Experimental strategies to discover and develop the next generation of psychedelics and entactogens as medicines

Research on classical psychedelics (psilocybin, LSD and DMT) and entactogen, MDMA, has produced a renais- sance in the search for more effective drugs to treat psychiatric, neurological and various peripheral disorders. Psychedelics and entactogens act though interaction with 5-HT 2A and other serotonergic receptors and/or monoamine reuptake transporters. 5-HT, which serves as a neurotransmitter and hormone, is ubiquitously distributed in the brain and peripheral organs, tissues and cells where it has vasoconstrictor, pro-inflammatory and pro-nociceptive actions. Serotonergic psychedelics and entactogens have known safety and toxicity risks. For these drugs, the risks been extensively researched and empirically assessed through human experience. However, novel drug-candidates require thorough non-clinical testing not only to predict clinical efficacy, but also to address the risks they pose during clinical development and later after approval as prescription medicines. We have defined the challenges researchers will encounter when developing novel serotonergic psychedelics and entactogens. We describe screening techniques to predict clinical efficacy and address the safety/toxicity risks emerging from our knowledge of the existing drugs:


Introduction
The last 20 years have been a lean time for the discovery and development of new drugs for use in psychiatry. For many years there was no real progress in finding antidepressants that were faster-acting and more effective than the SSRIs (selective serotonin reuptake inhibitors) which were introduced in the 1980s. Disillusionment and multiple failures in clinical trials prompted many of the major pharmaceutical companies to discontinue R&D in the CNS indications. In the last 5 years, however, optimism that this goal may be achievable has been rekindled based on clinical findings with the NMDA antagonist, ketamine, and the serotonergic psychedelics. In terminal cancer patients, psilocybin produced rapid and maintained reductions in anxiety (Grob et al., 2011), anxiety and depression Griffiths et al., 2016;Ross et al., 2016) and similar benefits in patients with treatment-resistant depression (Carhart-Harris et al., 2016 and major depressive disorder (Carhart-Harris et al., 2021). Ayahuasca, which is a plant derived psychoactive mixture containing N,N-dimethyltryptamine (DMT) and harmala alkaloids (potent inhibitors of monoamine oxidase A [MAO-A]) (Brito da Costa et al., 2020) has been reported to have a rapid antidepressant effect in major depressive disorder (van Oorsouw et al., 2022), treatment-resistant depression  and to reduce suicidality (Zeifman et al., 2019).
The tool compounds used in these and other clinical trials have been the so-called "classical psychedelics", i.e., psilocybin and DMT, and LSD (lysergic acid diethylamide) and the entactogen, MDMA (3,4-methylenedioxymethamphetamine). As shown in Fig. 1, these serotonergic drugs come from 3 major structural classes, i.e., the tryptamines, ergolines, and phenethylamines. Members of all 3 structural classes can be hallucinogens, but to date, entactogens come exclusively from the phenethylamine class. With a few notable exceptions, most are naturally occurring molecules produced by plants, fungi or animals ( Fig. 1), and in many cases, although their pharmacological mechanism was not characterised, their psychedelic properties have been exploited by humans for centuries.
These serotonergic compounds offer great promise as medications but there are serious questions over their toxicity and safety. Although thousands of articles have been published on the psychedelics and MDMA, they have never undergone the rigorous non-clinical pharmacological characterization and safety evaluation which is the accepted regulatory path for novel CNS drug-candidates, and therefore, substantial gaps in our knowledge exist. This situation is not the result of any deficiency on the part of drug regulators. Drug-candidates like Abbreviations 2C-I 2,5-dimethoxy-4-iodophenethylamine 2C-C 2,5-dimethoxy-4-chlorophenethylamine 2C-D 2,5-dimethoxy-4-methylphenethylamine 2C-E 2,5-dimethoxy-4-ethylphenethylamine 2C-T-2 2,5-dimethoxy-4-ethylthiophenethylamine 3-MT 3-Methoxytyramine; 5-HIAA, 5-hydroxyindoleacetic acid 5-HT 5-hydroxytryptamine 5-   D.J. Heal et al. psilocybin and MDMA are being developed to be used in a single session or a few sessions spread over weeks or months. This treatment regimen reduces the level of non-clinical toxicity and safety testing required for registration. Although this approach adequately addresses the risks to patients taking these drugs under medical supervision in accordance with the Product Label or Summary of Product Characteristics, it does not cover risks to individuals who self-medicate, misuse or abuse them. The Office for National Statistics reported that deaths involving MDMA in England and Wales increased from 8 individuals in 2010 to 82 in 2020 (Office for National Statistics [ONS] website). There has also been an explosion in the emergence companies illegally selling magic mushroom infused foods. Although MDMA and the 5-HT 2A agonists offer great promise in treating a range of psychiatric disorders, it is also important to appreciate that they produce powerful psychotropic effects and these risks should be mitigated through suitable product labelling and controlled drug scheduling. Progress in the field will inevitably involve the development of new drugs that offer greater efficacy, safety and benefits than the current psychedelics and entactogens, and an exploration of their potential as treatments for a wide range of psychiatric, neurological and peripheral disorders. Given the multifaceted pharmacology of the psychedelics and entactogens, defining the specific mechanism, or combination of mechanisms, that are responsible for efficacy in each disorder or disease will be a significant challenge. Moreover, the difficulty of the challenge increases substantially if the psychedelic or entactogen works through a combination of mechanisms because it infers that maximizing benefit/risk will be dependent on achieving an optimum balance of pharmacological effect across all relevant targets. The serotonergic psychedelics and entactogens are not the only drugs helping to revolutionise pharmacotherapy for psychiatric disorders. NMDA (n-methyl-D-aspartate) antagonists have also made a major impact, when reports emerged almost 2 decades ago showing that single or repeated ketamine infusions produced rapid improvements in treatment-resistant depression (Zarate et al., 2006;Ibrahim et al., 2012;Murrough et al., 2013), accelerated the antidepressant effect of electroconvulsive therapy (Okamoto et al., 2010;Yoosefi et al., 2014) and reduced suicidal ideation (Price et al., 2009). Although racemic ketamine is used off-label to treat depression, the S-enantiomer of ketamine, esketamine, has been developed and approved as a nasally administered, adjunctive treatment for treatment-resistant depression and major depressive disorder with suicidal ideation or behaviour (Spravato®). Ketamine is also being evaluated clinically in a number of other psychiatric indications including as a potential treatment for various substance use disorders.
There are some interesting similarities between the rapid and prolonged antidepressant effect of the NMDA antagonists, ketamine and esketamine, and the serotonergic psychedelics, psilocybin and DMT. For example, NMDA antagonists produce hallucinations (Powers et al., 2015;Mori and Suzuki, 2018), augment 5-HT 2A receptor signalling (see section on rodent head-twitches); both effects are relevant to the induction of a mystical experience and expanded consciousness are key factors in their therapeutic actions (Dakwar et al., 2014;Mollaahmetoglu et al., 2021;Rothberg et al., 2021).
In this review, we will interrogate current knowledge about the pharmacology of the serotonergic and NMDA antagonist psychedelics to identify and propose potential mechanisms for their therapeutic actions. We will describe and discuss the techniques which could be used to screen for novel psychedelic drug-candidates and explore some of the safety concerns that will need to be addressed when developing the next generation of serotonergic psychedelics.

Pharmacological characterisation and predictors of efficacy
Experiments conducted in human subjects with LSD and the 5-HT 2A receptor antagonist, ketanserin, have established that this receptor mediates its psychedelic, subjective, and neural effects (Preller et al., 2017(Preller et al., , 2018(Preller et al., 2019) Holze et al., 2020. It has also been established that the therapeutic effect of the serotonergic psychedelics to alleviate depression and anxiety is associated with the intensity of the mystical experience that subjects undergo after taking these drugs (MacLean et al., 2011;Griffiths et al., 2016;Roseman et al., 2016;Palhano-Fontes et al., 2019;Russ et al., 2019;van Oorsouw et al., 2022). Together, these observations provide substantial evidence to implicate the 5-HT 2A receptor as a key mediator of their efficacy in treating affective disorders. However, comprehensively elucidating the therapeutic mechanism of the psychedelics is more complex because they functionally interact with a multitude of 5-HT receptors in addition to the 5-HT 2A subtype. Even if one accepts that efficacy is dependent on 5-HT 2A agonism, it does not eliminate the possibility that interactions with other 5-HT receptor subtypes are significant contributors to their antidepressant and anxiolytic actions. Furthermore, the psychedelics differ from one another with respect to their relative affinities for and functional interaction with individual 5-HT receptor subtypes leading to the conclusion that their therapeutic applications and efficacy may vary as a consequence. As discussed below, there are conflicting reports about the ability of the 5-HT 2A agonists to alter the synaptic concentrations of 5-HT and other neurotransmitters thereby indirectly extending their spectrum of pharmacological effects. Although most microdialysis studies with hallucinogens like LSD, 5-MeO-DMT (5-methoxy-N, N-dimethyltryptamine), DOI (2,5-dimethoxy-4-iodoamphetamine) and psilocin (the psychoactive metabolite of psilocybin), report no effect on the extracellular levels of 5-HT or dopamine in various brain regions (Bogdanov et al., 1991;Gudelsky et al., 1994;Sakashita et al., 2015), others have observed increases in dopamine in the nucleus accumbens (Sakashita et al., 2015;Miliano et al., 2019) and prefrontal cortex (Miliano et al., 2019). Our view is these increases may be statistically significant but are too small to contribute the primary pharmacology of the hallucinogens, and probably reflect secondary changes in monoaminergic signalling in response to 5-HT 2A receptor activation. Attempting to identify the therapeutic mediators of MDMA and its metabolite, MDA (3,4-methylenedioxyamphetamine), is far more challenging because as 5-HT releasing agents, both compounds are potentially indirect agonists of every 5-HT receptor subtype. Since MDMA and MDA also release dopamine and noradrenaline (Nash and Nichols, 1991;Kankaanpää et al., 1998;Rothman et al., 2001;Starr et al., 2012;Brandt et al., 2020), this brings in a raft of other potential contributors to their therapeutic actions. The balance of MDMA's potentiating effect on individual monoamines is likely, therefore, to be an important factor in defining its efficacy and adverse event profile. Studies in vitro and in vivo have suggested that MDMA and MDA produce similar increases of 5-HT and dopamine efflux (Kankaanpää et al., 1998;Rothman et al., 2001;Brandt et al., 2020). Such observations can be highly misleading because they often do not sufficiently take account of dose, which can markedly change the balance of effects across the monoamines; for example, many phenethylamine releasing agents exert a much greater effect on dopamine as the dose is increased (see Heal et al., 2009Heal et al., , 2013. The basal concentration of a neurotransmitter is a reliable indicator of neuronal density and tonic activity. Given that the basal efflux of dopamine is ~200x greater than 5-HT in the striatum and ~20x greater in the prefrontal cortex (Rowley et al., 2014), it is erroneous to conclude that a drug which produces a similar percentage increase in the basal efflux of these two monoamines will exert the same potentiating effect on their CNS function. Many microdialysis experiments comparing the relative effect of MDMA or MDA on different monoamines are based on a single brain region, often the striatum or nucleus accumbens (Nash and Nichols, 1991;Kankaanpää et al., 1998;Brandt et al., 2020), which are probably not relevant to their therapeutic effect in post-traumatic stress disorder (PTSD).
Finally, it is important to appreciate that part of the development of a novel CNS drug-candidate will be to screen it against a broad panel of targets, e.g., receptors, modulatory binding sites, ion channels, transporters and enzymes. This step defines the spectrum of the drug-candidate's pharmacological actions and predicts its potential off-target adverse events. Although one or more of the serotonergic psychedelics may have undergone such screening, we could find no published reports leaving the possibility that these compounds may possess unknown pharmacological properties that could be beneficial or harmful.
From the perspective of developing a next generation of serotonergic psychedelics and entactogens, some aspects of pharmacological characterisation are relatively straightforward; for example, there are highly reliable in vitro screening techniques to identify which 5-HT receptor subtypes and monoamine reuptake transporters are the pharmacological targets of the NCE (new chemical entity) and functional assays exist to elucidate the nature of those interactions. Broad-spectrum in vitro screens covering hundreds of drug targets are also available to identify and predict potential side-effects before the drug-candidate enters clinical development.
If the induction of a psychedelic experience is essential for the therapeutic effect of the 5-HT 2A agonist psychedelics, a key challenge will be the availability of reliable non-clinical models to predict hallucinogenic effects in humans. Discovering a next generation serotonergic entactogens poses a different challenge which will be to identify which monoamine neurotransmitters are essential for efficacy and what is the relative balance of effects to maximise efficacy and minimise sideeffects.

Experimental techniques to characterise mechanism of action
The compounds are divided into two distinct classes, psychedelics and entactogens. The pharmacological mechanism responsible for the efficacy and safety of the psychedelics, (predominantly 5-HT 2A agonism), is very different from that of the entactogens, (predominantly neuronal release of 5-HT and dopamine). A critical first step in the development of novel serotonergic psychedelics and entactogens is to define the pharmacology of the lead compound or drug-candidate. Fig. 2 illustrates our proposed screening cascade to characterise the mode of action novel serotonergic compounds and also identify key pharmacological properties which are potential flags for drug safety.
The subtypes of 5-HT 2 receptors have close homology and as discussed earlier the serotonergic psychedelics have affinity for and functional activity at all of them. In vitro screening can be used to define affinity and function of the compound at 5-HT 2A receptors to predict efficacy. Functional screening against 5-HT 2B and 5-HT 2C receptors is strongly recommended because 5-HT 2B agonism can mediate cardiac valve damage as discussed in more detail later in this review, e.g., fenfluramine/norfenfluramine (Rothman et al., 2000;Setola et al., 2005), and 5-HT 2C agonism has the potential to decrease appetite/bodyweight, e.g., lorcaserin (Thomsen et al., 2008;Hess and Cross, 2013).
The entactogens act via interactions with the monoamine reuptake transporters and screening for affinity at the transporters for 5-HT (SERT), dopamine (DAT) and noradrenaline (NET) is important; defining whether the molecule serves as a competitive inhibitor or substrate is essential. Many of the psychedelics are phenethylamines or tryptamines and close structural homologues of the catecholamines and 5-HT, respectively. Some, e.g., MDMA and MDA, are amphetamines and well-known transporter substrates. These compounds have much more profound potentiating effects on monoaminergic signalling than reuptake inhibitors, and because of their intraneuronal site of action, have the potential to produce neurotoxicity (O'Hearn et al., 1988;Heal et al., 1998Heal et al., , 2013Cheetham et al., 2000). Molecular size dictates whether a molecule is a transporter substrate or blocker; the latter are physically too large to be transported into the presynaptic terminal (see Heal et al., 1998Heal et al., , 2013Heal et al., , 2014. Monoamine reuptake transporters are promiscuous in the amines they will transport and substitution is a critical determinant of the spectrum of action. Adding steric hindrance groups to the para-position on the phenyl ring reduces transport by DAT, but not SERT. The rank order of monoamine release of D-amphetamine (no substitution) is dopamine > 5-HT, MDMA (methylendioxy group) dopamine ≈ 5-HT and D-fenfluramine (trifluoromethyl group) dopamine « 5-HT (see Heal et al., 2013).

The rodent head-twitch model
Head-twitches in mice and wet dog shakes, their equivalent in rats, are widely assumed to be functional models of 5-HT 2A receptor agonism (see reviews by Green and Heal, 1985;Heal et al., 1992) and for this reason have been used increasingly as a behavioural screen to predict whether 5-HT 2A agonists have hallucinogenic properties. In our previous review, we provided a comprehensive list of the fungal, botanical, zoological and synthetic serotonergic drugs that have been tested in the model (Heal et al., 2018) which revealed that almost all known 5-HT 2A agonists and partial agonists evoke this behaviour in rodents. A strong correlation has been observed between the potency of phenylalkylamine and tryptamine 5-HT 2A agonists to induce head-twitches and to induce hallucinations in man . Hence, the head-twitch model can not only detect hallucinogenic activity in novel compounds, but in the case of the 5-HT 2A agonists, also predict their likely potency as hallucinogens in man; important information when planning clinical trials with powerful CNS drug-candidates.
There is a general consensus that the therapeutic effects of the serotonergic psychedelics like psilocybin and DMT are dependent on their hallucinogenic properties. Therefore, it is important to understand that head-twitch behaviour in mice was originally proposed as a model to identify drugs that produce hallucinations in man, not as a test for 5-HT 2A agonism (Corne and Pickering, 1967). Hence, non-serotonergic hallucinogens including the NMDA receptor antagonists, phencyclidine and ketamine, produce head-twitches in mice (Corne and Pickering, 1967;Yamaguchi et al., 1987;Rivera-García et al., 2015;Liu et al., 2006;Kim et al., 1999). The latter results are relevant because ketamine produces rapid and prolonged effects in major depressive disorder (Zarate et al., 2006;Ibrahim et al., 2012;Murrough et al., 2013) via a mechanism closely associated with its hallucinogenic effects (Sos et al.,

Fig. 2.
Proposed early-phase screening to characterise the pharmacology of novel serotonergic psychedelics and entactogens. 1 5-HT 2A agonists typically selected as a reference discriminative cue are 5-MeO-DMT or DOI; 2 Initial screening for affinities at the most relevant 5-HT receptor subtypes and reuptake transporters; 3 The hippocampus has good microdialysis characteristics for 5-HT, dopamine and noradrenaline. If dual-probe microdialysis is used to pharmacologically characterise novel drug-candidates, a dopamine signal in nucleus accumbens shell provides useful information on its potential for abuse; more relevant to entactogens than hallucinogens. 4 Tetrodotoxin reverse-dialysed via the probe will block increases of extracellular monoamines produced by reuptake inhibitors (transporter blockers) but not monoamine releasing agents (transporter substrates).SERT: serotonin reuptake transporter; DAT: dopamine reuptake transporter; NET: norepinephrine (noradrenaline) reuptake transporter; DA: dopamine; NA: noradrenaline: TTX: tetrodotoxin. 2013; Dore et al., 2019;Mathai et al., 2020;Sumner et al., 2021).
A further link between the antidepressant mechanism of the NMDA antagonists and serotonergic hallucinogens comes from the observations that NMDA antagonists potentiate 5-HT-mediated head-twitches (Kim et al., 1998(Kim et al., , 1999 and head-twitches produced by ketamine or phencyclidine are blocked by 5-HT 2A antagonists (Yamaguchi et al., 1987;Kim et al., 1999).
The head-twitch response is, therefore, a useful in vivo model for predicting hallucinogenic activity in a wider range of drugs than simply the 5-HT 2A agonists (Corne and Pickering, 1967; see also Heal et al., 2018). In addition, findings from head-twitch experiments revealed there is 5-HT 2A agonist component to the pharmacology of the NMDA antagonists, providing a tentative common link between their shared rapid onset and prolonged antidepressant effects.
Although head-twitches are induced by various CNS drugs which lack hallucinogenic activity, e.g., the benzodiazepines (see Heal et al., 2018), the model does not throw up many "false positives". Its major disadvantage is head-twitches can be profoundly suppressed by activating other neurotransmitter systems including noradrenaline, dopamine and GABA (see reviews by Green and Heal, 1985;Heal et al., 1992Heal et al., , 2018. Therefore, the head-twitch model has limited value when screening novel hallucinogens with pharmacological actions which include the activation of neurotransmitter systems that suppress head-twitches. This point is exemplified by the observation that the selective 5-HT releasing agent, fenfluramine, induces head-twitches (Joshi et al., 1983;Gada et al., 1984;Darmani, 1998), whereas MDMA, which releases 5-HT, dopamine and noradrenaline (Nash and Nichols, 1991;Kankaanpää et al., 1998;Rothman et al., 2001;Starr et al., 2012;Brandt et al., 2020) does not induce head twitches (Fantegrossi et al., 2005;Heal, unpublished observations).
In summary, the head-twitch model is a well-established and validated predictive test of hallucinogenic activity. When screening psychedelic drugs with multiple pharmacological actions, these compounds may appear as "false negatives". In this situation, screening novel serotonergic compounds for hallucinogenic activity using the headtwitch response should be supplemented by evaluation in 5-HT 2A agonist-cued drug-discrimination testing.

5-HT 2A agonist cued drug-discrimination
When discussing CNS drug-candidates, drug-discrimination is often considered solely in terms of its role as one of the models in the Safety Pharmacology assessment of abuse and dependence risks  , 2017]). However, drug-discrimination is an excellent screening tool to identify research compounds that produce an interoceptive cue that is similar to the discriminative drug the animal has been trained to recognise. Dr Nichols and his colleagues used LSD-cued drug-discrimination as a screen for novel hallucinogens starting in the 1980s (see Oberlender et al., 1984;Nichols et al., 1986Nichols et al., , 1994Schultz et al., 2008).
There are intriguing parallels between the head-twitch response and LSD discriminative cue as revealed by the finding that in a group of rats trained to discriminate salvinorin-A (a κ-opioid receptor psychedelic) from vehicle, LSD generalised to the salvinorin-A cue in 3/5 rats and showed 60-65% generalisation in 2/5 rats (Peet and Baker, 2011). Moreover, ketamine also generalised to salvinorin-A (Peet and Baker, 2011).
In summary, the drug-discrimination model using either LSD or a more selective 5-HT 2A agonist like DOI or DOM is a good test for detecting hallucinogenic activity in novel serotonergic psychedelic drug-candidates. The LSD discriminative cue has a more multifaceted pharmacology than DOI or DOM which creates the possibility of detecting more "false positives" in the test. However, with a couple of notable exceptions, drugs that are not hallucinogenic only generalise partially to the LSD cue. The finding that there is partial and crosssubstitution of between the LSD cue and those elicited by other hallucinogens, e.g., ketamine, yohimbine, salvinorin A, provides an additional dimension to its application. In summary, LSD may be as close as we can get to having a screening tool to detect the "psychoactive signature" of hallucinogenic drugs.

Neurochemical characterization by intracerebral microdialysis
Intracerebral microdialysis is a valuable technique for neurochemically and pharmacologically characterising psychedelics and entactogens, which can provide insights on mechanism of action and raise flags about potential safety. The sophisticated analytical methods, i.e., hplc with electrochemical detection (hplc-ecd) or mass spectrometry (hplc-ms) that are now routinely used to analyse microdialysate samples simultaneously quantify multiple monoamines including 5-HT, dopamine and noradrenaline and the majority of their metabolites, e.g. 5-HIAA, DOPAC, 3-MT, and HVA, and in the case of hplc-ms, the concentration of experimental drug in the extracellular fluid that is being sampled by the microdialysis probe.
Strengths and breadth of the information that can be obtained by intracerebral microdialysis are summarised as: 1. The ability to study the effects of the drug-candidate in specific brain regions; they include those relevant to efficacy in the selected therapeutic indication, potential for abuse (dopamine measurements in the nucleus accumbens), and safety. 2. The use of dual-probe microdialysis permits investigation of drug action simultaneously in 2 separate brain regions. 3. Simultaneous measurement of the 3 key monoamine neurotransmitters, 5-HT, dopamine and noradrenaline, helps to define the spectrum of monoaminergic effects of the psychedelic drugcandidate. As a quantitative technique, it provides information not only about the drug-candidate's effect on each monoamine, but also the relative level of effect versus one another. 4. As a multiple sampling technique, microdialysis can be used to study the speed of onset of drug effect, rate of change and maximal effects on extracellular monoamine concentrations and duration of action. 5. The use of neuronal firing inhibitors and reuptake transporter blockers in microdialysis experiments can elucidate whether a drugcandidate has an intraneuronal site of action or binds to reuptake transporters on the external surface of the presynaptic terminal. 6. Microdialysis can be simultaneously combined with automated blood sampling in venous catheterised rats to explore the relationship between the plasma drug concentrations and neurochemical effects in specific brain regions.
7. Microdialysis can also be combined with automated locomotor activity monitoring in rats or mice to look for correlations between neurochemical effects in the brain and effects on behaviour.
As discussed previously, most studies suggest that serotonergic psychedelics have little or no pharmacologically relevant effect on extracellular monoamine concentrations (Bogdanov et al., 1991;Gudelsky et al., 1994;Sakashita et al., 2015;Miliano et al., 2019). Since many of these molecules are close structural analogues of the indolamine and catecholamine neurotransmitters, it indicates their pharmacological actions are restricted to agonism at 5-HT 2A and other monoamine neurotransmitter receptors. Comprehensively defining the pharmacological profile is critical to developing the next generation of psychedelic drug-candidates.
The entactogens, on the other hand, are monoamine reuptake transporter ligands where they can serve as either competitive inhibitors or substrates. As demonstrated by microdialysis, MDMA and MDA enhance the synaptic concentration of 5-HT, dopamine and noradrenaline (Nash and Nichols, 1991;Kankaanpää et al., 1998;Rothman et al., 2001;Starr et al., 2012;Brandt et al., 2020). This is typical of monoamine transporter substrates which are generally relatively promiscuous as substrates for SERT, NET and DAT (Heal et al., 2013). It has been demonstrated on a number of occasions that localised perfusion of tetrodotoxin or removal of Ca ++ from the dialysate abolishes the neuronal firing-mediated increase of extracellular monoamines by reuptake inhibitors but does not attenuate the increase produced by high doses of transporter substrates like MDMA and fenfluramine and amphetamine (Carboni and di Chiara, 1989;Bonanno et al., 1994;Gudelsky and Nash, 1996;Heal et al., 2017). The intraneuronal action of MDMA and fenfluramine has been linked with serotonergic neurotoxicity in animals (Cheetham et al., 2000;Baumann and Rothman, 2009;Song et al., 2010;Halpin et al., 2014) and humans (McCann et al., 2005;Gouzalis-Mayfrank and Daumann, 2009;Benningfield and Cowan, 2013;Vegting et al., 2016;Müller et al., 2019) which has important safety implications for the development of novel entactogens. Reuptake inhibitors, on the other hand, bind to reuptake transporters on the of action on neurones, they are non-neurotoxic, and they are able to prevent the neurotoxic effects of reuptake transporter substrates by blocking their entry into presynaptic terminals (McCann and Ricaurte, 1993;Cheetham et al., 2000;Goñi-Allo et al., 2006;Piper et al., 2008). If a novel entactogen-like molecule has SERT affinity, determining whether it is a substrate or blocker is important because it has implications for safety and dosing regimens, i.e., level, frequency and duration.
Psychedelics and entactogens produce behavioural activation and other profound psychoactive effects that will impact on tolerability and safety that need to be taken into consideration when developing the next generation of drugs. Microdialysis with automated activity monitoring and blood sampling combined with drug quantification in plasma and microdialysate samples is an extremely valuable tool to investigate the PK/PD (neurochemical and/or behavioural) relationships for novel compounds, e.g., see Rowley et al. (2012Rowley et al. ( , 2014 and Gill et al. (2016).

Summary of pharmacological characterization and prediction of efficacy
These serotonergic compounds comprise two sub-classes with very different pharmacological characteristics, i.e., the hallucinogenic psychedelics and entactogens, and potentially very different clinical applications in treating CNS and peripheral disorders. The extensive body of information from human experience and thousands of non-clinical and rather fewer clinical studies provides an adequate platform to assess the benefit/risk for their use as therapeutic agents. However, quantity does not compensate for quality, and because the psychedelics directly or indirectly engage multiple receptor targets, there are almost certainly gaps in our knowledge about their therapeutic mechanism(s) of action. It is also probable that new drug-candidates will emerge with hybrid profiles that do not fit into either the "hallucinogen" or "entactogen" boxes. Fenfluramine is a 5-HT releasing agent that is structurally and pharmacologically related to MDMA that induces head-twitches (Green and Heal, 1985;Heal et al., 1992) and at supratherapeutic doses is hallucinogenic in humans (Levin, 1973;Griffith et al., 1975). More recently, novel substituted phenethylamines have appeared as "designer drugs" which possess both 5-HT 2A agonist and monoamine releasing properties, e.g., 2,5-dimethoxy-4-chlorophenethylamine (2C-C), 2,5-dimethoxy-4-methylphenethylamine (2C-D), 2,5-dimethoxy-4-ethylphenethylamine (2C-E), 2,5-dimethoxy-4-iodophenethylamine (2C-I), 2,5-dimethoxy-4-ethylthiophenethylamine (2C-T-2) and 2, 5-dimethoxy-4-chloroamphetamine (DOC) (Eshleman et al., 2014). This finding confirms the feasibility of combining these two pharmacological actions in a single molecule. Therefore, early-stage pharmacological characterization is an essential pre-requisite to predicting the psychotropic properties of the drug-candidates and to assist in directing them to the most appropriate therapeutic indications. Our recommendation is to use this early-stage screening as a platform to characterise novel psychedelic chemical series or compounds prior to evaluating them using in vitro or in vivo models predictive of therapeutic activity.

Assessment of potential for human abuse
The serotonergic psychedelics are almost without exception, controlled substances that reside in Schedule 1 (C-I) in the USA, Europe and many other countries. C-I is the schedule for compounds that pose a risk for abuse and/or dependence in humans and have no approved medical or veterinary use. It is often mistakenly assumed that C-I substances carry the highest risk for abuse and/or dependence. It is the lack of an approved medical or veterinary that use dictates C-I, and this schedule includes drugs that cover the spectrum from moderate abuse/ dependence risk, e.g., cannabinoids, through to highly addictive opioids, e.g., heroin is C-I in the USA.
With "real world" experience and evidence from extensive research in animals and humans, the abuse and dependence risks posed by serotonergic psychedelics like psilocybin, DMT, LSD and MDMA are well understood (see Johnson et al., 2018;Calderon et al., 2018Calderon et al., , 2022  The next generation of serotonergic psychedelics and entactogens will, however, need to undergo the same program of non-clinical and clinical evaluations to assess their abuse and dependence risks which apply to any other novel CNS-active, drug-candidates (Calderon et al., 2018;Heal et al., 2018).
The non-clinical, safety pharmacology evaluation focuses on two aspects of abuse potential, i.e., (i) exploring the psychoactive effects produced by the drug-candidate and (ii) its rewarding effects that could lead to psychological dependence and drug-seeking, and one on dependence, i.e., determining whether it produces physical dependence on withdrawal. We have recently published a review which comprehensively describes the experimental procedures for evaluating the abuse and dependence potential of serotonergic hallucinogens and entactogens, as well as other types of psychedelics, i.e., NMDA antagonists and κ-opioid agonists (Heal et al., 2018). In addition, we have briefly revisited the topic including the challenges of conducting human abuse trials with hallucinogens (Henningfield et al., 2022b). Hence, we will only provide an overview of the topic in this review together with revisions to the procedures based on evolving developments and opinions that have emerged in the period since writing our last major review on this topic 4 years ago. Our revised testing programme to evaluate the abuse and withdrawal-induced physical dependence potential of novel psychedelic drugs is shown in Fig. 3.

General considerations
The non-clinical evaluation of a drug-candidate's potential for abuse and dependence is part of the Safety Pharmacology assessment and should be conducted in accordance with GLP (CHMP/EMA, 2006; CDER/FDA, 2017; Heal et al., 2018). Therefore, results obtained from non-GLP, 5-HT 2A agonist or MDMA cued drug-discrimination testing in Early-stage Pharmacological Screening will not meet the regulatory standard required for a Safety Pharmacology assessment of abuse potential. The European and US guidelines advise that unless there are specific scientific reasons to use primates, all non-clinical testing should be conducted in rodents (CHMP/EMA, 2006; CDER/FDA, 2017); in essence, it means rats because mice are rarely, if ever, used. Non-GLP screening against a broad panel of abuse-related molecular targets in vitro is a prerequisite to ensure the design of the GLP abuse and dependence experiments are correctly aligned with the pharmacology of the drug-candidate. Complementary pharmacological information obtained from Early-stage Pharmacological Screening can serve as a tool to Fig. 3. Non-clinical testing program to investigate the abuse and dependence potential of novel psychedelics and entactogens. As this assessment forms part of the Safety Pharmacology package, all testing should be conducted according to GLP (Good Laboratory Practice). Using MDMA as a training cue in drug-discrimination will detect the abuse potential of entactogens and hallucinogens (see Heal et al., 2018). If non-GLP, 5-HT 2A agonist-cued drug discrimination has been used as a mode of action screen, drug-discrimination testing using MDMA rather than a 5-HT 2A agonist as the training cue would increase the scope of the information on the compound. Based on evidence obtained with the legacy psychedelics, it is probable, but not certain, that novel 5-HT 2A agonist hallucinogens will not serve as positive reinforcers in animals. Novel entactogens, which release dopamine as well as 5-HT, have a strong probability of serving as positive reinforcers, cf MDMA. An assessment of the potential to induce withdrawal-induced physical dependence is a mandatory part of the Safety Pharmacology assessment. Evidence from the legacy psychedelics predicts that novel hallucinogens and entactogens will not produce withdrawal-induced physical dependence. assist in the experimental design process. It is important to appreciate that the sole purpose of this part of the Safety Pharmacology assessment is to predict the drug-candidate's potential for abuse and dependence; it is not to perform additional pharmacological characterization of the molecule. The use of positive control (abused substance) and negative control (vehicle/placebo) and reference comparators (if included) are only there to validate the model and provide context to the safety findings.

Drug-discrimination
Drug-discrimination answers the question whether the interoceptive cue ("psychoactive signature") of the drug-candidate is identical or similar to that of an existing controlled substance of abuse.
MDMA is a versatile drug for use as a training cue because rats trained to discriminate between MDMA and vehicle can detect the abuse potential of serotonergic drug-candidates that are either hallucinogens or entactogens (see Heal et al., 2018). Practical advantages include MDMA's ease of synthesis and lack of tolerance to its psychoactive effects; a significant asset in a test where training involves frequent exposure to the drug over many weeks. MDMA's disadvantage as a cue is due to the fact that hallucinogenic psychedelics and entactogens generalise fully to MDMA although the test detects psychoactive effects that could lead to abuse, it is not possible to predict whether the drug-candidate is an entactogen or a hallucinogenic psychedelic. The use of a 5-HT 2A agonist as the discriminative cue is a more appropriate model when developing a psychedelic drug-candidate because the result should simultaneously address the question of its potential for abuse and confirm earlier predictions concerning its hallucinogenic properties. 5-HT 2A agonist-cued drug-discrimination is more "hit and miss" for detecting an abuse potential signal in novel entactogen drug-candidates. Various MDMA and MDA isomers failed to show generalisation when investigated in drug-discrimination experiments where DOM, mescaline or LSD were used as the training cue (Glennon et al., , 1988Callahan and Appel, 1988;Baker and Taylor, 1997). Evaluating the cue elicited by a novel drug-candidate with mixed 5-HT 2A agonist/monoamine releasing effects is even more problematic because compounds with this pharmacological profile do not consistently generalise either to 5-HT 2A agonists or MDMA. Based on the results reported by Eshleman et al. (2014), most would, however, achieve >60% generalisation to the cues produced by MDMA or a 5-HT 2A agonist and, therefore, exceed the threshold which the FDA sets for compounds sharing some pharmacological similarity to the training substance of abuse (CDER/FDA, 2017).

Intravenous self-administration
Intravenous self-administration experiments in rats or primates evaluate whether the drug-candidate produces reinforcing psychoactive effects that could lead to drug seeking ("craving"). It has been shown in many laboratories including our own that entactogens like MDMA and the cathinones, mephedrone and methylone, serve as positive reinforcers in rats and primates (Beardsley et al., 1986;Fantegrossi et al., 2002;Creehan et al., 2015;Vandewater et al., 2015;Heal et al., 2018).
Any novel psychedelic or entactogen drug-candidate will automatically carry a "red flag" for abuse, and consequently, should have its potential reinforcing effect determined by intravenous selfadministration testing. This mandate applies irrespective of whether it is an entactogen, mixed entactogen/hallucinogen or a hallucinogenic psychedelic. Our experience with entactogens is they generally produce moderate reinforcing effects in rats (Heal et al., 2018), and therefore, the experimental conditions of the intravenous self-administration test need to be adjusted to ensure the model has the relevant sensitivity to detect a drug-candidate's abuse signal. When using rats as preferred by the FDA and EMA (CHMP/EMA, 2006; CDER/FDA, 2017), we recommend that the animals be trained using a powerful reinforcer on a low fixed ratio (FR) schedule (FR3 or FR5) before saline extinction on the same reinforcement schedule (to eliminate false responders). The novel psychedelic can then be substituted in the model on the same reinforcement schedule to determine whether it serves as a positive reinforcer. Testing should be conducted with the objective of achieving stable levels of self-administration for each unit dose and this requires a minimum of 5 test sessions. To keep the experiment within reasonable time-frame, we advise a maximum of 10 test sessions; variable self-administration responses are meaned across all test sessions to give a reliable estimate of the magnitude of the reinforcing signal. The unit doses of the drug candidate selected for evaluation should produce plasma concentrations in the test animals that are fractions of the human exposure at the maximum therapeutic dose. This is necessary to avoid creating false negative results due to drug saturation occurring after the animals have taken very few intravenous infusions.
The general consensus is hallucinogenic psychedelics do not robustly or consistently serve as positive reinforcers in animals. While it may be correct, a search of the literature revealed very few publications to support this conclusion (Fantegrossi et al., 2004;Goodwin, 2016). This assumption needs, therefore, to be treated with a degree of scepticism for the following reasons. When dealing with weak reinforcers, the use of high drug reinforcement schedules can create false negatives, and as revealed from experience with colonies of drug-experienced primates, only a small minority of animals will robustly self-administer weak reinforcers like benzodiazepines. Hence, it is possible that the non-reinforcing effect of the hallucinogens is a false negative created by a lack of sensitivity and experimental rigour. This point is exemplified by our experience in demonstrating the positive reinforcing effects of various weak reinforcers including benzodiazepines, CB1 agonists and novel CNS drug-candidates Heal et al., 2018;Gray et al., 2022;Synan et al., 2022).

Withdrawal-induced physical dependence
The withdrawal-induced dependence test evaluates whether abrupt termination of dosing produces signs of dependence. It is important to note that this is physical, not psychological dependence. This may be a rather prescriptive view of the test because it involves behavioural dependence end-points as well as physiological and physical signs of dependence (Goddard et al., 2017;Heal et al., 2018). Psychological dependence is indirectly assessed in the intravenous self-administration test, and its presence can be confirmed by the reinstatement of drug-seeking provoked by stress, drug-associated salient cues, drug priming, or combinations thereof in experiments where lever-pressing by the animals results only in the delivery of saline infusions (Heal and Smith, 2019;Smith and Heal, 2020). In the case of opiates and benzodiazepines, pharmacological tolerance is a progenitor to withdrawal-induced physical dependence and, therefore, this aspect of the pharmacology of drug-candidates is also explored in the test. However, it is important to state that inducing pharmacological tolerance does not imply a drug will produce withdrawal-induced dependence; many CNS drugs require dose-titration to produce tolerance to their side-effects without adversely influencing their long-term therapeutic effect or inducing dependence on withdrawal.
The typical design for a withdrawal-induced physical dependence experiment is to administer the drug-candidate at doses producing 2-3x multiples of the clinical exposure (supratherapeutic dose). Dosing should be for a minimum of 28 days to allow sufficient time for the drugcandidate to produce neuronal adaptation. At this point, dosing is abruptly terminated and spontaneous behavioural, physiological and physical signs of dependence are then monitored for 7 days (≥5 drugcandidate half-lives). When evaluating novel psychedelic drugcandidates, the positive control would be an opiate or benzodiazepine because they produce robust physical dependence signs, and the negative control would be the vehicle. Based on current knowledge, it is likely that novel serotonergic psychedelics and entactogens will induce pharmacological tolerance, but it is unlikely that they will produce physical dependence on withdrawal.

Safety and toxicity concerns specific to the next generation of psychedelics
In addition to discussing abuse potential and how it needs to be evaluated for new psychedelic and entactogen drug-candidates, the following sections consider the potential toxicities of these molecules from a clinical safety perspective. This assessment has drawn on previously identified toxicity targets for the serotonergic compounds (Fig. 4) with extrapolation to new potential toxicity targets by examining 5-HT receptor subtype localisation on various cell types and target organs. The application of Preclinical Lead Optimisation Technology (PLOT) screening to minimise drug pipeline attrition rate and improve human safety profiling will also be discussed in relation to the potential toxicity targets. The implications and 'toxicological hazard' data will be covered in general terms in this part of the review and no particular or detailed distinction will be made between acute and chronic animal safety profiling or reported clinical adverse effects. The regulatory animal and in vitro preclinical studies prescribed for progressing molecules from IND (Investigational New Drug) to NDA (New Drug Application) status in pharmaceutical development, or the necessary toxicokinetic algorithms for assessing risk in terms of human safety margins will not be covered in any detail in this section (see Section 4.4 for more detail).
For the preclinical (and clinical) risk assessment of psychedelic NCEs associated with central or peripheral nervous systems, or peripheral cell 5-HT release (as presynaptic SERT/VMAT [vesicular monoamine transporter] co-transporter substrates) and/or 5-HT receptor subtype stimulation, there are three major classes of target organ toxicity to evaluate and potentially investigate. Two 'classes' (A -Neurotoxicity and B -Cardiotoxicity) are derived from historical findings with known serotonergic compounds and a third 'class' (C -Other Cell Types/Organs) is predicted from the localisation of various 5-HT receptor subtypes on cells, which could manifest as toxicological targets, the affinity of known molecules for these proteins, and includes the clinical 'Serotonin Syndrome'. The three safety categories are described in detail below.

Neurotoxicity
For a detailed discussion of serotonergic neurotoxicity and scientific data see O'Callaghan & Miller, 1994;Mueller et al. (2012);McCann et al. (1994McCann et al. ( , 1996McCann et al. ( , 2000. It is key in preclinical drug safety evaluation for the human risk assessment (in the CNS) to differentiate between irreversible/structural neurotoxic damage as detected neuropathologically, and reversible, reactive, neurobehavioural and neurochemical changes, as well as differentiating between developmental and adult neurotoxicological effects.
There has been a plethora of studies and debates on this topic between different groups (see above) providing somewhat equivocal conclusions between MDMA and fenfluramine/dexfenfluramine neurotoxicity across species and a paucity of conclusive clinical data to substantiate animal findings. In general terms, fenfluramine, dexfenfluramine, MDMA and other substituted amphetamines produce sustained, but potentially partially reversible 5-HT depletion and CNS 5-HT 'overload' mechanistically through their action on SERT and VMAT proteins (and dopamine systems in the case of MDMA in rodents and nonhuman primate models with associated loss of axonal/terminal density and SERT sites). However, there is a lack of unequivocal evidence for lasting neuropathic axonal damage to the long projecting serotonergic neurones from the median raphe nuclei region. The body of evidence suggests that MDMA (and possibly MDA versus fenfluramine/ dexfenfluramine does in fact produce irreversible axonal neuropathies in some species as seen through neuropathological histological stains (e. g., silver staining or fluro-jade B), neurochemical sequelae and large elevations of GFAP (glial fibrillary acidic protein), and reactive astrogliosis (astrocytic hypertrophy and hyperplasia) (Herndon et al., 2014).
Findings supporting neuropathic damage by MDMA show that excitotoxic processes can be involved and that metabolites of MDMA can generate free-radical species, e.g., peroxynitrite/superoxide, resulting in lipid peroxidation and oxidative stress to 5-HT neurones. In some cases, there has been evidence for dopaminergic lesions caused by MDMA. The variety of neurotoxicological possibilities with these classes of psychedelics thus mitigates heavily towards bespoke design of both PLOT screens in late stage drug-candidate selection and preclinical regulatory animal studies in drug-candidate development. These would incorporate serotonergic neuro-behavioural parameters, detailed neurochemical Fig. 4. Historical safety and toxicity events associated with the classical psychedelics and entactogens. The serotonergic hallucinogens and entactogens have pharmacological properties that have been linked with central and peripheral adverse events. The hallucinogens and entactogens have different safety risk profiles depending on their pharmacological mechanism of action. As an example, MDMA is associated with neurotoxicity because it is a SERT substrate which enters serotonergic neurons (McCann et al., 1996(McCann et al., , 2000Rothman and Baumann, 2002), but causes cardiac valve damage because it is a 5-HT 2B agonists (Setola et al., 2003). The individual risks posed serotonergic psychedelics are discussed in the review.
analyses of brain regional serotonergic (and other monoaminergic) components and peripheral biomarkers as well as comprehensive staining for neuropathological damage via conventional staining and parallel immunocytochemical stains for GFAP. The process would also assess 'lesion' reversibility, species-specificity and accurate toxicokinetic multifactorial modelling across species to determine and set meaningful safety margins for dosing in clinical trials. Concerning factors around MDMA particularly and its potential neurotoxicity in humans is its ability to produce neurotoxic 'lesions' after relatively short dosing/exposure periods and the exacerbation of its effects by hyperthermia (a clinical manifestation of MDMA use) and the possible enhanced vulnerability of long-projecting neurones to lasting neuropathies.

Cardiotoxicity/valvulopathy
There is a well-recognised toxicological liability for some serotonergic molecules to cause cardiotoxicity (Hutcheson et al., 2011;Elangbam, 2010) in the form of valvulopathy in both animals and humans. This process occurs mechanistically through overstimulation of the 5-HT 2B receptor subtype on valvular interstitial cells (VIC), by partial agonist activity (together with possible local 5-HT release and/or systemic 5-HT overload). Carcinoid heart disease was one of the first valvular pathologies studied showing that 5-HT produced by oncogenic gut enterochromaffin cells was the likely causative agent. In the late 1990's it was shown that the then popular anorexic diet drug, Fen-Phen (fenfluramine/phentermine combination), produced similar valvular pathologies. Due to the known mitogenic properties of the 5-HT 2 receptor subfamily, a number of groups (see Rothman et al., 2000) compared the in vitro pharmacology of norfenfluramine (the active metabolite of fenfluramine) with that of known inducers (5-HT, methysergide) of HVD (heart valvular disease) and found that HVD-associated compounds were potent 5-HT 2B agonists. In 1997-2007, the anti-obesity drugs, fenfluramine and dexfenfluramine, and the anti-Parkinson drug, pergolide (dopamine and 5-HT 2B receptor agonist), were withdrawn by the FDA of the USA for drug-induced valvulopathy safety concerns. MDMA and MDA are also 5-HT 2B agonists and produce prolonged mitogenic responses in human VIC's in vitro (Setola et al., 2003), and significant valvular regurgitation has been reported in MDMA users with similar valvular pathologies (Droogmans et al., 2007). As discussed in Section 2.3, the classical psychedelics are not selective for 5-HT 2A receptors and are also 5-HT 2B partial agonists. While this property is not a concern when they are administered in one or two high-dose psychedelic sessions, cardiac toxicity could become a risk if they are taken frequently as low-or micro-doses.
VICs, which are the most prevalent cells in all three layers of the valvular leaflet, maintain the integrity and stability of normal valvular function and regulate repair processes during disease and injury. In addition to various intracellular signalling molecules such as phosphokinase-C (PK-C) controlling mitogenesis, there are other cell membrane receptor proteins, e.g., transforming growth factor-β (TGF-β) participating alongside the 5-HT 2B receptor and are involved in the carcinoid syndrome-mediated VIC mitogenesis. The net toxicological result of this drug-induced 5-HT 2B mediated VIC 'overgrowth' (valvulopathy, leaflet thickening of potentially mitral, tricuspid and aortic valves) is valvular inefficiency, leading to cardiac blood leakage with functional deficit sequelae including pulmonary aortic hypertension. Despite the initial withdrawal of fenfluramine as an anorexic agent, fenfluramine as Fintepla® was subsequently approved for clinical use under supervision for epilepsy treatment with echocardiographic monitoring. It has been estimated that 10-15% of individuals taking fenfluramine can present with adverse drug reactions and, of course, overdose/overuse of MDMA has potentially more serious toxicological consequences.
In conclusion, as well as cultured human VICs providing a potential PLOT platform for screening new psychedelics for 5-HT 2B -mediated valvulopathy, parallel in vivo regulatory investigative toxicology studies in new psychedelic molecule development phases (and clinical monitoring in vivo) are both necessary and important as some findings suggest 5-HT 2B agonist potency alone may not provide a correlative prediction of degree of VIC mitogenic responses.

Other potential toxicity targets
The third more speculative, but evidence-based, category addresses other potential cellular and molecular toxicity targets for serotonergic drug-candidates causing either 5-HT release/overload outside the CNS (potential binding to SERT, tryptophan hydroxylase [TPH], monoamine oxidase [MAO]) and/or acting as full agonists/partial agonists at 5-HT receptor subtypes on various cells and organs in the body). An attempt to predict 'idiosyncratic toxicities' by extrapolating such information should help to preclude/exclude unexpected and unwanted toxicities during regulatory preclinical and clinical development phases. Additionally, as new psychedelic treatment strategies in the form of e.g., psilocybin and psilocin have recently evolved for intractable depression (by using low dose or microdosing procedures) and about which currently there is very little published toxicological information or regulatory background. Although renal failure has been reported in rare cases, the toxicity of psilocybin-containing mushrooms is reported to be low with an acute toxicity LD 50 value of 280 mg/kg in rodents. It is hoped that this short section may provide some advance insight into possible problem areas, pitfalls and barriers in conventional regulatory, preclinical and clinical drug development studies involving sub-chronic and chronic dosing regimens, which would possibly apply to future generations of molecules developed for long-term clinical administration.
One general and other clinical safety concern is that of the 'serotonin syndrome' in humans resulting from CNS and systemic serotonin overload. Serotonin syndrome occurs within hours of taking a new drug, or combination of drugs, producing systemic 'serotonin overload' (which could mechanistically include psychedelic NCEs), and a range of symptoms including agitation, cardiovascular activation, loss of coordination, headache and shivering, with life threatening signs including high fever, seizures, erratic heartbeat and unconsciousness. Lists of associated drugs include SSRIs (selective serotonin reuptake inhibitors, e.g., citalopram, fluoxetine, paroxetine), SNRIs (selective noradrenaline reuptake inhibitors, e.g., duloxetine, venlafaxine) and triptans (e.g., sumatriptan, rizatriptan). In view of the psychedelic potential to produce systemic and/or peripheral 5-HT release (from large 5-HT 'stores'), this is, therefore, a potential concern for psychedelic drug-candidates (alone or when taken in combination with known causative agents).
Other potential targets for new psychedelic molecules to elicit toxicity at high doses and exposure levels in regulatory preclinical development toxicology studies, or at low 'pharmacological doses/exposures' in early clinical development studies include the cellular systems of which the major published findings are in Table 2. These include mast cells (and the neuroimmune axis), cells of the immune system, blood cells, vascular endothelial cells, blood cell types including platelets (clotting mechanisms), neuroendocrine and neurodevelopmental effects (via 5-HT receptors in hypothalamic-pituitary and other CNS areas e.g., hippocampus) and enterochromafin cells of the mechanosecretory gastric systems (Herr et al., 2017;Purcell and Atterwill, 1995). Of particular note is the enterochromaffin cells, which store large quantities of 5-HT and are thought to be involved in the serotonin syndrome, and mast cells, which are involved clinically in adverse drug reactions via degranulation, potentially producing local skin and systemic anaphylactic shock sequelae. Toxicologically, human mast cells store mainly histamine which is a key mediator of these phenomena whereas rat cells store mainly 5-HT.
Although esketamine has recently been approved in Europe and the USA as a rapid-onset antidepressant nasal spray, there have been uncertainties regarding its potential urothelial toxicity, particularly after D.J. Heal et al. repeated application. In the context of rising recreational use, severe side-effects affecting the human urinary tract involving bladder epithelial cell sloughing-induced cystitis have been reported. Parallel studies in cultured human urothelial cells have shown that ketamine induces apoptosis by a direct NMDA receptor-independent pathway characterised by mitochondrial stress (Baker et al., 2016) in contrast to the NMDA receptor-antagonist classification of ketamine. The urothelial cell ketamine-induced apoptosis involves an intracellular signalling process with cytochrome-C release from mitochondria and caspase-9 and caspase-3/7 activation. The potential for ketamine, its enantiomers and derivatives to induce apoptosis in human bladder cells should thus be an additional important safety and toxicological preclinical and clinical development factor consideration in future psychedelic drug development of this molecular class.
Lastly, we consider the serotonergic psychedelics in clinical settings, including psilocybin and psilocin. Table 2 depicts their binding affinities to various 5-HT receptor subtypes as predictive tools for likely toxicity targets (Bauer, 2019), even though with current low micro-dosing regimens in humans these might only result in idiosyncratic adverse drug reactions in certain predisposed subjects through genetic variability in e. g., Hepatic P450 subtype variations and other ADME (absorption, distribution, metabolism, excretion) factors. As discussed above, fenfluramine is known roughly to double the risk of cardiac valve disease via 5-HT 2B activation after 90 days of treatment at 30 mg/day and has a reported Ki of ~30 nM. Both LSD and psilocin have higher affinities (Table 2) of 4-5 nM and even though a typical micro-dosing regimen involves a magnitudes lower dose than 30 mg/day, repeated dosing of these molecule classes in conventional preclinical pharmaceutical development settings points to a potential cardiotoxic 'hazard and risk' scenario for toxicological development. These risk estimates are based on affinity values and for cardiac valvulopathy and other adverse events the toxicity of the psychedelics is dependent on their level of intrinsic efficacy as receptor agonists. In Table 2, the highest level of risk has been set at "Caution" for high toxicity potential which is based on current dosing regimens for psychedelics which are either exposure to high doses in one or two sessions or the repeated administration of very low doses (micro-dosing). The magnitude of the risk will increase if dosing regimens of the psychedelics expand to repeated administration at moderate/high dose levels. It is already known that the safety/toxicity risk is increased when abusers self-administer psychedelics at high dose levels. For novel psychedelics which may have higher 5-HT receptor subtype affinities, it is important to determine not only their affinity for 5-HT receptor subtypes, but also their receptor function because risk predictions need to be based on potency and intrinsic efficacy.

Summary of safety and toxicity concerns
In addition to the assessment of drug abuse liability there are clearly several well-defined potential targets for new psychedelics and entactogens, which in the case of MDMA or substituted amphetamine Table 2 Distribution of 5-HT receptor subtypes on diverse cell types of the immune, blood and vascular systems together with 5-HT receptor subtype binding affinities of several psychedelic molecules of recent clinical interest. Ki ≤ 10 nM = High Affinity (High Toxicity Potential), Ki: 10-50 nM = Intermediate Affinity (Moderate Toxicity Potential), Ki > 50 nM = Low Affinity (Low Toxicity Potential). These estimates of risk are based on affinity values. It should also be noted that the toxicity risk posed by the psychedelics is influenced by their level of intrinsic efficacy as receptor agonists and should therefore be viewed with caution. DMT = Dimethyltryptamine; LSD = Lysergic acid diethylamide. a Data taken from Herr et al. (2017). b Data taken from Bauer (2019). c Data taken from Blair et al. (2000). d Non-human data. e Potential Cautionary flag (in pharmaceutical preclinical & clinical development). Current proposals for dosing regimens of psychedelics entail exposure to high doses in one or two sessions or the repeated administration of very low doses (micro-dosing). The magnitude of the risk will increase if psychedelic dosing regimens include repeated administration at moderate/high dose levels. It is already known that the safety/toxicity risk is increased when abusers self-administer psychedelics at high dose levels.
derivatives and fenfluramine analogues includes fairly hard evidence for potential CNS neurotoxicity to serotonergic and linked neuronal circuits in the form of either irreversible neuropathies or partly reversible functional deficits. The new NeuroDeRisk neurotoxicity in silico predictive database (NeuroDeRisk Database,) developed as an international consortium (Innovative Medicines Initiative 2) will provide additional weight to PLOT strategies for lead molecule optimisation ahead of regulatory toxicology strategies in drug development. Either way these pose challenges to drug development and clinical safety. Likewise, there is well established pathophysiological and molecular evidence for cardiotoxic risk in the form of valvulopathy and VIC mitogenesis for other newer psychedelics with 5-HT 2B agonist or partial agonist potential, including LSD and psilocin derivatives. For all psychedelics with SERT affinity and 5-HT synaptic or peripheral cellular-releasing ability, there is the additional risk of triggering a 'serotonin syndrome' in extremis and/or exacerbating this risk in co-medicated subjects on other types of antidepressant therapy. The range of 5-HT receptor subtypes on cells of e.g., the immune system, cardiovascular cells or mast cells also mitigates for careful consideration of these toxicity targets in drug design and discovery, including other factors such as the role of 5-HT/5-HT 2B in CNS cellular development and effects on mast cell degranulation producing potential clinical anaphylactic shock. In all scenarios the toxicokinetic determination of human safety margins and toxicity cellular and molecular mechanisms from NOAEL (no adverse effect level)-based data from conventional regulatory toxicology studies is, of course, imperative and a necessary regulatory mandate when and if new psychedelic molecules come into preclinical/clinical pharmaceutical development in the future. Fig. 5 depicts a summary of a possible multifaceted and pragmatic strategy in drug discovery and early-stage toxicokinetic and safety pharmacology studies to facilitate lead drug candidate attrition in the pharmaceutical development pipeline and the registration of safe novel therapies for the treatment of depression. These new approaches in regulatory clinical chemistry and pathology, immunotoxicology, haematology, neuropathology and cardiotoxicity, will need to operate in parallel to conventional development studies to assess the potential clinical safety versus efficacy for new psychedelic molecular entities.

Summary/conclusions
The serotonergic psychedelics and entactogens have led a renaissance in the development of new and more effective drugs to treat psychiatric disorders, and in addition, are being evaluated in various peripheral conditions, e.g., pain and inflammation, that would not have received consideration a decade ago. 5-HT serves as both a neurotransmitter and a hormone and is ubiquitously distributed in the brain and in many peripheral organs, tissues and cells of the human body. The serotonergic psychedelics have been harnessed as medicines by humans for thousands of years and, following the synthesis of LSD by Albert Hofmann in 1938 and his experiences after taking the drug in 1943, their powerful psychotropic properties became common knowledge. The counter-culture experimentation with psychedelics in the 1960s and their illicit use by a later generation as "designer drugs" in the "rave scene" has provided a wealth of knowledge about their psychopharmacology, psychopathology and other safety risks. Many of these potential harms have been side-stepped by administering psychedelics to patients in a single or a small number of sessions in highly structured and controlled clinical settings. However, there will almost certainly be clinical conditions which require daily or weekly treatment with psychedelics, and furthermore, we must not be blind to the potential public harms that may arise from their diversion and abuse. The wealth of accumulated knowledge on these classical serotonergic molecules provides a unique opportunity to design and propose a new screening strategy for drug discovery that will reset the balance between benefits and harms to deliver more effective and safer novel psychedelics and entactogens for clinical use.

Disclosure of funding support
The authors and DevelRx received no financial or material support from any agency in the public, commercial, or not-for-profit sectors when preparing this review. Funding open-access publication of this article for the Neuropharmacology Special Issue was through a gift from the Steven & Alexandra Cohen Foundation to the Foundation for the National Institutes of Health. Fig. 5. Strategic non-clinical safety and toxicity objectives for "de-risking" novel psychedelic and entactogen development-candidates. ADME = absorption, distribution, metabolism, and excretion; CNS = central nervous system PLOT = Preclinical Lead Optimisation Technology; PNS = peripheral nervous system; R/F = range-finding; TK = Toxicokinetics; VIC=Valvular interstitial cell (cardiac tissue).

Declaration of competing interest
David Heal, Sharon Smith and Jane Gosden are shareholders and employees of DevelRx Ltd. Christopher Atterwill is a consultant who is affiliated to DevelRx Ltd. DevelRx is a consultancy company which advises the pharmaceutical industry on the discovery and development of CNS drugs. The opinions expressed in this review are exclusively those of the authors and have not been influenced by any public, commercial, or not-for-profit organisation.

Data availability
No data was used for the research described in the article.