lilillllflllll Will 1293 01090 909251 i ' p ' ' 111111 This is to certify that the dissertation entitled Location and Structure of the Substrate Hexose Binding Site of Rat Brain Hexokinase presented by Douglas M. Schirch has been accepted towards fulfillment of the requirements for Ph.D. degree in Biochemistry /% é” 2&2“ Major professor Date November 11, 1986 MS U is an Affirmative Action/Equal Opportunity Institution 0-1277 1 MSU LIBRARIES -,‘-. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. LOCATION AND STRUCTURE OF THE SUBSTRATE HEXOSE BINDING SITE OF RAT BRAIN HEXOKINASE by Douglas M. Schirch A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Biochemistry 1986 ABSTRACT LOCATION AND STRUCTURE OF THE SUBSTRATE HEXOSE BINDING SITE OF RAT BRAIN HEXOKINASE by Douglas M. Schirch A glucose analog, N-(bromoacetyl)-D-glucosamine (GlcNBrAc), previously used to label the glucose binding sites of rat muscle Type II and bovine brain Type I hexokinases, also inactivates rat brain hexokinase (ATP: D- hexose 6-phosphotransferase, EC 2.7.1.1) with pseudo-first order kinetics. Several methods of kinetic analysis indicate that the affinity label binds reversibly to the glucose binding site prior to inactivating the enzyme. However, nonspecific (i.e., without prior complex formation) inactivation also occurs, and equations are derived to describe this behavior. Inactivation is dependent on the deprotonation of a residue with an alkaline pKa consistent with the modified residue being a sulfhydryl group as reported to be the case with the bovine brain enzyme. The affinity label modifies three residues at indistinguishable rates. Amino acid analysis of the labeled protein indicates that each of these residues is a cysteine. However, only one of these cysteines appears to be critical for activity. Peptide mapping techniques have permitted localization of the critical residue, and thus the glucose binding site, in a 40 kDa domain at the C-terminus of the enzyme. This is the same domain recently shown to include the ATP binding site. Thus, catalytic function is assigned to the C-terminal domain of rat brain hexokinase. Peptide mappping of hexokinase modified by radiolabeled affinity label yields three labeled peptides, two of which are protected by inclusion of Glc or GlcNAc during reaction with GlcNBrAc. The sequences of these two peptides show considerable homology to portions of the sequences of yeast hexokinase isozymes A and B. One of these peptides is homologous to a region of yeast hexokinase which is clearly in a position to be labeled by the bromoacetyl arm on the 2- carbon of GlcNBrAc. An essential serine residue in yeast hexokinase, implicated in glucose binding, is also conserved in rat brain hexokinase. This provides strong evidence that the 40 kDa region of rat brain hexokinase Type I shares a common ancestry with yeast hexokinase. This is dedicated to my parents, for giving me the freedom and encouragement to pursue my own goals, and for being such good parents that their goals have become my goals. iv ACKNOWLEDGMENTS Above all I would like to thank John Wilson for serving as my advisor, as well as being patient with my graduate school plans when they changed from time to time. Asking him to serve as my advisor is probably the only decision I made in graduate school that I never second guessed. Much thanks also go to Al Smith, who, either as a technician or a graduate student, taught me almost every technique I mention in this dissertation. For the major portion of my time in the lab he also filled my insatiable appetite for hexokinase, for which he reminded me more than once. I am indebted to Dave Schwab also, for writing computer programs for me when I needed them, as well as teaching me a thing or two about pool. A final thanks to Paul Polakis, Ken Finney, Mohsen Nemat-Gorgani, Tracey White, Guochen Xie, Annette Thelen and Laura Chiarantini for making room 301 the most enjoyable lab in the building. TABLE OF CONTENTS Page List of Tables... ....................................... 1x List of Figures ................... . ...................... x List of Abbreviations .......... ............... .......... xv Introduction ............................................. 1 Literature Review.... .............. .. ...... . ............. 4 Yeast hexokinases.. ......... . ............. ..... ..... 4 Mammalian hexokinases ............................... 6 Postulated evolution of hexokinases. ...... . ......... 8 Sulfhydryl groups in hexokinases........... ...... ..10 Sulfhydryl groups at enzyme active sites.... ....... 14 Materials and Methods................ ................... 17 Reaction of GlcNBrAc with hexokinase.. ............. 18 Measurement of incorporation of [14C]-GlcNBrAc ..... 20 SDS-PAGE and flourography..... ..................... 21 Reduction and carboxymethylation of [14C]- GlcNBrAc labeled hexokinase ........................ 23 Exhaustive tryptic digestion of [14C]-GlcNBrAc labeled hexokinase ................................. 23 Primary HPLC of tryptic peptides.............. ..... 24 Chapter I. Location of the Glucose Binding Site ........ 29 GlcNBrAc inactivation of hexokinase in the presence of various ligands ........................ 29 Kinetics of GlcNBrAc inactivation of hexokinase ............... ... ....................... 38 vi GlcNBrAc as a reversible competitive inhibitor ..... 42 Protection by Glc.......... ........ . ......... . ..... 43 pH dependence of inactivation................ ...... 50 Amino acid analysis of GlcNBrAc labeled heXORinase.......OOOOOOOOOOO0.00.0.0...O ........... 51 Stoichiometric relationship between incorporation of GlcNBrAc and inactivation.. ....... 56 Tryptic cleavage of GlcNBrAc modified hexokinase............................. ...... ......65 [14C]-GlcNBrAc labeling of tm-hexokinase... ........ 71 [14CJ-IAM labeling of denatured tm-hexokinase ...... 72 Two dimensional peptide mapping of [1 C]-GlcNBrAc labeled tm-hexokinase..... .......... 77 Vicinal sulfhydryls in rat brain hexokinase........80 Discussion..............................................85 Derivations.............................................90 Footnotes ................ . ..... .... ........... . ......... 93 Chapter II. Structure of the Glucose Binding Site......94 Peptide mapping of [14C1-GlcNBrAc labeled hexokinase............................... ..... .....94 Effects of Glc and GlcNAc on alkylation of Peptides I-III........................ ........... .102 Amino acid compositions of Peptides I-III.........107 Mass spectrometry of Peptide III. ................. 110 Amino acid sequence of Peptides I-III. ......... ...110 Discussion.................................. ........ ...115 Location of Peptides I and III within the overall sequence of brain hexokinase and comparison with location of homologous peptides in yeast hexokinase......... ......... ....115 viii Proposed locations of Peptides I and III within the tertiary structure of the 40 kDa catalytic domain of rat brain hexokinase .......... 122 Location and nature of "critical" residues in brain hexokinase ........... . ..... . ............. 123 Structural homology between yeast and rat brain hexokinases ................................. 130 Structural relationships in monomeric mammalian hexokinases compared with the dimeric yeast hexokinase.................... ...... 132 List of ReferenceSOOOOOOO......OOOOOOOOOOOOOOO ....... 0.141 Table LIST OF TABLES Page Amino acid analysis of GlcNBrAc labeled rat brain heXOkinaseOOO......OOOOOOO0.0.0.000000000054 Oxidation of Vicinal sulfhydryls in rat brain hexokinase.. ........ ...... ..... ...........82 Summary of reactions of various hexokinases with glucose- and galactose-derived alkylating reagents.............................86 Amino acid composition of Peptides I—III.......108 Sequence analysis of Peptide I.................111 Sequence analysis of Peptide II................112 Sequence analysis of Peptide III ............... 113 ix LIST OF FIGURES Figure Page 1 Protection by ligands against inactivation by GlcNBrAc......... ............ . ..................... 31 2 Protection by various hexose-6-phosphates against inactivation by GlcNBrAc ............ . ...... 34 3 Protection by various hexoses against inactivation by GlcNBrAc...........................37 4 Inactivation by varying concentrations of GlcNBrAc ........................................... 41 5 Dixon plot for GlcNBrAc as a reversible inhibitor of hexokinase ................ ............45 6 Protection by varying concentrations of Glc against inactivation by GlcNBrAc...................48 7 10 ll 12 13 14 xi Effect of pH on the rate of inactivation of hexokinase by GlcNBrAc ............................. 53 Stoichiometric relationship between incorporation of radioactivity from [14CJ-GlcNBrAc and residual activity ............... 58 Stoichiometric relationship between incorporation of radioactivity from [14CJ-GlcNBrAc and residual activity in the presence of added Glc or GlcNAc ............. 63 Limited tryptic digestion of [14CJ-GlcNBrAc labeled hexokinase .................. 68 Schematic representation of relevant proteolytic cleavage sites in rat brain hexokinase.... .............. ..... .................. 70 Labeling of tm-hexokinase with [14C]-GlcNBrAc ...... 74 Quantitation of sulfhydryls in rat brain hexokinase domains ....... ....... ................... 76 Fluorographs after two-dimensional peptide mapping of tm-hexokinase labeled with [14C]-GlcNBrAc .................................... ..79 15 16 17 18 19 20 21 xii Effect of Vicinal sulfhydryls on inactivation by GlcNBrAc ........................... 84 Primary HPLC fractionation of a tryptic digest of [14C1-GlcNBrAc labeled hexokinase ........ 96 Secondary HPLC fractionation of Peptides I-III ...... . ....................................... 98 HPLC fractionation of a tryptic digest of [14CJ-GlcNBrAc labeled 40 kDa domain.. ............ 101 Effects of Glc or GlcNAc on the labeling of Peptides I-III by [14c1-G1cNBrAc .................. 104 Correlation between protection by Glc and GlcNAc against [14C]-GlcNBrAc inactivation and the labeling of Peptides I—III ................ 106 Amino acid sequence comparisons between yeast hexokinase isozymes A and B and selected peptides from rat hexokinase isozymes I and III...... ..... . ........ . .......................... 117 - xiii 22 Locations of Peptides I and III within the 40 kDa domain of rat brain hexokinase compared to the locations of homologous sequences in the yeast isozymes ................... 120 23 Region of yeast hexokinase active site to which Peptides I and III from rat brain hexokinase are homologous ......................... 125 24 Structure of yeast hexokinase dimer, proposed to approximate the structures of the 40 kDa and 50 kDa domains of rat brain hexokinase ........ 135 25 Generalized diagram of rat brain hexokinase structure .................... . ............... .....140 DTNB GalNBrAc GlcNAc GchrAc GlcNHexBr HPLC IAM PAGE PMSF PTC PTH SDS TFA TNB tm TPCK LIST OF ABBREVIATIONS 5,5-dithiobis-(2-nitrobenzoate) N-(bromoacetyl)-D-galactosamine N-acetylglucosamine N-(bromoacetyl)-D-glucosamine N-(N-bromoacetyl)-6-amino-hexanoyl-glucosamine high performance liquid chromatography iodoacetamide polyacrylamide gel electrophoresis phenylmethylsulfonyl flouride phenyl thiocarbamyl derivatives of amino acids phenylthiohydantoin derivatives of amino acids sodium dodecyl sulfate trifluoroacetic acid thionitrobenzoate anion trypsin modified L-l-tosylamide-z-phenylethyl chloromethyl ketone xiv INTRODUCTION Hexokinase catalyzes the conversion of Glc and Mg-ATP to Glc-6-P and Mg-ADP. This phosphorylation of Glc represents the first step in a number of metabolic pathways, including glycolysis and the hexose monophosphate shunt. In mammalian brain tissue, where Glc represents the principal energy substrate (1), hexokinase assumes a major role in metabolic regulation (2). Rat Type I hexokinase (ATP:D-hexose 6- phosphotransferase, EC 2.7.1.1), the predominant isozyme in mammalian brain, is composed of a single polypeptide chain with Mr 100,000 (3). Studies with limited tryptic digestion of the enzyme suggest that the polypeptide chain is folded into three major structural domains (4). The approximate molecular masses of these domains are, from the N-terminus, 10 kDa, 50 kDa, and 40 kDa. Investigations of other enzymes have shown that discrete structural domains can often be associated with specific functions (e.g.,5,6), and so it is with rat brain hexokinase. Thus, this enzyme specifically and reversibly binds to the outer mitochondrial membrane (7), and the mitochondrial binding function has been localized to the 10 kDa N-terminal domain (8). More recently, the binding site for the nucleotide substrate, 1 ATP, has been shown to be contained within the 40 kDa C- terminal domain (9). A glucose analog affinity label, N-(bromoacetyl)—D- glucosamine (GlcNBrAc), has been used to alkylate a sulfhydryl at the glucose binding sites of rat muscle Type II hexokinase (10) and the Type I isozyme from bovine brain (11); in both cases, derivatization with GlcNBrAc caused a stoichiometric irreversible loss of enzyme activity. A similar alkylating reagent, GalNBrAc, was also reported to function as an affinity label for yeast hexokinase isozyme B, again reacting with a critical sulfhydryl (12,13). In the present work, GlcNBrAc has been used to label the Glc binding site of rat brain hexokinase. Though three sulfhydryls were modified, only one appeared to be critical for activity. Subsequent peptide mapping of the enzyme, derivatized with radiolabeled GlcNBrAc, permitted localization of the glucose binding site in the C-terminal 40 kDa domain previously shown to include the ATP binding site (9). This conjunction of the binding sites for both substrates in the 40 kDa domain makes it reasonable to assign catalytic function to the C-terminal domain of Type I hexokinase from brain. In the present work, three tryptic peptides derived from the rat brain enzyme and containing the GlcNBrAc modified sulfhydryls have been isolated and partially sequenced. Two of these peptides are homologous to regions of the yeast hexokinase sequence which, based on X-ray crystallographic studies (14-16), are located near the glucose binding site of this enzyme. The demonstration of homology in a functional region of these two hexokinases supports earlier suggestions (17-23) that mammalian hexokinase isozymes I-III (Mr 100,000) may have evolved by duplication and fusion of an ancestral gene similar to that coding for yeast, wheat germ, and mammalian Type IV hexokinases (Mr 50,000). The homology in the active site regions of these hexokinases provides a reasonable basis for suggesting that similarities in the secondary and tertiary structure may also exist, and it will be shown that the properties of Type I mammalian hexokinase can be reasonably interpreted in terms of the X-ray structure of the yeast enzyme. LITERATURE REVIEW Yeast Hexokinase. Originally it was believed that there were three hexokinase isozymes in yeast, designated A, B and C in order of elution from a DEAE-cellulose column (24). It was later determined that hexokinase C is a native enzyme which, without proper precautions, is modified toType B during purification (25). The two isozymes are now referred to as A and B. Some authors refer to isozymes A and B as PI and PII, respectively (26). Hexokinase A is composed of two identical alpha subunits. Likewise, two beta subunits comprise the B isozyme (27). The molecular weight of each subunit is approximately 50 kDa. The subunits are dissociated by increased Glc concentrations, ionic strength, pH or temperature (26). Some data had suggested that the enzyme is active in the dimer form (28,29). However, more recent evidence from measurements of the molecular weight of isozyme B under reacting conditions (30) demonstrates that the active enzyme is monomeric. Studies of isozyme B mutants indicate that it is involved in Glc repression (31), a system regulating the repression of various enzymes involved in gluconeogenesis and other catabolic pathways. The mechanism for involvement of the B isozyme in Glc repression is not yet understood, although mutants have been prepared in which the catalytic 4 activity of the enzyme is unaffected, but the ability to regulate Glc repression has been lost (31). The structures of the A and B isozymes have been refined from X-ray crystallography to resolutions of 3.0 A (32) and 2.5 A (14), respectively. A high degree of structural similarities between the two isozymes is evident (33). The predominant structural feature of yeast hexokinases is a cleft which separates the enzyme into two lobes (34), a property shared with several other kinases (35). The locations of the Glc (15,16) and ATP (33) binding sites inside the cleft has also been determined. In the B isozyme several residues have been identified as essential for binding Glc (15,16). The ATP binding site has been less accurately elucidated (33). The structure of the dimer has also been determined, showing the subunits in a nonsymmetrical arrangement (36). The amino acid sequences for both isozymes, as determined from the nucleotide sequence of the cloned cDNA, has recently been completed (37,38). Both isozymes contain 485 amino acids, of which 76% are homologous, a figure lower than isozymes for many other glycolytic enzymes (38). These nonhomologous positions are located in discrete clusters of the polypeptide chain (37,38). A refined three dimensional structure of yeast hexokinase, integrating the recently determined sequence data with the X-ray crystallographic structures, is in progress (37). Mammalian Hexokinases. Four different hexokinase isozymes are found in mammalian tissues (39,40). The isozymes are designated I-IV in order of increased electrophoretic mobility toward the anode of a starch gel (41). Alternatively, the isozymes are also referred to as A-D (39), corresponding to the same order as I-IV. There are several properties which make isozymes I—III distinct from the Type IV isozyme. The KG for Glc for Types I-III is much lower (10'4 to 10'6 M) than that observed with Type IV (10"2 M) (39-41). Each of the isozymes is composed of a single polypeptide chain, but the molecular weight of Type IV, approximately 50 kDa, is half that of Types I-III (reviewed in (7)). The isozymes also differ in their tissue distribution (7). Type I is the predominant isozyme in the brain, erythrocytes, kidney, lung, spleen and heart tissues. Skeletal muscles contain predominantly the Type II isozyme, while Type III hexokinase exists only in minor amounts in several tissues. Type IV is the primary hexokinase in the liver. Amino acid composition comparisons between hexokinases from yeast and mammalian sources indicate it is likely that they share a common evolutionary origin and are structurally similar (21,42). Recent demonstrations of antigenic cross reactivities between all four isozymes in mammals (42,43) also indicate structural homologies. Hexokinase isozymes I-III have four major functions: catalysis, regulation of catalysis by Glc-6-P, reversible binding to the outer mitochondrial membrane, and regulation of mitochondrial binding by Glc-6-P. Although several aspects of the mechanism of catalysis are unknown, a ternary complex between the enzyme, hexose and Mg-ATP is generally agreed upon (7). Glc-6-P is also a potent inhibitor of activity (17). This inhibition is believed to be mediated through an allosteric regulatory site (44), although some authors do not rule out the possibility that Glc-6-P exerts its influence by competition between its phosphate and the gamma-phosphate of ATP for the same binding site (45). Reversible binding to the outer mitochondrial membrane, presumably at porin (46), the pore-forming protein, gives hexokinase preferential access to ATP generated in the mitochondria (47,48). Glc-6-P solubilizes the enzyme from the mitochondria (49), which decreases the affinity of the enzyme for ATP (50,51) and increases its affinity for Glc-6- P (51). Thus, high levels of Glc-6-P induce a less active enzyme species, as well as directly inhibiting the enzyme. This is believed to be a major mechanism of hexokinase regulation in vivo (52). Because the characteristics of Glc-6-P inhibition and the ability of Glc-6-P to solubilize the enzyme from the mitochondria are similar in many respects (7), a single Glc-6-P binding site for both functions seems likely. Hexokinase can exist in several discrete conformations which are induced by the binding of various ligands (53—55). These conformations have been roughly compared by determining the protection afforded by each conformation against inactivation of the enzyme by chymotrypsin, glutaraldehyde, heat and DTNB. These comparisons indicate that the suitability of carbohydrates to serve as substrates or inhibitors depends on their ability to induce necessary conformational changes (53). For example, mannose induces the same conformation as glucose and it is also phosphorylated by the enzyme when Mg-ATP is present. GlcNAc, on the other hand, is neither phosphorylated nor able to induce the same conformation as Glc, although it does act as a competitive inhibitor of Glc. Likewise, for a hexose-G-phosphate to inhibit the enzyme it must not only bind to the Glc-6-P regulatory site, but also induce the necessary conformational change. Postulated Evolution of Hexokinases. Because mammalian hexokinase isozymes I-III, Mr 100 kDa, are twice the size of yeast hexokinase and mammalian Type IV hexokinase, it has often been suggested (17-23) that all hexokinases evolved from the same ancestral gene coding for a 50 kDa enzyme. Evolution of mammalian hexokinase isozymes I-III could have occurred after duplication and fusion of a hexokinase gene coding for a 50 kDa enzyme. This would lead to two catalytic sites in the "fused" enzyme, one of which could have evolved into the Glc-6-P allosteric site evident in mammalian isozymes I-III, but not present in yeast hexokinase or mammalian isozyme IV. Circumstantial evidence for this evolutionary relationship is provided by the amino acid composition comparisons discussed earlier (21,42). Analysis of hexokinases using a peptide mapping technique for peptides containing tyrosine also suggested structural homology between yeast hexokinase and rat hexokinase isozymes II and IV. The first direct evidence of homology between yeast and mammalian hexokinases was provided recently by Marcus and Ureta (56), who determined the amino acid sequence of seven peptides from rat Type III hexokinase, several of which showed significant homology to yeast hexokinase. Gene fusion is believed to have occurred in the evolutionary development of several other proteins. A series of fusions between genes coding for smaller, distinct enzymes is believed to have led to the variety of fatty acid synthases (57) apparent in several organisms, as suggested by the partial sequences of these enzymes. The entire amino acid sequence for yeast DNA topoisomerase II has been determined and subsequently its N- and C-terminal halves shown to be homologous to the B and A subunits of bacterial gyrase (58). Similar to the proposed evolution for hexokinases, the amino acid sequence of rabbit muscle phosphorylase, compared with that of a phosphorylase in Escherichia coli, suggests 10 that the rabbit enzyme arose after gene duplication and fusion, with one of the phosphoryl binding sites developing into an allosteric regulatory site (59). Yet another example of this type of protein evolution is found in rabbit muscle phosphofructokinase (60), which has two halves that are both highly homologous to the smaller phosphofructokinase in Bacillus stearothermophilus. This homology has allowed a three-dimensional model of the rabbit enzyme to be constructed from detailed information of the prokaryotic enzyme. Like hexokinase, the mammalian enzyme is regulated by effectors which do not alter the activity of the prokaryotic enzyme. It has been possible to predict the locations of these effector sites in the rabbit enzyme by identifying amino acid substitutions at key positions between the rabbit and the Bacillus sequences. Sulfhydryl Groups in Hexokinases. Reaction of yeast hexokinase B with the alkylating reagent IAM indicated the presence of four sulfydryls (61), which has since been confirmed by sequencing of the cloned cDNA (37,38). Of the four sulfhydryls, two react rapidly with IAM, only above 300 C, and are accompanied by a loss of enzyme activity (61). Glc gave strong protection against inactivation, while Mg-ATP gave less significant protection. Reaction with GalNBrAc, a galactose derived alkylating reagent, labeled the same two sulfhydryls labeled by IAM (12,13). In the presence of Glc, one sulfhydryl, described 11 as critical for activity, could be blocked from reaction with GlNBrAc and enzyme activity was retained. GlcNBrAc, structurally more similar to the enzymes's natural substrate that GalNBrAc, was not an affinity label. The critical sulfhydryl of the B isozyme was identified at a position 20% of the polypeptide length from the C- terminus. Anderson gt g1. (15) identified two cysteines in this region by X-ray crystallography, although in the three dimensional structure of the protein both cysteines were located about 20 A from the Glc binding site. From the X- ray data it was proposed (15) that a cysteine at position 243, of a sequence tentatively defined by X-ray crystallography (14), was in a position to react with GalNBrAc and may be critical for enzyme activity. However, from the recently published amino acid sequences for yeast hexokinase, as determined from the cDNA (37,38), it is now apparent that the amino acid at this position is actually an aspartate. The X-ray sequence (14) and the cDNA sequences (37,38) both show a cysteine at the adjacent position 244 (of the X-ray sequence), but the sulfur atom of this cysteine is located more than 10 A from the bound Glc molecule (15). Therefore, the location of the critical cysteine in yeast hexokinase remains unclear. Several mammalian hexokinases are also inactivated by sulfhydryl reagents. DTNB inactivates Type I hexokinases from rat (62) and bovine brains (63,64), as well as rat hexokinase Type IV (65,66). Glc and Glc-6-P protect either 12 the rat brain or bovine brain hexokinases from inactivation. Glc, but not Glc-6-P, blocks the inactivation of the rat Type IV hexokinase (65). The ability of Glc to protect these hexokinases from inactivation by DTNB suggests a conserved sulfhydryl at their Glc binding sites. Fourteen sulfhydryl groups are present in rat brain hexokinase, all of which are reactive toward DTNB in the protein's native conformation (62). Saturating levels of Glc, Glc-6-P, ATP and Pi each protect a different number of otherwise reactive sulfhydryls. Because the protection afforded by some ligands is substantial (Glc-6-P protects 10 sulfhydryls), it is highly unlikely that the ligands are physically covering the protected cysteines. The protective effects are probably mediated by the conformational changes induced by the ligands. The reaction rates of sulfhydryls with DTNB fall into three distinct categories, which are designated as fast (F), intermediate (I) or slow (S). For enzyme devoid of any protecting ligand, the breakdown of sulfhydryls into the F, I and S categories is 2, 3 and 9, respectively, The rate constant for inactivation is the same as the rate constant for the reaction of the I-type sulfhydryls (62), indicating that it is one of the I-type sulfhydryls (67) which is critical for activity. Similar sulfhydryl reactivities with DTNB are seen in bovine brain hexokinase (63,64). Of the enzyme's 11-13 sulfhydryls, 10-11 are reactive in the unfolded protein. The reactivities of the sulfhydryls were classified into 13 four categories. Only one sulfhydryl was implicated as crytical for activity. Using stoichiometric amounts of DTNB it was also possible to show that the critical sulfhydryl could be oxidized to a second, Vicinal sulfhydryl, forming an intramolecular disulfide bridge. Two such pairs of Vicinal sulfhydryls were demonstrated (64). Sulfhydryl oxidation has also been linked to inactivation of rabbit red blood cell hexokinase (68,69) and hexokinase II in ascites tumor cells (70). Magnani gt al. (68,69) demonstrated that the oxidation of sulfhydryls in rabbit red blood cell hexokinase may regulate enzyme activity in vivo. GlcNBrAc has been sucessfully used as an affinity label for bovine brain (11) and rat muscle Type II hexokinases (10). A single residue, most likely a sulfhydryl, is labeled in the muscle enzyme (10), while two sulfhydryls are labeled by GlcNBrAc in bovine brain hexokinase (11). The presence of Glc protected both enzymes from inactivation. In the bovine brain enzyme Glc protected only the sulfhydryl which was critical for activity, while the reactivity of the second sulfhydryl was unaffected. GlcNBrAc, although it binds to the rat Type IV hexokinase, does not inactivate the enzyme (71). A similar Glc derived affinity label, GlcNBrHex, with an extended alkylating "arm", did inactivate the rat Type IV isozyme (71). Two groups, one presumably a carboxyl and the other undetermined, were labeled. Glc selectively protected 14 against the reaction of the critical residue, assumed to be a carboxyl. GlcNBrHex is also an affinity label for rat Type II hexokinase (71). However, two groups, suggested to be a sulfhydryl and a carboxyl, appeared to be critical. They were the only two residues labeled. A table summarizing the reactions of hexose derived alkylating reagents with several hexokinases is included in the discussion section of Chapter 1. Also included are the results of GlcNBrAc inactivation of rat brain hexokinase. An ATP analog affinity label, 6-mercapto-9- -D- ribofuranosylpurine 5'-triphosphate, has also been used to inactivate bovine brain hexokinase (72). Although it appeared that an enzyme sulfhydryl was labeled, no oxidized enzyme-affinity label product could be demonstrated. Glc, and to a slightly lesser extent ATP, protected the enzyme against inactivation. It was proposed that the affinity label catalyzed the oxidation of one of the same pairs of Vicinal sulfhydryls, containing the critical residue, which was oxidized by DTNB. Sglfhydryl Groups at Enzyme Active Sites. Critical or "essential" sulfhydryls at binding sites of enzymes are commonly reported for many enzymes, as is indicated by a sampling of such reports during the first half of 1986 alone (73-76). It would seem that the presence of sulfhydryls at enzyme binding sites is a common occurrence in a wide variety of enzymes. Not as common, but 15 still a frequent occurrence, is the identification of Vicinal sulfhydryls at the binding sites of various enzymes (77-92). On the basis of their proximity to enzyme binding sites, sulfhydryls are often designated as "essential", implying that they play a direct role in substrate binding or catalysis. However, the location of a sulfhydryl at an active site should not be considered proof that the sulfhydryl is essential, as there have been cases were active site sulfhydryls, previously called "essential", have been modified with small blocking groups which leave enzyme activity intact (77,83). It seems likely that the labeling of sulfhydryls at the active sites of many other proteins may cause inactivation by introducing a steric bulk which the catalytic site cannot accommodate, rather than by chemically blocking the activity of an essential residue. Therefore, residues mentioned in this text whose chemical modification leads to enzyme inactivation will be termed "critical", unless there is evidence that the residue participates directly in catalysis and can be termed "essential". Some enzymes, such as papain or thiotransferases, have active site sulfhydryls whose roles are unambiguous. However, it would seem that the body of enzymes reported to have sulfhydryls at their active sites is much larger than the number of enzymes for which the roles of the sulfhydryls are understood. Given the susceptability of sulfhydryls to 16 oxidation, it would seem that their presence at enzyme active sites would be detrimental to enzyme activity. This would be especially true for enzymes possessing Vicinal sulfhydryls at their catalytic sites. That sulfhydryls do, nevertheless, often occur at catalytic sites suggests that there may be a common reason, not yet understood, for their presence. However, the evidence of functional roles for these sulfhydryls is limited. It has been proposed (84) that the oxidation or reduction of the sulfhydryls may regulate the activity of some enzymes, such as phosphofructokinase (84) and 1,6-diphosphatase (85), both of which are key enzymes of glycolysis and gluconeogenesis. Sulfhydryl oxidation has also been proposed to regulate the activity of other enzymes (86-88). MATERIALS AND METHODS Materials TPCK-treated trypsin and D(+)-glucosamine hydrochloride were purchased from Sigma Chemical Co. (St. Louis, MO). Staphylococcal aureus V8 protease was obtained from Miles Research Laboratories (Elkhart, IN). Bromoacetic anhydride was a product of Pfaltz and Bauer (Stanford, CT). D- Glucosamine hydrochloride [1-14C] and iodoacetamide [1-14C] were purchased from ICN Pharmaceuticals, Inc. (Irvine, CA). Safety-Solve liquid scintillation cocktail was a product of Research Products International Corp. (Mount Prospect, IL). Centricon 30 microconcentrators were obtained from Amicon Corp. (Danvers, MA). Molecular weight standards used for SDS-PAGE were from Bio-Rad Laboratories (Rockville Centre, NY). Ultra pure urea was a product of Schwarz/Mann (Cambridge, MA). HPLC grade ammonium acetate and Scinti Verse LC liquid scintillation cocktail were obtained from Fisher Scientific Co. (Fairlawn, NJ), and DC-Plastikfolien cellulose plates were purchased from American Scientific Products (Romulus, MI). Model 1750 sample cups were a product of Isco, Inc. (Lincoln, NE). 17 18 We Pupification of rat brain hexokinase. Rat brain hexokinase was prepared according to Polakis and Wilson (4). Protein and hexokinase assays. A molar extinction coefficient of 5.1 x 104 cm'1 M'1 at 280 nm was used to determine hexokinase concentrations (3). Hexokinase activity was coupled to Glc-6-P dehydrogenase activity and measured spectrophotometrically as described by Polakis and Wilson (89). Trypsin solutions were freshly prepared in 1 mM HCl and the concentration determined from absorbance at 280 nm based on an absorbance of 1.43 for a 1 mg/ml solution (90). Synthesis of N-(bromoacetvl)~D—glucosamine. GlcNBrAc was prepared from D-glucosamine and bromoacetic anhydride as described by Otieno g; a1. (12). Analysis by thin layer chromatography on silica gel G, with methanol-acetone (1:10, v/v) as solvent, gave a single symmetrical component at Rf 0.5. The melting point was 141- 1450 C with decomposition. [14C]-GlcNBrAc was prepared from [1-14C]-D-glucosamine and had a specific activity of 0.53 mCi/mmole. Reacpion of GlcNBrAc with hexokinase. Purified hexokinase was chromatographed on a Sephadex G—25 column equilibrated with 50 mM Tris-HCl, 0.5 mM EDTA, -pH 8.85, freeing the enzyme from Glc, P- 1' and thioglycerol 19 that were used to maintain enzyme activity during storage. When different buffers were used, this is indicated in the text. Prior to column equilibration, the buffer was degassed and purged with NZ to remove oxygen. After trypsin modification of native hexokinase (see below), a similarly equilibrated Sephadex G-75 column was used for removal of the protective agents, noted above; this also effected the removal of trypsin, which was retarded by the G-75 column. When necessary, samples were concentrated using a Centricon 30 microconcentrator. Inactivations were done at room temperature. GlcNBrAc was used at 1mM unless noted otherwise. Other ligands were used, where indicated, at the following concentrations: 10mM Glc, 10 mM GlcNAc, 1.0 mM Glc-6-P, 5.0 mM ATP and 5.0 mM Mg- ATP; these concentrations are at least 10-fold higher than the Kd values estimated for these ligands (7). When other ligands were used, their concentration is indicated in the text. Control samples (no GlcNBrAc) lost no more than 5% of the original activity during the time course of the experiments. Reactions were quenched either by addition of 10 mM 2-mercaptoethanol or by adding an aliquot of the reaction to a larger volume of reaction buffer which included 10 mM 2-mercaptoethanol, followed by incubation for at least 15 min to ensure complete reaction of residual GlcNBrAc. 20 Measurement of incorporation of fl£Cl-GlcNBrAc. At various stages of inactivation, aliquots were removed from the reaction mixture and quenched prior to addition to bovine serum albumin-coated polypropylene centrifuge tubes (91) containing 0.5 ml cold 10% (w/v) trichloroacetic acid. Bovine serum albumin (50 ug) was added as a carrier, and the samples centrifuged for 5 minutes in a microcentrifuge. The resulting pellet was washed twice with 0.5 ml cold 10% trichloroacetic acid and redissolved in 0.3 ml formic acid. This solution was added to 5.0 ml Safety-Solve scintillant for scintillation spectrometry. Samples were counted in a Packard 300 CD scintillation counter using an external standard and a quench curve to determine efficiencies. Trypsin modification of native hexokinase. This was done as previously described (4). Hexokinase was incubated with trypsin in 0.1 M sodium phosphate, 0.1 M Glc, 0.01 M thioglycerol and 0.5 mM EDTA, pH 7.0, at room temperature. The protease:substrate ratio was 1:5 (w/w) unless noted otherwise. Proteolysis was terminated by addition of 1.0 mM PMSF, followed by incubation for at least 15 minutes at room temperature. This trypsin modified native hexokinase is referred to as "tm-hexokinase" throughout this paper. Izypsin modificapion of [liQJ-GlcNBLAc labeled hexokinase. After reaction with GlcNBrAc, hexokinase samples were diluted 40 fold with the same pH 7.0 buffer used for trypsin 21 modification of native hexokinase. Samples were concentrated in a Centricon 30 microconcentrator to 50-100 ul. Digestion by trypsin was done as described above. LliCJ-IAM labeling of denatured tm-Hexokinase. Tm-hexokinase in 50 mM Tris, 0.5 mM EDTA, 1% SDS, pH 8.0, was incubated for 30 min with 5 mM [14C]-IAM. The reaction was quenched by addition of 10 mM 2- mercaptoethanol. The sample was subsequently run on an SDS- PAGE gel as described below. SDS-PAGE and fluoroqraphy. Both one and two dimensional separations were performed as described by Polakis and Wilson (4). Gels to be ' fluorographed were soaked in water for 30 minutes, followed by 30 minutes in 1.0 M sodium salicylate prior to drying the gels (92). Densitometric scanning of stained gels or fluorographs was done using a Gelman ACD-18 densitometer, which also integrated peak areas. Molecular weight standards (see Fig. 9) were lysozyme (14,400), soybean trypsin inhibitor (21,500), carbonic anhydrase (31,000), ovalbumin (45,000), bovine serum albumin (66,200) and phosphorylase B (92,500). Based on these Standaznds, the Mr of native hexokinase is approximately 115.000, somewhat larger than the value of 98,000 repeatedly f0und in previous work from this laboratory (e.g. 3,4,93) using both SDS-PAGE as well as other methods. As noted in the Introduction, and reviewed in (7) , many other investigators have also reported molecular weights of about 22 100,000 for mammalian hexokinases. We have no explanation for this recently encountered discrepancy.‘ However, for the sake of continuity, the molecular mass values for the 57 kDa, 67 kDa, and 33 kDa bands in Fig. 10, lane D, were calculated using the bands in lane C as standards with molecular weights as previously reported (4). Immunoblotting. The previously published method (4) was followed with the exception that the secondary antibody was used at a 1:3,000 dilution. Also, the blots were developed in the dark with a solution prepared immediately before use by adding 10 ml of ice-cold 0.3% (w/v) 4-chloro—1-naphthol in methanol to 50 ml of 0.015% (v/v) H202 in 20 mM Tris-HCl, 0.5 M NaCl, pH 7.5. The peroxidase reaction was quenched after 30-45 min by washing the filter in 0.02% (w/v) sodium azide. Nonlinear regresslop analysls. Data were analyzed by a Gauss-Newton algorithm (94) using a computer program written and kindly made available by Dr. S. Brooks (Biochemistry Dept., Michigan State University). Reaction of DTNB with hexokinase. Hexokinase was reacted with equimolar amounts of DTNB in 50 mM Tris, 0.5 mM EDTA, pH 8.5. DTNB stock solutions were prepared in the same buffer and their concentration determined by reacting an aliquot of the solution with an excess of 2-mercaptoethanol in the pH 8.5 buffer. The 23 concentration of the TNB anion was measured using an 1 cm'1 at an absorbance of extinction coefficient of 13600 M- 412 nm (95). Hexokinase oxidized by DTNB was assayed as described above, with the exception that thioglycerol was excluded from the assay solution. Reduction and carboxymethylation of [l3C]-GlcNBrAc labeled hexokinase. This was performed essentially as described by Crestfield g; pl. (96). Additional 2-mercaptoethanol was added to the quenched reaction solution to provide a final concentration that was loo-fold in excess of the hexokinase sulfhydryl content. Solid urea and a stock solution containing Tris-HCl and EDTA were added to yield final concentrations of 8.0 M urea, 0.2 M Tris-HCl, 5.0 mM EDTA, pH 8.5. After incubation for 30 minutes at 37° C, iodoacetamide was added in an amount stoichiometrically equivalent to the 2-mercaptoethanol, and incubation continued for an additional 15 min. Exhapstlve tryptic digestion of [AACJ-GlcNBrAc-labeled hexoklnase. The [14C1-GlcNBrAc-labeled and carboxymethylated-enzyme was extensively dialyzed against 1% ammonium bicarbonate, pH 7.8. Precipitated protein was redissolved by addition of solid urea to a final concentration of 8.0 M. The solution was diluted with a three fold excess of 1% ammonium bicarbonate and 10 mM glycylglycine, pH 7.8; the protein did not reprecipitate after dilution to 2.0 M urea. Trypsin was 24 added immediately at a trypsin to hexokinase ratio of 1:5 (w/w), and the sample incubated with stirring at 370 C for 45 minutes. Proteolysis was terminated by adding two drops of glacial acetic acid for each ml of digestion volume. Cyanate formed spontaneously in urea solutions (97) can modify the side chains of several amino acids, including lysine (98). Therefore, glycylglycine was added as a scavenger of cyanate (98), and a high trypsin:hexokinase ratio was used to keep the digestion time, and therefore possible cyanate modification, to a minimum. Primary HPLC of tryptic peptides. The digestion mixture was dried on a rotary evaporater with careful attention to prevent sudden degassing of C02. The dried residue was redissolved with 0.1% TFA to a final urea concentration between 4.0 and 8.0 M. The solution was centrifuged for 5 minutes at 38,000 x g prior to HPLC. The peptides were separated on an Altex Ultrasphere-ODS reverse phase column (4.6 mm 1.0. x 25 cm), monitoring absorbance at 214 nm with a Beckman Model 160 detector. Solvent A was 0.10% TFA in water and solvent B was 0.09% TFA in acetonitrile. The column was equilibrated in 5% solvent B and the peptides were eluted with gradients as indicated in Fig. 1 at a flow rate of 1.0 ml/min. Radioactivity in the effluent, after exiting the absorbance monitor, was detected by a Flo-One Model HS radioactive flow detector (Radiomatic Instruments and Chemical Co., Inc., Tampa, FL) using Scinti 25 Verse LC liquid scintillation cocktail at a flow rate of 3.0 ml/min. Secon ar HPLC of HW. Peptides were collected in silanized polypropylene centrifuge tubes immediately after exiting the UV detector, and lyophilized. Peptides were further purified by HPLC with the following changes from the above procedure. Solvent A was 5.0 mM ammonium acetate in water, pH 6.0, and solvent B was acetonitrile. Peptides I, II and III (see Fig. 1) were redissolved in 5%, 15% and 30% solvent B in solvent A, respectively, and loaded onto a column equilibrated with the same solvent mixture. An Altex Ultrasphere-I.P. reverse phase column (4.6 mm I.D. x 25 cm) was used for isolation of Peptides I and II, while the Altex Ultrasphere ODS column was used for Peptide III. Peptides were eluted with gradients as indicated in Fig. 2. Peppide mapping of the [lAQJ-GlgNBLAc labeled 49 kDa domaip. Tm-hexokinase (0.24 mg) was labeled with [14CJ-GlcNBrAc as described above. For subsequent tryptic digestion and separation of the peptides by HPLC it was necessary to reduce and carboxymethylate the protein at this point, also as described above, prior to separating the fragments by SDS-PAGE. The carboxymethylated protein was layered across the top of a SDS-polyacrylamide gel for preparative separation of the protein fragments. After electrophoreisis the protein bands were visualized by immersing the gel in cold 0.2 M KCl (99). The 40 kDa band was excised. A 26 representative vertical slice of the gel was stained in the usual manner to verify that the 40 kDa band had been correctly identified. The excised 40 kDa band was electroeluted from the gel slice using a Model 1750 sample cup. The buffer chambers were filled with the same Tris- glycine buffer used for SDS-PAGE, as referenced above. Electroelution was performed for four hrs at three watts. The 40 kDa protein was lyophilized and redissolved in 75 ul of water. SDS was removed by an ion pairing technique (100) using acetone:triethylamine:acetic acid (90:5:5). The precipitated protein was redissolved with 8.0 M urea and an aliquot counted for 14C content to determine the amount of hexokinase present for subsequent digestion with trypsin. After dialysis against 1% ammonium carbonate, the protein was digested by trypsin in 2.0 M urea and the peptides separated by HPLC as before. Amipp acid analyses and peptide seqpencing. Twenty-four hour hydrolysis and analysis of PTC- derivatized amino acids were performed by Dr. Y. M. Lee of the Macromolecular Structure Facility, Michigan State University, East Lansing, MI. A similar hydrolysis and amino acid analysis of underivatized amino acids was performed by Mr. Alan Smith of the Protein Structure Laboratory, University of California, Davis. Automated Edman degradation of peptides were performed by both facilities. The facility which performed a given analysis 27 or sequence determination are as indicated in the appropriate tables. Mass spectrometry of peptides. The molecular weight of Peptide III was estimated by fast atom bombardment mass spectrometry. Mass spectral data were obtained from the Michigan State University Mass Spectrometry Facility supported by a grant (RR-00480) from the Biotechnology Resources Branch, Division of Research Resources, NIH. Peptide III, 2-3 nmoles, was dissolved in a 5:1 dithiothreitol:dithioerythritol matrix. We are grateful to Dr. John Stults for assistance in this part of the work. Eggparatlon of N-carboxymethyllysine. Polylysine was alkylated by iodoacetate as described by Gundlach pp pl. (101) with the exception that the pH was maintained manually at 10.0 rather than with a pH stat. The reaction time was extended from 3.0 to 3.5 hours. The carboxymethylated polylysine was hydrolyzed for 24 hours and derivatized to the PTC amino acid prior to amino acid analysis. A single novel peak, corresponding to N- carboxymethyllysine, eluted at a position between the PTC derivatives of glutamate and S-carboxymethylcysteine. Preparation of S-[l5CJ-GlcNAc-cysteine. [14CJ-GlcNBrAc (50 nmoles) and cysteine (100 nmoles) in 15 ul of 66 mM Tris-HCl, pH 8.5, was allowed to react at room temperature for 2 hours. The reaction product was purified by thin layer chromatography on DC-Plastikfolien cellulose plates with ethanol-water (7:3, v/v) as solvent. 28 [14C]-GlcNBrAc had an Rf of 0.6, as determined by scanning the plate for radioactivity, while cysteine was detected by ninhydrin at an Rf of 0.4. A single ninhydrin positive and radioactive peak was detected in the reaction mixture at Rf 0.2. A scan of the radioactivity on the plate indicated a 90% conversion of [14c1-clcNBrAc to S-[14CJ-GlcNAc-cysteine. The product was eluted from the the plate with 200 ul of water for a final yield of 60%. Preparation of S-[li J-GlcNAc-glutathione and S-[lic - agetamido-glutathlone. Forty nmoles of glutathione were reacted with either 20 nmoles of [14CJ-IAM or [14C]-GlcNBrAc. Both reagents were 0.53 uCi/umole. Reactions were in 20 ul of 20 mM Tris, pH 8.5 for a duration of 20 min. The reaction mixtures were separated by HPLC using an Altex Ultrasphere-Octyl reverse phase column (4.6 mm I.D. x 15 cm). The column was equilibrated with 0.10% TFA in water at a flow rate of 1.0 ml/min. The reactants, glutathione, iodoacetamide and GlcNBrAc, eluted 5.5, 4.5 and 3.0 min after injection, respectively. Tris was not retained by the column and eluted at 2.0 min. Oxidized glutathione did not elute from the column unless an acetonitrile gradient was used. The l4C]-iodoacetamide or reaction products of glutathione with [ [14CJ-GlcNBrAc eluted at 4.5 and 3.5 min, respectively. Both products were hydrolyzed, derivatized to the PTC amino acid and subjected to amino acid analysis as described above. CHAPTER I LOCATION OF THE GLUCOSE BINDING SITE GlcNBrAc was evaluated as an affinity label for rat brain hexokinase by several methods of kinetic analysis. The results were consistent with the label binding reversibly to the substrate Glc site prior to inactivating the enzyme. The number and general locations of sulfhydryls labeled by GlcNBrAc were also determined. RESULTS GlcNBrAc Inactivation of Hexokinase in the Presence of Various Ligands. Reaction of rat brain hexokinase with GlcNBrAc causes pseudo-first order loss of enzyme activity (Fig. 1). Saturating amounts of Glc, Glc-6-P and GlcNAc give differing degrees of protection against inactivation, while ATP and MgATP slightly increase the rate of inactivation (Fig. 1). It was expected that Glc would give excellent protection against inactivation by GlcNBrAc. It is unclear why Glc-6-P, which is believed to bind to an allosteric regulatory site and not the Glc substrate site, would also give excellent protection. Yet protection by Glc-6-P against GlcNBrAc inactivation was also seen with the bovine Type I (11) and the rat Type II (10) hexokinases. It is unlikely, based on studies of substrate and inhibitor specificities (53,44,102) that GlcNBrAc would bind to the Glc-6-P regulatory site. It is also very unlikely that Glc- 29 30 Figure 1. Protection by ligands against inactivation of hexokinase by GlcNBrAc. Activity was monitored as a function of time after addition of 1 mM GlcNBrAc, and with no added ligand or with addition of the various ligands indicated beside the corresponding lines. Concentrations of added ligands are given in Methods. % Original Activity Figure 1 31 \No Ligand '\ Mg-ATP O \ ATP Time (min) 32 6-P competes with GlcNBrAc for the Glc substrate site, as Glc-6-P is a noncompetitive inhibitor yg. Glc (103). It is more likely that Glc-6-P protects against inactivation by virtue of the marked conformational change it induces (53,54) resulting in exclusion of ligands bearing bulky substituents at the 2-position from the hexose site. This would be consistent with previous results demonstrating synergistic binding of Glc-6-P and hexoses such as Glc, but no such effects with Glc-6-P and GlcNAc (53,54). That Glc-6—P protects against inactivation by virtue of the conformational change it induces, rather than by binding at the site of GlcNBrAc mediated inactivation, was tested by using other hexose-6-phosphates in conjunction with GlcNBrAc. Three hexose-6-phosphates, Gal-6-P, Man-6-P and 2-deoxyglucose-6-P, which bind to the same allosteric site as Glc-6-P, but do not induce the same conformation or inhibit hexokinase activity, were tested for their ability to protect against GlcNBrAc inactivation. Also tested was 1,5-anhydroglucitol-6-P, which is similar to Glc-6-P in respect to inhibition and ability to induce the same conformational change. Thus, if Glc-6-P protection is mediatated through the conformational change it induces, and not through its binding to the enzyme, then 1,5- anhydroglucitol-6-P would be predicted to also protect against inactivation by GlcNBrAc, while Gal-6-P, Man-6-P and 2-deoxyglucose would have little or no protective effects. This was seen to be the case (Fig. 2). Therefore, it is 33 Figure 2. Protection by various hexose-6-phosphates against inactivation by GlcNBrAc. This was performed as in Fig. 1, except that 50 mM glycylglycine, 0.5 mM EDTA, pH 8.5 was used as the buffer. Glc-6-P, 1,5-anhydroglucitol-6-P, and 2-deoxyglucose-6-P are indicated by the open circles, squares and triangles, respectively. The closed circles, squares and triangles represent no ligand used, mannose-6-P, and galactose-6-P. All ligand concentrations were 1 mM, as used previously in this laboratory (53). 34 22568:. ow oc m 659“. MgAnav IBU!5!JO % 35 possible to conclude that GlcNBrAc and Glc-6-P do not compete for the same binding site on rat brain hexokinase. A similar experiment was also conducted with hexoses other than Glc. Here it might be expected that hexoses which bind to hexokinase, regardless of whether they induce the same conformational change as Glc and can serve as substrates, would all give equal protection. However, this was not the case (Fig. 3). GlcNAc (also in Fig. 1) and fructose, which are competitive inhibitors yg. Glc, but do not induce the same conformational change, did not perform as well as Glc. While GlcNAc still provides a considerable degree of protection, the protection afforded by fructose is only slightly better than that given by Gal-6-P, Man-6-P and 2-deoxyglucose-6—P (Fig. 2). The concentration of GlcNAc used, 10 mM, was shown to be saturating by repeating the experiment with 100 mM GlcNAc, which did not increase protection against inactivation (results not shown). Glc and 2-deoxyglucose, which induce the same conformational change as Glc, give excellent protection against inactivation, as expected. The experiment in Fig. 3 is therefore a poor demonstration that GlcNBrAc inactivation is mediated through the Glc substrate site. Strong evidence that the Glc site is indeed targeted by GlcNBrAc will be presented later in ‘this chapter. The results from Fig. 3 do indicate, however, tlmat the excellent protection afforded by Glc is due not Ortly to competition with GlcNBrAc for the same binding site, 36 Figure 3. Protection by various hexoses against inactivation by GlcNBrAc. This was performed as in Fig. 1, except that 50 mM glycylglycine, 0.5 mM EDTA, pH 8.5 was used as the buffer. Glc (10 mM), 2-deoxyglucose (100 mM) and fructose (100 mM) are indicated by the open circles, triangles and squares, respectively. The closed circles, triangles and squares represent no ligand used, GlcNAc (10_mM) and mannose (10). All ligand concentrations are at least 10 fold higher than the Km values estimated for these ligands (102). 37 m 659”. see as: . om— om CV 28: oz /. 21 /a a../ o 46 KG KI ks _ ——____1 ¢——————- - . 1 E Glc,r_‘r___ E _____7_, E I-—————+E Glc I I k"n I kn I k'n E' E' E' Based on the latter scheme, the pseudo-first order rate constant for inactivation by GlcNBrAc in the presence of Glc is given by the following equation, analogous to equation 2: (5) kapp = k"nKI[I][Glcj/KG + kn KI[I] + k'n[I]2 + ks[I] [I] + KI(1 + [GlCJ/KG) where k"n is the second-order rate constant for nonspecific inactivation of the enzyme-Glc complex, and all other constants are as defined earlier. Inactivation of the enzyme in the presence of varying concentrations of Glc was analyzed, fitting the data by nonlinear regression to the reciprocal form of equation 5 (Fig. 6). The values of KI' kn, k'n and k5 were those obtained from inactivation in the absence of Glc and using equation 2, as indicated above. Best fit curves from three determinations gave mean values :50 of 0.14 :0-05 mM for KG and 0.0016 :0.0001 min'1 mM"1 for k"n° The value for KG obtained in the present work is ,similar to KG values reported previously, which are in the 47 Figure 6. Protection by varying concentrations of Glc against inactivation by GlcNBrAc. Hexokinase was inactivated by 0.5 mM GlcNBrAc. The line was determined by regression analysis of the data using the reciprocal of equation (5), given in k' the text. The values of KI, k and k5 used in the Il’ n equation are those determined from the plot in Fig. 4. The values :SD for KG and k"n determined from this set of data were 0.078 i0-004 mM and 0.0015 10.0002 mm'l min-1. 48 0.9 Cd ass: mom; ea .44 law m 959“. .66“ neee. eeex\_ .56? 1 com 49 range of 0.04-0.25 mM (54,106); the variation in the reported values may reflect the different methods used for determination of KG' It should also be noted that the present results were obtained at pH 8.85 rather than at pH 7-7.5 as in the earlier work. In this range pH appears to have little influence on the binding of Glc; Solheim and Fromm (107) have reported that the kinetic parameters of bovine brain hexokinase, including the Km for Glc, are not influenced by pH in this range. The value of k"n is considerably less than values of RD and k'n, given above. Therefore, binding of Glc appears to offer appreciable protection against nonspecific inactivation of the enzyme. It is likely that the inability of GlcNAc to fully protect against inactivation is because it can only protect against one "route" of inactivation, the .specific route. Thus, binding of GlcNAc blocks GlcNBrAc from binding to the enzyme and inactivating the enzyme via the specific route, while it does not protect against nonspecific inactivation. If this were the case, then it would be predicted that the rate of inactivation in the presence of GlcNAc would be the same as that determined to arise from nonspecific inactivation by GlcNBrAc in the absence of any ligand. The apparent pseudo-first order rate constant for inactivation in the presence of GlcNBrAc had a mean value, from three determinations, of 0.013 10.002 min'l. This is in close agreement with the apparent pseudo- 50 first order rate constant for nonspecific inactivation, 0.0092 10.0012, determined from k'n at 1.0 mM GlcNBrAc. This ability of Glc to induce a conformation change which "buries" a critical residue is in accord with previous results, which demonstrated that binding of glucose consistently provides protection against a variety of inactivating agents, including DTNB (53,54,62). pH Dependence of Inactivation. When the inactivation of an enzyme depends on the deprotonation of a single group, the relation between [H+] and the rate of inactivation are described by the following equation (108): (6) 1 = 1 + [11"] kapp kmax Ka kmax Ka is the dissociation constant for the critical group involved, and kmax is the rate constant for reaction of the dissociated group. Thus, a plot of l/k against [H+] is aPP expected to be linear, with extrapolation to the abscissa providing a value for Ka. Results obtained with the Type II isozyme from rat skeletal muscle (10) and with the Type I isozyme from bovine brain (11) conform to this expectation, and yield 9.1 and 8.9, respectively, as the pKa values for the groups reactive with GlcNBrAc in these two enzymes. However, the results obtained with rat brain hexokinase_were not in accord with 51 those predicted by equation 6; although the data did indeed define a line, the line appeared to extrapolate to the origin, or slightly below. The meaning of this is unclear, but it obviously precludes direct estimation of the pKa by this method. Nonetheless, a plot of the rate of inactivation as a function of pH (inset, Fig. 7) indicates that inactivation of rat brain hexokinase is dependent on deprotonation of a group with an alkaline pKa. The only three sidechains which are reactive with haloacetyl groups and have alkaline pKa's are those of cysteine, lysine and tyrosine (109,110). Of these, the sulfhydryl group of cysteine is the most reactive (111). Thus, a cysteine is likely to be the critical residue labeled by GlcNBrAc. Amino Acid Analysis of GlcNBrAc Labeled Hexokinase. Hexokinase was subjected to acid hydrolysis and amino acid analysis before and after labeling with GlcNBrAc (to 90% inactivation) to determine the nature of the derivitized residue(s). The expected hydrolysis product of cysteine derivitized with GlcNBrAc is S-carboxymethylcysteine (11). Amino acid analyses of two independently prepared GlcNBrAc labeled hexokinase samples indicated the presence of 3 1 1 moles of S-carboxymethylcysteine per mole of protein. The levels of amino acids susceptible to alkylation by haloacetyl groups, including lysine, tyrosine, aspartate, glutamate, and histidine, were not significantly altered by alklyation with GlcNBrAc. -The methionine content of the labeled and unlabeled samples were substantially lower than 52 Figure 7. Effect of pH on the rate of inactivation of hexokinase by GlcNBrAc. In the larger graph, the data are plotted in the format dictated by equation (6), given in the text. The inset shows the pseudo-first order rate constant for inactivation as a function of pH. 53 A 659”. cam 38 llx 54 Table 1. Amino Acid Analysis of GlcNBrAc labeled Rat Brain Hexokinase. Relative moles amino acida Analysis 1b Analysis 2b Amino Previously 5 Acid ReportedC Before After Before After GlcNBrAc GlcNBrAc GlcNBrAc GlcNBrAc ASX 5.36 + 0.11 5.55 5.58 5.65 5.69 THR 2.91 + 0.10 3.04 3.04 3.11 3.13 SER 2.84 + 0.05 3.27 3.20 2.88 2.88 GLX 5.09 + 0.11 5.26 5.25 5.29 5.22 PRO 1.44 + 0.03 1.37 1.35 1.27 1.28 GLY 4.41 + 0.08 4.90 4.81 4.57 4.61 ALA 2.84 + 0.08 3.01 3.03 2.71 2.69 VAL 3.49 + 0.06 3.39 3.58 3.61 3.59 MET 1.96 + 0.06 0.98 0.61 0.73 0.84 ILE 2.70 + 0.05 2.60 2.69 2.83 2.81 LEU 4.86 + 0.11 4.64 4.67 4.93 4.88 TYR 1.15 + 0.03 1.12 1.18 1.11 1.13 PHE 2.31 + 0.41 2.18 2.20 2.30 2.30 HIS (1.00) 1.25 1.25 1.20 1.24 LYS 3.47 + 0.05 3.40 3.44 3.52 3.47 ARG 3.08 + 0.07 2.97 3.01 3.20 3.18 cmCYS and 0.33 0.56 0.42 0.56 METSOZd TOTALa 48.9 48.9 48.9 48.9 48.9 aResults were previously reported as moles amino acid per mole histidine (3,112). Based on the known molecular weight of hexokinase, the presence of 14 moles of cysteine and 4 moles of tryptophan per mole of hexokinase, it was determined that the total number of moles amino acid per mole of histidine (48.9) represented one-eighteenth the total amino acid composition of the enzyme. However, the levels of histidine detected in the current study were considerably higher than those of previous analyses, making it misleading to represent the current results relative to the histidine content. So that the current results can be accurately compared to those of the previous studies, the amino acid compositions presented here are instead relative to one-eighteenth the total amino acid composition. bTwo samples were independently labeled, hydrolyzed and analyzed for amino acid composition. Both were performed by the University of Davis facility. cRef (112). 55 Table 1 (con't). dMethionine sulfoxide and S-carboxymethylcysteine both eluted at the same time, immediately before aspartate. The S-carboxymethylcysteine content of the labeled sample was therefore determined by subtracting the amount of methionine sulfoxide background detected in the unlabeled sample. 56 that detected previously (3,112) and demonstrated considerable variance. This is likely caused by the instability of methionine to acid hydrolysis. Nonetheless, because no unusual peaks were detected, included the positions of elution for N-carboxymethyllysine (101), O- carboxymethyltyrosine (113), and carboxymethylhomocysteine (the breakdown product of the carboxymethyl sulfonium salt of methionine (111)), the residues labeled by GlcNBrAc must therefore be only cysteines. Stoichiometric Relationship Between Incorporation of GlcNBrAc and Inactivation. In Fig. 8, residual activity is plotted as a function of the number of residues modified by reaction with radiolabeled GlcNBrAc. A linear relationship is evident throughout most of the inactivation. Extrapolation of the lines obtained in three such experiments gave a value of 3.4 10.1 GlcNBrAc incorporated at zero activity. This is in good agreement with the number of cysteines modified per hexokinase, as detected by amino acid analysis. Slight additional incorporation into the inactivated enzyme is also suggested by the departure from linearity at low activities. This is not unexpected since previous work (62) has shown that all 14 of the sulfhydryl groups in the enzyme are reactive with the general sulfhydryl reagent, DTNB, and thus may also be expected to be susceptible to alkylation by the bromoacetyl moiety of GlcNBrAc. 57 Figure 8. Stoichiometric relationship between incorporation of radioactivity from [14C]-GlcNBrAc and residual activity. The protein-bound radioactivity at various times during inactivation was determined as described in Methods. Fractional activity remaining (a) is plotted against the number of moles of [14C] -GlcNBrAc reacted per mole of enzyme. The line shows an extrapolation of the linear portion of the data to complete inactivation. The dotted» line describes the curvature expected if the specific and nonspecific modes of inactivation arose from the labeling of two different residues, as described in the text. 58 of 8.9: \ 2.5220 8.95 0.? od ON 0.? m 959“. mud omd mud .. 8.2 59 According to Tsou (114), a linear relationship between residual activity and the number of modified residues indicates that only one of the modified residues is critical for catalytic activity. However, an assumption in the treatment of Tsou (114) is that the critical groups, if more than one, are reacting at similar rates. In the present case, it is evident that both nonspecific and specific inactivation are occurring at appreciably different rates. It was conceivable that if different critical sulfhydryls were modified by the specific and nonspecific paths, curvature in the plot might not be evident because of the substantial difference in their reaction rates. Therefore, the following equation was derived, by a treatment analogous to that used by Tsou (114), describing the modification of two classes of residues reacting at appreciably different rates, and with the possibility of one or more critical groups in each class: 1 a. i+dj i+aj (7) m = n - pa - (n - p)a where m is the number of groups modified (per mole of enzyme) at any given time and a is the fraction of the original activity remaining at this same time, n is the total number of reactive groups, p is the number of groups in the first class (and thus n-p is the number of groups in the second class), i and j are the number of critical groups in the first and second classes, respectively, and at is the 60 ratio of the apparent rate constants for reaction of the second class of residues relative to that of the first.2 In the present case, the first class corresponds to sulfhydryls modified with the same ka as the critical sulfhydryl PP reacting with the bound inhibitor (i.e., the specific path), while the second class corresponds to sulfhydryls reacting at the slower rate of the nonspecific path. The value forcx is calculated using equation 2 by setting either kn = k'n = 0 , to give ka for the the first class, or kS = 0 to give PP kapp for the second class. Using the values for kn, k'n, ks, and KI given above, and at 1 mM GlcNBrAc, (xfi>*(xf‘j) , = Xfi+dj Or Xf = 3.1+“). at (I) and x = al+d3 S The fraction, x, of total reactive groups, n, at any given time will be nx = pr + (n-p)xS where p is the number of groups in the F class. Substituting from equation I, .1. at (II) nx = pai+""j + (n-p)ai+“'j The number of labeled groups per protein, m, at any given time can be determined from x=n-m or xn=n-m Substituting with equation II, or 92 1 at n _ m ___ pa1+°‘j + (n_p)ai+aj 1 a m = n _ pa1+¢j _ (n-p)al+°"3 FOOTNOTES 11h principle, k could be determined directly by observing n the rate of inactivation at very low concentrations of GlcNBrAc, at which inactivation via the "specific" pathway, involving formation of the E-I complex was negligible. However, the very low rates of inactivation under such conditions make this impractical. 2When j = 0, equation 7 simplifies to a form equivalent to equation 5 of Tsou (114). 93 CHAPTER II STRUCTURE OF THE GLUCOSE BINDING SITE Hexokinase labeled with [14C]-GlcNBrAc was exhaustively digested with trypsin and the peptides separated by HPLC. Three major radioactive peptides were found. The sequences of two of these peptides are highly homologous to amino acid sequences of yeast hexokinase which are near the Glc binding site. A serine residue in yeast hexokinase, identified as essential for a major conformational change necessary for catalysis, is also conserved in rat brain hexokinase. Several other structural parallels are made between yeast (and rat brain hexokinases. RESULTS Peptide Mapping of [liCJ-GlcNBrAc-Labeled Hexokinase. HPLC-peptide mapping of tryptic fragments generated from hexokinase labeled with [14C]-GlcNBrAc showed three major radioactive peptides (Fig. 16). Additional digestion with trypsin did not increase the yield of peptides I-III or produce new radioactive peaks. Peptides I and II were both found to have coeluted with one other major component in the primary HPLC separation, and were further purified by HPLC under modified conditions as described in Methods (Fig. 17). In contrast, further HPLC analysis of Peptide III (Fig. 17) showed no substantial contamination and it was subsequently prepared for amino 94 95 Figure 16. Primary HPLC fractionation of a tryptic digest of [14C]-GlcNBrAc labeled hexokinase. Top Panel: Profile of peptides eluted from HPLC column. The dashed line indicates the programmed elution gradient. The elution positions for radioactive Peptides I-III are indicated. Bottom panel: Profile of [14C]-GlcNBrAc labeled peptides eluted from HPLC column. 96 m sz>1_Om .x. insects-see 1111 TIME (min) 60 v F N woz 0 e H m. a m Gov 102 WWW- Of the three sulfhydryls labeled by GlcNBrAc, only one was believed to be critical and to be selectively protected by Glc or GlcNAc. To determine which peptide contained the critical sulfhydryl, enzyme was labeled with [14C]-GlcNBrAc in the presence of either Glc or GlcNAc and subjected to peptide mapping as in Fig. 16. Results were compared to a control sample which had been labeled by [14C]—GlcNBrAc in the absence of a protective ligand. The expectation was that the labeling of one of the peptides, containing the critical sulfhydryl, would be selectively decreased. However, the data from three experiments did not show such unambiguous results. The results from one such experiment are represented in Fig. 19. It is apparent that both Peptides I and III are protected by the inclusion of either Glc or GlcNAc during incubation with [14C]-GlcNBrAc. Labeling of Peptide II was quite insensitive to the addition of protective ligands, clearly indicating that this peptide can not include the critical sulfhydryl within its sequence. These results, as well as those from two other similar experiments, are diagramed in Fig. 20. The protection of both Peptides I and III approximates the level of protection expected if either was to contain the critical sulfhydryl. In every case, however, protection of Peptide III exceeded that of Peptide I. 103 Figure 19. Effects of Glc or GlcNAc on the labeling of Peptides I-III by [14C]-GlcNBrAc. Enzyme was incubated with [14C]-GlcNBrAc in the presence of Glc, GlcNAc or no protecting ligand at all, which served as a control. The reactions were terminated when the control sample was 90% inactivated. Each sample was subjected to peptide mapping in the same manner as the sample in Fig. 16. The control, GlcNAc and Glc protected samples are in the top, middle and bottom panels, respectively. The positions of Peptides I-III are indicated in each panel. 104 I I HI I . I In . l 500- .4 " 11 ‘ S 3 I o 500- ' I? - E 11 1. o._1d11n1mu L.iu1U MM. - 1- fl - 500- I [H - 0.1.1 1 MW- F 1 0 20 40 60 Time (min) Figure 19 105 Figure 20. Correlation between protection by Glc and GlcNAc against [14C]-GlcNBrAc inactivation and the labeling of Peptides I- III. The results displayed in Fig. 19, as well as those for two other such experiments, are represented in graphical form to compare the alkylation of each peptide to the extent of enzyme inactivation. Thus, in three experiments, enzyme was inactivated by [14CJ-GlcNBrAc in the presence of Glc, GlcNAc, or no protecting ligand. The latter sample served as the basis for comparison, with labeling of Peptides I-III and extent of inactivation expressed relative to that seen in the sample with no protective ligands. The reactions were initiated at the same time and terminated when the control sample was approximately 90% inactivated. The [14CJ-GlcNBrAc labeled samples were subjected to peptide mapping as in Fig. 16. The relative labeling of Peptides I-III was determined by comparing the area of the radioactive peak for each peptide to its corresponding peak in the control sample. Peak areas were determined by cutting and weighing. Each data point represents a radioactive peak from a sample protected by either Glc or GlcNAc, as indicated in the lower right hand side of the figure. The diagonal line indicates the 1:1 correlation expected between inactivation of the enzyme and labeling of a critical residue in that peptide. 106 r r I I 1.2 - - I 1.0 - ' - <3 I 2 0.8 - . -' .3: cu A o a. .1 2 Peak :3 III 111 0? GIC A o o " GlcNAc A II o Figure 20 0.6 0.8 1.0 Relative Inactivation 107 Amino Acid Compositions of Peptides I-III. Results of amino acid analyses of Peptides I-III are shown in Table 4. There are no particularly notable features such as relatively high content of acidic, basic, or hydrophobic amino acids. A single basic amino acid, lysine or arginine, would be expected in each peptide 1; tryptic cleavage were complete. This was the case with Peptide I but not with the larger Peptides II and III. Since, as mentioned above, continued digestion with trypsin did not result in further cleavage of these latter peptides, it seems that they may contain a potential cleavage site that is resistant to tryptic attack in the derivatized (and perhaps in the underivatized) protein. Swarup and Kenkare (11) apparently met similar resistance with the bovine brain hexokinase, using prolonged digestion times and still finding evidence of incomplete tryptic cleavage. Since haloacetyl derivatives can react with lysyl sidechains (109,110), modification of a lysine by GlcNBrAc is a possible explanation, although no derivitized lysine was detected after hydrolysis of the labeled protein (Chapter I). Furthermore, none of the peptides gave rise to a hydrolysis product that corresponded to N- carboxymethllysine. Other possible causes of the resistance to tryptic cleavage might include adjacent lysyl or arginyl residues (116), or steric hindrance of the cleavage site resulting from proximity to a sulfhydryl modified by GlcNBrAc. The latter being a causative factor is consistent' Table 4. Amino acid compositions of Peptides I-IIIa. Peptide I II III Asx Glx cmCysC Ser Gly His Arg [ Thr Ala Pro Tyr Val Met Ile Leu Phe Lys Trp O... \DOI—‘O‘QNU’IN A N \OOQkaOUIUIUIOONbI-‘me A .h v KJQtJFJmFJUJHorauiwiahautbcauap U ZOOI-‘OOi-‘OOI-‘OI-‘I-‘I-‘HOHN U U Estimated amino acids/ 17 40 36 peptide aExpressed relative to arginine, which was found in all three peptides in amounts stoichiometrically related to other amino acids present in significant amount. bNumbers in parentheses are nearest integer values. cS-carboxymethylcysteine. Although the cm-Cys contents are low, they are interpreted as indicating the presence of cysteine in the peptide (see text). dNot determined. 109 with the observation that the amount of Peptide III, which is well resolved from other digestion products seen in the 214 nm absorbance profile (Fig. 16), was markedly reduced in digests of enzyme that had been reacted with GlcNBrAc in the presence of a protective ligand, Glc or GlcNAc. This suggests that Peptide III had undergone further digestion in the unmodified enzyme, the mere absence of a GlcNAc moiety being considered unlikely to radically alter the retention time of such a large peptide (i.e., the apparent decrease was pp; simply due to altered retention of the corresponding peptide derived from the unmodified enzyme); if this is indeed the case, then it obviously implies that such digestion was hindered in the derivatized protein. As discussed in Chapter I, GlcNBrAc labels three cysteines in the enzyme. Thus, it is expected that each of the GlcNBrAc-labeled peptides must contain (minimally) one alkylated cysteine. Though S-carboxymethylcysteine, the expected hydrolysis product from cysteine derivatized with GlcNBrAc (11), was found in all three peptides, the yield was much less than this. Since S-carboxymethylcysteine is not a naturally occurring amino acid, its presence in each peptide clearly suggested that cysteine was, in fact, present. When samples of glutathione, labeled with either [14C]-GlcNBrAc or [14C]-iodoacetamide, were subjected to similar amino acid analysis, both derivatized peptides yielded the expected amounts of glycine and glutamate, based on 14C content of the samples prior to hydrolysis, but only 110 0-30% of the expected carboxymethylcysteine (results not shown). Thus the low amounts of S—carboxymethylcysteine detected in these analyses (Table 4) reflect the instability of this compound under the conditions used. Analysis of another preparation of Peptide III using conventional ion exchange methods, i.e., without derivatization to the PTC amino acid, yielded 0.8 moles of S-carboxymethylcysteine per mole of arginine; similar analyses were not performed on Peptides I or II. Mass Spectrometry of Peptide III. A more accurate analysis of the size of Peptide III was determined by Fast Atom Bombardment Mass Spectroscopy. A protonated molecular ion was detected at 3394.4, as well as a diprotonated molecular ion at 1697.8 (1697.2 is half the mass/charge ratio of 3394.4). A fragment ion detected at 3176 corresponds well with loss of a GlcNAc moiety (Mr 220) from the peptide. A peptide of Mr 3176, based on an average molecular weight of 105 per amino acid residue, calculated from the composition shown in Table 4, corresponds to a peptide 30 amino acid residues in length, slightly smaller than that estimated from amino acid analysis (Table 4). Amino Acid Segpepces of Peppides l-lll. Partial amino acid sequences of Peptides I-III are presented in Tables 5-7. As will be discussed below, Peptides I and III show considerable homology to sequences in yeast hexokinase isozymes A and B (37,38) and mammalian Type III hexokinase (56). No significant homology was 111 Table 5. Sequence analysis Cycle Residue 1 Ala (Ser, Gly) 2 Thr 3 Asp 4 —. 5 Glu 6 Gly 7 _ 8 Asp 9 Val 10 Ala 11d Ser 12d Leu of Peptide I.a Yieldb 175 (-°,38) 108 aThe contents of each cycle were analyzed by two non- identical HPLC systems, unless otherwise noted. This analysis was performed by the Michigan State University facility. inelds, when determined, are indicated in pmoles. cThe Ser was not quantitated, but it was a minor component. dCycles 11 and 12 were analyzed by only one HPLC system. 112 Table 6. Sequence analysis of Peptide IIa. Cycle Residue Yieldb 1 Ala 506 2 Ile 367 3 Leu 434 4 Gln 194 5 Gln 146 6 Leu 349 7 Gly 154 8 Leu 321 9 Asn 147 10 Ser - 11 Thr - 12 - - 11: Asp - 14C Asp - 15 Ser - aThe contents of each cycle were analyzed by two non- identical HPLC systems, unless noted otherwise. The analysis listed here was performed by the Michigan State University facility. Cycles 1 - 9 of an identical sample were also sequenced by the University of Davis facility and the same residues were detected at each step. inelds, when determined, are indicated in pmoles. CCycles 13-15 were analyzed by only one HPLC system. 1113 Table 7. Sequence analysis of Peptide III“. Analysis: 1 2 3 d Consensus Cycle AAb Yieldc AA Yield AA Yield 1 Ile 90 Gly (Hot) 548 (-) Hot (Gly) 607 (107) X 2 Pro - Pro 72 Pro 108 Pro 3 Leu 60 Lou (Val) 403 (-) Lou (Val) 504 (90) Leu 4 Gly - Gly 260 my: (Gly) 33: (52) oly‘ 5 Phe 267 Phe 464 Phe 6 Thr - - - Thr 7 Phe 232 Phe 447 Phe 8 Ser - Ser - Ser 9 Phe 187 Phe 378 Phe 10 Pro 151 - - Pro 11 X - - - X 12 His - - - His 13 Gln - clef - Gln 15 Mn - ”n: _ Mn 16 Leu 13 0 Leaf - Leu 17 Asp - Asp 18 X - X 19 Gly - Gly 20 X - X 21 Leu 95 Lou aThe contents of each cycle were analyzed by two non-identical HPLC systems, unless noted otherise. Analyses l and 2 were performed by the University of Davis facility. Analysis 3 was performed by the Michigan State University facility. . bAmino acid. cYields, when determined, are indicated in pmoles. dConsensus for each cycle. 'Although tyrosine was determined to be the major amino acid at this position of analysis 3, it was concluded to be a glycine on the basis of analyses 1 and 2, where no tyrosine was detected at this position, but only glycine. {These cylces were analyzed by only one HPLC system. 114 detected between the sequence of Peptide II and those reported for other hexokinases (37,38,56). In principle, the position of the modified sulfhydryl might be determined by monitoring radioactivity released in each cycle during sequencing of peptides prepared from enzyme labeled with [14C1-GlcNBrAc. However, no significant radioactivity was detected during sequencing of Peptide II. Nor was significant radioactivity left in the cup containing the unsequenced portion of this peptide. Thus it was apparent that the Glcnbrac labeled cysteine was unstable to the rather harsh conditions encountered during the sequencing operation, with the radioactive moiety being lost during one of the wash phases of the cycle. Previous workers (117) have encountered similar problems of stability in sequencing peptides containing cysteine derivatized with a related alkylating agent, 2-bromo-4'-nitroacetophenone. Moreover, when cysteine that had been derivatized by reaction with [14C]-GlcNBrAc was subjected to 5 reaction cycles in the sequencer, no detectable PTH amino acid and less than 5 % of the radioactivity was released as a PTH derivative in any cycle, nor was there significant radioactivity left in the cup. In view of these results, this approach was not pursued with peptides I and III. It is obvious, however, that the position of the derivatized sulfhydryl must correspond to one of the unidentified residues or to unsequenced portions of these peptides. DISCUSSION Location of Peptides I and III within the Overall Sequence of Brain Hexokinase and Comparison with Location of Homologous Peptides in Yeast Hexokinase. In Chapter I it was demonstrated that the sulfhydryl(s) labeled by GlcNBrAc and protected by GlcNAc are located within a 5 kDa region located near the N-terminus of the 40 kDa C-terminal domain of rat brain hexokinase. The present work shows that these sulfhydryls are contained within Peptides I and III. Since, together, Peptides I and III account for a sequence containing approximately 50 residues, it is apparent that essentially the entire sequence in this 5 kDa region must be represented by these two peptides. That Peptides I and III are derived from the same 5 kDa region of rat brain hexokinase is supported by their homology to amino acid sequences in an approximately 5 kDa region of yeast hexokinase (Fig. 21). If Peptides I and III are from the same region of rat brain hexokinase, it raises the possibility that Peptide I may be included in the larger Peptide III. Based on their homology to the yeast enzyme, Peptide I would represent the C-terminal portion of Peptide III. This would explain the similar protection of Peptides I and III against labeling by GlcNBrAc (Figs. 19 and 20). However, the results in Table I do not support this possibility since the alanine content of Peptide I exceeds that of Peptide III and hence the former cannot be contained within the latter. Additional evidence that Peptides I and 115 116 Figure 21. Amino acid sequence comparisons between yeast hexokinase and selected peptides from rat hexokinase isozymes I and III. The sequences are as follows: RI(I) and RI(III) are rat hexokinase isozyme I, Peptides I and III, respectively; RIII(4), RIII(S) and RIII(6) are rat hexokinase isozyme III, peptides 4, 5 and 6 (56); YA and YB are yeast hexokinase isozymes A and B, as determined by sequencing the cloned gene (37,38); YB(X) is the sequence of yeast hexokinase isozyme B, as tentatively determined by X-ray crystallography (14); T,Aand 8 designate amino acids thought to have two, three or four atoms in their side chain. Where the the crystallographic sequence agrees with the actual sequence, or where the correct number of atoms in the amino acid side chain were identified, the residues are underlined. Homologous regions between the yeast and rat isozymes are enclosed in boxes. An X denotes an unidentified residue in the sequence. 117 E 656:. Mo>oz_z_m_._za.oeospree 9 m .m mxm v.04am>>ozom>zm_OmOXhSIO—: <> IAAO>>QOOw>OmO 3:5... 4w<>ox0wx0h< CHE s._1~ox4>~ Nxdo<~sapeoaa>m o<< omega. com; a.omz.xzom 4.0w2_¥20mmuhu04m4h OXhZAaw XQJOhOIwQummEu—Ga Guam _I_ m a s u. o ..-a ..-m o omen—JR Gear 4x0x04-2x02xmumuhu04&x 3:25 118 III were discrete nonoverlapping species came from sequencing. For example, 3 leucines were found in the partial sequence determined for Peptide III, accounting for the 3 leucines seen in the amino acid analysis. Yet amino acid analysis of Peptide I also indicated the presence of 2 leucines which, based on sequencing, were clearly distinct from those of Peptide III. Thus, the sequence of Peptide I cannot be contained within that of Peptide III. Similar considerations also permit the conclusion that there is no overlap of these peptides. Thus, if the C- terminal segment of Peptide III overlapped with the N- terminal segment of Peptide I, it is apparent that the latter segment must contain a basic amino acid, susceptible to tryptic cleavage, that would correspond to the C-terminal residue of Peptide III. Thus, Peptide I would have to contain a lysine or arginine in addition to the basic residue that represents its own C-terminus. This is not the case (Table 4). Therefore, based on their homology with known sequences in the yeast enzymes (37,38), it can be deduced that Peptide III represents the N-terminal region and Peptide I the C-terminal region of this 5 kDa segment of rat brain hexokinase. Examination of the amino acid sequence of the yeast enzymes (37,38) indicates that there are 29 residues from the start of the segment homologous to Peptide III to the start of the segment homologous to Peptide I. This is virtually identical to the 30 residues in Peptide III, 119 estimated from its molecular weight determined by mass spectrometry. It is clear that the overall length of the region represented by Peptides I and III in the rat brain enzyme must be virtually identical to that of the corresponding region in yeast hexokinase. Thus, in rat. brain hexokinase, the nonoverlapping Peptides I and III must be contiguous, or nearly so, with no appreciable insertions or deletions in the sequence determined for yeast hexokinase. The placement of Peptides I and III within the overall sequence of the 40 kDa domain of rat brain hexokinase is shown in Fig. 22, and compared to the disposition of the homologous sequences within the overall sequence of yeast hexokinase. The similarity is striking, and in accord with the postulated evolutionary relationship of these enzymes (17-23) in which the 40 kDa catalytic domain (9, and the present work) of rat brain hexokinase can be considered to have evolved by a process of gene duplication and fusion from an ancestral hexokinase similar to the present-day yeast enzyme. The representation in this figure also suggests that deletion of approximately 80 amino acid residues, corresponding to the N-terminal region of the yeast enzyme, occurred in the course of evolution of the 40 kDa domain from the ancestral form. Because estimation of the number of residues deleted from the N-terminus depends on molecular weights determined by SDS-PAGE, a technique with a 5-10% margin of error (118,119), the figure of 80 120 Figure 22. Locations of Peptides I and III within the 40 kDa domain of rat brain hexokinase compared to the locations of homologous sequences in the yeast isozymes. The horizontal lines represent the relative length of the yeast enzyme and the 40 kDa C-terminal catalytic domain of the rat brain enzyme. N-termini are oriented on the left. The darkened regions represent the homologous sequences between the two proteins, positions 151-195 from yeast hexokinase and Peptides I and III from the rat brain hexokinase 40 kDa domain. Homologous regions are aligned. 121 mmv Xm H 6266a \. ./E 8:81 map Pmp 0 mm 61:9“. cases 0. av .x: 58m Em v.1 Ammo> 122 residues is only an approximation. As will be discussed below, this can readily be interpreted in light of the postulated evolutionary relationship and current knowledge of the structure of yeast hexokinase. Proposed Locations of Peptides I and III within the Tertiary Structure of the 40 kDa Catalytic Domain of Rat Brain Hexokinase. Prior to deduction of the yeast hexokinase amino acid sequence based on the nucleotide sequence of the cloned cDNA (37,38), Steitz and coworkers had attempted to determine this sequence using X-ray crystallographic methods (14). In subsequent discussions of their structural studies of yeast hexokinase, the Steitz group understandably referred to this proposed sequence, then the only one available. However, comparison of the sequence proposed on the basis of X-ray diffraction measurements (14) with that now known from nucleotide sequencing (37,38) discloses numerous discrepancies, some of which may be seen from the alignments shown in Fig. 21; in aligning these sequences, we have made insertions or deletions deemed appropriate for increasing agreement of the sequence proposed from X-ray data with the sequence determined from nucleotide sequencing. It will be apparent from Fig. 21 that the position of specific residues, expressed conventionally beginning from the N- terminus, will be different in the "X-ray" sequence and in the sequence deduced from the nucleotide sequence. This can obviously lead to confusion. In an attempt to minimize that 123 confusion in the following discussion, residue positions in the "X-ray" sequence will be given in quotes, whereas a position numbered without this indication may be considered to correspond to a residue as given in the sequence deduced from the cloned cDNA. The three dimensional structure of yeast hexokinase has been determined with considerable resolution (14- 16,33,36,120). Thus the location of those regions homologous to Peptides I and III can be identified within the yeast enzyme's tertiary structure. This is depicted in Fig. 23. It is immediately obvious that these peptides are in close proximity to the binding site for Glc, as proposed by Anderson e; el. (14-16). That this is true for Peptides I and III themselves (in brain hexokinase) seems likely, considering the identification of these peptides based on their reactivity with a glucose analog. This close similarity in both sequence and function can reasonably be taken to indicate similarity in tertiary structure (121,122). Thus, the structure of the glucose binding site of the 40 kDa domain of rat brain hexokinase is very likely to closely approximate that of yeast hexokinase, as will be discussed later in more detail. Logatlop and Nature of "Cyitlcal" Sulfhydryls in Brain Hexokinase. Results presented in Chapter I indicated that there was a single critical sulfhydryl in rat brain hexokinase that was modified by GlcNBrAc and which was protected against 124 Figure 23. Region of yeast hexokinase active site to which Peptides I and III from rat brain hexokinase are homologous. The structure of the active site was redrawn from Fig. 6 of Anderson ep. pl. (120). The cleft, which runs from the center of the figure to the upper right hand corner, is in the open conformation. The regions which are homologous to Peptides III and I are the solid and cross-hatched regions, respectively. Although the entire length of Peptides I and III were not sequenced, the length of each peptide, as estimated by amino acid analysis or mass spectrometry, is indicated in the figure. The locations of asn-"188", asp- "189", asn-"215", asn-"245", and gln-"277" are indicated. These residues have been identified as playing essential roles in the binding of hexoses (16). The position of ser- "138", a highly conserved residue playing a critical role in substrate binding and conformational changes required for catalysis (see text), is shown, as is the position of residue "155" (R-155), predicted to be in a position to be labeled by an appropriate glucosamine derivative (17). GlcNAc is oriented in the Glc binding site in the same orientation as that determined for Glc (120). The ATP binding site is located at the far right hand side of the figure (33)- 125 / / , P55“\88 ‘ , ’l " III I II I ’II, I, It, I ’l [I I! 1 L_—ll____ Figure 23 126 modification in the presence of Glc or GlcNAc, as had been found with the bovine Type I (11) and rat Type II (10) isozymes. It is evident from the comments above that the location of this sulfhydryl has not been defined by the present studies, and indeed even the existence of a giggle critical sulfhydryl may need to be questioned. Based on its proximity to the Glc binding site, it would seem that modification of a sulfhydryl within the sequence defined by Peptide III would be most likely to markedly affect catalytic activity. Indeed, Bennett and Steitz (16) predicted that an appropriate glucosamine derivative could be used to label the residue at position "155", which is included in the region corresponding to the unsequenced portion of Peptide III. Residue "155" is located on a loop of the polypeptide backbone (Fig. 23) identified as a region of high mobility in yeast hexokinase (16). When the cleft is in the closed position, this loop has moved several A closer in the direction of the bound Glc molecule. With the cleft closed, the fi carbon of residue "155" is located within 5 A of the 2-hydroxyl of the substrate hexose. Given that the bromoacetyl group of GlcNBrAc can rotate on a 4 A arm (123), it is also possible that one of the residues adjacent to position "155" could be labeled. Since this region was predicted to be labeled by a glucosamine derivative, and such a glucosamine derivative labeled a homologous peptide in rat brain hexokinase which contains this region, it is 127 very likely that the derivatized sulfhydryl of Peptide III is located within this region. An explanation of the kinetics of GlcNBrAc inactivation, as discussed in Chapter I, is possible if it assumed that the rat brain hexokinase Glc binding site is as depicted in Fig. 23 and the alkylated sulfhydryl corresponds to position "155" (or possibly an adjacent position). Alkylation of a sulfhydryl in this region of the molecule would block cleft closure, a step considered necessary for catalysis (124), and therefore inactivate the enzyme. Furthermore, the dynamics of the cleft closure have the characteristics described in Chapter I of the conformational change induced by Glc, but not GlcNAc, which makes the critical sulfhydryl inaccessible to alkylation. As mentioned above, binding of Glc closes the cleft and residue "155" is buried next to the bound Glc molecule. This action would protect residue "155" from alkylation by GlcNAc, either by a specific or nonspecific mechanism. The X-ray crystallography data indicated, however, that Glc derivatives with bulky substituents at the number 2 position, such as GlcNAc, would impede cleft closure, and thus block catalysis (16,124). This observation supported previous demonstrations that such Glc derivatives were competitive inhibitors, but not substrates, of yeast hexokinase (125). The same is true for mammalian Type I hexokinases (54,102). With GlcNAc bound at the active site, and the cleft remaining in the open position, it appears 128 that residue "155" would remain open to the solvent and GlcNAc bound to the enzyme would not be in a position to block nonspecific alkylation. Therefore, it seems likely, based on the X-ray crystallography data, the predicted structural homology between yeast and rat brain hexokinase, and the kinetics of GlcNBrAc inactivation, that the critical sulfhydryl in rat brain hexokinase is located at a position analogous to residue "155" in the yeast enzyme. Work is underway in this laboratory to deduce the rat brain hexokinase amino acid sequence based on the nucleotide sequence of the cloned cDNA. Completion of this project would determine the nature of the amino acid at the predicted position for the reactive sulfhydryl in Peptide III, and whether it indeed is the only cysteine in Peptide III. Based on the structure shown in Fig. 23, it is unclear why a sulfhydryl in Peptide I might also be labeled by this reactive substrate analog, with protection by competing ligands such as Glc. In the open cleft form of the yeast enzyme (Fig. 23), this peptide segment is located at an appreciable distance from the postulated Glc binding site, and this remains the case even in the closed form of the enzyme (16). However, this is not unheard of, as there have clearly been other cases when affinity labeling has occurred at amino acids far removed from the targeted binding site (110,126). Alternatively, it may be that there is some variance in the structure in this region of the molecule, 129 and that in the rat brain enzyme the segment represented by Peptide I is in closer proximity to the Glc binding site. If this were the case, then the protective effects of Glc and GlcNAc on labeling of Peptide I would be understandable, and it might be expected that modification of the sulfhydryl in this peptide segment, as in Peptide III, could lead to decreased catalytic activity. If the sulfhydryl in Peptide I is critical for activity, the possibility must be raised that GlcNBrAc labeled two sulfhydryls critical for activity. This suggests that the reasons for concluding there could only be a single critical sulfhydryl should be reevaluated. The previous conclusion that there was a single critical sulfhydryl was based on the assumption that modification of that sulfhydryl totally inactivated the enzyme. l; the modified sulfhydryls in Peptides I and III were not directly involved in catalysis, then it is conceivable that their modification might result in only partial loss of catalytic activity, e.g., by hindering but not totally preventing access to the active site or closure of the cleft. Hence, the conclusion that there was a single critical sulfhydryl might have been incorrect because the assumption on which it was based was incorrect, the relationship between modification and loss of activity being more complex than assumed. From the present work it is apparent that the critical sulfhydryl(s) in rat brain hexokinase can not be the same 130 critical sulfhydryl which is observed in yeast hexokinase. The yeast hexokinase sequence which corresponds to peptides I and III are totally devoid of cysteine. It also explains the inability of GlcNBrAc to inactivate the yeast enzyme (12). In view of the evidence that these hexokinases are fundamentally similar in their structure and catalytic mechanism, it is obvious that this absence of cysteines precludes their direct function in catalysis. This conclusion was reached earlier by Anderson e; el. (15), who, on the basis of X-ray crystallography data, stated that sulfhydryls are unlikely to play a role in substrate binding or catalysis by the yeast enzyme. u -_ o o ... =-twe- e.st a d Ra ; a'n t-xok'nases. Despite the evidence that there is no conserved, essential cysteine in hexokinases, it is possible to make several comparisons between yeast hexokinase and the rat brain hexokinase 40 kDa domain and see that several structural features must be conserved between the two enzymes. The strongest evidence for this comes from the extensive homology between the N-terminal portion of Peptide III and the sequence around serine "138" (serine 157 in the corrected sequence) of the yeast enzyme (Fig. 21). X-ray studies (16) have indicated that Ser-"138" is involved in hydrogen bonding to the 6-hydroxyl group of Glc, and that this interaction is critical for the Glc-induced closure of a cleft necessary for catalysis (124). This serine is present in all hexokinases examined thus far (Fig. 21), both 131 yeast and mammalian, clearly implying functional importance. This is evidence that not only is the secondary and tertiary structure of the immediate region likely to be conserved, but also the predominant structural features of yeast hexokinase: a cleft formed by two lobes of the enzyme connected by a flexible "hinge" region. If indeed it is the predicted C-terminal region of Peptide III which is labeled by GlcNBrAc, then this region of the cleft must also be similar between the two enzymes, as well as the orientation at which Glc is bound in the cleft. Evidence that the ATP binding site in the 40 kDa domain of rat brain hexokinase is also structurally similar to that of the yeast enzyme can be inferred from recent work with the ATP analog affinity label, 8-azido-ATP (9). In yeast hexokinase, the 8-position of the bound ATP is in close proximity to the residue at position "393" (33), i.e., a residue located toward the C-terminal region of the yeast enzyme and thus expected to occupy a comparable position in the 40 kDa domain of the brain enzyme (Fig. 22). As expected, labeling by 8-azido-ATP was shown to occur in this region of the 40 kDa domain (9). Binding of ATP does not involve residues in the regions defined by Peptides I and III (33) and hence would be predicted not to interfere with inactivation by GlcNBrAc, and this has been found to be the case, as discussed in Chapter I. 132 Stppctural Relationships in Monomeric Mammalian Hexokinases Qompaped witp ppe Dimeric Yeast Hexokinase. That mammalian hexokinase evolved from an ancestral gene shared with that in yeast is evident from the present study. Marcus and Ureta (56) also found such evidence, as well as proof of gene duplication from the existence of two highly homologous peptides from the Type III isozyme (peptides 5 and 6, shown in Fig. 21). Based on results with the Type I isozyme in the present work and comparison with the yeast hexokinase, it is now recognized that these peptides include the highly conserved Ser-"138" that functions in binding of hexose moieties and closure of the cleft (16,124). The conservation of this residue in two homologous but nonidentical peptides implies that mammalian hexokinases have pyg structurally related functional clefts whose opening or closing fulfill essential catalytic or regulatory roles. The role of this residue in catalytic function, now associated with the C-terminal domain of mammalian hexokinase, has been discussed above. According to the proposals for hexokinase evolution (17-23), duplication and fusion of an ancestral gene for hexokinase initially resulted in a fusion protein possessing duplicate catalytic sites. It is further proposed that one was retained for catalytic function while the other site evolved to serve a regulatory function, providing the basis for the potent allosteric regulation by Glc-6-P seen with the mammalian hexokinases but not with yeast hexokinase. 133 Identification of the 40 kDa domain at the C-terminus of brain hexokinase as serving catalytic function obviously implies that the modified catalytic site, now serving a regulatory function, must be associated with the N—terminal region of the enzyme, as discussed in Chapter I. According to this evolutionary scenario, the duplicate cleft can be expected in the N-terminal region, probably in the 50 kDa domain of brain hexokinase (4), and to be the site of binding for the allosteric effector, Glc-6-P. Binding of this ligand has been shown to induce closing of the cleft with the yeast enzyme (127). Given this evidence that rat brain hexokinase evolved by gene duplication and fusion of a gene similar to that in yeast, it can be postulated that the subunit structure of the yeast dimer is conserved in the binding between the 40 kDa and 50 kDa domains of rat brain hexokinase. In the X- ray crystallographic structure of the yeast dimer (36), reproduced in Fig. 24, there are no close contacts between the C- and N-termini of either subunit, indicating that structural alterations would be required to join the two subunits as a single polypeptide chain. The present work provides evidence of such an alteration leading to a "fused dimer" structure in mammalian hexokinase. As mentioned above and depicted in Fig. 22, alignment of homologous sequences from the region of the Glc binding sites in yeast hexokinase and in the 40 kDa catalytic domain of brain hexokinase suggests that approximately 80 amino 134 Figure 24. Structure of yeast hexokinase dimer, proposed to approximate the structure of the 40 kDa and 50 kDa domains of rat brain hexokinase. This is a duplication of Fig. 13 of Steitz e; el. (36). The subunits bind to each other nonsymmetrically. The "down" and "up" subunits are indicated. Both subunits have their N-termini at the top of the figure and their C-termini at the bottom. The region of the down subunit which is homologous to Peptides I and III is darkened. The numbering used in the figure was approximated from the total number of amino acids in the protein, and does not correspond to the sequence as determined by either X-ray crystallography (14) or sequencing of the cloned cDNA (37,38), as this figure was published prior to any of those sequence determinations. As discussed in the text, it is proposed that the rat brain hexokinase 40 kDa domain is analogous to the "down" subunit, minus the A and B helices, and the 50 kDa domain is analogous to the "up" subunit. The point at which the 40 kDa and 50 kDa domains are believed to be fused is indicated by a dotted line. 135 'Up' Subunit 'Down' Subunit Figure 24 9136 acid residues have been deleted in the course of the evolution of the 40 kDa domain from its ancestral 50 kDa precursor. The initial portion (approx. 25 residues) at the N-terminal region of yeast hexokinase could not be localized by X-ray diffraction, possibly due to localized disorder (36). Taking these residues into consideration, deletion of approximately 80 residues would correspond to loss of helices A and B, as defined in Fig. 13 of Steitz e; el. (36), and reproduced here in Fig. 24. After loss of these residues, the resulting N-terminus in the "down" subunit is in the same region as the C-terminus of the "up" subunit and it is evident that fusion at this point could occur with little disruption of structure. This proposed site of fusion would also correspond to a "surface loop", generally considered as favorable sites for such events, and according to the analysis of Craik e; el. (128) it is predicted that this may correspond to an intron-exon junction in the mammalian hexokinase gene. Though helix B provides some of the interactions maintaining the dimeric structure of the yeast enzyme (36), its loss is unlikely to be catastrophic since other interactions exist and might even be enhanced by loss of helix B. Thus, the structure of dimeric yeast hexokinase may represent a very close approximation to the structure of mammalian hexokinase, with the "down" subunit of the yeast enzyme corresponding to the C-terminal catalytic domain, and the evolutionary descendant of the "up" subunit being the 50 137 kDa domain associated with the N—terminal half of the brain enzyme and serving a regulatory function in binding of the allosteric effector, Glc-6-P. Binding of Glc-6-P to the cleft in the "regulatory" domain in the 50 kDa domain of mammalian hexokinases must reasonably be expected to-have a significant influence on the structure and function of the C-terminal catalytic domain if regulation is to be mediated by Glc-6-P. There is already direct evidence that these regions of the molecule are linked by intense noncovalent interactions (4). Previous data on the conformational states induced by Glc and Glc-6-P in conjunction with data from this work also allow predictions of the nature of these conformational changes. 0f the 14 sulfhydryl groups in rat brain hexokinase, Glc and Glc-6-P protect six and ten sulfhydryls, respectively, from reaction with DTNB (62). Since the present study has shown that six sulfhydryls are located in the putative 50 kDa regulatory domain, binding of Glc-6-P to the 50 kDa domain must therefore alter the reactivity of e; leeep half of the eight sulfhydryls in the catalytic 40 kDa domain. This clearly suggests a large conformational change in the 40 kDa domain. The most obvious conformational change which can take place in the 40 kDa domain is the closing of the cleft. This would concur with the ability of Glc-6-P to protect the critical sulfhydryl in the 40 kDa domain from labeling by GlcNBrAc by inducing a conformational change with the same effect as that induced 138 by Glc. Other evidence (53,54) has also indicated that the conformational changes induced by Glc and Glc-6-P are very similar. The ability of either Glc or Glc-6-P to induce a cleft closure in both domains would clearly explain the observed synergistic binding between the two ligands (53). The gene fusion scheme, which is believed to have led to the "fused dimer" structure of Fig. 24, does not provide any information about the evolutionary background of the 10 kDa domain at the N-terminus of the 50 kDa domain. It is possible that the 10 kDa N-terminal domain originated from another protein which must also bind to porin on the outer mitochondrial membrane, such as has been suggested for mitochondrial glycerol kinase (129). A description of the general rat brain hexokinase structure proposed in this study is diagramed in Fig. 25. Evidence presented here indicates that the 40 kDa domain contains the catalytic function, while the mitochondrial binding function has previously been assigned to the 10 kDa domain. It is apparent that the 50 kDa domain, similar in structure to the 40 kDa but proposed to contain the Glc-6-P binding site, is in contact with both domains whose activities it regulates, catalysis and mitochondrial binding. Hopefully, proof of the location of the Glc-6-P regulatory site within the 50 kDa domain will be forthcoming. 139 Figure 25. Generalized diagram of rat brain hexokinase structure. The three major domains and their molecular weights are indicated. The binding sites for Glc, ATP (P-P- P-Ad) and Glc-6-P are indicated. A second, low affinity ATP or ADP binding site is indicated by broken lines, which may represent the residual ATP-binding site described by Gregoriou e; el. (22). 140 mm 6591 max op 10. 11. 12. 13. 14. LIST OF REFERENCES Sokoloff, L., (1977) ip "Nutrition and the Brain" (R. J. Wurtman and J. J. Wurtman, eds.), Vol. 1, pp. 87- 139. Raven, New York. Bachelard, H. S., (1970) ip "Handbook of Neurochemistry" (A. Lajtha, ed.), Vol. 4, pp. 1-12. Chou, A. C., and Wilson, J. E. (1972) Ar h. Biochem. Blgpgy . 151, 48-55. . Polakis, P. G., and Wilson, J. E. (1984) Aych. Biochem. Bi ophys. gee, 341-352 . . Tucker, M. M., Robinson, J. B., Jr., and Stellwagen, E. (1981) Q. 5191. Qpem. ZSQ, 9051-9058. Gervais, M., Risler, Y., and Corazzin, S. (1983) Eur. Q. Biochem. 139, 253-259. Wilson, J. E., (1984) ip Regulation of Carbohydrate Metabolism (Beitner, R., ed.) pp.45-85, CRC Press, Boca Raton, Florida. Polakis, P. G., and Wilson, J. E. (1985) Arch. Biochem. Biophys. 236, 328-337. Nemat-Gorgani, M., and Wilson, J. E. (1986) Arch. Biochem. Biophys. in press. Connolly, B. A., and Trayer, I. P. (1979) Buy. 1. Eiochem. 2;, 375-385. Swarup, G., and Kenkare, U. W. (1980) Biocpemistry 12, 4058-4064. Otieno, S., Bhargava, A. K., Barnard, E. A., and Ramel, A. H. (1975) Biochemistry 11, 2403-2410. Otieno, S., Bhargava, A. K., Serelis, D., and Barnard, E. A. (1977) Biochem. 1e, 4249-4255. Anderson, C. M., Stenkamp, R. E., and Steitz, T. A. (1978) Q. Mel. 8101. 123, 15-33. 141 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 142 Anderson, C. A., Stenkamp, R. E., McDonald, R. C., and Steitz, T. A. (1978) 1. Mol. Biol. 123, 207-219. Bennett, Jr., W. S., and Steitz, T. A. (1980) Q. Mol. Bio . 140, 211-230. Colowick, S. P. (1973) in The Enzymes (P. D. Boyer, ed) 3rd ed., Vol. 9, pp. 1-48, Academic Press, New York. Easterby, J. 8., and O'Brien, M. J. (1973) Eur. Q. Bioghem. 18, 201-211. Rose, I. A., Warms, J. V. B., and Kosow, D. P. (1974) Arch. Biochem. Biophys. 164, 729-735. Holroyde, M. J., and Trayer, I. P. (1976) FEES Letp. 52. 215-219. Ureta. T- (1982) some. Biochem. Physiol. 115. 549-555. Gregoriou, M., Trayer, I. P., and Cornish-Bowden, A. (1983) Ben. Q. Eiochem. 134, 283-288. Manning, T. A., and Wilson, J. E., (1984) Biocpem. fiieppy_. Res. Qommun. 118, 90-96. Rustum, Y. M., Ramel, A. H., and Barnard, E. A. (1971) Epep. piechem. l, 309. Ramel, A. H., Rustum, Y. M., Jones, J. G., and Barnard, E. A. (1971) Eiochemistpy lg, 3499-3508. Schulze, I. T., and Colowick, S. P., (1969) Q. Biol. QAe_. 244, 2306-2316. Rustum, Y. M., Massaro, E. J., and Barnard, E. A. (1971) Biochemisppy 19, 3509-3516. Shill, J. P. and Neet, K. E. (1975) Q. Biel. Chem. 2 0, 2259-2268. Steitz, T. A., Anderson, W. F., Fetterick, R. J., and Anderson, c. M., (1977) q. Biol. Chem. 252, 4494-4500. Furman, T. C. and Neet, K. E. (1983) Q. fiiel. cpem. 258, 4930-4936. Entian, K.-D. and Frohlich, K. U. (1984) Q. Bacteriol. 158, 29-35. Bennett, W. 8., Jr., and Steitz, T. A. (1978) Proc. Negl. Acad. Sci. USA 15, 4848-4852. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 143 Shoham, M., and Steitz, T. A. (1980) Q. Mol. Biol. 40, 1-14. Steitz, T. A., Fletterick, R. J., and Hwang, K. J. (1973) 1. Mol. Biol. 1e, 551-561. Anderson, C. M., Zucker, F. H., and Steitz, T. A. (1979) Science 204, 375-380. Steitz, T. A., Fletterick, R. J., Anderson, W. F., and Anderson, C. M. (1976) Q. Mol. Biol. 104, 197-222. Stachelek, C., Stachelek, J., Swan, J., Botstein, D., and Konigsberg, W. (1986) Nuel. Acids Res. 11, 945-963. Kopetzki, E., Entian, K.-D., and Mecke, D. (1985) Gene Q2, 95-102. Gonzalez, C., Ureta, T., Babul, J., Rabajille, E., and Niemeyer, H. (1967) Biochemistpy p, 460-468. Katzen, H. M., Soderman, D. D., and Nitowsky, H. M. (1965) Biochem. Biophys. Res. Commun. l2, 377-382. Katzen, H. M. and Schimke, R. T. (1965) Proc. Natl. Aged. Sci. USA e1, 1218-1225. Lawrence, G. M. and Trayer, I. P. (1984) Comp. Biochem. Physiol. 79B, 233-238. Lawrence, G. M., Walker, D. G., and Trayer, I. P. (1983) Biochim. Biophye. Acta 743, 219-225. Crane, R. K., and $015, A. (1954) Q. Biol. Qhem. 210, 597-606. Fromm, H. J., (1981) ip "The Regulation of Carbohydrate Formation and Utilization in Mammals", (C. M., Veneziale, ed.), University Park Press, Baltimore, p.45. Felgner, P. L., Messer, J. L., and Wilson, J. E. (1979) Q. Biol. Chem. 254, 4946-4949. Gots, R. E. and Bessman, S. P. (1972) Biecpem. Bioppys. Res. Commun. 1e, 1249-1255. Inui, M. and Ishibashi, S. (1979) Q. Biochem. pp, 1151- 1156. Rose, I. A. and Warms, J. V. B. (1967) Q. piel. ghem. gee, 1635-1645. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 144 ' Kurokawa M., Tokuoka, S., Oda, S., Tsubotani, E., Ishibashi, S. (1981) Biochem. Int. ;, 645-650. Wilson, J. E. (1968) 1. Biol. Chem. 243, 3540-3647. Wilson, J. E. (1978) Trends Biochem. c1. S, 124-125. Wilson, J. E. (1979) Arch. Biochem. Biophys. 196, 79- 87. H Wilson, J. E. (1978) Arch. Biochem. Biophys. 99. 85, 88- Wilson, J. E. (1973) Arch. Biochem. Biophys. 59, 543- 549. Marcus, F., and Ureta, T. (1986) Biochem. Eiophys. Res. eommun. 139, 714-719. McCarthy, A. P. and Hardie, D. G. (1984) Irends Eiochem. Sci. S, 60-63. Lynn, R., Giaever, G., Swanberg, S. L., and Wang, J. C. (1986) Science 23;, 647-649. Palm, D., Goerl, R., and Burger, K. J. (1985) Nature 313, 500-502. Poorman, R. A., Randolph, A., Kemp, R. G., and Heinrikson, R. L. (1984) Natupe Sl, 467-469. Jones, J. GL, Otieno, S., Barnard, E. A. and Bhargava, A. K. (1975) Eieehemisppy 14. 2396-2403. Chou, A. C., and Wilson, J. E. (1974) Apeh. Eiocpem. EulugpL_. 163, 191-199. Redkar, V. D. and Kenkare, U. W. (1972) Q. Eigl. Chem. 247, 7576-7584. Redkar, V. D. and Kenkare, U. W. (1975) Eiochemistry 14. 4704-4712. Monasterio, 0., Orellana, 0., Heberlein, U., Niemeyer, H. (1981) Arch. Biol. Med. Exp. l1, 139-142. Niemeyer, H., Heberlein, U., Monasterio, 0. (1981) Agep. Eiel. Med. Exp. lg, 177-182. Ray, W. J., and Koshland, D. E., Jr. (1961) Q. Eiol. Qhe_- 23g, 1973-1979. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 145 Magnani, M., Stocchi, V., Ninfali, P., Dacha, M., Fornaini, G. (1980) Biochim. Biophys. Acta 615, 113- 120. Magnani, M., Stocchi, V., Dacha, M., Fornaini, G. (1984) eieri . Eiophy . Acta 804, 145-153. Rose, I. A. and Warms, J. V. B. (1982) Arcp. Biochem. Elephys. 213, 625-634. Darby, M. K., Connolly, B. A., and Trayer, I. P. (1981) Arch. Eiol. Med. Exp. l1, 163-170. Subbarao, B. and Kenkare, U. W. (1977) Arch. Biochem. EiOphys. 181, 19-25. Batra, S. P. and Colman, R. F. (1986) Eiocpemistgy ES, 3508-3515. King, G. C., Martin, C. T., Pham, T. T., and Coleman, J. E. (1986), Eiochemistgy ge, 36-40. Gonzalez, D. H., Iglesias, A. A., and Andreo, C. S. (1986) Arch. Biochem. Bioppy . 245, 179-186. Grubmeyer, C. T. and Cray, W. R. (1986) Biochemistry ES, 4778-4784. Alter, S. C. and DeLuca, M. (1986) Biochemistry ES, 1599-1605. Marshall, M. and Fahien, L. A. (1985) Apch. Eiochem. Biophys. 241, 200-214. Carlson, G. M., Colombo, G., and Lardy, H. A. (1971) Eiochemisppy l1, 5329-5338. Ray, M. and Bhaduri, A. (1980) Q. Eiel. Shem. 255, 10777-10781. Gavilares, F., Peterson, D., Bullis, B., and Schirch, L. (1983) Q. Elgl. Qgem. 2S8, 13155-13159. Olson, S. T. and Massey, V. (1980) Eioepemistzy l2, 3137-3144. Smith, D. J. and Kenyon, G. L. (1974) Q. Biol. Chem. 249, 3317-3318. Gilbert, n. F. (1982) 1. p191. Spem. ge7, 12086-12091. Nakashima, K., Pontremoli, S., and Horecker, B. L. (1969) Epee. Eapl. Aged. Sei. SSA SA, 947-951.. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 146 Tschesche, H. and MaCartney, H. W. (1981) Eur. ;. Biochem i20, 183-190. Pontis, H. G., Babio, J. R., and Salerno, G. (1981) EEOC. Natl. Acad. Sci. USA 18, 6667-6669. Doehlert, D. C. and Huber, S. C. (1985) Biochim. Biophy . Acta 830, 267-273. Polakis, P. G., and Wilson, J. E. (1982) Biochem. Bigphys. Res. Commug. i07, 937-943. Decker, L. A., ed. (1977) Worthington Enzyme Manual, p.221, Worthington Biochemical Corp., Freehold, NJ. Felgner, P. L., and Wilson, J. E. (1976) Anal. Biochem. 14, 631-635. Chamberlain, J. P. (1979) Anal. Biochem. 92, 132-135. Wilson, J. E. (1973) Arch. Bioghem. Biophys. 1 4, 332- 340. Fox, J. (1984) Linear Statistical Models and Related Models, pp. 204-213, John Wiley & Sons, New York. Ellman, G. L. (1959) Arch. Biochem. Biophys. 22, 70-77. Crestfield, A. M., Moore, 3., and Stein, W. H. (1963) J. Biol. Chem. 238, 622-627. Dirnhuber P., and Schutz, F. (1948) Biochem. Q. A2, 628-632. Stark, G. R. (1967) Methods. Enzym. ii, 590-594. Hagar, D. A. and Burgess, R. R. (1980) Anai. Bioghem. 192, 76-86. Henderson, L. E., Oroszlan, S., and Konigsberg, W. (1979) Anal. Biochem. 21. 153-157. Gundlach, H. G., Stein, W. H., and Moore, S. (1959) g. 5121- Chem. 234. 1754-1760- Sols, A., and Crane, R. K. (1954) ;. Bigl. ghem. 0, 581-595. Grossbard, L., and Schimke, R. T. (1966) J. Biol. Chem. 24;, 3546-3560. Kitz, R., and Wilson, I. B. (1962) J. Bioi. CAg . 37, 3245-3249. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 147 Dixon, M. (1953) Biochem. ;. 5;, 170-171. Chou, A. C., and Wilson, J. E. (1974) Arch. Biochem. Biophys. 165, 628-633. Solheim, L. P., and Fromm, H. J. (1980) Biochemistry i2, 6074-6080. Lindley, H. (1960) Biochem. J. 12, 577-584. Wilchek, M. and Givol, D. (1977) Meghods. Enzym. Ag, 153-157. Plapp, B. V. (1982) Methods Enzym. 21, 469-499. Gurd, F. R. N. (1972) Methods Enzym. 25. 424-438. Wilson, J. E. (1982) Arch. Biochem. Biophys. 218, 254- 262. Plummer, T. H. and Lawson, W. B. (1966) J. Bioi. Chem. 241, 1648-1650. Tsou, C. -L. (1962) Sci. Sin. 1, 1535-1558. Wilson, J. E., and Smith, A. D. (1985) J. Bigi. Chem. 260, 12838-12843. Kasper, C. B. (1975) in Protein Sequence Determination (S. B. Needleman, ed.), p. 132, Springer-Verlag, New ‘York. Parkinson, A., Ryan, D. E., Thomas, P. E., Jerina, D. M., Sayer, J. M., van Bladeren, P. J., Haniu, M., Shively, J. E., and Levin, W. (1986) J. Biol. Chem. 26;, 11478-11486. Weber, K. and Osborn, M. (1969) 2. Biol. Chem. 44, 4406-4412. Dunker, A. K. and Rueckert, R. R. (1969) J. Biol. Chem. 244, 5074-5080. Anderson, C. M., McDonald, R. C, and Steitz, T. A. (1978) J. M01. 3101. 123, 1-13. Weber, I. T., Takio, K., Titani, K., and Steitz, T. A. (1982) Eroc. Nati. Acad. Sci. USA 1g, 7679-7683. Rossman M. G., Liljas, A., Branden, C.-I., and Banaszak, L. J. (1975) i3 The Enzymes (P. D. Boyer, ed.), 3rd ed., Vol. 11, pp. 61-102, Academic Press, New York. 123. 124. 125. 126. 127. 128. 129. 148 Warren, J. C., and Mueller, J. R. (1977) Methods. Eflzxm- 46. 447-460- Shoham, M., and Steitz, T. A. (1982) Biochim. Biophys. Acta. 705, 380-384. Maley, F., and Lardy, H. A. (1955) J. Biol. Chem. 2 4, 765-773. Cheng, S.-Y., Wilchek, M., Cahnmann, H. J., and Robbins, J. (1977) Q. Biol. Chem. 252, 6076-6081. McDonald, R. C., Steitz, T. A., and Engleman, D. M. (1979) Biochem. 18. 333-342. Craik, C. S., Rutter, W. J., and Fletterick, R. (1983) Science 220, 1125-1129. Kaneko, M., Kurokawa, M., and Ishibashi, S. (1985) Arch. Biochem. Biophys. 237, 135-141.