Detection and phase I metabolism of the 7-azaindole-derived synthetic cannabinoid 5F-AB-P7AICA including a preliminary pharmacokinetic evaluation

In June 2018, a 'research chemica'l labeled 'AB-FUB7AICA' was purchased online and analytically identified as 5F-AB-P7AICA, the 7-azaindole analog of 5F-AB-PINACA. Here we present data on structural characterization, suitable urinary consumption markers, and preliminary pharmacokinetic data. Structure characterization was performed by nuclear magnetic resonance spectroscopy, gas chromatography – mass spectrometry, infrared and Raman spectroscopy. Phase I metabolites were generated by applying a pooled human liver microsome assay (pHLM) to confirm the analysis results of authentic urine samples collected after oral self-administration of 2.5 mg 5F-AB-P7AICA. Analyses of pHLM and urine samples were performed by liquid chromatography − time-of-flight mass spectrometry and liquid chromatography – tandem mass spectrometry (LC – MS/MS). An LC – MS/MS method for the quantification of 5F-AB-P7AICA in serum was validated. Ten phase I metabolites were detected in human urine samples and confirmed in vitro. The main metabolites were formed by hydroxylation, amide hydrolysis, and hydrolytic defluorination, though – in contrast with most other synthetic cannabinoids – the parent compound showed the highest signals in most urine samples. The compound detection window was more than 45 hours in serum. The concentration-time profile was best explained by a two-phase pharmacokinetic model. 5F-AB-P7AICA was detected in urine samples until 65 hours post ingestion. Monitoring of metabolite M07, hydroxylated at the alkyl chain, next to parent 5F-AB-P7AICA, is recommended to confirm the uptake of 5F-AB-P7AICA in urinalysis. It seems plausible that the shift of the nitrogen atom from position 2 to 7 (e.g. 5F-AB-PINACA to 5F-AB-P7AICA) leads to a lower metabolic reactivity, which might be of general interest in medicinal chemistry.


| INTRODUCTION
Synthetic cannabinoids (SCs) are a class of exceptionally widespread recreational designer drugs and new psychoactive substances (NPS), produced to mimic the effects of delta-9-THC and sold as alleged legal and safe cannabis alternatives. Since the first identification of SCs in 'Spice' products, which dates back to 2008, 1,2 the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) has reported and identified more than 179 compounds pertaining to the subgroup of SCs, with a continuous production of novel molecules, 3  3-carboxamide (semisystematic name: 5F-AB-PINACA), were discovered on the Japanese drug market. 4,5 These compounds were shown to be full agonists at the CB 1 cannabinoid receptor with high potencies (CB 1 receptor binding and activation occur at low nanomolar concentrations). [6][7][8] in vitro studies have also shown that AB-PINACA displays a greater efficacy than delta-9-THC in activating the G-coupled proteins at CB 1 receptors and that some of its metabolites retain high affinity and activity. 9 Even if in vivo pharmacology and toxicology data related to these compounds are still lacking, some intoxications and even death cases have recently been reported in association to AB-FUBINACA, AB-PINACA, and 5F-AB-PINACA consumption. [10][11][12] A structurally related SC, methyl-[2-(1-(5-fluoropentyl)-1H-indazole-3-carboxamido)-3,3-dimethylbutanoate] (semisystematic name: 5F-MDMB-PINACA or 5F-ADB), which bears an aminoalkylindazole structure, was reported in associations with about 10 fatal intoxications in Japan and detected for the first time in a death case in 2014. 13 It was described as one of the most dangerous SCs 11 and detected in several autopsies after that. [13][14][15][16] While the majority of SCs are characterized by an indole or indazole core structure, SCs with a 7-azaindole (7-AI) core structure have been synthesized and introduced into the NPS market most likely in response to regulations like the German act on the NPS (NpSG), and now represent about 3% of all monitored SCs 3 . Among these, a 7-AI analog of AB-FUBINACA, N-(1-amino-3-methyl-1-oxobutan-2-yl)-1-(4-fluorobenzyl)-1H-pyrrolo [2,3-b]pyridine-3-carboxamide (semisystematic name: AB-FUB7AICA; also referred to as AB-7-FUBAICA) entered the market as pure substance in powder form ('research chemical') and started to proliferate in online shops and NPS user forums. Recently, the 7-AI analog of 5F-ADB, methyl 2-(1-(5-fluoropentyl)-1H-pyrrolo [2,3-b]pyridine-3-carboxamido)-3,3-dimethylbutanoate (semisystematic name: 7'N-5F-ADB or 5F-MDMB-P7AICA) also appeared on NPS user forums. SCs of the 7-AI type usually present relatively low binding affinity and functional activity at the CB 1 receptor. 17,18 From an analytical point of view, these compounds may lead to additional difficulties for identification, due to the similarity in physicochemical properties and identical masses with the corresponding indazole analogs.
In both clinical and forensic settings, it is fundamental to prove an SC intake by the analysis of biological samples, with serum and urine samples being the most common matrices to screen for these compounds and their metabolites. In this context, the drug's postdosing detectability as well as a clearance model are of major concern in forensic toxicology, as the detection windows for SCs are often short after acute intake and significantly longer after chronic use. Due to extensive metabolism, for most SCs the parent compound is rarely found in urine samples. [19][20][21][22] To the best of our knowledge, no pharmacokinetic data of this substance have been published.
The aim of the present study was to characterize and identify human phase I metabolites in human urine samples by means of liquid chromatography−time-of-flight mass spectrometry (LC−qToF−MS) and liquid chromatography−triple-quadrupole mass spectrometry (LC-MS/MS). As no commercially reference standards of 5F-AB-P7AICA phase I metabolites were available, reference spectra had to be generated in vitro using a pooled human liver microsome assay (pHLM). Finally, the most reliable urinary marker metabolites were evaluated, taking into consideration the pharmacokinetic profile of the substance. A mix of SCs and one of SCs commercially available metabolites were also used. 5F-AB-P7AICA was purchased as a 'research chemical' named AB-FUB7AICA from an Internet shop. Identities and purities (>95%) of 5F-AB-P7AICA were confirmed by 1 H nuclear magnetic resonance (NMR) spectroscopy and gas chromatography-mass spectrometry (GC-MS). Stock solutions (1 mg/mL) were prepared in ACN and stored at −20 C until analysis.

| Chemicals and reagents
2.2 | Structure elucidation 2.2.1 | Sample preparation (5F-AB-P7AICA) 5F-AB-P7AICA was analytically confirmed by GC-MS, NMR, infrared and Raman spectroscopy as described in the following sections. Furthermore, a solution of the compound was studied by liquid chromatography and electrospray ionization quadrupole time-of-flight mass spectrometry (LC-ESI-qToF-MS). Briefly, 1 mg 5F-AB-P7AICA was dissolved in 1 mL of methanol. Then, 10 μL of this solution was evaporated to dryness at 40 C under nitrogen. Prior to the injection into the GC-MS system (injection volume: 1 μL), the sample was reconstituted in 100 μL of dry ethyl acetate.

| GC-MS method
The GC−MS system consisted of a 6890 N-series gas chromatograph combined with a 5973-series mass selective detector and a 683 B series injector. The software used was Chemstation G1701GA version D.03.00.611. Mentioned products were purchased from Agilent (Waldbronn, Germany). The detailed method used is described elsewhere. 23 F I G U R E 1 Chemical structures and semisystematic names of synthetic cannabinoids mentioned [Colour figure can be viewed at wileyonlinelibrary.com] Briefly, carrier gas was helium, injection port temperature was 270 C, flow rate 1 mL/min, oven temperature 100 C for 3 minutes, then ramped to 310 C at 30 C/min, 310 C were kept for 10 minutes.
Electron ionization (EI, 70 V) was used and the MS was operated in scan mode (m/z 40 to 550 amu). The obtained mass spectra were compared to commercially available EI−MS spectra libraries (Cayman Chemical, Wiley, MPW) and an in-house library of previously identified synthetic cannabinoids.

| Solid-state infrared spectroscopy (IR)
Nicolet 380 FT−IR spectrometer with Smart Golden Gate Diamond ATR. Software: OMNIC, Ver. 7.4.127 (Thermo Electron Corporation, Dreieich, Germany). Wavelength resolution: 4 cm −1 ; scan range 650-4000 cm −1 ; 32 scans/spectrum. IR spectra were recorded from salts and from free bases as neat film after following sample preparation procedure: For generation of the free bases, 2-5 mg of the salt were dissolved in demineralized water and were alkalized with one drop of NaOH (5% w/w). The solution was extracted with 1 mL diethylether, the ethereal phase was transferred in a new glass vial and the solvent was evaporated under a gentle nitrogen flow until the volume reached approximately 100 μL. The remaining fluid was aspirated with a glass pipette and transferred directly on the ATR crystal where the remaining diethylether was continuously evaporated.

| NMR analysis
NMR spectra of the pure research chemical 5F-AB-P7AICA were recorded in DMSO-d 6 (1D and 2D spectra) at room temperature with a spectrometer Bruker AVANCE III HD 500 with probe: Bruker 5 mm CryoProbe Prodigy BBO with z-gradient. Sample preparation: approx.

| LC-qToF-MS experiments
LC-qToF-MS analysis for metabolites was performed on an impact II™ qToF instrument coupled with an Elute HPLC system (both from Bruker Daltonik, Bremen, Germany). Chromatographic separation was achieved on a Kinetex ® C18 column (2.6 μm, 100 Å, 100 × 2.1 mm; Phenomenex, Aschaffenburg, Germany), protected by an equivalent Security Guard™ ULTRA catridge precolumn (Phenomenex, Aschaffenburg, Germany), applying gradient elution as follows: mobile phase B starting concentration was 20%, linearly increased to 25% in 2.5 minutes, further increased to 40% in 2.5 minutes, further increased to 65% in 1.5 minutes, further increased to 95% in 1.0 minutes, held for 1.0 minutes, decreased to starting conditions of 20% in 0.1 minutes and held for 1.4 minutes for re-equilibration. Total run time was 10 minutes. The total flow rate was set to 0.4 mL/min.
The autosampler was cooled down to 10 C. Column oven temperature was 40 C. The injection volume was 10 μL.
The qToF−MS was operated in positive ionization mode acquiring spectra in the range of m/z 30-600 at an acquisition rate of 4.0 Hz.

Full scan and broadband collision-induced dissociation (bbCID) data
were acquired in one run. The collision energy applied for bbCID was 30 ± 6 eV. Instrument parameters were set as described previously. 24 (Table S1).
For metabolite identification EPI scan experiments with the hypothetical masses of anticipated phase-I metabolites were conducted and the obtained spectra were compared with the EPI spectrum of the parent compound. All detected precursor ions were further characterized by EPI scans.

| In vitro and in vivo metabolite identification
Metabolites generated in the pHLM assay were tentatively identified and characterized by means of LC-qToF-MS. On the basis of previous studies regarding the metabolism of structurally related SCs, particularly of AB-PINACA and 5F-AB-PINACA, 26,27 a list of hypothetical metabolites, most probably formed in vitro and in vivo, was generated. Data were processed manually, applying the following criteria: MS peak area > 1 × 10 5 cps, mass error of the precursor ion Optimized MRM parameters of the parent compound (Table S1) were adopted for the selected ion transitions of the identified metabolites, assuming similar fragmentation behavior (Table S2 shows

| Method validation
The LC-MS/MS method for quantification of 5F-AB-P7AICA in serum samples was validated in accordance with the literature recommendations. 28 Analytical parameters validated included selectivity, linearity, accuracy, precision, limit of detection (

| In vitro metabolism
In total, 14 metabolites were detected in the pHLM assay samples. The following metabolic reactions were observed: mono-and di-hydroxylation, hydrolytic defluorination, elimination of hydrofluoric acid, dehydrogenation and amide hydrolysis. The metabolites were ordered from lower to higher retention times (RTs) and numbered accordingly. As expected, none of the previous were found in the negative control samples.
A qualitative ranking of the metabolites produced by in vitro phase I metabolism was obtained. As shown in Table S3, the mono-hydroxylated metabolites, particularly M07, M09, and M04, presented the most intense peaks, while amide hydrolysis (M13) and a di-hydroxylated metabolites (M01) were the lowest in the in vitro metabolite ranking.

| In vivo metabolism
The parent compound was detected in the urine samples after oral selfadministration, as the most intense peak in the majority of the urine samples ( Figure S17). This has never been described in the metabolism of SCs, since parent compounds are only seen in high concentrated urine and, normally, the peak intensity is far lower than that of metabolites. represented about 5% of the parent compound and the amide hydrolysis only 0.08%. In contrast, the most abundant OH-metabolite showed about 10% of the peak area of 5F-AB-PINACA and the amide hydrolysis product accounted for about 14%. So far, we were not able to identify authentic urine samples with confirmed uptake of 5F-AB-P7AICA (which is usually smoked) to confirm the results after oral administration and we are aware of the possibility that ratios might change in such samples due to the different route of administration or to the different sampling time from intake. Nevertheless, we hypothesize that the 7-AI core is responsible for the reduction of metabolic reactivity.
In a recent paper, Franz et al. demonstrated that the heterocyclic core structure has an impact on several metabolic reactions, and especially on the hydrolysis of terminal or secondary amide functionality.
Particularly, indoles were shown to be significantly less reactive than their indazole analogs. 34 The metabolism of the 7-AI analog of 5F-ADB, 7'N-5F-ADB, was also recently investigated. The parent compound presented relatively high peaks in authentic human urine samples, though being the most abundant only in one of four samples. 35 Contrarily, regarding the indazole analog, Yeter et al. 36 showed that in most of the authentic urine samples the concentration of the carboxylic acid metabolite M20 was much higher than the concentration of the parent compound 5F-ADB. This observation supports the hypothesis that the shift of a nitrogen atom from position 2 (indazole) to position 7 (7-AI) significantly lowers the metabolic reactivity.
Overall, the metabolic profile of 5F-AB-P7AICA was very similar to the metabolic profiles of AB-PINACA and 5F-AB-PINACA, the metabolites of which have been studied by means of pHLM and authentic human urine samples. 26,27 Several SCs, such as 5F-MDMB-PICA, AB-FUBINACA and AB-PINACA, 27,37,38 undergo extensive ester/amide hydrolysis. This is one of the main reactions in vivo, although the abundancy of ester/amide hydrolysis products, which are mainly produced by carboxylesterases, 39  in the in vivo sample #07. This is a further confirmation that a verification of pHLM results by analyzing authentic urinary samples, as done in the present work through oral self-administration, is necessary for final selection of reliable marker metabolites. As

| Method validation
In the selectivity testing, no interferences with 5F-AB-P7AICA occurred, neither in the blank serum nor in the blank samples containing IS or a mixture of SCs and SC metabolites.
Good linearity was shown for 5F-AB-P7AICA in serum in the con- Accuracy and precision are shown in Table 2, where the results of the matrix effect analysis are also given.

| Preliminary pharmacokinetic data
After validation, the proposed method was successfully applied for the quantification of 5F-AB-P7AICA in serum samples (n = 34) collected pre and after a controlled oral self-administration of 2.5 mg of the compound. In Figure 5, the concentration-time profile is displayed. None of the main metabolites of 5F-AB-P7AICA, as identified in the in vitro and in vivo urinary samples, was detected in the serum of the voluntary in its free form (phase II metabolites like glucuronidates were not targeted).
In the authentic urine samples collected after a single controlled oral administration the parent compound, M07 (rank #1), M05 (rank #2), and M13 (rank #4) showed the highest intensities at 4.67 hours post intake, remained above the threshold of MS peak area > 1 × 10 4 cps for 67 hours (peak areas were normalized for the creatinine levels), and was not detected anymore in a urine sample collected 75.67 hours after the intake. As visible in Figure 6, M07 was the most abundant metabolite in the first 5 hours, showed a plateau from 4.67 hours until 10 hours and then slowly decreased. M05 was in rank position #2 for approximately the first 5 hours, then rapidly decreased, with intensities lower than M13, which, on the contrary, passed from a #4 to a #2 rank position.
The relatively low amount of amide hydrolysis product (M13) when compared to the parent compound and the abundance of the latter could be partially explained by the oral intake of the substance.
Since hydrolysis products might be formed during smoking even before entering the body, 43,44 it is reasonable to assume that smoking results in a different ratio of metabolites. However, the amount of hydrolysis products is comparatively low with respect to metabolic hydrolysis. 43,44 Only a small number of studies investigating human pharmacokinetics of SCs are found in the literature. 45 Thus, data on distribution and elimination phases widely lacks, especially for new SCs. 45 In the present case, an oral intake was deemed most appropriate in order to avoid severe potential side effects in the absence of data on pharmacodynamics. Indeed, smoking self-experiments retrieved in the literature mostly cover substances which have already been listed as illicit in most of the European countries, and are therefore no longer present on the drug market. 44 A point of strength of the present study is the consideration of a substance with a relatively new core structure (7-AI derivative).
Although these results could be useful for understanding the pharmacokinetics of 5F-AB-P7AICA, a limitation that has to be kept in mind is that SCs are mainly consumed by smoking, and not by oral administration. Parenteral intake could results in stronger effects at the same dose and a differing kinetic profile. Particularly when analyzing urine, higher signal intensities for the amide hydrolysis product could be expected due to artefactual formation of the hydrolysis product during smoking 43,44 or to metabolism in lung, where carboxylesterases are also expressed. 39 However, as previously discussed and confirming data regarding other indole-indazole analogs, it is highly plausible that the shift of a nitrogen atom from position 2 (indazole) to position 7 (7-AI) significantly increases the metabolic stability. A second limitation resides in the fact that data refers to a single volunteer. It is well known that the pharmacokinetic behavior of drugs may vary within different subjects, thus studies on a larger sample are highly encouraged, even though limited by ethical reasons. In the attempt to confirm the in vivo results, an in vitro comparison between 5F-AB-P7AICA and 5F-AB-PINACA was performed and confirmed the qualitative data. Further, we did not target phase II metabolites in our study because the sample work-up for SCs screening usually involves β-glucuronidase treatment to enhance sensitivity. 38,46 4 | CONCLUSIONS The present study gives a further example of a wrongly labeled SC sold online as a 'research chemical'. The questionable identity of substances purchased on the Internet could lead to serious adverse events, particularly if drug potencies differ.
The investigated 7-AI analog of 5F-AB-PINACA only partially fit the expected pattern of urinary metabolites, as based on the metabolic profiles of other SCs of the pentyl indole/indazole type and their 5-fluoropentyl analogs. In contrast to previously investigated compounds, the parent 5F-AB-P7AICA should be monitored to confirm the uptake of this drug in routine urine screening in addition to the main metabolites. The hydroxylated metabolite M07 is suggested as a highly specific marker in authentic urine samples with the potential to clearly differentiate from closely related 7-AI derivatives sharing, for example, the same carboxylic acid or defluorinated metabolite. In addition, a validated method to quantify 5F-AB-P7AICA in serum samples was presented, together with preliminary data on pharmacokinetics after oral intake of the drug. From the presented data, it seems plausible that the shift of the nitrogen atom from position 2 (indazole) to position 7 (7-AI) leads to a lower metabolic reactivity, which might be of interest in terms of the development of medicinal drugs featuring similar structural elements.

DECLARATION OF INTERESTS
The authors declare no competing financial interests.