Modulation of Syrian Hamster 3-Hydroxy-3-methylglutaryl-CoA Reductase Activity by Phosphorylation ROLE OF SERINE 871*

Attenuation of Syrian hamster 3-hydroxy-3-methyl- glutaryl coenzyme A reductase (HMG-CoA reductase, EC 1.1.1.34) activity by in vitro phosphorylation was studied using AMP-activated protein kinase and wild-type and mutant forms of HMG-CoA reductase. The only residue of the wild-type enzyme phosphorylated was SeF". Sub- strates protected against kinase-mediated attenuation of activity, consistent with substrate-induced conforma- tional changes at the C-terminal region. Although close to the catalytic histidine Hism6, SePl appears to play no direct role in catalysis or substrate recognition. Mutant enzymes S871A, S871H, S871N, and S871Q exhibited from 62106% of wild-type activity and had wild-type K, values for HMG-CoA and NADPH. Replacement of SeF1 by aspartate or glutamate, but not by glutamine, aspar-agine, histidine, or tyrosine, severely attenuated activ- ity. Attenuation of catalytic activity that accompanies phosphorylation thus appears to result primarily from the introduction of negative charge, not merely steric hindrance. Other than the wild-type enzyme, only mu- tant enzyme S871T was phosphorylated, and phosphorylation was accompanied by attenuation of activity. The

NADP+, constitutes a major focus for control of the biosynthesis of polyisoprenoids and other mevalonate-derived metabolites (1). Knowledge of the mechanism of catalysis and of the regulation of the catalytic activity of HMG-CoA reductase thus is important for rational chemotherapy of certain forms of hypercholesterolemia. Regulation of higher eukaryote HMG-CoA reductases involves both changes in enzyme quantity and modulation of catalytic activity by post-translational processes that include phosphorylation (1-10). The protein kinases that phosphorylate rat liver HMG-CoA reductase in vitro, protein kinase C (111, calciudcalmodulin-activated kinase (121, and AMP-activated kinase (13), all phosphorylate the rat liver enzyme at serine 871 (13, 14).
* This work was funded by National Institutes of Health Grant HL 47113. This is journal paper 13828 from the Purdue University Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  [18][19][20]. No consensus has, however, emerged concerning whether the effects of phosphorylation are due to negative charge or to steric factors. The contributions of charge and bulk have been investigated by replacing phosphorylatable serines with bulky or negatively charged amino acids. For E. coli isocitrate dehydrogenase (18-20) and the product of the fos proto-oncogene (21), the introduction of negative charge mimics phosphorylation. However, unlike phosphorylation, the presence of acidic residues at the C terminus of the cyclic AMP-responsive element-binding protein did not abolish transcriptional activity (22), nor did substitution of glutamate for Ser477 and Ser478 of the epidermal growth factor-like protein ErbB mimic the ability of phosphorylation to suppress oncogenic potential (23).
SerS7l of Syrian hamster HMG-CoA reductase is located within 6 residues of His865, a residue functional in catalysis (24). By analogy to the recently solved crystal structure of

Pseudomonas mevalonii
HMG-CoA reductase (251, Hiss65 should reside in a flexible domain that approaches the active site when substrates are bound. We investigate here whether SeP1, the site of phosphorylation in the rat liver enzyme (13), is essential for catalysis. We also replaced with charged, neutral, and bulky residues to investigate whether the effects of phosphorylation are due to negative charge or steric properties. In addition, we have investigated the residue specificity of the AMP-activated protein kinase, the kinase of prime physiologic importance for regulation of HMG-CoA reductase (26), by exploiting the presence of the protein kinase recognition sequence around S e P 1 of hamster HMG-CoA reductase.
Site-directed Mutagenesis and DNA Sequencing-Molecular biological procedures followed Sambmk et al. (30). Isolation of singlestranded pKFT7-21 DNAused as template in site-directed mutagenesis employed helper phage M13K07 and the Promega procedure (31). Mutagenesis employed an Amersham site-directed mutagenesis kit (32) and mutant oligonucleotides (Table I). All mutations were verified by double-stranded sequencing of phagemid DNA using the Sanger dideoxynucleotide chain termination method (33) as modified by Hsiao (34). Sequencing employed a Sequenase" kit, [d5S1dATP, and the se- Autoradiography and Measurement of Incorporation of 32P-Following resolution of samples by SDS-PAGE (30), gels were soaked in 10% methanol, 10% acetic acid, dried under vacuum (60 "C, 30 mid, and exposed to x-ray film a t -70 "C. Major radioactive bands, which corresponded in mobility to the HMG-CoA reductase subunit, were excised and counted in scintillation fluor in a Beckman model LS 1801 scintillation spectrometer.

Homogeneity and Physical Integrity of Wild-type and Sef171
Mutant Enzymes-Wild-type enzyme and mutant HMG-CoA reductases S871A, S871D, S871E, S871H, S871N, S871Q, and S871T, purified as previously described for the wild-type enzyme (28) were over 95% homogeneous as judged by SDS-PAGE (Fig. 1). All purified enzymes exhibited wild-type chromatographic behavior during molecular sieve high-performance liq- uid chromatography on a Bio-Rad Bio-Si1 TSK-250 column eluted with 200 mM Na,P04, pH 6.8, implying that each possessed wild-type quaternary structure. Incorporation of 32P Is Accompanied by Attenuation of Catalytic Activity-MgATP plus the AMP-activated protein kinase attenuated the activity of wild-type HMG-CoA reductase in a time-dependent manner (Fig. 2). No attenuation was observed when MgATP, kinase, or both were omitted. Attenuation of activity accompanied incorporation of radioisotope from [Y-~~PIATP (Fig. 3).
HMG-CoA or NADPH Decreases the Susceptibility of HMG-CoA Reductase to Attenuation of Its Activity-HMG-CoA or NADPH decreased the rate at which the protein kinase attenuated HMG-CoA reductase activity. The time required to attenuate activity to half its initial value increased from 43 min (no additions) to 120 min (+ NADPH) or to 250 min (+ HMG-CoA) (Fig. 4). These changes reflect effects on HMG-CoA reductase, not on the protein kinase. Kinase activity measured using the SAMs peptide, 10.2 2 0.1 nmoVmidmg, was unaffected by either HMG-CoA or NADPH, even under conditions when the kinase was rate-limiting.
SeP7' Is Not Essential for Catalysis or Substrate Binding-SeP7I cannot be essential for catalysis since the specific activities of mutant enzymes S871A, S871N, and S871H approached wild-type values ( Table 11). SeP71 also does not appear essential for substrate binding since K, values for both substrates approximated wild-type values (Fig. 5, Table 111).
Susceptibility of SeP71 Mutant Enzymes to Phosphorylation and Attenuation of Their Activity-The AMP-activated kinase neither affected the activity of mutant enzymes S871A, S871D, S871E, S871H, S871Q, or S871N (Fig. 61, nor catalyzed their phosphorylation (Fig. 7). The kinase thus recognized and phosphorylated only Sers71 of hamster HMGCoA reductase.
AMP-activated Protein Kinase Can Phosphorylate Threonyl Residues-The presence in HMG-CoA reductase of the recognition sequence for the AMP-activated protein kinase permitted us to ask whether this kinase, not previously reported to catalyze phosphorylation of residues other than serine, could phosphorylate threonyl or tyrosyl residues. Mutant enzyme S871T, but not mutant enzyme S871Y or any other S871 mutant enzyme, was indeed phosphorylated (Fig. 7). Phosphorylation of threonine was confirmed by the detection of 32Plabeled phosphothreonine in an acid hydrolysate of phosphorylated mutant enzyme S871T (Fig. 8).
Phosphorylation of a Threonyl Residue Attenuates HMG-CoA Reductase ActivitySince the protein kinase can phosphorylate Sef171, we asked whether phosphorylation of T h F 1 was accompanied by the attenuation of HMG-CoA reductase activity. This was indeed the case (Fig. 6). Replacement of SeP7' by Aspartate or Glutamate Partially Mimics the Effect of Phosphorylation-Mutant enzymes S871D and S871E had 10 and 16% of the activity of the wild-type enzyme, respectively (Table 11). Lowered activity is unlikely to reflect impaired ability to bind substrates, since for both mutant enzymes K, values approximated wild-type values (Fig. 5,  Table 111). Low activity also did not appear to be due solely to steric effects since substitution of the cognate amides (glutamine and asparagine), of a positively charged residue (histidine), or of a neutral but bulky residue (tyrosine) had less effect. Mutant enzymes S871N, S871Q, S871H, and S871Y exhibited 106,62,100, and 55% of wild-type activity, respectively. The presence of negative charge, and not steric hindrance, thus appears to be a key factor in the attenuation of activity that accompanied phosphorylation of Sers71.  1.6 10 DISCUSSION Reversible phosphorylation has long been known to attenuate the catalytic activity of the HMG-CoA reductase of higher eukaryotes (3,6,10). We have investigated whether Sef171, the putative site of phosphorylation of hamster HMG-CoA reductase, functions in catalysis and have assessed the relative importance of charge and bulk at position 871 in attenuating catalytic activity.
As was first shown for rat liver HMG-CoA reductase (131, the AMP-activated protein kinase phosphorylated the catalytic domain of the Syrian hamster enzyme exclusively at S e F . Although not far from active site residue Hisss5, Sers71 plays no role in catalysis since replacement of Sers71 by alanine, asparagine, or histidine had no significant effect on catalytic activity. We also consider it unlikely that participates in substrate recognition since the K, values of NADPH and HMG-CoA approximated wild-type values for all eight Sef171 mutant enzymes. While as for E. coli isocitrate dehydrogenase the regulatory serine is present at the active site, unlike for isocitrate dehydrogenase, Sers7* of HMG-CoA reductase does not appear to participate in substrate binding. While substrates protect HMGCoA reductase against various inhibitors (37-39), the effect of substrates on the susceptibility of HMG-CoA reductase to phosphorylation has not previously been reported. The rate at which the activity of wildtype HMG-CoA reductase was attenuated by phosphorylation was decreased by factors of 3 and 6 by NADPH or HMG-CoA, respectively. These substrates do not inhibit the kinase itself since kinase activity measured using the synthetic SAMs peptide was unaffected by HMG-CoA or NADPH.

FIG. 5. Double-reciprocal plots for the dependence of initial
Shown are data for (RS)-HMG-CoA (Zefi) or NADPH (right) for wildtype and the indicated mutant enzymes. All assays were conducted at pH 6.5. Yv is the reciprocal of the specific activity (pnol of NADPH oxidizedmidmg of protein). Where NADPH concentration was vaned, the fixed concentration of (RS)-HMG-CoA was 270 p (12 times the wild-type IC,,,). Where (RS)-HMG-CoA concentration was vaned, the fixed concentration of NADPH was 270 p (6.5 times the wild-type K,,,).

K,,, values for wild-type and SerS7' mutant enzymes
The data are mean values of triplicate or quadruplicate determinations * the standard deviation from the mean. Sef171 of Syrian hamster HMGCoA reductase is located within 6 residues of Hiss65, a residue functional in catalysis (24).
By analogy to the recently solved crystal structure of l? meualonii HMG-CoA reductase (25), His866 and SerS7l should reside in a flexible domain at the C-terminal region that approaches the active site when substrates bind. A decreased rate of inactivation in the presence of either substrate thus probably reflects decreased availability of S e P 1 to the kinase consequent to a substrate-induced conformational change in the C-terminal region of the enzyme. The susceptibility of HMG-CoA reductase to phosphorylation thus might in future be exploited to study conformational changes at the C-terminal region of the enzyme.
For isocitrate dehydrogenase, replacement of Ser113 by glutamate or aspartate mimicked the effects of phosphorylation.
Substitution of tyrosine, a residue similar in size to a phosphorylated serine, also attenuated activity, but less severely (18,19). These investigators concluded that the effects of phos- phorylation resulted primarily from electrostatic repulsion (18,19). HMG-CoA reductase behaved in an analogous fashion. Replacement of Sers71 by aspartate or glutamate attenuated catalytic activity, while replacement by tyrosine attenuated activity less severely. Replacement of Sers71 by asparagine, glutamine, or histidine had far less effect. While neither glutamate nor aspartate attenuated activity as profoundly as phosphorylation of SeP71, these acidic residues introduce only a single charge, which probably is positioned somewhat differently from that of a seryl phosphate. We conclude that, while steric hindrance may contribute to some extent, the major factor responsible for the attenuation of activity that accompanies phosphorylation of SerS7l is the negative charge of the phosphate group. To our best knowledge, all known phosphorylation sites for the AMP-activated protein kinase are seryl residues (40, 41). However, this kinase catalyzed phosphorylation of mutant enzyme S871T, and phosphorylation was accompanied by attenuation of activity. Since no other sites in the wild-type or mutant enzymes other than S871T are phosphorylated by this kinase, and since phosphothreonine was detected in a hydrolysate of mutant enzyme S871T, we infer that ThP7I was phosphorylated. The AMP-activated protein kinase, which fulfills a regulatory role in lipid metabolism (261, thus might well phosphorylate threonyl residues of other proteins. Finally, we note that, while no known HMG-CoA reductase has a threonine at the position corresponding to position 871 of the Syrian hamster enzyme, the activity of HMG-CoA reductase can be attenuated as well by the phosphorylation of a threonyl as of a seryl residue.  Fig. 3, except for a 10-pl incubation that contained 0.46 pg of wild-type or mutant enzyme, 2 pl(4.6 pg) ofAMP-activated kinase in Buffer K, and 5.5 pl of Buffer T. Following incubation a t 37 "C for 60 min, samples were subjected to SDS-PAGE and autoradiography. Lanes -K and -R are incubations in which either the kinase (-K) or HMG-CoA reductase (-R) was omitted. Arrows indicate the positions of molecular weight standards of the indicated size (in ma).

S87 1T
no. 8. Phosphorylation of mutant enzyme 9871T. Incorporation of 32P into 50 pg of wild-type HMGCoA reductase or mutant enzyme S871T was conducted as in Fig. 3. A control incubation lacked HMG-CoA reductase. Incubation a t 37 "C for 2.5 h was terminated by adding 25 pl of 50% trichloroacetic acid. Following centrifugation, the precipitate was washed successively with 10% trichloroacetic acid (2 times) and acetone (2 times), dried in uacuo, dissolved in 200 p1 of 6 N HCI, maintained at 110 "C in a closed Eppendorf tube for 4 h, dried, and dissolved in 30 pl of electrophoresis buffer (2.5% formic acid, 7.8% acetic acid). Electrophoresis (4) on a plastic-backed cellulose TLC sheet was for 4 h a t 500 V. Shown is an autoradiograph of the electropherogram of hydrolysates of wild-type enzyme ( WT), mutant enzyme S871T (S871T), and wildtype enzyme plus mutant enzyme S871T (WT+S871T). Arrows indicate the positions to which non-radioactive phosphothreonine (P-Thr) and phosphoserine (P-Ser) standards, detected with ninhydrin, migrated.