.

Unless stated otherwise, kinase activity assays were set up in triplicates in a total volume of 50 μL. All reactions contained 1 mm MgCl₂, 1 mm ATP and 0.5 mm substrate (either 4‐HT or psilocin). Reactions were started by adding PsiK (0.1 μm final concentration) and incubated at 40 °C. After 4 min, reactions were stopped by flash‐cooling in liquid nitrogen. The material was then lyophilised, resuspended in 50 μL methanol and placed in an ultrasonic bath for 3 min, followed by centrifugation.

Chromatographic analysis was performed on an Agilent Infinity II UHPLC–MS instrument, equipped with a 6130 single quadrupole mass detector, and by electrospray ionisation in positive mode, and a Phenomenex Luna Omega polar C₁₈ column (50 × 2.1 mm, 1.6 μm particle size). Solvent A was 0.1% (v/v) formic acid in water, solvent B was acetonitrile. To detect tryptamines, the following gradient was applied: 0 min, 1% B; within 3 min to 5% B; within another minute to 100% B; held at 100% B for 2 min. The flow rate was 0.5 mL·min⁻¹.

The X‐ray structure of PsiK, crystallised in the absence of any substrate, was determined at a resolution of 2.5 Å (Fig. 2, Table 1). Consistent with sequence‐based predictions, the enzyme adopts the typical bilobed structure found in eukaryotic protein kinases (Fig. 2A). The small N‐terminal lobe (comprising residues 1–118) consists of a five‐stranded antiparallel β‐sheet and two α‐helices with αβββαββ topology, while the larger C‐terminal lobe (residues 126–362) is almost entirely α‐helical. The N‐ and C‐terminal lobes are connected by a short linker (residues 119–125).

The two PsiK molecules in the asymmetric unit of the crystal (ASU) are related by 2‐fold non‐crystallographic symmetry (NCS) and can be superimposed with a C_α‐RMSD of 0.13 Å for 363 matched atoms. Intriguingly, the A and B chains appear to form a homodimer, mainly through contacts between the N‐terminal lobes (Fig. 2B). This mode of association is unusual and does not resemble protein–protein interactions previously observed in known homodimeric kinases. The interaction between PsiK molecules is mediated to a large extent by a polyethylene glycol 550 monomethyl ether (PEG 550 MME) molecule and three sulphate ions from the crystallisation buffer, which are sequestered at the interface between the neighbouring protein chains. Analysis using the PDBePISA server [35] suggests that the homodimeric assembly is not stable in the absence of PEG 550 MME and sulphate, with a negative ΔG^diss of −6.2 kcal·mol⁻¹. It therefore seems likely that the dimerisation of PsiK in our structure is a crystallisation artefact. This is consistent with the results of gel filtration experiments, which indicate that PsiK exists as a monomer in solution [14].

Dali [36, 37] analysis identified the methylthioribose (MTR) kinases from Bacillus subtilis (PDB code 2PUN↗, see Fig. 3A) and Arabidopsis thaliana (PDB code 2PYW↗) as the most similar PDB entries [38, 39]. The Dali Z‐scores for these structures are 27.0 and 26.6, respectively, with C_α‐RMSD values of 3.1 and 3.2 Å (307 superimposed atoms). PsiK shares several structural features with the MTR kinases that were previously identified as distinguishing the latter from true protein kinases [38]. Like the MTR kinases, PsiK lacks the characteristic DFG motif typically found at the beginning of the activation loop of protein kinases. Moreover, an atypical DXE motif (residues 249–251, equivalent to residues 250–252 in the MTR kinase of B. subtilis) is likely to bind the catalytically important Mg²⁺‐ion that interacts simultaneously with the β‐ and γ‐phosphates of ATP. In sharp contrast to PsiK and the majority of protein kinases, however, the two MTR kinases are known to form stable homodimers in solution [38, 39].

Superposition of PsiK and the MTR kinase of B. subtilis reveals that in the former the small and large lobes are separated by an additional ~ 7 Å (Fig. 3A). Thus, PsiK adopts an ‘open’ nucleotide‐free conformation, reminiscent of the inactive conformation of regulatory protein kinases in eukaryotes [40, 41]. This observation is somewhat surprising, as the more closely related MTR kinases have been proposed to exclusively exist in the active (or ‘closed’) conformation, even in the absence of nucleotides and substrates [38]. It remains possible, however, that the open conformation seen in PsiK is imposed by crystal packing forces.

The MTR kinase of B. subtilis mainly recognises its substrate by means of an acidic residue (Asp233), a twin arginine motif (Arg340, Arg341) and the so‐called W‐loop (residues 64–79) [38]. Superposition of the crystal structures of the MTR kinase and PsiK shows that Asp233 in the former corresponds to Asp224 in the latter (Fig. 3B). This suggests that the negatively charged side chain of Asp224 in PsiK likewise participates in substrate binding, for instance by recognising the protonated ethylamine moiety of 4‐HT and psilocin. In apparent agreement with this hypothesis, a D224A mutant of PsiK was found to be entirely inactive [14]. The twin arginine motif of the B. subtilis MTR kinase is not conserved in PsiK, but appears to be structurally equivalent to a pair of solvent‐exposed hydrophobic residues (Met315 and Trp316) that could interact with the indole ring of the substrate. Together with Thr180, Leu184, Trp226, Leu252 and Phe319, these residues form a hydrophobic pocket that in our X‐ray structure sequesters part of a PEG molecule from the crystallisation buffer (Figs 2 and 3). The W‐loop of the B. subtilis MTR kinase does not have an equivalent in PsiK. Instead, the corresponding residues in the protein sequence extend helix α2, which in PsiK is several turns longer than in the MTR kinases.

In order to confirm the location of the putative substrate‐binding site in PsiK we produced six alanine mutants, targeting hydrophobic side chains that we predicted to play a role in ligand recognition (F319A, Y264A, W316A, W226A, L184A and L252A). The acidic residues of the DXE motif, expected to interact with the catalytically important Mg²⁺‐ion, were mutated to alanine in two further variants (D249A and E251A). The activity of each mutant was measured in two independent in vitro assays, one with the biosynthetic precursor 4‐HT and the other with the hydrolysis product psilocin (Fig. 4A,B, respectively). Irrespective of the substrate, the D249A and E251A variants were completely inactive, reflecting the critical role of the DXE motif in positioning the catalytic Mg²⁺‐ion. All of the other substitutions also affected phosphorylation of the 4‐HT substrate, particularly W316A, W226A, L184A and L252A. The negative effect of the F319A and Y264A substitutions was smaller but still statistically significant. Thus, all of the hydrophobic residues that we mutated are likely to play a role in substrate recognition.

For most variants, activity towards psilocin mirrored the results from the 4‐HT assay. In contrast, PsiK^Y264A was found to be fully active when phosphorylating psilocin, while PsiK^F319A showed a significant increase (2.5‐fold, with n = 6 and P = 0.001) in turnover compared to PsiK^wt. These substrate‐dependent effects may indicate that residues Y264 and F319 are in direct contact with the ethylamine moiety, an interaction likely to be affected by the presence of the methyl groups in psilocin.

The crystallographic structure of PsiK and the corresponding X‐ray diffraction data have been deposited in the PDB with accession number 9ETO↗.