The amino acid, glycine, is an important signalling molecule in the central nervous system (CNS). As an inhibitory neurotransmitter, glycine binds to the strychnine-sensitive glycine-A site on ionotropic glycine receptors and activates an inward chloride current through the integral anion channel of the glycine receptor complex which leads to the hyperpolarization of the postsynaptic membrane. This fast neuronal inhibition is the predominant function of glycine receptor found predominantly at inhibitory synapses in the spinal cord, brainstem, and retina in the adult brain. The existence of non-strychnine-sensitive high affinity binding sites was revealed by [³H] glycine autoradiography because such sites were not labeled by [³H] strychnine [20]. The distribution pattern of the non-strychnine-sensitive, glycine-B sites, corresponds well to the binding profile of NMDA receptors [21]. The physiological interaction between glycine and NMDA receptors was demonstrated by Johnson and Ascher [22]. They showed that glycine could augment NMDA receptor-mediated electrophysiological responses. Kleckner and Dingledine [23] subsequently demonstrated that glycine is a prerequisite for the activation of the NMDA receptor and coined the term “co-agonist” to describe its essential role at the NMDA receptor. It is known that glycine is not the only endogenous ligand at the glycine-B site. The amino acid d-serine can also bind to the glycine-B site and serves the function of an obligatory co-agonist [24–26]. Hence, the concomitant binding of glycine or d-serine at the glycine-B site is a prerequisite for the activation of postsynaptic NMDA receptors by glutamate released from presynaptic terminals. The distinction and cooperation between the regulatory function of d-serine and glycine at NMDA receptors, however, remain to be fully delineated. One suggestion is that d-serine might be preferentially involved in the regulation of synaptic NMDA receptors, whereas glycine appears to be more critically involved in the regulation of extra-synaptic NMDA receptor excitability—a demarcation that can be linked to the differential distributions of synaptic versus extra-synaptic NMDA receptors, which can be distinguished by the presence of NMDA receptor (NR) subunit NR2A and NR2B, respectively.

Occupancy of the glycine-B site is not only required for initiating signaling through the NMDA receptor, but it also primes the NMDA receptor for internalization [27]. Saturation of the glycine-B sites is expected to trigger the internalization of NMDA receptors, and thereby would be expected to weaken the signals mediated by NMDA receptors. This could take place when the extra-cellular glycine concentration is sufficiently high. It has been shown that the magnitude of NMDA receptor-dependent long-term potentiation can be enhanced by increasing concentration of ambient glycine up to 10 μM, but signs of impairment can appear at ≥100 μM concentrations [28]. Therefore, disturbance of the homeostatic regulation of extracellular glycine concentration in the brain could have a profound effect not only on the excitability of individual NMDA receptors but also on their overall expression levels.

The ambient concentration of glycine in the extracellular space near NMDA receptors may be exploited as a new avenue to facilitate NMDA receptor neurotransmission [29]. One obvious means to increase extracellular glycine availability is the direct augmentation through dietary intake of glycine. However, this proves to be ineffective because of the poor penetration of glycine through the blood–brain-barrier (see section “Proof of principle I”). Alternatives include glycine-B site partial agonist, d-cycloserine, which is able to penetrate into the CNS. A potential breakthrough hinges on the homeostasis (re-uptake, release, and metabolism) of glycine in the brain.

It was initially believed that the levels of glycine in the extracellular space were sufficiently high so that the glycine-B sites of the NMDA receptors would be saturated under physiological conditions [30, 31] and therefore not relevant in the normal regulation of NMDA receptor function. It soon became apparent that glycine transporters provide active removal of extracellular glycine such that the glycine concentration in the synaptic cleft is maintained at sub-saturating levels [32–34].

The two principal subtypes of glycine transporters are GlyT1 [35, 36] and GlyT2 [37]. They belong to the sodium-dependent intracelluar solute carrier family 6 of transporters but differ in terms of 1) regional and cellular expression patterns in the CNS, 2) Na⁺:glycine stoichiometries, and 3) the ability to reverse-transport intracelluar glycine into the extracellular space [38–42]. Five variants of GlyT1 (a–e) and three variants of GlyT2 (a–c) as a result of alternative splicing and promotor usage have been identified [40, 43–45].

As illustrated in Fig. 25.2, GlyT2 expressed in glycinergic terminals assumes the critical role in the reuptake and recycling of glycine released by the pre-synaptic terminal into the synaptic cleft, and together with astrocytic GlyT1 contribute to the termination of the stimulation of glycinergic receptors in the post-synaptic membrane [40]. The coordinated activities of GlyT1 and GlyT2 are essential for the vital functions that depend on inhibitory neurotransmission such as respiratory regulation. The loss of pre-synaptic GlyT2 drastically curtails the refilling of glycine vesicles by the presynaptic cell and severely undermines inhibitory glycinergic neurotransmission [42, 46, 47]. On the other hand, the loss of astrocytic GlyT1 severely undermines the clearance of glycine released from the synaptic cleft leading to sustained neuronal inhibition over brain stem respiratory centers [48]. In fact, constitutive homozygous deletion of either the GLYT1 or GLYT2 gene is lethal in mice [46, 48].

In contrast, GlyT1 expressed in pre- and post-synaptic sites of glutamatergic synapses plays the pivotal role in preventing the saturation of the glycine-B site on NMDA receptors, with GlyT1 expressed in adjoining astrocytes as well as extrasynaptic sites playing a supplementary role [33, 34, 49–52]. GlyT1 is therefore the obvious target of choice to modify NMDA receptor function. Inhibition of GlyT1 should elevate baseline glycine-B site occupancy and thereby increase the probability of NMDA receptor activation by presynaptic release of glutamate (Fig. 25.1). The result is an ‘on-demand’ facilitation of NMDA receptor activation which amplifies the glutamatergic signals. This is unlike the elevation of the tonic activity of NMDA receptors produced by direct NMDA receptor agonists. GlyT1 inhibition is therefore expected to confer therapeutic potential for diseases such as schizophrenia, in which deficient signaling via forebrain NMDA receptors is implicated. We have shown that the deletion of neuronal GlyT1 in the forebrain alone is sufficient to enhance selectively the NMDA receptor current; and the knockout mice have exhibited antipsychotic-like as well as pro-cognitive phenotypes [53, 54].

Schizophrenia could be the first neuropsychiatric disorder to benefit from this novel pharmacological approach. The recent phase II clinical trials of the GlyT1 inhibiting drug, bitopertin, [4-(3-Fluoro-5-trifluoromethylpyridin-2-yl)piperazin-1-yl]-[5-methanesulfonyl-2-((S)-2,2,2-trifluoro-1-methylethoxy)phenyl]methanone] (a.k.a RO4917838 and RG1678) developed by Hoffmann-La Roche and its subsidiary Chugai Pharmaceutical have yielded promising outcomes for the treatment of schizophrenia negative and cognitive symptoms [55–58]. The compound had the potential to become the first successful translation of the glutamate hypofunction hypothesis of schizophrenia into a clinic-ready pharmacotherapy since the hypothesis’s inception in the 1980s [59] before the hope was dashed finally with the release of the disappointing results. First, two trials were reported to have failed to meet the primary endpoint in improving negative symptoms (www.​roche.​com/​med-cor-2014-01-21-e.​pdf). Subsequently, more information were released in abstract form in three international conferences [60–62]. Out of the six phase III trials, only one treatment arm (10 mg/kg adjuvant biotopertin) in the study, which had targeted positive symptoms, had yielded significant improvement the primary endpoint relative to placebo adjuvant treatment. The evidence was too weak to justify further commercial development of bitopertin, and Roche has since shelved its development as an anti-schizophrenia agent, although it is still being pursued as an adjuvant treatment for obsessive compulsive disorder (NCT01674361).

The development of potent and highly selective GlyT1 inhibitors has been initiated with the synthesis of the sarcosine derivatives, Org 24598 and N[3-(4’-fluorophenyl)-3-(4’-phenylphenoxy)propyl]sarcosine (NFPS; also known as ALX 5407 since 2001) [63, 64]. A variety of compounds have since been synthesized, including compounds that are not based on a sarcosine backbone such as SSR504734, SSR103800, and bitopertin [53, 65]. As proof of mechanism, GlyT1 inhibitors have been demonstrated to elevate brain glycine levels, potentiate NMDA receptor-mediated transmission and synaptic plasticity, and antagonize the behavioral effects of NMDA receptor blockade [53].

The first preclinical evidence of antipsychotic-like potential of glycine re-uptake inhibition can be traced back to the study by Toth et al. [66], which showed that the glycine derivative, glycyldodecylamide, was much more effective than glycine itself to reverse the hyperlocomotor response induced by the NMDA receptor antagonist, phencyclidine. However, it took over a decade until the connection with glycine transporter was identified by Javitt and colleagues. They demonstrated that behaviorally effective doses of glycyldodecylamide were highly effective in inhibiting forebrain glycine re-uptake [67, 68]. That early finding marked the beginning of the preclinical evaluation of many GlyT1 inhibitors in animal models of schizophrenia ranging from the readily available compound sarcosine to highly potent synthetic compounds developed by diverse pharmaceutical companies. As summarized in Table 25.1, the results obtained in preclinical animal models have been largely encouraging.

Reversal of the hyperlocomotor activity induced by NMDA receptor blockers such as phencyclidine and dizocilpine (MK-801) is most commonly used as antipsychotic potential test (Table 25.1). In an effort to more directly measure the in vivo efficacy of a given compound to enhance the occupancy of the NMDA receptor glycine-B sites, Alberati et al. [69] instead measured a given compound’s ability to reverse the motor disturbance (head weaving, body rolling, hyperlocomotion) induced by l-687,414. As a partial agonist, l-687,414 is less efficacious than the endogenous agonists, glycine or d-serine, at the co-agonist glycine-B site; and therefore l-687,414 can induce behavioral effects resembling direct NMDA receptor antagonists. Hence, a drug that can effectively raise the levels of the endogenous co-agonist, glycine, could reverse the behavioral effects of l-687,414 because the increase in glycine directly competes with l-687,414 for binding to the glycine-B site. The test is uniquely suited for the detection of potential antipsychotic drugs that act to increase endogenous glycine-B site occupancy. At doses that were devoid of notable behavioral effects, glycine and five potent GlyT1 inhibitors (ALX 5407, ORG24598, RO4543338, RO4840700, and SSR504734) dose-dependently reversed l-687,414-induced hyperlocomotion. Additionally, the assay is sensitive to conventional antipsychotics, but not to other CNS drugs such as the analgesic morphine, the antidepressant fluoxetine, or the anxiolytic drug diazepam [69].

It is worth noting that the ability of GlyT1 inhibitors to antagonize the motor stimulant effect of the indirect dopamine agonist, amphetamine—a model of the positive symptoms associated with hyperdopaminergia [70, 71], is less consistent than their ability to reverse the hyperlocomotor effects effect of NMDA receptor blockade (see Table 25.2). One competitive GlyT1 inhibitor, SSR504734, has even been reported to exacerbate the hyperlocomotor response of amphetamine [72] and the sensorimotor gating deficits in the prepulse inhibition paradigm induced by apomorphine (a direct dopamine receptor) [73]. These drug–drug interactions are consistent with a report that SSR50473 potentiated the release of dopamine in the nucleus accumbens triggered by direct electrical stimulation of the amygdala [74] and increased basal dopamine activity in the prefrontal cortices [75].

It is not certain, however, if similar interactions with the dopamine neurotransmission are shared by other GlyT1 inhibitors. If so, this may predict weak efficacy against positive symptoms and provides a rationale for GlyT1 inhibition as an add-on to be combined with conventional antipsychotic drugs that are known to block dopamine D₂ receptors [76–78]. On the other hand, the positive effects of SSR504734 on mesocortico-limbic dopamine may contribute to its efficacy to enhance working memory and problem solving in normal animals [79, 80] because stimulation of prefrontal dopamine activity could enhance cognitive performance in healthy humans and ameliorate cognitive deficits in schizophrenia patients [81–85]. However, this is unlikely the sole mechanism for the pro-cognitive potential of GlyT1 inhibitors. Several other GlyT1 inhibitors have also been shown to ameliorate the learning and memory deficits induced by NMDA receptor blockade across multiple tests (see Table 25.1).

The prepulse inhibition (PPI) and latent inhibition (LI) tests are two highly translational paradigms with face, construct, and predictive validity [86]. PPI measures a form of sensory gating that regulates stimulus access to higher cognitive resources. It is considered as a pre-attentional filtering mechanism whereby ongoing information processing is protected from intrusion by spurious environmental stimuli [87]. LI, on the other hand, refers to a specific form of selective attention whereby one learns to pay less attention to stimuli that are evidently irrelevant (i.e., lacking biological significance) in the past. This is an important component of associative learning, which suppresses the formation of spurious contingency between events in one’s environment, and thereby favors the learning of reliable contingency between environmental events to guide future behavior [88]. Schizophrenic patients exhibit deficits in PPI as well as in LI [89, 90]; and antipsychotic drugs can strengthen both phenomena [91–100]. With few exceptions, GlyT1 inhibitors exhibited antipsychotic potential in the PPI test (Table 25.1). All three compounds (ALX 5407, SSR504734, and SSR103800) that have been evaluated in the LI paradigm are effective in enhancing LI in normal animals [101, 102], as well as antagonizing the LI impairment induced by the NMDA receptor antagonist MK801—a specific form of LI deficit suggested to mimic attentional dysfunction linked to negative symptoms [103].

Additional tests that might point to an efficacy against affective symptoms in schizophrenia include enhanced social recognition memory by ALX 5407 [104], and antidepressant-like property in the Porsolt forced swim test by SSR103800 [105]. Indications for potential anti-anxiety effects of GlyT1 inhibitors include the suppression of contextual fear learning by SSR504734 [106] and the facilitation of fear extinction by ALX 5407 [101].

One of the first clinical trials evaluating the antipsychotic potential of glycine augmentation therapy was an open-label pilot study of orally administered glycine (10.8 g/day) as an add-on medication to conventional antipsychotics drugs [107]. The study yielded inconclusive results, and another study done by the same group using milacimide, a prodrug for glycine, failed to reveal any benefits [108].

These early negative findings were followed by a series of placebo-controlled clinical trials of glycine add-on therapy with substantially higher doses (0.4–0.8 g/kg/day) designed to overcome poor brain penetration of orally administered glycine. At the highest dose, which was associated with a six-fold increase in the plasma levels of glycine, glycine add-on therapy significantly ameliorated the negative symptoms and improved global functions in the patients. However, the required high doses are considered impractical for long-term clinical use [109] because of potential gastrointestinal disturbances [110, 111].

Subsequent add-on studies with the glycine-B site agonist, d-serine, revealed similar beneficial effects against the negative and cognitive symptoms at significantly lower doses (2–8 g/day) [112, 113]. Nephrotoxicity was a concern at the highest dose (8 g/day), but could be avoided at lower doses (≤4 g/day) that were still clinical effective [112]. As an alternative to overcome the poor brain penetrance of glycine, and to a lesser degree d-serine, the partial glycine-B site agonist, d-cycloserine, has also been evaluated as an add-on medication in clinical trials [114–118]. When added to conventional antipsychotics, d-cycloserine significantly reduced negative symptoms over a relatively small dose range with optimal therapeutic efficacy at 50 mg/day [114, 117, 119]. However, the clinical potential of d-cycloserine has been limited by the decrease in therapeutic efficacy over time, the modest effect size, and narrow effective dose range [116].

These studies have provided important evidence that adjunctive glycine augmentation therapy directed at increasing glycine-B site occupancy could improve negative and cognitive symptoms of schizophrenia. At the same time they have identified obvious limitations that justify the alternative approach to inhibit GlyT1. Targeting GlyT1 is an attractive strategy considering that the potential of designing new compounds that can selectively mimic the action of small molecules such as glycine and d-serine on the glycine-B site is limited because their simple backbone structures leave little room for possible structural modifications [113].

The first proof-of-concept studies were carried out with the naturally occurring GlyT1 inhibitor, sarcosine (N-methyl glycine). Like glycine, sarcosine is a natural amino acid that is generated as an intermediate in the synthesis and degradation of glycine. Sarcosine is well tolerated and has no known toxicity as indicated by the lack of adverse effects in sarcosinemia, a rare congenital condition caused by dysfunction of sarcosine metabolism that leads to the accumulation of sarcosine in plasma and urine [120]. A number of randomized, double blind, placebo-controlled trials have shown that sarcosine add-on treatment improves positive and negative symptoms as well as general functions in patients stabilized on conventional antipsychotic medication [121–124]. The beneficial effects are not limited to patients with stable positive symptoms but are also seen in patients in the acute phase of the disease [122]. It should be emphasized that all these studies administered sarcosine as an add on, and so far only one small-scale non-placebo controlled trial had evaluated sarcosine as a monotherapy [124] and the tentative trend of negative symptoms reduction reported needs to be substantiated by standard controlled trials with larger sample size.

However, adjunctive sarcosine appeared ineffective when combined with clozapine [125], which is similar to what had been learned from combining glycine or d-serine with clozapine [126]. Adjunctive glycine treatment might worsen the positive symptoms in patients maintained on baseline clozapine [126], while a significant exacerbation of the negative symptoms has been observed when d-cycloserine is combined with clozapine [118]. The reason for the unique drug–drug interaction is not fully understood. It is suspected that clozapine may already increase synaptic glycine levels through as yet unknown mechanism [127]. Hence, baseline clozapine medication should be avoided; this has been recognized in all subsequent trials with synthetic GlyT1 inhibitors.

The encouraging outcomes with sarcosine have spearheaded the development of a variety of synthetic GlyT1 inhibitors. Many of them have been evaluated in clinical trials as a potential new class of antipsychotic drugs (Table 25.2). The first generation of synthetic compounds were substituted sarcosine derivatives characterized by irreversible and non-competitive inhibition of GlyT1 such as ALX 5407 and Org24589 [128]. They were associated with motor and respiratory side effects resulting from excessive glycinergic inhibition in the brain stem and cerebellum [128–130]. This has led to the development of a second generation of non-sarcosine-based compounds exhibiting reversible and competitive inhibition of glycine transport such as SSR504734 [128] which have been associated with fewer side effects. However, this view has been challenged by Kopec and colleagues [131] who contended that the induction of motor side effects may instead be better predicted by the target residence time, which refers to the dissociative half-life of the compound–target complex.