Rat Liver Serine Dehydratase

A pCW vector harboring rat liver serine dehydratase cDNA was expressed in Escherichia coli. The expressed level was about 5-fold higher in E. coli BL21 than in JM109 cell extract; the former lacked two kinds of proteases. Immunoblot analysis revealed the occurrence of a derivative other than serine dehydratase in the JM109 cell extract. The recombinant enzyme was purified to homogeneity. Staphylococcus aureus V8 protease and trypsin cleaved the enzyme at Glu-206 and Lys-220, respectively, with a concomitant loss of enzyme activity. Spectrophotometrically, the nicked enzyme showed a ∼50% reduced capacity for binding of the coenzyme pyridoxal phosphate and no spectral change of circular dichroism in the region at 300–480 nm, whereas circular dichroism spectra of both enzymes in the far-UV region were similar, suggesting that proteolysis impairs the coenzyme binding without an accompanying gross change of the secondary structure. Whereas the nicked enzyme behaved like the intact enzyme on Sephadex G-75 column chromatography, it was dissociated into two fragments on the column containing 6 m urea. Upon the removal of urea, both fragments spontaneously refolded. These results suggest that serine dehydratase consists of two folding domains connected by a region that is very susceptible to proteases.

region 100 -130 amino acid residues downstream of the PLP binding lysyl residue (9). The importance of this motif in the interaction with the coenzyme was assessed by the finding that substitution of the glycine residues with aspartic acid residues impairs PLP binding to E. coli D-serine deaminase (11,12). These two conserved sequences suggest that SDHs have evolved from a common ancestral protein (9).
PLP catalyzes a variety of enzyme reactions such as transamination, decarboxylation, isomerization, elimination, and so on (see Ref. 13 for a review). Thus far, the crystal structures of more than 10 PLP enzymes are available (14). These enzymes are mainly from bacterial sources, with the exception of aspartate aminotransferases from chicken and pig livers. On the other hand, PLP enzymes are classified into at least three types of ␣, ␤, and ␥ families on the basis of their primary sequences (15). SDH, which catalyzes ␣,␤ elimination, belongs to the ␤ family, whereas tryptophanase (14) and tyrosine phenol-lyase (16), although using similar catalyzing reaction mechanisms, are affiliated with the ␣ family (15). Thus, classification based on the sequence alignment does not always conform to that based on the reaction mechanism. For insight into this problem, we feel that it is vital to accumulate information about the crystal structure of the ␤ family members, which are known to include tryptophan synthase (17) and E. coli threonine deaminase (18). A crucial step toward this goal is to obtain pure enzyme for crystallization. Purification of SDH from rat liver was extremely hard because of its relatively low abundance and the sacrifice of numerous animals (19 -23). In this work, we have developed a bacterial expression and purification procedure and characterized the recombinant enzyme. Along with crystal data on other PLP enzymes, our key finding that SDH is specifically cleaved into two fragments by various proteases that can independently refold after denaturation strongly suggests that this enzyme is composed of at least two folding domains.

EXPERIMENTAL PROCEDURES
Materials-Male Wistar strain rats (200 g) and male Japanese white rabbits (2.5 kg) were purchased from Sankyo Labo Service. Biochemical reagents were commercially available and used without further purification.
Plasmid Construction-Plasmid pCWOriϩ was described previously (24). The initial portions of E. coli mRNAs are usually rich in A and U, and the expression of foreign DNAs in the bacterium is often facilitated by making the relevant regions rich in A or T. Thus, we introduced three silent mutations in codons 2-5 (i.e. from the native sequence 5Ј-GCTGCCCAGGAG to the mutated sequence 5Ј-GCTGCTCAAGAT (the introduced changes are underlined)) via polymerase chain reaction mutagenesis. Plasmid pCWOriϩ has a NdeI restriction site (CA͉TATG; ͉ indicates the NdeI cutting site) coincident with the initiation ATG codon * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and multicloning sites of XbaI, SalI, PstI, and HindIII downstream of the NdeI site. A foreign DNA is to be inserted between the NdeI site and any of the cloning sites. SDH cDNA had no NdeI site coincident with the initiation ATG codon and also had no usable restriction site in the 3Ј noncoding region. Therefore, a NdeI site and a HindIII site were created by polymerase chain reaction mutagenesis. For this purpose, oligonucleotides 5Ј-TGGCCTGCTCAAGATTCCCTGCACGTG-3Ј (the underlined sequence is part of the NdeI site) and 5Ј-GGATAAAGAAGCTTGGGCCACT-GTC-3Ј (the underlined sequence is the HindIII site) were synthesized as the 5Ј and 3Ј primers, respectively. The latter sequence is derived from the native antisense strand sequence corresponding to positions 35-59 downstream of the TGA stop codon (5Ј-GACAGTGGCCCACCCTTCTTTATCC-3Ј) (3). With these two primers, the SDH cDNA containing the mutations was amplified by polymerase chain reaction. The polymerase chain reaction product was then digested with HindIII to produce a HindIII cut site at the 3Ј end. Before ligation of the DNA, pCWOriϩ was digested with NdeI, and the NdeI site was filled with the Klenow enzyme. This linearized plasmid was further digested with HindIII to remove the original insert, and the resulting plasmid was ligated to the modified SDH cDNA. This construct was designated pCW-SDH.
Enzyme Assay-SDH activity was determined by the dinitrophenylhydrazine method (19). The complete reaction mixture consisted of 50 mM borate-KOH (pH 8.3), 50 mM serine, and 50 M PLP/enzyme in 0.25 ml. The absorption coefficient of hydrazone is 11.6 mM Ϫ1 , as determined with authentic sodium pyruvate. One unit of enzyme activity is defined as the amount that catalyzes the formation of 1 mol of pyruvate per minute at 37°C.
Purification of Recombinant SDH-All operations were carried out at a temperature of 0°C to 4°C unless otherwise stated. E. coli carrying pCW-SDH were cultured in 2YT medium (25) containing ampicillin (50 g/ml) at 37°C. IPTG was added to make a final concentration of 0.5 mM when the cell turbidity measured at 600 nm reached about 0.4, and the culture was continued for an additional 14 h. The cells suspended in 20 ml of 20 mM Tris-HCl (pH 8.0), 2 mM EDTA, and 5 mM 2-mercaptoethanol received 10 mg of lysozyme. They were kept at 0°C for 15 min, followed by Ϫ80°C for 30 min. After sonication at 200 watts for 1 min, the suspension was clarified by centrifugation at 10,000 ϫ g for 15 min. The supernatant from 1 liter of cultured cells was then put on a DE-52 column (Whatman; diameter, 2 cm; height, 6 cm) prewashed with 10 mM Tris-HCl (pH 8.0). The column was washed with 100 ml of 50 mM Tris-HCl (pH 7.8), 1 mM EDTA, and 5 mM mercaptoethanol. Ammonium sulfate was added at 28 g/100 ml to the flow-through fraction. After 1 h, the precipitate was collected by centrifugation at 12,000 ϫ g for 30 min and dissolved in 5 ml of 10 mM Tris-HCl (pH 7.5). After a brief centrifugation, the clarified solution was applied to a column of Sephacryl S-200 (Pharmacia; 3.2 ϫ 97 cm) equilibrated with 10 mM potassium phosphate (pH 7.8, 1 mM EDTA, 50 M PLP, and 5 mM 2-mercaptoethanol. Fractions with a high specific activity were pooled and put on an AH-Sepharose column (Pharmacia; 2 ϫ 7 cm) equilibrated with 10 mM potassium phosphate (pH 7.8). SDH appeared as a single peak at about 40 mM potassium phosphate when eluted by a linear gradient of 100 ml each of 10 mM and 100 mM potassium phosphate (pH 7.8) containing 0.1 mM EDTA and 1 mM dithiothreitol. The fractions judged pure by SDS-PAGE were collected and concentrated by a Collodion bag. The enzyme was kept in portions on ice or at Ϫ80°C without loss of activity for at least a month.
Isolation of Large and Small Fragments-Trypsinolysis (L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin; Worthington) was done in 0.1 M Tris-HCl (pH 8.0) at the substrate:protease ratio of 1,000:1 (w/w) and a temperature of 25°C. The reaction was terminated by adding a 5-fold molar excess of leupeptin over trypsin. The digest was subjected to 13.5% SDS-PAGE, and the gel was soaked in 1 M KCl. The visualized large and small bands were cut out and electroeluted in TAE (40 mM Tris acetate, 1 mM EDTA) buffer (25). Alternatively, the digest (5 mg) was brought to 6 M urea and size-fractionated on a Sephadex G-75 column (1.8 ϫ 98 cm) equilibrated with 10 mM potassium phosphate (pH 6.8), 0.1 mM EDTA, 1 mM dithiothreitol, and 6 M urea. Two-ml portions were fractionated by monitoring the absorbance at 280 nm. The fractions containing the large or small fragments were dialyzed against 10 mM potassium phosphate (pH 6.8), 0.1 mM EDTA, and 1 mM dithiothreitol overnight and concentrated by a Collodion bag.
Amino Acid Sequencing-The amino acid sequence of the polypeptide (100 pmol) was determined by automated Edman degradation on a Shimadzu PPSQ-10 gas-phase sequencer.
Spectral Analyses-Absorption and CD spectra were measured using a Hitachi 320 spectrophotometer and a Jasco J-500C spectropolarimeter, respectively. The ␣-helix content was estimated by the method of Chen and Yang (26).
Other Methods-Protein was determined by the method of Bradford (27) with bovine serum albumin as the standard. SDS-PAGE (28) and immunoblotting (29) were performed as described previously. The monospecific IgG to rat liver SDH was described previously (2) and was used to identify recombinant SDH in Fig. 1. The antibody against the recombinant enzyme was raised in rabbits, further purified by recombinant SDH-coupled agarose gel chromatography as described previously (30), and used for the experiments shown in Figs. 3 and 6. Densitometry was examined using NIH Image software.

RESULTS
Expression of SDH cDNA-E. coli JM109 cells transformed with the recombinant vector pCW-SDH or the control pCW with no insert were cultured in the presence of IPTG. The IPTG-induced SDH activity in a crude extract was found to be about 2 units/mg. The specific activity of SDH purified from rat liver was in the range of 150 -989 units/mg (19 -23); thus, a 75to 500-fold purification appeared to be necessary for homogeneity. A plausible cause of the low expression was the instability of the enzyme in this strain. We then resorted to E. coli BL21, which lacks both ATP-dependent Lon protease (31) and OmpT outer membrane protease (32). As expected, the BL21 cell extract was found to have an enzyme activity 5-fold higher than that of the JM109 extract. Fig. 1A shows SDS-PAGE of the extracts. The IPTG-treated JM109 cell extract had a M r 35,000 protein corresponding to the subunit of rat liver SDH. This band was more intense in BL21 cells than in JM109 cells (lanes 2 and 4). Immunoblot analysis with the IgG to rat liver SDH indicated that JM109 contained not only the M r 35,000 band but a faint band of M r 24,000. This band became more evident in an aged preparation (Fig. 1B, lane 4) or in a preparation from JM109 cells grown at 31°C (data not shown), but it was not found in the BL21 extract. Thus, it is thought that the low expression of the enzyme in JM109 cells results from extensive proteolysis. . B, immunoblot analysis of E. coli JM109 and BL21 extracts. Ten g of the extracts from JM109 and BL21 cells cultured with or without IPTG were electrophoresed as described above, blotted onto a nitrocellulose membrane, and probed with anti-rat liver SDH IgG (2). An ECL TM detection kit from Amersham was used. Purification and Some Properties-SDH was purified from E. coli BL21 by the conventional procedure as described under "Experimental Procedures." Approximately 15 mg of pure enzyme were obtained from a 1-liter culture. The enzyme activity emerged as a single peak on AH-Sepharose or DEAE-cellulose column chromatography. The chromatography deprived PLP of SDH, as found with a colorless preparation. The addition of PLP to the apoenzyme allowed absorption maxima to be restored at 330 and 415 nm, as reported previously for rat liver enzyme (20). Amino acid analysis indicated that the N terminus of the recombinant enzyme was alanine, whereas that of the liver enzyme was acetylalanine (4,5). Thus, in E. coli, the N-terminal methionine residue is removed from the enzyme by posttranslational modification, but acetylation of the new terminal residue does not occur. The native and subunit molecular weights were about 66,000 by gel filtration and 34,000 by SDS-PAGE, respectively, indicating that the recombinant enzyme is a dimer like the liver enzyme. The K m values for serine and threonine of the recombinant enzyme were 67 and 50 mM, respectively, which were almost comparable to those reported for the liver enzyme (20). The Susceptibility of SDH to Proteases- Fig. 1 suggested that SDH was vulnerable to attack by proteases. Then the purified enzyme (the apoenzyme form) was examined for susceptibility to various proteases. The reactions were done at the substrate: protease ratio of 1,000:1 to 30:1 by w/w. As shown in Fig. 2, trypsin and lysyl endopeptidase produced two fragments with molecular weights of 22,000 (large fragment) and 12,000 (small fragment). Staphylococcus aureus V8 protease also brought two fragments with molecular weights of 21,000 and 13,000. Subtilisin cleaved the enzyme into two fragments comparable to the tryptic fragments in size. However, the enzyme was virtually or considerably resistant to arginyl endopeptidase, chymotrypsin, and thermolysis at the concentrations examined. Similar results were obtained with the holoenzyme (data not shown). These results suggest that PLP does not protect against proteolysis.
Kinetics of Limited Proteolysis-To clarify the relationship between proteolysis and inactivation, an apoenzyme was incubated with trypsin as described above, and aliquots of the reaction mixture were withdrawn over time. Trypsin abolished the enzyme activity following pseudo-first order kinetics (Fig.  3A). SDS-PAGE showed the appearance of M r 22,000 and M r 12,000 bands with time (Fig. 3B, left panel). There was a stoichiometry between the loss of enzyme activity and the disappearance of the parent band by proteolysis (Fig. 3B, right  panel). The monospecific IgG to the recombinant enzyme predominantly reacted with the M r 22,000 band (Fig. 3C). The Sites Digested by Trypsin and S. aureus V8 Protease-Next we determined the proteolytic site. After treatment with trypsin, the digest was subjected to SDS-PAGE, and the large and small fragments were recovered from the gel. Edman degradation demonstrated that the first 10 amino acid sequences of the large and the small fragments were consistent with the N-terminal sequence of the intact protein and the amino acid sequence of a peptide from Ala-221 to Gln-230, respectively. Likewise, the cleavage site by S. aureus V8 protease was identified to be between Glu-206 and Gly-207.
Proteolysis Destabilizes PLP Binding-In this section, we studied the spectral change of a nicked enzyme. For this, the recombinant enzyme (the holoenzyme form) was digested with trypsin until an over 95% loss of the enzyme activity occurred and was then dialyzed extensively. The treated enzyme exhibited about 50% reduced absorptions at 330 and 415 nm and showed almost no difference in absorption at 280 nm compared with the control enzyme (Fig. 4A). Likewise, the apoenzyme was digested with trypsin, followed by an incubation with 200 M PLP and dialysis. This enzyme also showed about 50% decreased absorptions at 330 and 415 nm compared with the PLP-reconstituted enzyme (data not shown). CD is a good tool to explore the secondary structure of protein. It is known that PLP itself gives no CD spectrum between 300 and 480 nm. When bound to the apoenzyme, PLP could induce a CD spectrum with positive ellipticities at 330 and 415 nm (Fig. 4B). Interestingly, the holoenzyme previously nicked with trypsin gave no appreciable signal in this region. Similarly, the apoenzyme previously nicked and then reconstituted with PLP did not display a positive ellipticity at either 330 or 415 nm (data not shown). On the other hand, both the intact and nicked holoenzymes exhibited similar CD spectra in the far-UV region; their ␣-helices were estimated to be 54% and 51%, respectively (Fig. 4C). These results show that proteolysis fairly impairs coenzyme binding without accompanying a gross change in the secondary structure. We further addressed the question of whether the two fragments are separable under nondenaturing conditions. To test this, the trypsin-treated enzyme was applied to a Sephadex G-75 column. The nicked enzyme appeared at the position identical to that of the intact enzyme as monitored with the absorption at 280 nm, suggesting that both fragments still associate with each other.
Refolding of Large and Small Fragments-Accumulating crystal data have revealed that many PLP enzymes constitute two (or three) folding domains (13). If SDH is made up of distinct folding domains, these domains may be capable of refolding independently under renaturing conditions after denaturation. Thus, a nicked enzyme was denatured with 6 M urea and subjected to separation on a Sephadex G-75 column containing 6 M urea. The isolated fragments were confirmed by SDS-PAGE (Fig. 5A). Urea was removed by extensive dialysis, and the CD spectra thereof were measured in the far-UV region (Fig. 5B). As a reference, an uncut enzyme was processed in a similar manner. The renatured protein had a specific activity of 144 units/mg and 41% ␣-helix, corresponding to 72% and 76% of the specific activity and ␣-helix of the unprocessed enzyme, respectively. Under these conditions, the large and small fragments independently refolded, and the calculated ␣-helices were 14% and 37%, respectively (Fig. 5C). Although it is unknown at present to what extent the ␣-helices are included in the large and small segments of the SDH polypeptide, the ␣-helix of the refolded large fragment seems to be less than that of the small fragment or the native protein. These results do not exclude the possibility that the large fragment could not refold more efficiently than the small fragment.
Liver Enzyme Is also Susceptible to Trypsin-It has been proposed that protein folding in eukaryotes is cotranslational, whereas that in prokaryotes is posttranslational (33). It is possible that the folding of the recombinant enzyme may be different from that of liver enzyme or that the absence or presence of N-terminal blocking may be responsible for the FIG. 4. UV absorption and circular dichroism spectra of native and trypsin-treated SDH. SDH (holoenzyme) was digested with trypsin until a Ͼ95% loss of enzyme activity and was dialyzed against two changes of 1 liter of 10 mM potassium phosphate (pH 7.8), 0.1 mM EDTA, and 0.1 mM dithiothreitol at 0°C over a period of 14 h. A, absorption spectra of native (solid curve) and trypsin-treated (dashed curve) SDH. The concentration of SDH used was 1 mg/ml in each case. CD spectra were measured with a 0.5-mm-thick cell in the 280 -480 nm region (B) and with a 0.2-mm-thick cell in the 200 -280 nm region (C). Thick, thin, and dotted curves represent the CD spectra of native SDH, nicked SDH, and buffer solution, respectively. The protein concentration used was 0.9 mg/ml in B and 0.2 mg/ml in C.  Fig. 4 legend. As a reference, a holoenzyme was denatured with 6 M urea and renatured by dialysis in the same manner as the tryptic fragments (uncut SDH). different folding pattern in the recombinant and liver SDHs. To test this possibility, liver enzyme was partially purified through ammonium sulfate fractionation and gel filtration (Fig. 6A) and subjected to limited proteolysis followed by immunoblot analysis. The preparation that was not treated with trypsin exhibited two bands of M r 22,000, and M r 24,000 other than an intact band (Fig. 6B, lane 4). Because no protein inhibitor was included in the course of purification, these subbands were considered to be degradation products. Trypsin treatment increased the M r 22,000 band accompanying the disappearance of the parent band (Fig. 6B). The result suggests that the liver enzyme is also susceptible to trypsin. DISCUSSION SDH is widely spread in nature, but its physicochemical properties are considerably different from species to species. For example, rat (20) and sheep liver enzyme (34) is a dimer with Michaelis-Menten kinetics with respect to the substrate, whereas yeast and E. coli biosynthetic threonine dehydratase, the first enzyme in the isoleucine synthesis pathway, is a tetramer and is feedback-inhibited by isoleucine and heterotropically activated by valine (8). E. coli catabolic threonine dehydratase induced anaerobically in tryptone-yeast extract medium is a tetramer and is allosterically activated by AMP (9). However, E. coli D-SDH is a monomer (35,36). (An appreciable amount of D-serine is present in mammalian brain, which is produced by the racemization of L-serine. The occurrence of D-SDH remains to be determined (37)). The importance of the glycine-rich sequence in SDH was first verified with D-SDH (11,12), and its preliminary crystal data were reported (38). Recently, non-PLP-dependent, sulfur/iron-dependent SDH was found in some microorganisms (39). It is obvious that this new type of enzyme has no sequence homology to authentic SDH including the sequence around the PLP-binding lysyl residue and the glycine-rich motif (39), but it is unclear whether such an enzyme occurs in eukaryotes.
We have previously succeeded in isolating more than 80 mg of glycine methyltransferase, a rat liver enzyme, from a 1-liter culture with a combination of the pCW vector and E. coli JM109 (40), and this led to crystallographic results (41). However, the same vector did not work effectively on SDH. This was due largely, if not entirely, to the high sensitivity of recombinant enzyme to endogenous protease(s). The E. coli BL21 strain circumvented this problem. Even in the absence of IPTG, a significant amount of SDH was seen in this strain relative to JM109 (Fig. 1). This finding gave us a chance to analyze a higher order structure of SDH by limited proteolysis. S. aureus V8 protease and trypsin nicked at Glu-206 and Lys-220 of the purified enzyme, respectively. Although the cutting sites by subtilisin and lysyl endopeptidase remain undetermined, they are expected to be near Lys-220, because similar fragments were liberated by these proteases (Fig. 2). Moreover, a fragment appearing in the JM109 cell extract was about M r 1,500 larger than the tryptic large fragment, indicating that the cleavage site would be around Leu-234. PLP did not protect against proteolysis, suggesting that a stretch of about 30 residues susceptible to protease does not lie in the PLP binding site but rather is exposed to a solvent phase.
It is said that proteins that consist of two independently folding domains connected by a hinge peptide are particularly susceptible to proteolysis (e.g. see Ref. 42 for a review). It is tempting to speculate that the protease-sensitive region of SDH acts as a hinge for the two domains. For substantiating this notion, two tryptic fragments were separated under denaturing conditions and dialyzed extensively. CD measurements revealed that the two fragments could spontaneously refold (Fig. 5B). Thus, these results support the notion that SDH is composed of at least two domains. In fact, most PLP enzymes are known to have multiple folding domains; the two-domain structure is found in aspartate aminotransferase (43), tryptophanase (14), tyrosine phenol-lyase (16), dialkylglycine decarboxylase (44), and tryptophan synthase (17), whereas the three-domain structure is seen in glutamate-1-semialdehyde aminomutase (45) and cystathionine ␤-lyase (46). Fig. 7 compares the diagrammatic structure of SDH with those of the ␤ subunit of tryptophan synthase (17) and the biosynthetic threonine deaminase of E. coli (18). SDH is considered to be close to the tryptophan synthase ␤ subunit because (i) in both enzymes, the PLP binding lysyl residue is located relatively early in the primary sequences and the distance between this residue and the glycine-rich sequence is about 130 residues, and (ii) the reaction mechanisms of ␤ replacement and ␤ elimination are relatively similar. Tryptophan synthase ␤ subunit was nicked at Lys-272, Arg-275, and Lys-283 by various proteases to result in the large and small fragments, and the isolated fragments could spontaneously refold under renaturing conditions (47). The x-ray crystallography revealed a 54-residue stretch containing the residues susceptible to proteases located at the N-terminal side of the C-domain (17). Threonine deaminase was found to have the N-terminal catalytic and C-terminal regulatory domains that are connected with a neck region consisting of 12 amino acid FIG. 6. Liver SDH is also susceptible to trypsin. One g of a purified recombinant enzyme and 20 g of a liver preparation (Sephacryl S-200 fraction) were digested with 1 and 40 ng of trypsin, respectively, at 25°C for 10 or 20 min and subjected to 13.5% SDS-PAGE followed by Coomassie Blue staining (A) or immunoblotting with anti-recombinant SDH IgG (B).
FIG. 7. Schematic representation of the structures of the Salmonella typhimurium tryptophan synthase ␤ subunit, rat liver SDH, and E. coli threonine deaminase. The structures of S. typhimurium tryptophan synthase ␤ subunit and E. coli threonine deaminase are taken from Refs. 17 and 18. Arrows indicate the location of the lysyl residue capable of forming a Schiff base with PLP and the glycinerich sequence that interacts with the phosphate group of PLP.`, the region susceptible to proteases. The f in threonine deaminase represents the "Neck" according to the nomenclature of the authors. o, the C-domains assigned by x-ray analysis. A portion of the N-terminal region contributes to the structure of the C-domain of tryptophan synthase ␤ subunit. The structure of the C-domain of threonine deaminase resembles that of the serine binding domain of 3-phosphoglycerate dehydrogenase.
residues (18). The folding pattern of the N-domain resembles that of the tryptophan synthase ␤ subunit, but the structure of the regulatory domain is rather similar to that of the regulatory serine binding domain in allosteric 3-phosphoglycerate dehydrogenase (48). It remains unclear whether the neck region is susceptible to proteases like SDH and tryptophan synthase ␤ subunit. At any rate, these similarities are supporting evidence that SDH is comprised of two domains. Although large differences exist in structural and kinetic properties between E. coli threonine deaminase and rat liver SDH, the first indication of the x-ray structure of the former among various hydroxyamino acid dehydratases would be informative for the future study of rat liver SDH.