Physiological activation of the NMDARs requires ligation of both the glutamate binding site (on NR2 subunits) and what has been termed the "glycineB
" site (on NR1 subunits). There is considerable evidence that the dominant agonist for the NR1 site in most parts of brain is not glycine but D-serine [3
]. As the only known enzyme to synthesize this novel neurotransmitter in vivo, SR has drawn increasing attention for its role involved in regulation of D-serine synthesis and modulation of glutamatergic synapse activity. In this study we describe the relationship between structural and functional characterization of SR.
Gel-filtration chromatography and cross-linking experiments demonstrated that the major oligomeric structure of recombinant human SR is a dimer. Cross-linking experiments in cell culture also suggested that SR dimer is present in intact cells. One surprising discovery is that recombinant SR had different oligomeric compositions in Tris- and phosphate-based buffers. In phosphate buffer SR is predominantly dimer with a small amount of tetramer, whereas in Tris buffer the major species is monomer with a small dimer peak. By contrast, Smith et al.
] reported that recombinant SR based on the human and rat sequences were dimeric during gel-filtration chromatography with a Tris-based buffer. This distinction may involve differences in the SR primary sequence. Smith et al.
introduced missense mutations (Cys2
→ Arg and Cys8
→ Arg) in order to improve solubility because their production yield was low. Because cysteine residues are so important for SR activity and structure-particularly quaternary structure-those mutants are not comparable to the wild-type sequence we used. In another case, SR was extracted and sequentially purified from mouse brain with an anion-exchange column in Tris-HCl buffer, a Q-sepharose column in Tris-HCl buffer, and hydroxylapatite column in phosphate buffer [13
]. The enzyme was then resolved with a gel-filtration column in Tris-HCl buffer, which indicated a species with an apparent size of 55 kDa. Because 55 kDa is between the sizes expected for monomer (37 kDa) and dimer (74 kDa), this finding is difficult to interpret. Though the reason for the buffer's effect is still unknown, this phenomenon permitted a comparison of activity across conditions that favor monomer versus dimer formation. The results support the conclusion that dimer has higher activity as phosphate buffer resulted in both more dimer and more activity. In addition, phosphate buffer is probably closer to physiological conditions than is Tris. Thus, phosphate buffers would seem more appropriate than Tris buffers for biochemical analysis of SR.
The effects of DTT on SR's activity are prominent. The influence of DTT on enzyme's structure includes two possibilities: First, DTT may keep the thiol group of critical cysteine residue(s) in a reduced state. The critical residue(s) may reside in the active site of the enzyme and be important in catalysis of enzyme reactions or reside outside the active site and be critical in modulating the enzyme reactions allosterically. Second, DTT may prevent or break randomly formed disulfide bonds between cysteine residues outside of the active center, thereby helping the protein to fold correctly during synthesis/purification in this recombinant system.
Blocking of free sulfhydryl groups by cystamine dramatically decreased SR activity, which indicates a key role for a cysteine residue(s) in the active site and/or a modulatory site. One previous study [18
] mentioned that murine SR protein with Cys217
→ Ser mutation is basally inactive, which may indicate that Cys217
is located in the SR active site and involved in catalysis. Also, Cys113
in mouse SR was reported to be S-nitrosylated by NO to form S-nitrosothiol (SNO), leading to a dramatic inhibition of enzyme activity [18
]. Structural analysis reveals that the ATP-binding site is in close proximity to Cys113
, and nitrosylation of Cys113
interferes with ATP binding to SR [18
]. If the structure of the free thiol group of Cys113
is critical for ATP binding to SR, other types of covalent modification of Cys113
including a disulfide bond between Cys113
and any other cysteine residue is very likely to have the same effect on ATP binding as nitrosylation of Cys113
are both conserved evolutionarily, suggesting that disulfide bonds between these residues, either intra- or intermolecularly, accounts for the cross-species similarities documented by Mustafa et al.
vis-à-vis our results.
In addition to S-nitrosylation, cysteine residue can undergo a broad range of post-translational redox modifications by diverse reactive oxygen species (ROS) and reactive nitrogen species (RNS). The ability to cycle between different redox states and the specificity toward oxidative reactions endow the cysteine residue with properties consistent with it being a participant in redox-based physiological signal transduction or modulation [29
]. Hydrogen peroxide can oxidize the cysteine residue in the catalytic center of protein tyrosine phosphatases (PTPs) to sulphenic acid (S-OH), abolishing the catalytic activity of the cysteine residue as a phosphate acceptor [30
]. RNS such as NO and peroxynitrite are reportedly able to produce sulphenic acid in the oxidation of thiol compounds [31
]. In many cases, sulphenic acid reacts rapidly with other cysteine thiols to form disulfide bonds (S-S). These disulfide bonds can be formed either intramolecularly or intermolecularly, the latter connecting two subunits in one complex or two different quaternary complexes.
As the major covalent connection between peptides, the disulfide bond plays a very important role in protein folding, polymerization, and activity regulation [32
], which was also tested in SR here. Under some circumstances, a mixed disulfide bond is formed between cysteine residues in proteins and the small molecule glutathione, an event known as S-glutathionylation (protein-SSG). Accumulating evidence implicates S-glutathionylation in many pathological conditions such as neurodegenerative diseases, cancer, lung diseases, cardiovascular diseases, and diabetes [33
]. Under conditions of a robust oxidative burst, sulphenic moieties of cysteine residues can be progressively oxidized to sulphinic acid (R-SO2
H) and sulphonic acid (R-SO3
H). While S-nitrosothiol (SNO), sulphenic acid (S-OH), disulfide bonds (S-S), and S-glutathionylation (protein-SSG) can be reduced by glutathione and/or thioredoxin in biological system, formation of sulphinic acid (R-SO2
H) and sulphonic acid (R-SO3
H) is believed to be irreversible. Hence, these two hyperoxidized thiol moieties are not considered to be involved in reversible redox signaling [34
This study also supports the second potential effect of DTT on the enzyme's structure: creating conditions permissive for optimal flexibility of the structure. DTT facilitated release of non-covalent dimer from large aggregates and tetramers, suggesting that during substrate binding and/or catalysis, SR enzyme may need to change its conformation to make the active center more suitable for substrate and the positions of critical amino acids more accurate for catalysis. This is confirmed by recent findings from a SR crystallization study [28
], which demonstrated that the tertiary structure of SR has large and small domains connected by a loop region. During the racemization reaction, the small domain appears capable of a large degree of flexibility to correctly orientate the substrate L-serine close to active site for catalysis. Compared to non-covalent dimer, disulfide-linked dimer is less likely to change conformation freely, which may explain its lower catalytic ability. Thus, the flexibility of the enzyme protein appears to be important for its function.
We previously found that inflammatory activation of microglia elevated the levels of a stable SR dimer and D-serine production [26
]. This finding suggested that an apparent covalent dimer was responsible for enzymatic activity. Activation of microglia typically involves a programmatic superoxide production from NADPH oxidase, and this dramatically enhances the oxidation conditions in the cells. It is possible that these conditions result in oxidative crosslinking of the SR subunits. This may involve disulfide bonds, but the species induced by microglial activation was stable to DTT and β-mercaptoethanol. Moreover, the data reported here indicate that disulfide-linked dimers are actually less active than noncovalent dimers, arguing against any connection between formation of the stable dimer and elevated D-serine production in activated microglia. It seems more likely, therefore, that the stable dimer observed in activated microglia results from some chemical conjugation of amino acids other than cystine disulfides. For example, superoxide dismutase has been found to aggregate through several forms of covalent bonds arising from a tryptophanyl radical [36
]. It is possible that microglial activation initially results in an elevation of SR expression (via the AP-1 transcription factor, as described [27
]), a substantial fraction of which would be active, noncovalent dimer. The subsequent oxidative conditions may then crosslink these dimers via as yet undefined reactions.