Different Enzymes and Processes for Making Psilocybin in Inocybe and Psilocybe Mushrooms

BlastP analysis of the IpsD amino acid sequence (Table ) shows a standard binding pocket for the prosthetic group pyridoxal‐5′‐phosphate (PLP). Native IpsD shares 63% identical amino acids (aa) with the aromatic amino acid decarboxylase CsTDC of the mushroom.In contrast, PsiD, the gateway enzyme offormation inspecies, is phylogenetically entirely unrelated to IpsD, as the former belongs to the PLP‐independent phosphatidylserine decarboxylase family that features a self‐cleavage mechanism to generate the prosthetic group necessary for Schiff base formation (Scheme).,Thegene was heterologously expressed into produce an N‐ and C‐biterminally tagged fusion protein, purified by metal affinity chromatography (Figure), and assayed for activity in vitro. Assuming decarboxylation represents the entry step forbiosynthesis inas well, we first tested IpsD with 1 mas substrate and analyzed for product formation by LC‐MS. Surprisingly, tryptamine () formation was not observed, and not even traces were detected (Figure ; Scheme ), even after 16 h of incubation. This finding implies that IpsD does not acceptand, hence, cannot initiate the pathway.

In the previously elucidatedpathway, hydroxylation of the indole moietydecarboxylation. In contrast, mammalian serotonin formation involves ring hydroxylationthe decarboxylation step.,,We therefore tested whetherbiosynthesis infollows the mammalian order of events and added 1 m4‐hydroxy‐‐tryptophan () to the IpsD assay. The chromatographic analysis showed near‐quantitative turnover to(= 2.1 min,/177.1, [+H]Figure ; Scheme ), indicating thatbiosynthesis may take a dissimilar course than inspecies (Scheme ). Still, decarboxylation must precede methylation in the cascade, as a primary amine is necessary for the decarboxylase catalytic cycle, during which an intramolecular hydrogen bond is formed between PLP and the substrate to establish a tautomeric system (Scheme).

Kinetically, IpsD followed a typical Michaelis–Menten‐type response. We determined avalue of 66 µand aof 0.44 swhich translates into a catalytic efficiency of/ = 6.58 smwithas substrate (Figure). This efficiency is more than two‐fold lower than that ofPsiD (17.4 sm)but comparable to that of CsTDC (/ = 7.56 sm).

To date, only a low number of fungal tryptophan decarboxylases have been characterized for their substrate specificity. The PLP‐independent monomeric PsiD is unspecific and tolerates various ring substitutions.,,Likewise, the known PLP‐dependent aromatic amino acid decarboxylases CsTDC,and IasA*,are flexible and tolerate at least 5‐hydroxy‐‐tryptophan as well. In contrast, adecarboxylase rejected 5‐hydroxy‐‐tryptophan but curiously accepted phenylalanine and tyrosine.

A decarboxylase that strictly rejects unsubstituted, such as IpsD, is remarkable. To shed more light onto this phenomenon, we superimposed an IpsD model, created with AlphaFold 3 (Figure ), on the crystal structure 6EEW of CrTDC, the‐tryptophan decarboxylase of rose periwinkle ().In contrast to monomeric PsiD,aromatic amino acid decarboxylases and the IpsD model form obligate dimerswith buried interface surfaces of ∼6500 Å.

The substrate binding pockets of CrTDC and IpsD are formed by residues from both protomers of the dimer (Figure). The PLP‐linked lysine residue in IpsD and the exactly superimposed CrTDC PLP‐lysine accurately anchor the superposition of the models and thus allow a plausible assessment of the binding sites given the boundin the CrTDC structure (Figure ). The substitution of CrTDC F100 with IpsD Y79 offers a plausible explanation as to why IpsD binds to 4‐hydroxylated but not to unsubstituted: the additional hydroxy group generates a polar environment that allows, with structural relaxation, the formation of acceptor/donor interactions with the 4‐OH group, while at the same time disrupting the hydrophobic environment necessary to accommodate the aromatic indole group of the unsubstituted(Figure ). In summary, IpsD likely catalyzes the second biosynthetic step (Scheme ) and contrasts thedecarboxylase PsiD, which serves as a gatekeeping enzyme by catalyzing the initial step.

The second enzyme in the cascade familiar fromspecies (Scheme ) is the cytochrome P‐dependent monooxygenase PsiH that regioselectively introduces an oxygen atom to C‐4 of. However, this verified monooxygenaseand IpsH (Table), a putative cytochrome Penzyme as well, share only low sequence similarity (29.6% identical aa).

Like other Pmonooxygenases,IpsH possesses an N‐terminal helical membrane anchor which is clearly visible in an AlphaFold 3 in silico model (Figure ). The prediction model of IpsH reveals that, in addition to the ∼30 residue long helical membrane anchor common to cytochrome Penzymes,an unstructured, approximately 13 residue long N‐terminal extension is present. Superposition with 8YZ8, a peroxygenasein complex with adenine as a placeholder for a potentialbinding site, reveals an unusual insertion at G310 in the likely binding site that bulges out from a helix which is continuous in most Penzymes (Figure ). This insertion covers the likely binding site and thus could play a role inbinding. Future crystal structures are required to confirm these predictions.

Most Pmonooxygenases are insoluble.Hence, IpsH was not available for in vitro assays, yet mechanistic reasons allow its placement as the gateway enzyme of the Ips pathway: Dissimilar to thepathway, the substrate specificity of IpsD precludes hydroxylation after the decarboxylation step. Furthermore, hydroxylation must precede phosphate ester formation, and decarboxylation cannot follow methylation, as the removal of the carboxy group requires a primary amine in the substrate (Schemes and).

The phosphoryloxy group ofis a very rare structural feature among natural products. In, the phosphate ester is introduced by the kinase PsiK that falls into the thioribokinase family.,IpsK (Table) shows a very low degree of sequence identity with PsiK of only 22.4% identical aa. A search in UniProtwith IpsK as a query against characterized kinases did not return any further hits besides PsiK.

At this point, a first picture of the possible sequential enzymatic orders had emerged of howmay assemble(hydroxylation toby IpsH, followed by IpsD‐catalyzed decarboxylation to[Scheme ]). The question remained howformation is completed. To elucidate the late pathway steps, N‐terminally hexahistidine‐tagged IpsK (Figure) was assayed in vitro in TRIS‐HCl buffer, pH 7.5, with ATP (2 mas well as(1 m) as a phosphate acceptor substrate. In stark contrast to PsiK, LC‐MS analysis showed only minute amounts ofas an IpsK product (= 1.1 min, Figure ). To identify the authentic IpsK acceptor substrate after this surprising result, we next investigated other potential substrates. In separate assays, we investigated whether IpsK phosphorylates,,, and(Figure ).

Product formation was observed with all substrates except. However, onlyled to a quantitative turnover to(= 3.2 min,/285.1 [+H]Figure ). This particular reaction is catalyzed by the kinase PsiK of thepathway as well, yet does not fulfill a biosynthetic function there. Rather, it is considered a mechanism to protect the cells against the presence of free, which may form oligomers that may unspecifically inactivate proteins.,With, not even traces of a product were detected. Yet, the products(= 1.7 min,/271.1 [+H]) and aeruginascin (,= 2.9 min,/299.2 [+H]) resulted fromandas respective substrates (Figure ; Scheme ). Taking the pathway branch toward eitherorinto account, the kinetic investigation of IpsK was carried out with either, i.e., the dimethylated phosphate acceptor, oras the monomethylated acceptor substrate. IpsK showed a clear preference for the dimethylated substrate, evident by avalue of 20.5 µforversus 336.7 µfor. Thevalues were 0.69 sforand 0.37 sfor, resulting in catalytic efficiencies of/ = 33.51 and/ = 1.09 sm, respectively, forand(Figure).

To explain the dissimilar substrate requirements of PsiK and IpsK, we modeled the IpsK structure in silico. Given that two X‐ray structures of PsiK are available, both were included in our comparison. The first PsiK structure that was reported (9ETO) is a polyethylene glycol (PEG)‐mediated dimerwhile a subsequently published model, 8ZIC contains ADP and(TSS).The structure superposition of the two experimentally determined PsiK structures and the AlphaFold 3 model of IpsK is illustrated in Figure . Both IpsK and PsiK phosphorylateto, but IpsK exclusively requires methylated substrates. PsiK is indifferent to the methylation state, and its primary function in vivo is the phosphorylation ofto(Scheme ). Analysis of the binding pocket as to why IpsK hardly accepts the former with its primary amine side chain is difficult due to the ambiguous modelling ofin 8ZIC, which likely does not provide a good template for reliable modelling of thesubstrate. The electron density for the, modeled with partial occupancy in 8ZIC, is ambiguous. An alternative model of the ligand in 8ZIC with a PEG fragment placed into the same electron density (Figure) can be refined with higher real space correlation coefficients.PEG was present in high concentration in the crystallization cocktailand is then commonly observed in ligand binding sites.

In summary, our biochemical results suggest IpsK as the final biosynthetic enzyme, which implies a reversed order of methylation and phosphotransfer compared to the situation in, where iterative methylation by a single enzyme, PsiM, completes the biosynthesis (Schemes and ).

All known biosynthetic gene clusters inspecies code for only one methyltransferase, PsiM.,,In contrast, the published genomic sequence ofencodes two near‐identical methyltransferases, IpsM1 and IpsM2 (93.1% identical aa, Table). These are members of the methyltransferase family 25 and are, hence, neither phylogenetically related tomethyltransferases PsiM (forbiosynthesis) of family 10 nor to TrpM (,‐dimethyl‐‐tryptophan synthase), which falls into family 33.AlphaFold 3 models of IpsM1 and IpsM2 show the typical Rossmann fold harboring the methyl source SAM. However, as a consequence of the phylogenetic distance, structural alignment with experimentally determined portions beyond the conserved SAM binding core is poor.

Neither the superposition with‐bound PsiM (8PB4) nor a putative phosphoethanolamine‐methyltransferase of(4MWZ, unpublished) allowed meaningful predictions of substrate binding site details. Crystal structures of the enzyme complexes will be necessary to explain the differences between the methylation of the 4‐phosphorylated substratesandby PsiM versus the 4‐hydroxylated substratesand(IpsM2, IpsM1).

We addressed whether both enzymes are active and whether they catalyze a single or two consecutive methyl transfers. The heterologously produced and purified enzymes IpsM1 and IpsM2 (Figure) were assayed separately in vitro, again in TRIS‐HCl buffer, pH 7.5. The primary aminewas only poorly turned over by the kinase IpsK. Therefore, this compound appeared to be a more likely substrate for the two methyltransferases, which implies at least the first methyl transfer to precede the phosphotransfer. We offered, but also, as substrates (1 min separate reactions. Both IpsM1 and IpsM2 were active, as shown by LC‐MS analyses (Figure ), and the detected products pointed to similar catalytic activities. Both enzymes acceptedand catalyzed two consecutive methyl transfers, as evident by the productsand(Scheme ). However, we noticed quantitative differences of their ratio (Figure). Whilewas the major product of IpsM1 (= 8.6 min,/205.1 [+H]), we detected mainlyin the IpsM2 assays (= 7.5 min,/191.1 [+H]).

We compared the kinetics of these two methyltransferases, using bothandas methyl acceptors. Even though thevalues were comparable (101.0 µforand 138.2 µfor), thevalues (0.08 sforand 0.19 sfor) and the catalytic efficiencies (/ = 0.83 smforand 1.40 smfor) indicate a preference of IpsM1 for(Figure ). In contrast, the kinetic properties of IpsM2 showed an opposite preference: itsvalues were 64.5 µforbut 433.2 µfor. Likewise, we determined a = 0.11 sforand 0.08 sfor, which leads to catalytic efficiencies of/ = 1.72 smforand/ = 0.18 s mfor.

Subsequently, we sought to confirm that IpsM1/IpsM2 reject phosphorylated substrates. We offered,, and, i.e., the phosphorylated analogs of,, and, in separate single‐enzyme assays as substrates.was assayed for possibleformation (Scheme ). None of these substrates led to product formation (Figure). This finding is congruent with the previous biochemical results on these two methyltransferases and provides additional evidence for the dissimilar sequential order ofassembly in, compared to the pathway order inspecies (Scheme ).

To gain a more profound understanding of if and how IpsM1 and IpsM2 cooperate, we ran multi‐enzyme assays, in all cases withas substrate plus the required cofactors and co‐substrates PLP, ATP, SAM, and MgCl. A one‐pot reaction containing the four enzymes IpsD, IpsK, IpsM1, and IpsM2 led toformation (Figure ), as expected from the outcome of the single‐enzyme assays.

Minor amounts of intermediates,, and, along with the shunt product, were also detected. Likewise, the triple‐enzyme combination IpsD, IpsK, and IpsM1 led toformation as the main product as well, accompanied by.

To some degree, the methyltransferases are redundant in that they turn over, i.e., the chemically most unstable compound in the pathway.Still, following these in vitro results, the missing second methyltransferase (IpsM2) led to a quantitatively decreased and qualitatively shifted pathway product profile (Figure), pointing away from a simple scenario of two redundant methyltransferases fulfilling equal functions. This notion was supported by the opposite assay with IpsD and IpsK, but only IpsM2 for methyl transfer. In this combination,became the major product, whilewas only present in very modest titers (Figure ). Intriguingly,was found in equal or even higher titers thanin prior analytical works on the metabolite profile of.,,Of note, the pathway organization in, involving two semi‐redundant methyltransferases, makes bothandend products of a branched biosynthesis (Scheme ). Contrasting tertiary amine formation duringbiosynthesis in, the two methyl transfer steps are catalyzed sequentially and iteratively by one single methyltransferase, PsiM, as the ultimate step in. Unlike in,represents the direct precursor ofin, even though this second methyl transfer step is kinetically less favored than the first one.

The term "convergent evolution" describes the independent appearance of similar morphological or physiological phenotypes in distantly or unrelated groups of organisms. For primary metabolism, a textbook example for this phenomenon is the assembly of‐lysine in fungi along a pathway which is entirely unrelated to that in bacteria and plants.In the area of natural products of flowering plants, the pyrrolizidinesor the benzoxazinoidsare made along independently evolved pathways that rely, at least in part, on unrelated enzymes yet lead to the identical end product. Nitrogen‐containing betalains, or the pre‐anthraquinone pigments, emerged through parallel evolution in flowering plants or molds (ascomycetes) versus in mushrooms (basidiomycetes).,,

However, convergently evolved pathwaysthe mushrooms—and even within the same mushroom order—have not been biochemically verified yet but surely raise the question of which particular environmental pressure made the mushrooms evolve a metabolic pathway toand whetherandspecies had been exposed to the same selective pressure. As a symbiont, the former follows a different lifestyle thanspecies, which are saprotrophic wood‐ or dung‐inhabiting fungi that, consequently, inhabit different ecological niches. Previous research unambiguously established horizontal gene transfer as a major strategy to confer the capacity to produceonto mushrooms.,,Other established mechanisms to evolve new metabolic capacities in fungi are vertical gene duplicationand subsequent neofunctionalization, or de novo gene birth from previously non‐coding DNA.However, the evolutionary origin of thecluster remains unknown.

Building upon the hypothesis by Awan et al.,our results contribute biochemical evidence for a parallel‐pathway evolution by recruiting and evolving a set of distantly or unrelated enzymes. Of note, outside the mushrooms,was detected in cicada‐parasitizing evolutionary basal fungi (species).In these fungi unrelated to mushrooms, the biosynthesis genes have not been identified yet and may represent a third line of parallel evolution ofbiosynthesis and the first instance outside the mushrooms.

Mutually non‐exclusive theories on the ecological function ofhave been put forward and include modulation of the behavior of mycophagic predators, assuming neuroactive monomericis the ecologically relevant compound. Alternatively, oligo‐ and polymeric, formed after mycelial damage, may represent an induced defense to deter predators due to protein‐inactivating effects.,,Intriguingly, the twopathways found in mushrooms do not share any common reaction. While proceeding throughas the sole common intermediate, decarboxylation precedes regioselective cytochrome P‐mediated introduction of an oxygen atom at‐4 in, while these two early steps are reversed in. The two late steps follow an inverse sequence as well. Notably, the substrate spectrum of the kinase IpsK comprises, making it less available for the second methyl transfer reaction (Schemes and ) but favoring accumulation ofininstead, as both methyltransferases reject phosphorylated compounds.

The enzyme triple of IpsK, IpsM1, and IpsM2 may hence constitute a finely balanced system that leads toas long as both methyltransferases are active, yet allows the producerto shift the pathway towardby downregulating (or otherwise attenuating) IpsM1. Our findings explain earlier reports thatwas consistently and independently detected infruiting bodies, besides.,,Furthermore, the combined kinetic data support the concept of a kinetically regulated pathway branch: with the dimethylated substrateto yield, IpsK's catalytic efficiency is about 30 times higher than with monomethylatedwhich is converted to(Scheme ). IpsM1 is primarily catalyzing the second methyl transfer reaction and, thus, competes with IpsK for substrate. The higher catalytic efficiency of IpsM1 with, compared to IpsK, favors the branch toward. In return, IpsK shows a higher catalytic efficiency withthan IpsM2. Consequently, in the absence of IpsM1, i.e., the second methyltransferase, the kinase IpsK may potentially outcompete IpsM2, andwould preferentially be formed. Therefore, the preference of IpsM1 for the monomethylated substrateappears key to securingproduction and is congruent with the findings with the coupled assays (Figures and). Furthermore, the apparent role of IpsM2 is to supplyas a precursor.

The indolethylamine profile ofhas been exhaustively investigated.,,At the time of the first analyses,,had not been described yet, whereas in the latest study, it was detected in traces.The substrate preferences of both methyltransferases and the IpsK kinase, experimentally determined in this work, are in part compatible with these prior analytical observations: IpsK phosphorylatedto. However, based on our in vitro assays and using pure enzyme, this appears to be a non‐physiological reaction, as neither methyltransferase carries out a third methylation to establish a quaternary side chain ammonium necessary to provide IpsK with this particular substrate (Figures and). Yet,was detected in the eponymous sister speciesin higher quantities.Further research is warranted to address the biosynthetic enzymes, the number of methyltransferases, and their substrate specificities in the above‐mentioned species, for which genomic sequence data is not available yet, to elucidate the origin of this quaternary compound.

The indole alkaloidis under consideration as a drug to treat therapy‐resistant depression.Advanced clinical trials and possibly approval and introduction into clinical use in various countries will entail an increasing demand. The discovery of thegenes inspecies set the stage for heterologous production in surrogate hosts in vivo and in scalable cultures.,,,,In parallel, an in vitro procedure with immobilized enzymes constituting a reusable set of catalysts was recently devised.Regardless of the approach, the Ips enzymes, characterized in this work, contribute new enzymes to producebiotechnologically and sustainably, which adds an applied component to our work.