Chemicals and reagents

Pure standards of dipyanone, fentanyl, metonitazene, methadone-d9, SNC-80, and U-4880 were purchased from Cayman Chemical (Ann Arbor, Michigan, USA), diclofenac, while ammonium acetate, β-glucuronidase and dimethyl sulfoxide (DMSO) were bought from Sigma Aldrich (Milan, Italy). The standards were prepared as 1 mg/mL solutions in methanol and stored at – 20 °C until analysis. Carlo Erba (Cornaredo, Italy) supplied LC–MS grade methanol, acetonitrile, water, and formic acid. William’s medium E, HEPES buffer (2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic acid), and L-glutamine were obtained from Sigma Aldrich. Supplemented William’s Medium E (SWM) was prepared by adding HEPES (2 mmol/L) and L-glutamine (20 mmol/L) to William’s medium E. The solution was stored at 4 °C until use in incubation experiments. Thawing medium, 0.4% trypan blue, and ten-donor-pooled cryopreserved human hepatocytes were supplied by Lonza (Basel, Switzerland). Human MOR, DOR, and KOR membrane and the GTP Gi Binding assay kit were purchased from Revvity (Milan, Italy). The kit included the following reagent and stock solutions: GTP Eu Cryptate reagent, GTP d2 antibody, GDP, magnesium chloride (MgCl2) GTPγS, Gi protein control and Stimulation Buffer.

In silico metabolite prediction

The molecular structure of dipyanone was represented using SMILES (Simplified Molecular Input Line Entry System), a line notation system that encodes chemical structures as compact text strings. These SMILES strings, generated using ChemSketch (Advanced Chemistry Development, Inc.; v. 2020.1.2), were then utilised in GLORYx, an open-access software collaboratively developed by the University of Vienna, Austria and the University of Hamburg, Germany (Carlier et al. 2022). GLORYx was employed to predict phase I and phase II human metabolites of dipyanone. The software assigns a prediction score to each metabolite, indicating its likelihood of formation. Metabolites with a prediction score of 25% or higher were selected for further analysis. To simulate additional metabolic transformations, these selected metabolites were reprocessed through GLORYx, generating “second-generation metabolites”. The final score of each second-generation metabolite was calculated by multiplying its score with that of its corresponding first-generation metabolite. Again, only those with a final score of 25% or higher were retained. The selected metabolites were added to the LC-HRMS/MS inclusion list, and their corresponding metabolic transformations were included in the list of predicted transformations for data mining. This two-step prediction process aimed to more comprehensively model the potential metabolic pathways of dipyanone in the human body, accounting for both immediate and subsequent metabolic transformations. This approach aided in identifying a wider range of potential metabolites that might be formed during drug metabolism, which is crucial for understanding the drug’s behaviour and potential effects in the body.

Hepatocyte incubations

Dipyanone incubations with human hepatocytes were carried out as previously described (Carlier et al. 2022; Taoussi et al. 2024). Briefly, the hepatocytes were thawed in 50 mL of thawing medium at 37 °C. After centrifugation (100 g, 5 min), the supernatant was discarded and the pellet was resuspended in 50 mL of SWM at 37 °C. Following a second centrifugation (100 g, 5 min), the supernatant was removed, and the pellet was resuspended in 2 mL of SWM at 37 °C. Cell viability was determined using the trypan blue exclusion method, and the SWM volume was adjusted to reach a concentration of 2 × 106 cells/mL. In sterile 24-well culture plates, 250 µL hepatocyte suspension was gently mixed with 250 µL of dipyanone (20 µmol/L in SWM). The plates were incubated at 37 °C, and reactions were stopped after 0 or 3 h with 500 µL ice-cold acetonitrile, followed by centrifugation (15,000 g, 10 min). Negative (without SWM, hepatocytes, or dipyanone) and positive controls (diclofenac incubation) were incubated for 0 and 3 h under the same conditions to exclude nonspecific reactions and ensure proper metabolic activity. Incubates were stored at – 80° until analysis.

Authentic samples

Biological samples from two dipyanone-positive forensic cases were analysed to confirm the in vitro metabolite predictions.

In case #1, postmortem femoral blood and urine were collected at the autopsy. Dipyanone concentrations in blood and urine were 720 and > 1000 ng/mL, respectively. No other substances of toxicological interest were detected.

In case #2, dipyanone concentrations in autoptic heart blood and urine were 80 and 5,500 ng/mL, respectively. Other substances of toxicological interest that were detected in blood included: 2-fluoromethamphetamine (96 ng/mL), 2-fluoroamphetamine (24 ng/mL), deschloroketamine (1.0 ng/ml), 2-fluoro-deschloroketamine (< 1.0 ng/mL), deschloro-N-ethylketamine (46 ng/mL), mitragynine (160 ng/mL), and 7-hydroxymitragynine (4.7 ng/mL). In urine: 2-fluoromethamphetamine (not quantified), deschloroketamine (23 ng/mL), 2-fluoro-deschloroketamine (> 50 ng/mL), deschloro-N-ethylketamine (> 50 ng/mL), 2-fluoroamphetamine (not quantified), mitragynine (> 200 ng/mL), and 7-hydroxymitragynine (> 200 ng/mL).

Sample preparation for metabolite identificationIncubates

A 100 µL volume of incubate was mixed with 100 µL of acetonitrile and centrifuged for 10 min, 15,000 g at room temperature for protein precipitation (acetonitrile:SWM, 3:1, v/v). The supernatant was dried under a nitrogen stream at 37 °C. The residue was reconstituted in 100 µL of a mixture containing 95% mobile phase A (0.1% formic acid in water) and 5% mobile phase B (0.1% formic acid in acetonitrile). Again, the solution was centrifuged (10 min, 15,000 g) at room temperature. The resulting supernatant was transferred into vials with a glass insert, and 10 µL was injected into the chromatographic system for analysis.

Urine samples

Samples were thawed at room temperature. A 100 µL aliquot was mixed with 200 µL of acetonitrile and centrifuged for 10 min at 15,000 g at room temperature. The supernatants were evaporated to dryness under nitrogen at 37 °C. The residues were reconstituted in 100 µL of a mixture containing 95% mobile phase A and 5% mobile phase B (v/v). After centrifugation under the same conditions, the supernatants were transferred to autosampler vials with glass inserts. A 10 µL volume was injected into the chromatographic system for analysis.

To investigate glucuronic acid conjugation, 100 µL of urine was mixed with 10 µL of 10 mol/L ammonium acetate (pH 5.0) and 100 µL of β-glucuronidase (5000 units), then incubated at 37 °C for 90 min. Subsequently, 400 µL of ice-cold acetonitrile was added to the mixtures, which were centrifuged for 10 min at 15,000 g at room temperature. The supernatants were evaporated to dryness under nitrogen at 37 °C and reconstituted in 100 µL of a mixture containing 95% mobile phase A and 5% mobile phase B (v/v). After centrifugation under the same conditions, the supernatants were transferred to autosampler vials with glass inserts. A 10 µL volume was injected into the chromatographic system for analysis.

Instrumental conditions for metabolite identification

The analyses were conducted by liquid chromatography-high-resolution tandem mass spectrometry (LC-HRMS/MS) with a DIONEX UltiMate 3000 liquid chromatographer coupled to a Q Exactive quadrupole-Orbitrap hybrid high-resolution mass spectrometer equipped with a heated electrospray ionisation (HESI) source from Thermo Scientific (Waltham, Massachusetts, USA).

Liquid chromatography conditions

The compounds were separated using a Kinetex Biphenyl column (150 × 2.1 mm, 2 μm) from Phenomenex (Torrance, California, USA). The separation was achieved using mobile phases A and B at a flow rate of 0.4 mL/min. The gradient elution program was as follows: 2% B was held for 2 min, then increased to 25% B over 12 min, followed by a rapid increase to 95% B within 2 min, and held for 4 min. Subsequently, the initial conditions were restored within 0.1 min and maintained for 3.9 min. The total chromatographic run time was 24 min. Throughout the analysis, the column oven temperature was maintained at 37 ± 1 °C, while the autosampler temperature was set to 10 ± 1 C.

Mass spectrometry conditions

All samples were analysed in both positive- and negative-ion modes, requiring two separate injections while maintaining the same HESI conditions. These conditions were as follows: spray voltage, ± 3.5 kV; sheath gas and auxiliary flow rates, 50 a.u. and 10 a.u., respectively; capillary temperature and auxiliary gas heater temperature, 300 °C; and S-lens radio frequency level, 50 a.u.; Notably, the sweep gas flow rate was not utilised. Prior to each analytical session, mass calibration was performed using certified calibration solutions in both positive and negative ion modes. To enhance accuracy, a lock mass list was compiled for positive- (m/z 279.0933, 144.9821, 146.9803) and negative ion modes (m/z 265.1479, 162.9824, 248.9604).

The mass spectrometer acquired from 1 to 20 min of the chromatographic run in full-scan HRMS (FullMS)/data-dependent MS/MS (ddMS2) mode. The FullMS settings were as follows: range, m/z 100 to 650; resolution at full width at half maximum at m/z 200, 70,000; automatic gain control target, 1 × 106; and maximum injection time, 200 ms. For ddMS2, the settings were: automatic gain control target, 2 × 105; maximum injection time, 64 ms; isolation window, m/z 1.2; resolution, 17,500; and stepped normalised collision energy, 40, 70, and 90 a.u. A maximum of five ddMS2 scans were triggered for each FullMS scan, with a minimum intensity of 104 and dynamic exclusion of 2.0 s..

The data-dependent acquisition relied on an inclusion list of putative metabolites (Supplementary Table S1) based on in silico predictions (Supplementary Table S2) and extrapolation from the metabolism of structural analogues (Ferrari et al. 2004; Manier et al. 2024).

Additionally, ions not included in the inclusion list could also trigger ddMS2 scans, albeit at a lower priority (using the “pick others if idle” option). Furthermore, an exclusion list was compiled based on background noise, as evaluated during the injection of blank control samples (A:B 95:5 v/v).

Data mining for metabolite identification

LC-HRMS/MS data were processed with Thermo Scientific Compound Discoverer in a single analysis. Following a previously described workflow (Berardinelli et al. 2024; Taoussi et al. 2024), the detected ions were compared to a list of theoretical metabolites. This list was generated according to the settings displayed in Supplementary Table S3, with an intensity threshold of 5 × 103 and an HRMS mass tolerance of 5 ppm. Additionally, the HRMS/MS spectra and theoretical elemental compositions of the ions were compared to mzCloud (Drugs of Abuse/ Illegal Drugs database), ChemSpider (Cayman Chemical, DrugBank), and HighResNPS online databases. For these comparisons, an intensity threshold of 105; an HRMS mass tolerance of 5 ppm, and an HRMS/MS mass tolerance of 10 ppm were applied. For final identification, the intensity threshold was set at 1% of the signal of the most intense metabolite in the corresponding sample.

HTRF-based GTP Gi binding assay

MOR, KOR, and DOR activation were evaluated through a HTRF®-based GTP Gi binding assay. In this assay, a drug is incubated with a membrane preparation from cells that express recombinant or endogenous receptors, containing G protein-coupled receptors (GPCR), specifically MOR, KOR, or DOR. The activation of the GPCR through agonist binding leads to the replacement of the GDP nucleotide in the receptor G alpha subunit by a non-hydrolysable GTP coupled to a fluorescent europium cryptate donor (Eu-PGT). When a d2-labelled anti-Gαi monoclonal antibody acceptor is in proximity to the donor, a fluorescence resonance energy transfer (FRET) signal is emitted. This signal, proportional to the Gαi activation state, can be measured at a specific wavelength (Koval et al. 2010; Rozwandowicz-Jansen et al. 2010).

The total assay volume was 20 µL, containing a supplemented stimulation buffer with optimised GDP and magnesium chloride concentrations, dipyanone or a reference compound (MOR, fentanyl; KOR, U-50488; DOR, SNC-80), a detection reagent mix of equal volumes of europium cryptate and d2-labelled antibody, and human MOR, KOR, or DOR membrane preparation. Non-specific binding was evaluated using a non-hydrolysable GTPγS at a saturation concentration (25 µmol/L) to measure the assay background signal. A positive control was prepared using a recombinant Gαi subunit to which both the Eu-GTP analogue and the d2-antibody bind, allowing control of the detection reagent. Membranes were incubated overnight at room temperature. Dipyanone and the reference compounds were dissolved in a mixture of dimethyl sulfoxide and supplemented stimulation buffer (90:10, v/v), with final concentrations ranging from 10–5 to10−11 mol/L. Each concentration was tested in duplicates, and experiments were performed in triplicates (n = 3). The FRET signal was measured using a Multilabel Plate reader (PerkinElmer), calculating the fluorescence ratio of 665 to 620 nm to remove photophysical interference (delay: 100 μs; total time window: 200 μs). All values were normalised to the maximal signal of the corresponding reference receptor agonist, set to 100%. Concentration–response curves were fitted using GraphPad Prism (v. 10.2.3) with a three-parameter fit to determine potency (EC50) and efficacy (Emax).

Sample preparation for dipyanone quantificationBlood samples

Liquid–liquid extraction was performed on blood samples using acetonitrile and ammonium formate. The process began by fortifying 100 µL of blood or serum with an internal standard solution (methadone-d9, 10 ng/mL). Subsequently, 100 µL of ammonium formate (10 M) and 1.0 mL of ice-cold acetonitrile ( – 20 °C) were added and mixed for 5 min using an overhead shaker. The samples were then centrifuged at 4000 rpm for 10 min. Following centrifugation, the organic phase was transferred to an autosampler vial. This phase was evaporated to dryness under a stream of nitrogen at 40 °C and reconstituted in 100 µL of mobile phase (C/D, 90/10, v/v). Mobile phase C consisted of a 2 mM ammonium formate buffer with additives (0.1% formic acid and 1% acetonitrile), while mobile phase D was acetonitrile-based with additives (2 mM ammonium formate buffer and 0.1% formic acid).

Urine samples

Urine samples underwent an enzymatic conjugate cleavage prior to liquid–liquid extraction. Initially, 100 µL of urine was fortified with the same internal standard solution used for blood samples. Then, 100 µL of phosphate buffer and 10 µL of glucuronidase-arylsulfatase solution were added and incubated at 45 °C for 60 min. The subsequent extraction followed a similar procedure to that of serum, with adjusted extractant volumes: 200 µL of ammonium formate (10 M) and 1.5 mL of ice-cold acetonitrile ( – 20 °C).

Instrumental conditions for dipyanone quantification

For the quantification of dipyanone, a semi-quantitative toxicological method validated for serum was used (Giorgetti et al. 2024). Briefly, analytical procedures were conducted using a QTRAP 5500 mass spectrometer from Sciex (Darmstadt, Germany) equipped with a Shimadzu Nexera X2 UHPLC-30AD system (Duisburg, Germany). Chromatographic separation was achieved on a Kinetex® F5 column (100 × 2.1 mm, 2.6 µm), along with a matching pre-column from Phenomenex (Aschaffenburg, Germany). The autosampler was set to maintain a temperature of 10 °C, and 10 µL of sample was injected into the system. The flow rate was set to 0.5 ml/min. Mobile phases consisted of an aqueous 2 mM ammonium formate buffer (with 0.1% formic acid and 1% acetonitrile) for phase C, and an acetonitrile-based eluent (with 2 mM ammonium formate buffer and 0.1% formic acid) for phase D. The gradient elution program was as follows: 0–1 min at 5% D, increase to 22.5% D from 1 to 4.5 min, increase to 32.5% D from 4.5 to 10.75 min, increase to 95% D from 10.75 to 13.5 min, hold at 95% D from 13.5 to 15.5 min, return to 5% D from 15.5 to 16 min, and hold at 5% D from 16 to 19.5 min. The total runtime was 19.5 min. Dipyanone was detected at 11.9 min (60 s detection window), while the internal standard methadone-d9 had a retention time of 10.6 min (90 s detection window). Mass spectrometry analysis was performed in positive electrospray ionization (ESI) mode with the following settings: ion spray voltage at 4500 V, curtain gas at 40 psi, collision gas set to medium, ion source gases at 60 psi (gas 1) and 70 psi (gas 2), and a source temperature of 500 °C. Data acquisition was performed using scheduled multiple reaction monitoring (sMRM). The sMRM parameters for dipyanone were as follows: Q1 m/z 336.0, transition 1 (Q3 m/z 265.0, collision energy (CE) 20 V), transition 2 (Q3 m/z 105.0, CE 35 V), transition 3 (Q3 m/z 219.0, CE 35 V). The sMRM parameters for methadone-d9 were: Q1 m/z 319.2, Q3 m/z 268.2, CE 20 V. All transitions were measured with an entrance potential of 10 V, a cell exit potential of 13 V and a declustering potential of 60 V, except for methadone-d9, where a declustering potential of 90 V was applied. Data analysis was performed using Analyst® software (version 1.5.1, Sciex, Darmstadt, Germany).