The reaction mixture included 780 µL distilled water, 100 µL 0.5 M potassium phosphate buffer pH 7.4, 10 µL solution B, and 10 µL 100 µmol/L ADB-FUBINACA in methanol. After vortexing, HLM suspensions (20 mg/mL) were thawed at 37ºC, 50 µL added to the reaction mixture and gently mixed. The suspension was pre-incubated at 37ºC for 3 min, and the reaction initiated by adding 50 µL solution A. Samples (100 µL) were collected after 0, 3, 8, 13, 20, 45 and 60 min incubation and the reaction quenched with an equal volume of ice-cold acetonitrile. Samples were stored at – 80ºC until analysis.

HLM samples were thawed and diluted 100-fold with mobile phase A before injection onto the LC-HRMS. Liquid chromatography (LC)-HRMS parameters for HLM incubations are described below.

HLM LC-MS analysis was performed on a 3200 QTRAP^® mass spectrometer (triple quadrupole-linear ion trap) equipped with an TurboV ion source operated in positive electrospray mode from Sciex (Foster City, CA, USA) and coupled with an LC-20ADXR HPLC system from Shimadzu (Columbia, MD, USA). Data were acquired with analyst TF software version 1.5 (Sciex).

Separation was performed using a Kinetex^® C₁₈ column (100 x 2.1 mm, 2.6 µm) combined with a 50 x 2.1 mm guard column of identical mobile phase from Phenomenex (Torrance, CA, USA). Elution was achieved within 15 min with a gradient mobile phase composed of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) at a flow rate of 0.3 mL/min. Injection volume was 5 µL. Autosampler and column oven temperatures were 4 and 40ºC, respectively. Gradient conditions started at 5% B for 0.5 min, increased to 95% B in 9.5 min, were held for 2.5 min, returned to initial conditions in 0.1 min and re-equilibrated for 2.4 min.

The mass spectrometer was operated in multiple reaction monitoring (MRM) mode. Three transitions were monitored between 6.5 and 8.5 min (accumulation time, 0.5 s): transition m/z 383.2 > 253.3 with a 33 V collision energy (CE) and a 4 V collision cell exit potential (CXP), transition m/z 383.2 > 109.0 with a 57 V CE and a 4 V CXP, and transition m/z 383.2 > 366.3 with a 17 V CE and a 6 V CXP. The first transition was for quantification and the last two for identification confirmation. Source parameters were: gas 1, 45 psi; gas 2, 70 psi; curtain gas, 30 psi; temperature, 500ºC; ion spray voltage, 4 kV; declustering potential, 31 V.

In vitro microsomal half-life (T_1/2) and clearance (CL_{int, micr}), predicted intrinsic clearance (CL_int), predicted human hepatic clearance (CL_hep), and extraction ratio (ER) were calculated based on Baranczewski [35] and McNaney’s [36] models, without consideration of plasma protein binding. T_1/2 was calculated based on the percentage of ADB-FUBINACA MS signal remaining at each time point.

Hepatocytes were thawed at 37ºC, washed twice with InVitroGRO™ HT medium and KHB, and centrifuged at 100 g for 5 min. Supernatant was removed and the pellet re-suspended in 2 mL KHB, yielding a 2 x 10⁶ cells/min concentration. Cell viability, assessed with trypan blue exclusion dye (0.4%, v/v), was ≥80%. ADB-FUBINACA was dissolved in methanol and diluted in KHB to a final 20 µmol/L concentration. The reaction mixture was prepared with 250 µL hepatocyte suspension and 250 µL drug solution and incubated in a Forma™ Steri-Cycle™ CO₂ incubator at 37ºC (Thermo Scientific). Samples (500 µL) were incubated for 0, 1 and 3 h and the reaction quenched with an equal volume of ice-cold acetonitrile. Diclofenac was incubated in a separate mixture under the same conditions and 4’-hydroxydiclofenac and acyl-β-D-glucuronide diclofenac monitored to ensure hepatocyte metabolic activity. In addition, a negative control with ADB-FUBINACA in buffer without hepatocytes was incubated to rule out non-enzymatic reactions. Samples were stored at – 80ºC until analysis.

Hepatocyte samples were thawed and centrifuged at 4ºC, 15,000 g for 10 min, to remove cell debris. Supernatants were diluted 5-fold with 0.1% formic acid in water (mobile phase A) before injection.

LC-HRMS analysis of hepatocyte incubations was performed on a 5600⁺ TripleTOF^® quadrupole-time of flight mass spectrometer equipped with a DuoSpray source (Sciex) operated in positive electrospray mode, coupled with an LC-20ADXR HPLC system from Shimadzu. Data were acquired with Analyst TF software version 1.6 (Sciex).

Separation was performed on an Ultra Biphenyl column (100 x 2.1 mm, 3 µm) combined with a 10 x 2.1 mm guard column of identical phase from Restek^® (Bellefonte, PA, USA). Elution was achieved within 15 min with a gradient mobile phase composed of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) at a flow rate of 0.5 mL/min. Injection volume was 15 µL. Autosampler and column oven temperatures were set at 4 and 30ºC respectively. Gradient conditions started with 20% B for 0.5 min, increased to 95% B in 10.5 min and then held for 2 min, returned to initial conditions in 0.1 min and re-equilibrated for 1.9 min. Efflux was diverted to waste before 2 and after 13 min.

The mass spectrometer operated in full-scan TOF-MS and information-dependent acquisition (IDA) MS/MS modes. One MS cycle was composed of one full-scan TOF-MS survey scan followed by up to eight IDA MS/MS scans (accumulation time, 0.5 s). The eight ions detected with the most intense signal in the survey scan (intensity threshold, 100 cps), triggered an MS/MS event with fragmentation (isolation window, m/z 1.5). IDA mode included a list of expected metabolites (Supplementary material, Table S1) that were fragmented in priority, regardless of their intensity (mass tolerance, 50 mDa). Full MS scans were acquired from m/z 100 – 1000 with high resolution and a 0.1 s accumulation time. IDA MS/MS scans were acquired from m/z 30 – 1000 with high resolution; collision energy was 35 ± 15 V and accumulation time 0.05 s. Source parameters were: gas 1, 60 psi; gas 2, 75 psi; curtain gas, 45 psi; temperature, 650ºC; ion spray voltage, 4 kV; declustering potential, 80 V.

Data from hepatocyte incubations were analyzed with MetabolitePilot™ software (v. 1.6; Sciex) for ADB-FUBINACA metabolites identification. Processing settings are described in Table 1.

M14 and M16 (m/z 399.1827, 5.45 and 5.63 min respectively) were hydroxyl metabolites of ADB-FUBINACA as suggested by the 15.9949 Da increase from parent mass (+O).

M14 MS/MS spectrum included the fragments m/z 269.0721, 286.0986, 354.1612, and 382.1561 that were also 15.9949 Da larger and also included the fluorobenzylium ion (m/z 109.0448), indicating that the hydroxylation occurred on the benzene part of the indazole ring. It is noteworthy that no water loss was detected during M14 fragmentation as the delocalized electrons of the benzene ring strengthened the hydroxyl bond. M6 (m/z 575.2148, 3.93 min) likely is the corresponding glucuronide of M14. M4 (m/z 591.2097, 3.72 min) was formed by further hydroxylation of ADB-FUBINA hydroxyindazole-glucuronide on the benzene ring.

M16 MS/MS spectrum included fragments m/z 354.1612 and 382.1561 that were 15.9949 Da larger than those of the parent. Fragments m/z 109.0448 and 253.0772, also present in the parent MS/MS spectrum, further suggested that M16 hydroxylation occurred on the dimethylpropane chain. Fragment m/z 324.1507 was produced by a carboxamide and CHOH loss, perhaps the result of a hydroxylation on a methyl group of the dimethylpropane chain. Further hydroxylation of M16 generated M17 (m/z 413.1619, 5.65 min).

Metabolites generated by amide hydrolysis eluted later than non-hydrolyzed metabolites, as observed with other SCs with the same terminal carboxamide group [37].

M22 (m/z 384.1718, 7.17 min) was the result of amide hydrolysis as suggested by a 0.9840 Da increase (-NH₂, +OH). The MS/MS spectrum showed fragments m/z 109.0448, 253.0772, 338.1663, and 366.1612, also present in parent spectrum, demonstrating that ADB-FUBINACA was not transformed except amide hydrolysis on the terminal amine function. Similarly, M19 (m/z 400.1667, 6.06 min) presented a 0.9840 Da increase from M16 mass (ADB-FUBINACA hydroxydimethylpropyl) and the same fragmentation pattern, indicating that both amide hydrolysis and dimethylpropane hydroxylation occurred.

M23 (m/z 382.1561, 5.92 min) was formed by amide hydrolysis and dehydrogenation as suggested by a 1.0317 Da decrease (-NH₂, +OH, -H₂). Given the late RT, M23 could not possibly be an artifact derived from a water loss of a hydroxylated metabolite. MS/MS spectrum presented fragments m/z 109.0448 and 253.0772, indicating that the dehydrogenation occurred on the aminodimethylbutanamide portion of ABD-FUBINACA, leading to an imine formation. Fragment m/z 324.1507 was produced by the loss of the carboxylic acid and one methyl group (– 58.0054 Da).

M15 (m/z 397.1670, 5.60 min) was formed by ADB-FUBINACA hydroxylation and dehydrogenation as suggested by the 13.9792 Da decrease (+O, -H₂). Fragments m/z 109.0448, 253.0772, 270.1037, and 324.1507 indicate imine formation and methyl hydroxylation. Fragment m/z 379.1565 confirmed hydroxylation (water loss).

M20 (m/z 395.1514, 6.48 min) was generated by hydroxylation and a di-dehydrogenation, as suggested by the 11.9636 Da increase (+O, -2H₂). Fragments m/z 109.0448 and 253.0772, also detected in ADB-FUBINACA’s mass spectrum, indicated that reactions occurred on the amino-dimethylbutanamide moiety. Fragment m/z 324.1507 indicated dehydrogenation to the imine function.

M13 (m/z 286.0986, 4.76 min) was formed by dimethylbutanamide cleavage and a hydroxylation, as suggested by the 97.0892 Da decrease (+O, -C₆H₁₁NO). Few fragments were produced as M13 signal intensity was low in the Full-MS scan after 1 h incubation and not detectable after 3 h incubation. Fragment m/z 109.0448 corresponds to unchanged methylenefluorophenyl and indicated that the reaction occurred on the indazole ring benzene group. Fragment m/z 269.0720 was formed by amine loss.

M7 (m/z 417.1933, 3.95 min) was a dihydrodiol, as suggested by a 34.0055 Da increase (+2O, +2H). Fragments m/z 287.0826, 304.1092, 372.1718, and 400.1667, were 34.0055 Da larger than the corresponding ADB-FUBINACA fragments while fragment m/z 109.0448 was still present, indicating that the reactions did only affect the indazole ring. We hypothesized that M7 was formed by epoxidation of the benzene moiety of ADB-FUBINACA’s indazole ring, followed by hydrolysis of the newly formed epoxide. This metabolic reaction was previously observed in several SCs containing indole or indazole ring metabolic pathways [33, 37].

HRMS is a powerful tool in metabolite identification studies as it theoretically allows capturing every compound in a single injection, and facilitates molecular formula determination of metabolites and fragments. Nitrogen atoms of the indazole core and the two amide functions are easily charged in acidic conditions, making positive-ion mode ionization suitable for ADB-FUBINACA metabolites’ detection. A 50 ppm MS mass tolerance was chosen during preliminary automated metabolite identification as it is broad enough that no possible metabolite with a high mass error could be missed, but narrow enough to predict elemental composition. MS/MS IDA mode scan speed allowed monitoring eight compounds at each MS cycle with a ramped collision energy fragmentation to maximize the production of different fragments and facilitate identification.

Samples were prepared by simple dilution as opposed to extraction before injection to increase detection of potential metabolites. However, matrix suppression could impede detection of metabolites with low signal intensity. Similarly, relative metabolite rank based on MS peak intensity could be confounded by matrix effect and ionization efficiency.

The suggested ADB-FUBINACA metabolic pathway is described in Fig. (4). M16 (ADB-FUBINACA hydroxy-alkyl), M15 (ADB-FUBINACA hydroxydehydroalkyl) and M14 (ADB-FUBINACA hydroxylindazole) are the three metabolites recommended as ADB-FUBINACA intake markers. Several glucuronidated metabolites (M1, M2, M4 to M6, M8, M10, M12, and M18) suggest that hydrolysis of biological samples (e.g. urine, blood) prior to extraction could increase non-glucuronidated metabolites’ concentrations and facilitate their detection.

Major metabolic pathways of AB-FUBINACA, ADB-FUBINACA’s analog (Fig. 1), in hepatocyte incubations included amide hydrolysis and alkyl and indazole hydroxylations [37]. Alkyl dehydrogenations, dihydrodiol formations, and glucuronidations also were observed to a lesser extent. ADB-FUBINACA metabolism in the present experiments was consistent with these results. ADB-FUBINACA N-dealkylated metabolites formed by dimethylbutanamide cleavage (M2 and M13) also were detected in AB-FUBINACA metabolism [37]. The same two metabolites could theoretically be formed after MDMB-FUBINACA metabolism as it presents the same indazole- methylenefluorophenyl structure. Similarly, metabolites formed by N-dealkylation (M3 and M9) could theoretically be formed after MAB-CHMINACA, ADB-CHMINACA, ADB-PINACA and 5-F-ADB-PINACA metabolism, as they present the same indazole-dimethylbutanamide structure. Metabolites generated by amide hydrolysis, M21, M22 and M23, could theoretically be formed after MDMB-FUBINACA O-demethylation. However, MDMB-FUBINACA, MAB-CHMINACA, ADB-CHMINACA, ADB-PINACA and 5-F-ADB-PINACA metabolism has yet to be elucidated. Given the constant emergence of new substances and depending on the legislation, being able to precisely distinguish one synthetic cannabinoid from another is critical. As M2, M3, M9, M13, M21, M22 and M23 may not be specific for ADB-FUBINACA intake, they are not recommended as ADB-FUBINACA consumption markers. Noteworthy, AB-FUBINACA carboxylic acid, the corresponding M22 ADB-FUBINACA metabolite, is specific for AB-FUBINACA metabolism, as there are no other compounds known that can produce it [37].

This in vitro hepatocyte marker profile should be confirmed with human urine samples after ADB-FUBINACA intake, as post-liver metabolism processes also can occur (e.g. re-absorption, extrahepatic metabolism, enterohepatic circulation). Unfortunately, authentic urine or blood specimens following ADB-FUBINACA consumption were not available despite our efforts to obtain them. It is important to verify in vitro hepatocyte metabolism results with human urine samples, although we have observed good correlation with authentic urine specimens and hepatocyte incubations for 10 µM FDU-PB-22, FUB-PB-22 [40], AB-PINACA [32] and AB-FUBINACA [37]. Castaneto et al. observed good agreement between metabolites identified during AB-FUBINACA hepatocyte incubations and authentic urine specimens following AB-FUBINACA intake, including the most intense signals for AB-FUBINACA carboxylic acid, AB-FUBINACA carboxylic acid-glucuronide, and AB-FUBINACA carboxylic acid-hydroxyalkyl metabolites [37].