SR is a rational target for the treatment of diseases described in section 7. However, in spite of its relevance, investigations carried out towards the development of potent inhibitors or effectors are surprisingly rather limited. This fact might be dependent on at least two factors: i) the yield and stability of the enzyme are low, making inhibition studies difficult, and ii) inhibitors should be very specific and cross the blood-brain barrier. In order to generate specific inhibitors, the structural and functional features of SR active site have been exploited as well as the well known reactivity of the coenzyme. SR inhibitors have been developed to reduce the excitoxic effects of high levels of D-serine. Specifically, these compounds are aimed at reducing SR activity, thus decreasing D-serine concentration, and/or at increasing the b-elimination operating on D-serine, thus speeding up the depletion of the neurotransmitter.

SR effectors that are under development can be grouped based on their mechanism of action:

1. PLP targeting with reversible modifications;

2. PLP targeting with irreversible modifications;

3. Active site targeting;

4. Allosteric effectors;

5. Modulators of SR-protein interactions.

To our knowledge, inhibitors so far identified belong exclusively to the first three groups. In Table 2 the structure and binding affinities of the most potent inhibitors are reported.

Table Structure and activity of the most potent inhibitors of SR identified to date

Inhibitors targeting PLP		Structure	K_i (mM)	Species	Reference
Reversible	Glycine		0.36 1.64	hSR mSR	(51) (22)
Reversible	L-erythro-3-hydroxyaspartate		0.01 0.04	hSR mSR	(51) (22)
Irreversible	Succinodihydroxamic acid		0.003	mSR	(109)
Irreversible	L-aspartic acid b-hydroxamate		0.10	mSR	(109)
Reversible Competitive	Malonate		0.03-0.06 0.07 0.06	hSR mSR rSR	(17, 51) (22) (17)

PLP targeting with reversible modifications

L- and D-amino acids were assayed as substrate analogs and inhibitors (20, 50, 56). Among them the stronger inhibitor was the NMDAR ligand glycine (56) that exhibits a K_i of 366 and 1640 mM for hSR and mSR, respectively (51). Asparagine, aspartate and oxolacetate also bind to the enzyme indicating that neither an amine group nor a negative charge in the lateral chain are essential for inhibition, and that compounds with bulkier groups can enter in SR active site (56). Sulfhydryl containing amino acids, capable of reacting with PLP to form thiazolidine derivatives, are also SR inhibitors (50, 56). Furthermore, L-serine derivatives, such as L-serine O-sulfate, were found to be effective uncompetitive SR inhibitors in the absence of ATP (50). L-serine O-sulfate undergoes a b-elimination reaction to sulfate, ammonia and pyruvate. This finding indicates that an a-aminoacrylate species is first formed, and successively hydrolyzed. However, there is not yet an explanation for the observed uncompetitive inhibition.

A comprehensive analysis of compounds derived from serine and L-Ser-O-sulfate led to the identification of the so far strongest mSR competitive inhibitor, (2S,3R) L-erythro 3-hydroxyaspartate, exhibiting a K_iof 43 mM (22) (Table 2). Interestingly, the K_i for hSR is almost four fold lower than for mSR, 11 mM (51). This compound is a four carbon amino acid analog that reacts with PLP, forming a stable external aldimine. Accordingly to the docked conformation proposed by Jiraskova-Vanickova et al. (12), one carboxylic group interacts with the backbone nitrogen of Ser84 and His87, the other with Arg135 and Asn174, while the hydroxyl moiety contacts the side chain of Ser84. SR is a close relative of bacterial L-threo 3-hydroxyaspartate dehydratases and L-threo-3-hydroxyaspartate (2S,3S) is a substrate of the murine (22) and human enzymes (12) that degrade it to oxaloacetate and ammonia. In contrast, D-threo-3-hydroxyaspartate (2R,3R) is only a weak inhibitor (22).

Further studies identified b-haloalanines as SR inhibitors (56), similarly to that observed for other PLP-dependent enzymes belonging to fold type II, as tryptophan synthase (14, 107) and O-acetylserine sulfhydrylase (15, 108). This class of compounds reacts with PLP forming an external aldimine that, given the favourable b-leaving group, undergoes a b-elimination reaction with formation of an a-aminoacrylate derivative.

PLP targeting with irreversible modifications

The first reported SR inhibitor of this class was aminooxoacetic acid, an aspecific PLP-dependent enzyme inhibitor that reacts with the Schiff base linkage between PLP and the enzyme, forming an aldoxime (50). A series of small aliphatic hydroxamic and dihydroxamic acids were assayed as potential mSR inhibitors (109). It was found that the most effective were malonodihydroxamic acid, succinodihydroxamic acid, glutarodihydroxamic acid and L-aspartic acid β-hydroxamate. The K_i value calculated for succinodihydroxamic acid was about 3 mM, the most potent SR inhibitor so far identified (Table 2), whereas L-aspartic acid β-hydroxamate exhibited a K_iof about 100 μM. A clear disadvantage of hydroxamic acid derivatives as SR inhibitors and potential drugs is their lack of specificity, as they react with several PLP-dependent enzymes (109).

Cell migration assays were used for the identification of SR inhibitors. It was found that phenazine, phenazine methosulfate, and phenazine ethosulfate are effective inhibitors of SR with IC₅₀ of 3 and 5 μM for the methosulfate and ethosulfate derivatives, respectively (66). However, no information on the assay conditions were reported. Furthermore, much higher IC₅₀ values for these compounds were reported by Kovalinka and coworkers (12).

Active site targeting

This approach is aimed at identifying compounds that do not react with PLP and exploit for binding their tight functional and geometric complementarity with the active site. A recent example is the interaction between OASS and pentapeptides mimicking the C-terminal of serine acetyltransferase (110, 111).

Several dicarboxylic acids were tested as SR active site inhibitors. It was found that malonate is the best binder with a K_diss of 33 mM for hSR (22) (Table 2). A slightly different value, 59 mM, was reported by Smith et al. (17), for hSR, and 111 mM for a mutant hSR in which Cys2 and Cys6 were substituted with Asp residues. A larger discrepancy for the affinity of malonate to mSR was found with a K_i of 71 mM (12, 22, 51), and a Ki of 568 mM to rSR and 1599 mM to the mutant (17). Malonate is a dicarboxylic acid that is orthosteric with L-serine. Its ability to interact with SR is due to the small dimension and the carboxylate groups that make strong ionic bonds with active site residues. As reported in Figure 2, one carboxylate of malonate interacts with the nitrogen backbone of Ser84, Asn86 and His87, with the side chain of Ser83 and with one water molecule, while the other forms a salt bridge type ionic interactions with Arg135, and also contacts Ser84 and Ser242, in addition to a pair of water molecules. As suggested, Arg135 might play a similar role to the arginine finger identified in the active site of G-proteins, involved in the stabilization of the reaction transition state and in the enhancement of the reaction rate (112, 113). On the other side, the N-terminal loop, containing Ser83, Ser84, Gly85, Asn86 and His87, is recognized as the asparagine loop, located in the PLP O3’ side, and acting as the recognition site for the ligand in the closed enzyme conformation (25). Interestingly, Arg135, which appears fundamental for the stabilization of the enzyme-inhibitor complex is not conserved in other PLP-dependent enzymes belonging to the fold-type II (18). A series of malonate derivatives were synthesized and tested (12). No significant improvement with respect to the parent molecule was observed. Among a series of dicarboxylic and tricarboxylicacids meso-diaminosuccinic as well as meso-tartaric exhibits a degree of rSR inhibition comparable with malonate.

A different approach towards the identification of SR inhibitors was followed by Dixon et al (114). A 74088 tripeptide library bound to individual beads was screened with a fluorescent-labelled human SR. Sixty peptide-beads were found to be positive and twenty five of them were randomly selected, identified, synthesized and assayed for SR inhibition. It was found that most of the identified peptides contained either histidine or phenylpropionic moieties. The best binders exhibit K_i values between 300 and 600 mM with the inhibition mechanism involving a fast interaction with the enzyme followed by a slow binding.

A still different method is pursued by Spyrakis et al. (quoted in (1)). In silico screening using both the open and the closed conformational states of SR, docking, scoring and best hits experimental evaluation was aimed at identifying active site SR binders. Studies are still ongoing (Spyrakis et al., unpublished results).

Allosteric effectors and modulators of SR-protein interactions

No investigation has been so far carried out aimed at developing compounds that target the multiple sites that allosterically modulate SR activity, i.e. the ATP binding site with the reactive Cys113, the cation binding site and the sites of the interaction between SR and effector proteins, GRIP, PICK1, and GOLGA3. This is a fascinating and challenging approach that might allow to generate effectors that either activate or inhibit SR activity.