Tyrosine 121 moves revealing a ligandable pocket that couples catalysis to ATP-binding in serine racemase

Human serine racemase (hSR) catalyses racemisation of L-serine to D-serine, the latter of which is a co-agonist of the NMDA subtype of glutamate receptors that are important in synaptic plasticity, learning and memory. In a ‘closed’ hSR structure containing the allosteric activator ATP, the inhibitor malonate is enclosed between the large and small domains while ATP is distal to the active site, residing at the dimer interface with the Tyr121 hydroxyl group contacting the α-phosphate of ATP. In contrast, in ‘open’ hSR structures, Tyr121 sits in the core of the small domain with its hydroxyl contacting the key catalytic residue Ser84. The ability to regulate SR activity by flipping Tyr121 from the core of the small domain to the dimer interface appears to have evolved in animals with a CNS. Multiple X-ray crystallographic enzyme-fragment structures show Tyr121 flipped out of its pocket in the core of the small domain. Data suggest that this ligandable pocket could be targeted by molecules that inhibit enzyme activity.


5-x0501
(1.82Å) Tyrosine 121 in pocket in subunits A and B, electron density in subunits C and D suggests half occupancy compound with carboxamide in pocket mimicking hydroxyl of Tyr 121?

DISC1
Stabilises SR and prevents its degradation

Supplementary Figure 11. ATP dependent activity data on Wild-type, Y121A, Y121F and Y121G mutants
The activity of the the wild-type and the three Y121 mutants: Y121F, Y121A and Y121G, was measured at six ATP concentrations (1mM, 0.5mM, 0.25mM, 0.125mM, 0.062mM and 0.031mM) -see Methods for details. Error bars show standard error. The compounds were all assayed in 16.7% DMSO. The maximum compound concentration used in the assay was 16 5.55mM, 1.85mM, 0.62mM, 0.21mM, 0.069mM, 0.023mM and  0.0077mM). Error bars show standard error. The mechanism suggests that Arg 135 could protonate the hydroxyl of the substrate serine; the substrate serine protonates Ser 84, which in turn protonates the Ca of the substrate. Arg 135 is conserved in serine racemase sequences but is not present in serine dehydratases (see Supplementary Fig. 15). Figure 15. Serine dehydratase and serine racemase sequence alignment. SR amino acid sequences are as described in Supplementary Fig. 7. Structures for all sequences are available in the PDB (see Supplementary Table 2). Serine dehydratase (SDH) sequences for H. sapiens (SDHL_HUMAN) and R. norvegicus (SDHL_RAT) (isoform 2) were obtained from UniProt 34 . Conserved residues are indicated by * underneath alignment. Residues highlighted in red on the hSR sequence (T52, K56, H82, S84, Q89, Y121, R135, E136). Note the conservation of S83, G85, and N86. In the CNS, SR can localise to specific regions in cells and its activity is tightly spatially and temporally regulated by a number of protein interactions [24][25][26][27]31,[36][37][38][39][40] and posttranslational modifications 28,35,36,41,42 . The emerging model suggests that astrocytes normally synthesize L-serine, which is then transported to neurons where D-serine is synthesised and utilized at synapses 24,40,43,44 . However, in the diseased or damaged brain, astrocytes can synthesize D-serine and the extra-synaptic activity of D-serine is neurotoxic 44 . SR interacts with postsynaptic proteins such as PICK1 26,45 , GRIP 25,46 , and PSD-95 31 to increase production of D-serine through unspecified structural mechanisms (see Supplementary  Table 6 for a list of physiological modulators of SR). SR ubiquitination and degradation is tempered by its interaction with Golga3 27 and DISC1 32,40,47,48 , the latter of which is linked to diminished D-serine and a schizophrenic phenotype when mutated. Understanding how SR activity is regulated in cells may facilitate optimisation of therapeutic inhibitors and activators of SR. While it is clear that serine racemase and NMDA receptors play important roles in animals with a central nervous system, from C. elegans to man 49 , more research remains to be done to understand exactly how SR activity is regulated in vivo.

Supplementary
How serine racemase activity evolved is also an intriguing question, one possibility is that SR evolved from a progenitor protein that had serine dehydratase activity but lacked racemisation activity. Human serine dehydratase (SDH), the closest human homologue of hSR with which it shares a common ancestor, is about 75-fold more active at elimination than human SR 9 . However, human SDH has no detectable elimination activity against Dserine. Racemisation is believed to have emerged as a side-reaction of the elimination function 50 . Comparison of structures of rat SDH with hSR ( Fig. 8) suggested that Ser 84 is the crucial catalytic residue for racemisation 3 , and this was confirmed by site-directed mutagenesis experiments in which the activity of a S84A mutant was determined 9 . The hSR S84A mutant had no detectable racemase activity against L-serine or D-serine. The S84A mutant also had no detectable elimination activity against D-serine while its activity against L-serine was comparable to wild-type 9 . However, the A65S mutant of SDH ( Supplementary  Fig. 15 for sequence alignment) had no detectable racemase activity 9 but it did acquire belimination activity against D-serine. Wild-type SDH only efficiently catalyses one reaction (the breakdown of L-serine into pyruvate and ammonia) of the four reactions on L-or Dserine catalysed by SR (L to D-Ser, D to L-Ser, L-Ser to pyruvate and ammonia, D-Ser to pyruvate and ammonia). The mutation A65S in SDH gives the enzyme some ability to catalyse the breakdown of D-serine to pyruvate and ammonia 9 , supporting the generally accepted roles of Ser 84 and Lys 56 in hSR as the Si and Re-face acid/bases ( Supplementary  Fig. 13). However, if SR did indeed evolve from a progenitor of SDH, the acquisition of racemisation activity seems to require more than a simple acquisition of a serine at position 84 (=65 in rSD; Fig. 8).
In yeast S. pombe SR (spSR), L-serine racemisation activity is over 20-fold lower than its elimination activity 5 . Mouse and human SR have comparable efficiencies for both reactions, with L-serine racemisation efficiency increased by over 10-fold relative to yeast SR while L-serine elimination is mostly unchanged 11,51 . Further, the in vitro rate of human SR racemisation is about 100-fold lower than that of alanine racemase, an archetypical amino acid racemase in bacteria 52 , which supports the theory racemisation evolved later as a result of catalytic promiscuity 53 . Note, however, that in vivo activity is regulated by various interactions (see Supplementary Table 6) and could differ appreciably from in vitro activity.
The roles of Ser 84 and Lys 56 in abstracting or donating a proton from the central Ca to convert L-to D-serine (or vice versa) is quite well understood (Supplementary Fig. 13). However, two alternate mechanisms have been proposed for the origin of the hydrogen in the b-elimination reaction. Smith et al. 3 state that the elimination reaction requires removal of water with the hydroxyl group from the Cb of L-serine and hydrogen from the lysyl-NH3 as leaving groups. Another mechanism proposed for SDH states that the phosphate group of PLP acts as a general acid to donate a proton to the leaving hydroxyl group of the serine 7 . A proposed elimination mechanism (Fig. 9d) has the phosphate indirectly protonating the leaving hydroxyl, via a water. Unexpectedly in a recent neutron study of a PLP-dependent enzyme 54 a deuterium was observed equidistant between the Schiff base and the Cterminal carboxylate of the substrate. How protons move in the catalytic cycles of PLP dependent enzymes is not fully understood. The structure of yeast SR with a lysino-D-alanyl residue (Fig. 8) 5 showed that the catalytic lysine can become covalently attached to the substrate Cb side-chain 5 . Yamauchi et al. 5 reported that incubation of mouse SR with Lserine caused an enzyme to form with a higher mass than the native enzyme, which may indicate the formation of the lysine-substrate complex even in mammalian SRs. The broad range of possible protonation and tautomeric states 55 of the substrate/PLP conjugate ( Supplementary Fig. 1) suggests that further studies to experimentally determine hydrogen positions as the enzyme moves through the racemisation or elimination catalytic cycle would be of interest. Close inspection of our 1.6 Å malonate/ATP structure suggests that the acidity of Ser 84 may be enhanced by Arg 135 (Supplementary Fig. 14), consistent with the conservation of Arg 135 in serine racemase but not serine dehydratase (Supplementary Fig.  15). However, in 2017, Nelson et al. (36) proposed that Lys 114 might play an important role in enhancing the ability of Ser 84 to protonate the substrate and thus produce D-Serine. The proposed mechanism (Fig. 9c) adequately explains the protonation of a planar quinonoid, although other possible mechanisms exist (e.g. Supplementary Fig. 14). However, neither mechanism (Fig. 9c, Supplementary Fig. 14) provides a ready explanation for the role of Ser 84 in abstracting the proton from the Ca of a D-serine substrate (Supplementary Fig. 13).
The sensitivity of SR to ATP activation varies among the SR orthologues with a conserved tyrosine. In slime mould (D. discoideum) SR, Mg-ATP increases racemisation by 2.5-fold and elimination by 6-fold, indicating the presence of an ATP site 56 . Yeast SR is minimally sensitive to ATP as both reactions increase by less than 2-fold 4 . In human SR, L-serine elimination is most profoundly affected, increasing by 31-fold upon addition of ATP, followed by D-serine elimination which increases by 4-fold and finally L-serine racemisation by almost 2-fold 57,58 . Thus in vitro, the L-serine elimination reaction is several times more efficient than the L-serine racemisation reaction 51,57,58 . SDH does not have a conserved tyrosine or an ATP site and is not regulated by nucleotides 9 , suggesting ATP binding is coupled to SR's evolutionary divergence from dehydratases.
The role of the Tyr 121 flip appears to be most relevant for the mammalian enzyme, where the complexity of the nervous system necessitates stricter maintenance of D-serine homeostasis in neurons. SR is thought to alternately synthesise and degrade D-serine in brain regions lacking D-amino acid oxidase (DAO), the main degradative enzyme for Damino acids 59 that is restricted to the cerebellum and brainstem. It has been reported that enhanced astrocytic D-serine synthesis underlies synaptic damage after traumatic brain injury 60 and that neurotoxic astrocytes synthesize SR in Alzheimer's disease 61 . SR mutants deficient in elimination activity accumulate D-serine 62 , therefore the elimination function of mammalian SR is critical with respect to preventing D-serine related neuropathologies. Several small molecules and post-translational modifications reduce SR activity in an ATPdependent manner, including NADH 30 , the membrane-localised inhibitor PIP2 24 , and Snitrosylation 28 . The Tyr 121 flip may serve to regulate serine metabolism by preventing basal degradation or synthesis of D-serine when ATP has been displaced by external regulators. When bound, the position of ATP appears to restrict how the small domain moves relative to the large domain, which may explain why ATP enhances racemisation less potently than L-serine elimination.