The off-target activity profiles of seven SCRAs (AMB-FUBINACA, XLR11, PB-22, AKB-48, AB-CHMINICA, CUMYL-PINACA, and 4F-MDMB-BUTINACA) were assessed using the PathHunter β-arrestin assay, with subsequent interrogation of the activity of AMB-FUBINACA against several off-target hits. Overall, there were few hits in agonist mode at targets beyond CB1 and CB2 receptors, but many hits in antagonist mode against a wide range of targets. These hits included the oxytocin receptor and chemokine receptors, among many others, indicating substantial inhibitory promiscuity at 30 μM.

However, literature exploration of these hits revealed few obvious toxigenic targets, with most having either well-characterized and well-tolerated antagonists, or known non-toxic knockout phenotypes (Table 1). In some cases the situation was more complex; for instance, one “hit” was HCRTR1 (orexin OX1 receptor), where some KOs are more susceptible to pharmacologically-induced seizures (32); however, pharmacological inhibition of the receptor does not appear to have the same effect (33). Indeed, the relationship between the OX1 receptor and seizures is multifaceted, such that OX1 antagonists are being investigated for their therapeutic potential in some types of epilepsy (34), thus, it is not known whether or how this susceptibility interacts with CB1-receptor mediated seizures. Moreover, follow-up concentration response experiments using AMB-FUBINACA revealed low potency at this target (IC₅₀ 8.59 μM; Figure 4).

The follow-up concentration-response experiments demonstrated appreciable inhibitory activity of AMB-FUBINACA at histamine H₁, and adrenoceptors α2B and α2C, even at lower 1 μM concentrations. Interestingly, H₁, α2B, and α2C, are all involved in cardiac function, and AMB-FUBINACA has been associated with cardiotoxicity and appears to carry additional risk in people with underlying heart disease (19). However, H₁, α2B, and α2C antagonists and inverse agonists exist as approved and well-tolerated therapeutic drugs (i.e., common antihistamines, alpha-blockers). Thus, it seems unlikely that these targets are responsible for the cardiotoxicity associated with AMB-FUBINACA. Other targets, such as CCR9, MCHR1, and DRD2L were only inhibited at the highest 30 μM concentration, raising questions concerning the physiological relevance of these ‘hits' for typical SCRA intoxication. Post-mortem tissue analysis in cases where SCRAs have been linked to death has found typical SCRA concentrations <10 ng/mL in blood, occasionally reaching as high as 200 ng/mL (35), corresponding to <25 nM and occasionally up to 500 nM for the SCRAs around 400 g/mol molecular weight, some 2–3 orders of magnitude below our highest 30 μM concentration. A recent study in rats found AMB-FUBINACA concentrations to be ~5-fold higher in brain than plasma, suggesting that brain concentrations may reach as high as 2.7 μM (36).

We selected a high concentration (30 μM) for the initial single-point screen to minimize false-negatives that could occur at lower concentrations (see Figure 4). Nevertheless, it is possible that some false-negatives may have occurred if this high concentration produced cell death or otherwise disrupted cellular signaling. We cannot also rule out the possibility of assay interference due to colloidal aggregation given this concentration (37); however, one would expect to observe pan-inhibition across receptors rather than selected inhibition. This is a fundamental limitation of single-point screens, and it was not feasible to conduct full concentration-response curves for every target with every compound.

Disrupted signaling was evident with medium to high concentrations (>1 μM) of AMB-FUBINACA on CB1-expressing cells (Figure 3). The disrupted signaling could be indicative of CB1-mediated cytotoxicity at high doses; yet the results of our cytotoxicity assay (Figure 6) suggest this is not due to overt cell death. Although LDH release was significantly greater in CB1-expressing cells treated with AMB-FUBINACA, this 18% increase observed only at 1 h does not suggest all the cells are non-viable, and thus does not account for the complete lack of β-arrestin recruitment observed at this concentration. It should be noted, however, that this study was conducted in a different cell type (AtT20 vs. PathHunter CHO cells), and that we only used a single marker of membrane integrity disruption.

Instead, the loss of signal might be explained by a downregulation or degradation of CB1 receptor signaling. Previous studies of investigating β-arrestin recruitment in CB1, including those that have observed recruitment with AMB-FUBINACA, typically have used a much shorter observation window (20 min), with peak β-arrestin-2 recruitment observed with 10 μM AMB-FUBINACA in <10 min (23). Ligand- and concentration-dependent differences in rate of CB1 internalization, particularly at concentrations >10 nM, have also been previously reported (23). Of note, a 15 min application of another potent SCRA, HU210, at 10 nM caused near maximal internalization of CB1, that was not recovered up to 5 h later (38). Encouragingly, we did not observe similar concentration-dependent disruption for any other target selected for follow-up concentration-response experiments with AMB-FUBINACA (Figure 4).

Disrupted signaling of CB1 at high-concentration may be indicative of on-target toxicity for AMB-FUBINACA. CB1 receptor signaling is an integral component of retrograde endocannabinoid inhibition of presynaptic neurons; disabling this retrograde “feedback loop” could produce aberrant neural activity. Seizures have been observed preceding human fatalities involving AMB-FUBINACA (19), and SCRAs can reliably produce convulsant activity in rodent models. Seizures and pro-convulsant neural activity in rodents can be blocked by CB1 receptor antagonist/inverse agonist SR141716 (Rimonabant) or AM-251 (39, 40), which also restores surface CB1 expression following agonist-induced internalization (38). Peripherally, HU-210-induced cardio-depression can be similarly reversed by SR141716 in rats (41), suggesting CB1 receptor involvement, which is potentially of relevance to the bradycardia and hypotension observed during SCRA intoxication in humans (42).

Thus, although we set out to investigate off-target, non-cannabinoid receptor mediated toxicity, the present dataset highlights that high potency and efficacy, to the point of disabling CB1 receptor signaling at high concentrations, may be the predominant GPCR-mediated pathway underpinning the SCRA toxidrome. The underlying cellular mechanism(s), such as receptor internalization, recruitment of β-arrestins, or other signaling modifications, remain to be fully elucidated.

While we investigated the widest range of GPCRs in any SCRA study to date, other receptor classes like ion channels remain to be explored. For instance, several indole- and indazole-carboxamide type SCRAs are appreciably similar in structure to the 5-HT₃ receptor antagonists alosetron and granisetron. JWH-030, an early-generation partial CB1 and CB2 receptor agonist, inhibits hERG channel current, albeit with relatively low affinity, and may prolong cardiac QT intervals in rats, leading to arrhythmia (43). Moreover, some putative SCRAs related to MEPIRAPIM inhibit voltage-gated T-type calcium (Ca_V3) channels (44). Other actions such as enzyme induction and inhibition also require further consideration and investigation, particularly in combination with common medications that may be susceptible to adverse pharmacokinetic interactions. Numerous SCRAs of varying subclasses possess inhibitory activity at several cytochrome P450 and glucuronosyltransferase enzymes (45). Finally, other modes of toxicity, such as genotoxicity or toxicity arising from repeated dosing (46–49), and the potential pharmacodynamic and pharmacokinetic contributions of SCRA metabolites or thermal degradants (24, 50), are not encapsulated by the present study.

In sum, broad single-point screening of 7 representative SCRAs against 168 G-protein coupled receptors, with follow-up concentration-response investigations using AMB-FUBINACA, failed to reveal any obvious, standout off-targets that are likely to readily produce toxicity matching the SCRA toxidrome, particularly at concentrations observed in post-mortem tissue. Conversely, a substantial decrease in CB1-mediated signaling, but not overt cytotoxicity, was observed at high concentrations of AMB-FUBINACA, suggesting an on-target mechanism of toxicity whereby cannabinoid receptor signaling is impaired. Nevertheless, other large classes of potential off-targets are yet to be systematically investigated (e.g., ion channels), and toxicity arising from pharmacokinetic interactions or from the route of administration are potential contributing factors that require further investigation.