Direct Phosphorylation of Psilocin Enables Optimized cGMP Kilogram-Scale Manufacture of Psilocybin

The investigators began by adapting a literature-based multigram route to psilocybin and systematically identified scale-up failure modes. Key process development goals were to minimise persistent byproducts (notably tetramethyloxamide), reduce the need for large excesses of reducing agent, develop a rapid purification of psilocin without chromatography, implement a direct phosphorylation of psilocin to psilocybin, and avoid high-temperature aqueous recrystallisation that promoted hydrolysis. Process changes introduced in the second-generation route included: formation of acyl chloride 2 from 4-acetoxyindole using 1.2 equivalents of oxalyl chloride in methyl tert-butyl ether (MTBE) at –10 to 0 °C, followed by heptane precipitation and washes monitored for residual oxalyl chloride via aniline derivatisation by HPLC; amidation of 2 with 2 M dimethylamine in THF and triethylamine as base, with recrystallisation from isopropanol and water washing to yield ketoamide 3; and a scaled reduction of 3 to psilocin (4) using LiAlH4 in 2-methyltetrahydrofuran (2-Me-THF) with the improved reverse-addition protocol (slurry of 3 added to LAH at >60 °C) and a modified quench using THF/H2O plus silica gel and sodium sulfate followed by filtration through celite and silica to provide a colourless solution and precipitation into heptane/diisopropyl ether. For the final phosphorylation step, the team screened phosphorylating agents and found direct phosphorylation of psilocin with POCl3 in THF to be viable. The optimised conditions used 1.5 equivalents of POCl3, inclusion of celite to avoid formation of a sticky, insoluble mass, and tight control of reaction time (ideally ≤2 h). The crude phosphorodichloridate intermediate was hydrolytically quenched into cold (≤0 °C) 30% aqueous THF containing six equivalents of triethylamine, then extracted and precipitated by addition of isopropanol and distillation to isolate crude psilocybin. Final purification consisted of sequential reslurries in methanol and warm (45–55 °C) water to improve purity and to obtain the crystalline trihydrate, which was then dried to an anhydrous polymorph under controlled temperature to produce the final API. Analytical and in-process controls reported include HPLC/UPLC monitoring of reaction consumption and impurity profiles (including microquench-HPLC during phosphorylation), aniline derivatisation to detect residual oxalyl chloride in washes, AgNO3 precipitation to visualise residual triethylamine·HCl in filtrates, XRPD to confirm solid forms, and preparative reversed-phase HPLC plus MS/NMR for impurity characterisation when needed. Several steps included specific temperature ranges, hold-time constraints, and wash/solvent-exchange sequences to limit degradation or side-product formation.

First-generation, literature-adapted processing provided a functional multistep route but displayed multiple scale-up issues. Telescoping the oxalyl chloride step into the amidation introduced residual oxalyl chloride that reacted with excess dimethylamine to form tetramethyloxamide, a persistent impurity. This impurity necessitated the use of large excesses (10 equivalents) of LiAlH4 in the reduction stage, generating substantial aluminium salts that complicated work-up and exposed psilocin to oxidative degradation. The literature phosphorylation employed tetrabenzylpyrophosphate (TBPP) and cryogenic n-butyllithium chemistry to install benzyl-protected phosphate groups, followed by kinetically variable benzyl migration during a solvent swap and an exhaustive catalytic hydrogenolysis to remove benzyl groups. This sequence had poor atom economy and, on scale, produced problematic ultrafine precipitates that hindered filtration; overall it afforded only 99 g of final psilocybin from a 0.65 kg intermediate, corresponding to an overall yield of about 5% and 99.8% HPLC purity. The second-generation process delivered substantial improvements at scale. The acyl chloride 2 was isolated in 92% yield and >98% HPLC area purity after MTBE/heptane precipitation and washes. Ketoamide 3 was obtained in 83% yield and >98% area purity using triethylamine rather than pyridine and a recrystallisation protocol. The LiAlH4 reduction to psilocin was improved by reverse addition into LAH in 2-Me-THF, heating to reflux to reduce persistent β-hydroxy intermediate, and by changing the quench to THF/H2O with silica gel and sodium sulfate followed by filtration through celite and silica; on kilogram scale this step provided >70% yield and 94.4% area purity (3.09 kg isolated, 73% reported in one run, 94.4% by HPLC). Direct phosphorylation of psilocin with POCl3 (1.5 equivalents) in THF, aided by addition of celite, gave rapid conversion to a labile phosphorodichloridate intermediate (7), typically near completion within 10 minutes. Side reactions developed with prolonged reaction hold times, so the authors recommend proceeding to hydrolytic quench within about 2 h. Hydrolysis was achieved by quenching into 30% aqueous THF containing six equivalents of triethylamine at ≤0 °C; the crude aqueous phase was worked up and psilocybin precipitated with isopropanol and distillation to afford crude psilocybin in about 50–55% yield and ~98% purity, with psilocin being the principal impurity. Sequential reslurries—first in methanol, then in warm water (45–55 °C)—raised purity to ca. 98.9% after methanol and to about 99.9% after warm-water reslurry in development batches up to 90 g. XRPD showed isolation as a crystalline methanol solvate and, after water work-up, as a crystalline trihydrate; controlled drying at 45–55 °C for 24 h converted the trihydrate to an anhydrous polymorph A, whereas higher temperatures or longer drying induced polymorphic changes. An unidentified impurity at relative retention time (RRT) 0.58 exceeded the ICH Q3A unidentified-impurity threshold (0.11%) on kilo-scale. Preparative isolation and spectroscopic analysis identified this impurity as a pyrophosphate species (structure 10), plausibly formed when excess POCl3 promoted pyrophosphate generation. The impurity hydrolysed back to psilocybin under acidic aqueous conditions, indicating it could be purged by adjusting aqueous hold times. Implementing these controls, the authors completed a first cGMP production run producing 1.21 kg of API with 17% overall yield from 4-acetoxyindole, a 31% yield for the direct phosphorylation step, and 99.7% UPLC assay purity.

The investigators interpret their results as demonstrating that a purpose-designed, second-generation route can overcome the reproducibility and scalability limitations encountered when adapting literature procedures to cGMP kilogram manufacture. Key improvements cited include isolation and control of the acyl chloride intermediate to avoid oxalyl chloride carryover and tetramethyloxamide formation, a modified LiAlH4 reduction strategy that minimises persistent intermediates and product exposure to degrading conditions, and the novel direct phosphorylation of psilocin using POCl3 that obviates the benzyl-protecting-group strategy and the subsequent hydrogenolysis step that had poor atom economy. The authors acknowledge several process sensitivities and limitations encountered during development. Psilocin is oxidation-prone and degrades rapidly in solution, so short processing times and controlled quench/work-up are essential. The POCl3 phosphorylation is kinetically fast and develops unwanted side reactions if held too long; hydrolytic quench timing and temperature are therefore critical. An unidentified pyrophosphate impurity was encountered at kilo scale but was characterised and found to be hydrolysable to psilocybin, allowing process adjustments to meet ICH Q3A impurity limits. The long-term stability of acyl chloride 2 beyond 24 h was not established in the extracted text. Solid-form control also required attention—reslurry and drying conditions were used to access a consistent anhydrous polymorph A, while avoiding temperatures or drying times that induce polymorphic conversion. Taken together, the authors state that the route’s scalability, in-process controls, and avoidance of inefficient protecting-group chemistry render it suitable to meet current clinical demand for cGMP psilocybin and to serve as a viable basis for potential future commercial manufacture. They note that certain impurity- and stability-related parameters require active control in production-scale runs.

The second-generation synthesis enabled manufacture of 1.21 kg of cGMP psilocybin with 99.7% UPLC assay purity and an overall 17% yield from 4-acetoxyindole; the direct phosphorylation step proceeded in 31% yield. The redesigned route addressed variability and control issues observed with literature procedures when scaled, by implementing an optimised three-step conversion to psilocin with enhanced in-process controls and a novel direct phosphorylation that avoids benzyl protecting groups. The authors conclude that the procedure is scalable, controllable, and reproducible, providing sufficient API for current clinical needs and a practical approach for future commercial manufacture.