M‘”: .333}l}.}} p} , ‘ 3 1‘3}3}3“}}}}}}}}},,..,1},3“ . 'I}}}} 133} }}‘}} 3., ..WI333 LII 3 3"}313731‘3‘?‘ 3'3} }I}II.I l’I'I 1}“ “35:13:23 111}. },...,.},},,} “3‘ r33}‘}‘n .}‘3‘3}‘ I. 2‘1}, ‘3}1'13} }}I‘;‘ 133} Ld}}.}}'}:}3}}}_}3}3 .__,_.,.. w... ‘ W:- “fl ..._._ .. . , ”£5..- ”.1. £1- "‘1‘ y... .— v—v 7-1:- ‘ W“W 3313 }}“III 33‘ }3 .M’L ..g‘. “W {T ..fi__._.._ - .. W ‘ “- W-¢m . > W _ _. H, —v w an— - 32% .=:~—._:?..:~; vi...“ ,2 “‘ ..TAS. }}3}I ‘33:“ ‘1}"33‘3‘ ““““ I‘ 3;. .“ “‘II}VI‘II : 1111» ' $.- J.- ———'—4—P_ 'l'L—E ”if ‘-—- --- fir...— WW“? mi" v.0“? M ..d W— —' ”C:- } I }}}}} ‘. } } }3} . .“ } 3363“ u} I'. 2 W“:;‘3‘I:‘"§. 133} 1'. 3.1331131 . . 11 . “II- ' }‘II ..I... ~z I‘I“.‘3:‘I ‘I . . ‘ ‘ ‘ .11.“ , 1. ,3 , }}I_.,}3}1}}.}}1 }1} I“}‘}‘}.I.}33I 1‘} I}}"}3}‘I‘. . .I , Hg} . , }..}, ‘ £23.}: 3' . ,. .1... . :. I1} .1 1.33.1} 1. I. . . . . .1». . ~ I};".‘}I‘.}3‘}}}3}}H}I3}‘}} }13} }}}33} }}}}}3}}}}}}| }}}}}}3} .I}I,}} }I}}},}} . .. “..I P3333.“ II . I I , -,,. '.- ;., '. » ‘ ; 1-1.1 ‘3}I‘I‘. }.Iw“‘3}33 1}} 1.1111133 .I 1‘}"}‘I‘}3 1} 1}}3“ I "“‘3 .'- :;',,' - I, I‘; j }E 3? . .. ,.}1 }}}1}}!. I 3 ..I 3 .}I 11.. }I }. 33 . 1%.} 113I1I1.Iw“IIIIIdW3’IIMI.II3 '" ~.1£ .. .III-IIIIIIIIIII IIIIIIIIIII1IIIIIIIIII.111111 IIIIII . }}1}}1 . .}.11, .1 1.1...1III,3 }111} }}}}'}.}} .I}}}I’}}.,II}1111}}}}’}I-I 311,113,} 13.17: . 33313,}... I 97‘. 7-‘213..Eo:v[2_‘;‘ }}1,‘}3}}}};.‘ I 2}} “.‘,‘I‘."“I .. II‘I. I3I}II.} }133 I'}I‘}‘,,‘I“II}‘ 3}3“I3I3I}} 3313}}}}‘}3}‘}.33‘33“.‘I;7‘II3‘3 I} 33‘I333._"‘ ' ' .. 1 ~ ‘ IIUII . I..I%13313%InhTIMI.. ' 'II I; I ‘I II 'II I I " I II ...I... 11..- .}. II. I 2335353133333333133"3333313 I331 I3 3 II ' = ..“II . III .1 1111 It I. 1. II “. "III...11333131333331.3.13. . I““}“I“}‘,.“‘ . I‘I‘I I‘ ,‘1 [13 . “}I'I;“I}‘\ I. 51.33 3"}? 31.3233 ‘13}:1‘ ,‘ 1"‘3, 3... I“l‘ '|““.' }.33‘ .}‘}‘}‘I}.} 1‘)“ {13'3"} in?“ 1)} 33‘1" 1.3"“. ‘ “‘1“ 1““ ‘ 13. I311 }..: .‘3‘ ‘3“ ‘3 . .. _ . .} .}I If} ‘1'}1"“"" } }}1,,,‘ l3:‘, WI} . .'.I, 3‘1" JV}, ‘3‘ , 3| ,“..'.' ., .}I} 3,11}. If" }‘3‘}I} """ }I" 3113,.3‘3.}}‘,3 I} ..I'-' I}I~',.I.v.‘ I." , . ., "“‘“‘III.I.II . , .}.}. 1.2311. }..}... EII }.3 3 ,. " ,, . ' }3 I”“.}‘3}}?,',‘,‘.I.‘}‘.I,I‘ ‘Il‘}}}3}‘3‘I33‘}“;‘}‘.‘}‘,‘}:‘_}I}3} 1‘3“.“ }}I‘I }}}} .3,3 :3“ }}}}I} II I }.I }.. .'1 I‘ “I ' . ‘ 3-}} \‘I"‘ I. 'I“}'«II I}? ‘3 “3} - . ‘1 ' . ""‘ '31‘ "‘II »“ “ .I‘c“I3 ‘ . .“‘ .‘I “13‘1‘1 1"!"nggt‘ I 4:1: Efi:‘-‘ ‘ ‘I I I .1. I I." IIIII, . III-:11." I :31). “33.11131; ’13.-IIII3.I:‘.II‘ ‘33.. ""3 III" I‘ }‘II}I.'I"I .5“. ' " " '- .,- I" “. II ‘ ‘I I“. .‘ I.I IIIIII‘.'I-‘ I1" {.13‘..:?-‘ , 3‘3“? '. '. I:I ‘ 3 3.“ I'.‘ w“"f“ ‘.‘I“‘." 3531‘ .' ?I ".3“ ‘..1'1.‘1i“...‘.‘..".‘-,‘-‘.‘~.I,III;..“IJHI' 2.11.13. ‘1 ., ““3“ 1}}},.3 3 :‘ 3'I..3..' .H 3} ”‘:‘-}II "I ‘-‘. “1'. ‘ I ‘““““‘““““‘““““‘“,“:3““ .3 3:1“ I331 .333 ‘I “3} - ‘ "3‘1‘ “ ‘ “‘ I ‘I‘I‘ ‘ ‘ II “ I“ ‘I‘ ‘I‘ ‘II3‘ ‘ ‘ ‘ I , I13" ,II‘I‘3I“} ‘ ‘, “ , 131.3, '13" , 2"‘3‘1 .I‘" .}I'I'L» 5’“ } . .. {I 3.3,3'.“"'1‘I“..I. . j_ ‘3 3,1 .1}.. ' 3 ‘ 1' ‘ ‘- . “3“ ““1"1'.‘3‘IT““I.‘II““ 3 “3‘“‘“}“ 1“} }}‘I ‘33‘ “‘I“‘ I ‘ . I . 1 “ " .1'I1I.’,".I.'§ I. "'I'I.'Il.’.I'I‘I I . _. 3 3'3" “‘333,‘ 315;: ‘l|}‘ 'I, . 3| ‘ 1}}‘21 13,,3' 1‘3 33“ '} I} H . """ ‘III .m1IHIM3 i‘-‘ I‘. 3“ .3 "‘ "“ ‘ . 31 9} .~ ‘ ,.I‘ ITM ;1 I“ .. ,.... ..WW3»1 ‘ ‘ ‘ “ “““”‘““"“"‘“ “‘ I“'3‘ “.H.‘ }.I},I33133}3“‘3}3 "‘ ‘ I I‘3“ ‘-"I3"|3'|‘ ‘ ..I‘IL‘. 1113333MNI313M‘JJJQE‘ ‘ WMUI!‘ I32? 3.31»IILIas.‘=.'“"" JIM LIBRA R Y Michigan State University This is to certify that the thesis entitled LYOTROPIC SALTS AND THE HEXOKINASE MEMBRANE INTERACTION: PURIFICATION, RECONSTITUTION AND CHARACTERISTICS OF THE OUTER MITOCHONDRIAL MEMBRANE BINDING SITE FOR HEXOKINASE presented by PHILIP LOUIS FELGNER has been accepted towards fulfillment of the requirements for f" ‘3' Aggreein li‘dt‘fifi-I.‘ IV; 7% 5‘? Via“ Date X’lé7; 0-7 639 C I . .‘.C- N {o_ .‘Qaijfl ‘5 I 23‘3”» 3&{7 f. ' ‘_ {‘1‘ r. >d ' OVERDUE FINES: K per duper ttu mamas usmv MTERIAL : N Place In book return to law we fro. circulation recon LYOTROPIC SALTS AND THE HEXOKINASE MEMBRANE INTERACTION: PURIFICATION, RECONSTITU- TION AND CHARACTERISTICS OF THE OUTER MITOCHONDRIAL MEMBRANE BINDING SITE FOR HEXOKINASE By Philip L. Felgner AN ABSTRACT OF A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Biochemistry I978 ABSTRACT LYOTROPIC SALTS AND THE HEXOKINASE MEMBRANE INTERACTION: PURIFICATION, RECONSTITU- TION AND CHARACTERISTICS OF THE OUTER MITOCHONDRIAL MEMBRANE BINDING SITE FOR HEXOKINASE By Philip L. Felgner Lyotropic Salts and the Hexokinase/Membrane Interaction-- Neutral salts can be ranked according to their relative abilities to disrupt or stabilize the tertiary structure of macromolecules. This ranking has been called the Hofmeister or lyotropic series, with the more lyotropic salts being more effective at disrupting the native structure of macromolecules than non-lyotropic salts. Some salts in this series, in order of increasing lyotropicity, are Na2504 < NaF < NaCl < NaBr < NaI < NaClO4 < NaSCN. A molecular interpretation of the effect of neutral salts on macromolecular structure was developed that accommodates two general properties of neutral salts on macromoleculars in aqueous solution. (I) Salts can disrupt ionic linkages (including hydro- gen bonds) by binding to charged (or partially charged) moieties. The binding constants, though small, can be measured. (2) Salts can promote hydrophobic interactions by changing the solution properties of water such that hydrophobic molecules are less adil cifi pnfll‘ 5 '::J:' HIEFO' I.’ ‘I. IN :‘F‘y. 5 U l Philip L. Felgner readily hydrated. These two effects oppose one another with respect to whether or not the tertiary structure of a macromolecule will be stabilized or destabilized; i.e., breakage of ionic linkages destabilizes the tertiary structure, and promoting hydrophobic interactions tends to stabilize tertiary structure. Lyotropic salts have a large destabilizing component and a small stabilizing component. Conversely, non-lyotropic salts have a relatively smaller destabilizing component and a large stabilizing component. A system of equations has recently been derived by Melander and Horvath (l) which allow the quantitation of the relative con- tributions of electrostatic and hydrophobic forces between two interacting molecules. Their development was applied to the hexokinase/membrane interaction and it was estimated that about half of the surface of the enzyme molecule is involved in hydro- phobic interactions with its binding site on the membrane surface. Purification, Reconstitution and Characteristics of the Mitochondrial Binding Site for Hexokinase--Very pure outer mitochondrial membranes (OMM) have been obtained from rat liver that contain a binding site for rat brain hexokinase. The specific activity for this membrane binding site (units of hexokinase bound/ mg membrane protein) is 40 fold higher than in either microsomes, erythrocytes, or inner mitochondrial membranes. Glucose-G-P at low concentrations (1 mM) specifically elutes the enzyme from ‘aric '5‘ 'I 19":5 1.. . CIA“: v.0“. -hlrl he . IE ll Philip L. Felgner these membranes as it does from intact mitochondria from either liver or brain. As a means of gaining insight into the nature of the OMM binding site, numerous attempts were made to chemically or enzy- matically modify the binding site. All these approaches were met with equivocal results. The enzyme would not bind to any liposome preparations in the glucose-6-P sensitive manner that is charac- teristic of the intact binding site. The OMM proteins can be depleted of their lipid content by treatment with low concentra- tions of detergent. This treatment modifies the binding site in a partially reversible manner. Treatment of the OMM with the non-ionic detergent octyl glucoside preferentially extracts a single protein of mol. wt. 31,000 that has been tentatively identified as the hexokinase binding protein. Removal of the detergent from this solubilized membrane protein by dialysis results in the formation of lipid containing membrane vesicles that contain a glucose-6-P sensitive hexokinase binding site. ACKNOWLEDGMENTS I am certainly pleased to finally be able to acknowledge the very capable technical (as well as intellectual) help of Jan Messer and Suzanne Murrmann. I would also like to acknowledge John Wilson who is primarily responsible for training these two individuals and who's ability at preparing responsible laboratory help I have always admired. Thanks are extended to David Alessi who's data appear on Table 13 of this thesis and to Pat Kelly who provided much of the data on Table 12. Judy Kao's expert electron microscopy skills and willingness to help are also very much appreciated. And thanks are extended to Clarence Suelter who's enzymology class prompted my thinking about salt effects and hydrophobic effects on macromolecules. I acknowledge the assistance of my committee members who are Dr. Wilson, Dr. Suelter, Dr. Ferguson-Miller, Dr. K. Schubert and Dr. Tien. I owe special thanks to Dr. Schubert for his very thorough reading of my thesis and to Dr. Ferguson-Miller for stimulating discussions which I am sure will continue past the publication of this thesis. And I am grateful to John Boezi, Dave McConnell and John Wilson for supporting me through some difficult times. Obvious additional appreciation is due to Dr. Wilson for Providing an admirable personal and professional example over the Past several years. 11' stag-c. espeC' This thesis is dedicated to my parents who provided constant support, to my brothers who were always good examples and especially to Jean, who would rather I became a minstrel. TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES PREFACE . Chapter I. LYOTROPIC SALTS AND THE HEXOKINASE MEMBRANE INTERACTION . . Introduction . . The Molecular Basis for Neutral Salt Effects on Macromolecules . . Description of the Hofmeister (or Lyotropic) Series Neutral Salt Binding Affinities to the Peptide Bond Dipole . . The Effect of Neutral SaIts on the Surface Tension Increment. . Physical Description of the Surface Tension Increment . What is the Relationship Between Surface Tension, o, and the Hydrophobic Effect? Summary . Theoretical Treatment of the Salt. Effects Data on the Hexokinase/Membrane Interaction . II. SOLUBILIZATION, RECONSTITUTION AND CHARACTERISTICS OF THE OUTER MITOCHONDRIAL BINDING SITE FOR HEXOKINASE . . Introduction . Materials and Methods Chemicals . Hexokinase Assay Preparation of Mitochondria . . Glucose-G- -Phosphate Solubilized Enzyme . Preparation of Purified Outer Mitochonrial Membranes . . % Sodium Dodecyl Sulfate-PonacryIamide Disc Gel Electrophoresis iv Page vii ix xi (boom 10 IS 24 26 35 35 38 38 39 39 39 4O 41 Chapter Page Mitochondrial Binding Assay. . 42 Determination of Glucose- --6 P Dependent Squbili- zation of Hexokinase from Purified Outer Mitochondrial Membranes . . 43 Explanation of the Glucose- 6- P/Galactose- 6- P Ratio and Rationale For Using This Ratio as a Means of Assessing the Intactness of the Hexokinase Binding Site . . . . . 44 Results . 46 Specificity of Hexokinase Binding to the Outer Mitochondrial Membrane . . . 46 Glucose- 6- P Dependent Solubilization of Hexo- kinase from Purified Outer Mitochondrial Membranes . . 46 The Effect of Exogenous Protein and Polyethylene Glycol on the Hexokinase Membrane Inter- action . . . . . . . . 51 Hexokinase Binding to Liposomes . . . 57 The Role of Protein in the Binding Site;. Attempts to Modify the Binding Pro- perties Chemically and Enzymatically. . . 61 Protease digestion . . . . . . 61 Heat inactivation of the Glucose- 6- P dependent solubilization . . . . . . 70 Miscellaneous . . A 70 Lipid extraction from outer membranes with Emulgen 913 . . . 74 The effect of lipid extraction on glucose- -6- P dependent solubilization . 78 Restoration of g1ucose-6- P dpendent solubili- zation by adding back OMM lipids to Emulgen extracted membranes . . 78 Comparison between outer mitochondrial mem- branes and microsomes with respect to detergent solubilization . . . 87 Solubilization, Reconstitution and Purification of a Putative Hexokinase Binding Protein from the Outer Mitochondrial Membrane . 90 Solubilization of the OMM with octyl gluco- side and reconstitution of the hexoki- nase binding properties by dialysis . . 90 Electron microscopy of the reconstituted vesicles . . . 99 Protease treatment of OMM followed by reconstitution . . . . 106 Subunit molecular weights and tentative identification of some of the pro- tein bands on $05 gels of OMM . . . . 114 V Page ADDENDUM . 119 REFERENCES 122 APPENDICES 127 vi Table II 12 I3 14 LIST OF TABLES Approximate Salting-out Constants for Acetyletetra- glycine Ethyl Ester at 25. 00a . . . . . . Molal Surface Tension Increments of Various Salts The Hydration Radii of Several Ions . Specificity of Hexokinase Binding to Outer Mitochondrial Membranes . . Glucose-6-P Dependent Solubilization of Hexokinase from Purified Outer Mitochondrial Membranes Hexokinase Binding to Liposomes The Effect of Protease Treatment on the Outer Membrane Binding Site for Hexokinase Proteoloysis of Intact OMM with .Trypsin and Chymotrypsin Effect of Periodate on Hexokinase Binding . Heat Inactivation of the Glucose-6-P Dependent Solubilization Effect . . . . . Heat Inactivation of the Glucose- 6- P Dependent Solubilization Effect . . Lipid Extraction of OMM with Emulgen 913 Partial Restoration of the Glucose-6-P Dependent Solubilization Effect to Emulgen Treated OMM, Using Lipids Extracted from Microsomes . . . Solubilization and Reconstitution of the OMM Binding Site for Hexokinase; Detergent Concentration Dependence . . . . . . . . vii Page II 23 47 50 58 64 66 69 71 72 75 81 91 Table 15 16 I7 18 Solubilization and Reconstitution of the OMM Binding Site; the Effect of Preincubation of Solubilized Membranes at 25°C Prior to Centrifugation . Hexokinase Binding Properties of 1X and 2X Reconsti- tuted OMM . . . . . Chymotrypsin and Trypsin Treatment of OMM Prior to Reconstitution . Trypsin Treatment of Octyl Glucoside Solubilized OMM, Followed by Reconstitution . . . viii Page 93 96 I07 110 Figure IO ll 'I2 1.3 111 155 LIST OF FIGURES The Relationship Between the Solubility Coefficient and the Surface Tension Increment The Relationship Between P and 6 . . Plots using Equation 6 with Assumed Values for All Constants . . Plots Using Equation 6 with Assumed Values for All Constants; the Effect of Different Salts SDS Gels of Purified Outer Mitochondrial Membranes With and Without Added Hexokinase . . The Effect of BSA on the Hexokinase/OMM Association The Effect of Polyethylene Glycol on the Hexokinase/ OMM Association . . Magnesium Requirement for Binding to Lysozyme Coated Liposomes . . . . . SDS Gels of Protease Treated Membranes . Inactivation of Purified Hexokinase by Periodate Thin- layer Plates of the Chloroform/Methanol Extracts of OMM, Emulgen Extracted OMM, and Reconstituted Emulgen Extracted OMM . The Effect of Emulgen Treatment of OMM on Glucose- 6- P Dependent Solubilization . Restoration of Glucose- 6- P Dependent Solubilization from Emulgen Extracted OMM by the Addition of OMM Lipids . Solubilization of Outer Membranes and Microsomes with Emulgen 913 and Cholate SDS Gel Electrophoresis of Reconstituted OMM . ix Page 12 I9 29 32 48 52 55 59 62 67 76 80 83 88 94 Figure 16 17 18 19 20 21 22 SDS Gels of 1X and 2X Reconstituted OMM; Comparison of Gels With and Without Bound Hexokinase . Negative Staining, Electron Microscopy of Intact and Reconstituted OMM . . . . . . Thin Sectioning, Electron Microscopy of Intact and Reconstituted OMM . . . . . Thin Sectioning, Electron Microscopy of Intact and Reconstituted OMM . . . . SDS Gels of Protease Treated/Reconstituted OMM SDS Gels of Trypsin Treated Reconstituted OMM Tentative Identification and Molecular Weight Determination of the Proteins in Native OMM Page 98 100 102 104 108 112 115 PREFACE The requirements of the Graduate School limited the length of the Abstract that appears in the front of this dissertation. This preface is intended to be an expanded versioniyfthat Abstract. It is hoped that this will aid in the reading of this material. Chapter I Lyotropic Salts and the Hexokinase/Membrane Interaction-- Neutral salts can be ranked according to their relative abilities to disrupt or stabilize the tertiary structure of macromolecules. This ranking has been called the Hofmeister or lyotropic series, with the more lyotropic salts being more effective at disrupting the native structure of macromolecules than nonlyotropic salts. Some salts in this series, in order of increasing lyotrOpicity, are NaZSO4 < NaF < NaCl < NaBr < NaI < NaC104 < NaSCN. A molecular interpretation of the effect of neutral salts CH1 macromolecular structure was developed that accommodates two general properties of neutral salts on macromolecules in aqueous Scflution. (1) Salts can disrupt ionic linkages (including hydro- gen bonds) by binding to charged (or partially charged)moieties. The binding constants, though small, can be measured. (2) Salts xi can promote hydrophobic interactions by changing the solution properties of water such that hydrophobic molecules are less readily hydrated. These two effects oppose one another with re- spect to whether or not the tertiary structure of a macromolecule will be stabilized or destabilized; i.e., breakage of ionic linkages destabilizes the tertiary structure, and promoting hydrophobic interactions tends to stabilize tertiary structure. With respect to the first point, several observations have been made. (1) Salts bind to the peptide bond dipole. (2) The binding affinity of the salts for the dipole varies according to the position of the salt in the lyotropic series, such that the more lyotropic salts bind more tightly. (3) Vicinal hydrophobic groups around the dipole are required to produce the observed variations in the binding affinities of different salts to the dipole. (4) Hydrophobic groups are surrounded by interfacial water that is different from bulk water. (5) Surface tension theory predicts that salts with a deeper hydration sphere, are repelled from interfacial water more than salts with smaller hydration spheres. (6) The depth of the hydration sphere around anions decreases as the lyotropicity increases. Considering all these points together, the general conclusion is that, the smaller 'the hydration sphere around the anion, the more tightly it will IJind to peptide bond dipoles containing vicinal hydrophobic groups. Binding to such a dipole could lead to destabilization (If the tertiary structure of a macromolecule by disrupting the (l-helix or other secondary structures. xii The second point regarding stabilization of hydrophobic interactions by salts is also supported by several observations. (1) To disrupt a hydrophobic interaction, a new hydrocarbon/water interface must be created that has characteristics similar to the air/water interface. (2) Since the value of the surface tension can be described as the extent to which a given solution prefers to minimize its interfacial area, anything which increases the surface tension will tend to decrease the interfacial area, favoring hydrophobic interactions. (3) The extent to which a given salt will increase the surface tension of an aqueous solution (the surface tension increment) is dependent on the salt used, and non-lyotropic salts increase the surface tension more than lyo- tropic salts. From consideration of these three points the general conclusion is that neutral salts tend to stabilize hydro- phobic interactions with non-lyotropic salts being more effective than lyotropic salts. So, to summarize, a given salt will either stabilize or de- stabilize the tertiary structure of a macromolecule depending on the interplay between two opposing forces, one tending to stabilize the overall structure and the other tending to destabilize it. Lyotropic salts have a large destabilizing component and a small stabilizing CCanonent. Conversely, non-lyotropic salts have a relatively smaller dEstabilizing component and a large stabilizing component. A system of equations has recently been derived by Melander at": Horvath (l) which allow the quantitation of the relative Contributions of electrostatic and hydrophobic forces between two xiii interacting molecules. Their development was applied to the hexokinase/membrane interaction and it was estimated that about half of the surface of the enzyme molecule is involved in hydro- phobic interactions with its binding site on the membrane surface. Chapter II Purification, Reconstitution and Characteristics of the Mitochondrial Binding Site for Hexokinase--Purified outer mito- chondrial membranes (OMM) have been obtained from rat liver mito- chondria that contain a binding site for rat brain hexokinase. The effective concentration for this membrane binding site (units of hexokinase bound/mg membrane protein) is 40 fold higher than in either microsomes, erythrocytes, or inner mitochondrial membranes. Glucose-6-P at low concentrations (1 mM) specifically elutes the enzyme from these membranes as it does from intact mitochondria of either liver or brain. Binding appears to be weaker in the OMM, however. Exogenous protein in the crude mitochondrial preparation, which is absent in the purified OMM, appears to play a role in strengthening the binding. The enzyme can be bound to positively charged liposomes, and to negatively charged liposomes that contain tightly bound lysozyme, IMJt in neither case is the binding sensitive to glucose-G-P. Hexokinase binding to the liposomes that contain bound lysozyme is M9(21 dependent as it is in OMM, but glucose-6-P causes tighter 2 Instead of weaker binding of hexokinase. Negatively charged xiv liposomes or liposomes that have been prepared from total lipid extracts of either OMM or microsomes do not bind any hexokinase. SDS gels of Chymotrypsin or trypsin treated OMM indicate that proteolysis modifies several proteins in the membrane, but the binding properties of the protease treated membranes are not appre- ciably modified. Periodate treatment of OMM removes 30% of the binding sites for hexokinase and the enzyme bound to this modified binding site does not show the glucose-6-P specific elution effect that is characteristic of intact OMM. While it was clear that periodate inactivated the binding site, no inferences about the involvement of carbohydrate residues in the binding site could be made, since it was demonstrated that periodate readily inactivates a protein (hexokinase) which contains no carbohydrate. Incubation of OMM at 37°C leads to losses in the number of binding sites as well as in the glucose-6-P sensitivity. Paradoxically, Mg++ and Ca++, as well as EDTA, have stabilizing effects. Phospholipase A2, intrinsic to the OMM, may play a role in this loss of the binding properties. Treatment of OMM with the non-ionic detergent Emulgen 913, extracts more than 90% of the phospholipids from the OMM while 5011flfilizing less than 40% of the membrane protein. The insoluble, ll'Pid depleted, membrane proteins remaining after Emulgen treat- mETTt, contain a hexokinase binding site but this binding site is "04: glucose-6-P sensitive. The addition of phospholipids extracted from OMM or microsomes to the lipid depleted proteins, with (soni- cation, restores the glucose-6-P dependent solubilization effect. XV The reason for this unusual ability of detergent to preferentially extract phospholipids from the OMM without solubilizing the proteins may be that the OMM proteins are unusually difficult to solubilize by detergents. Consistent with this view, it was observed that at least a 10 fold higher concentration of either cholate or Emulgen 913 was required to completely solubilize the OMM than was required to solubilize microsomes. Treatment of OMM with the non-ionic detergent octyl gluco- side under the appropriate conditions solubilizes primarily two. proteins which have molecular weights on $05 gels of 61,500 and 31,000. On the basis of its molecular weight the 61,500 M.W. protein has been tentatively identified as monoamine oxidase. When this solubilized material is dialyzed to remove the detergent, membranous vesicles are formed which retain about 50% of the ori- ginal glucose-6-P sensitive hexokinase binding properties. Electron microscopy confirms the membranous character of the reconstituted material. Resolubilization of these reconstituted OMM with octyl glucoside, followed by dialysis results in the for- mation of twice reconstituted vesicles which still contain 25-50% 0f the original glucose-6-P sensitive hexokinase binding sites. These twice reconstituted membranes contain primarily a single PYWDtein of molecular weight 31,000. When the octyl glucoside- So1ubilized membranes are treated with trypsin, the resulting h.Ydrolysate cannot be reconstituted into membranes containing an I'Ttact, glucose-6-P sensitive binding site for hexokinase. SOS 9€¥Is of the octyl glucoside-solubilized, protease-treated membranes xvi confirmed that the amount of the 31,000 M.W. protein was substan- tially reduced. These results strongly suggest that hexokinase binding is dependent on a single protein (M.Wt. 31,000) of the OMM, and that this protein can be selectively purified by solubilization of the OMM with octyl glucoside followed by reconstitution of membrane vesicles. xvii CHAPTER I LYOTROPIC SALTS AND THE HEXOKINASE MEMBRANE INTERACTION Introduction The effect of salts on macromolecules has been studied for nearly one hundred years and yet after all this time no clear explanation, in molecular terms, has appeared, despite a truly remarkable abundance of data on the subject. In this first chapter I will develop a working hypothesis, based on the literature and my own data, that describes, in molecular terms, the basis for the action of neutral salts on macromolecules. Is there any physiological relevance for studies of macro- molecular structure at high ionic strength, with neutral salts like KSCN or NaClO4 which are rarely found in vivo? In view of my bias, it's obvious that my answer to this question will be affirma- tive. We have found, for instance, that hexokinase in the Presence of mitochondria in vitro can be predominantly either Soluble or membrane-bound depending on the ionic strength in the range 0-0.25M. The physiological ionic strength is within this r‘alrlge. The fact that the intracellular saltconcentration is (presumably) a fixed value does not argue against performing experiments outside that range. For example, one way to shed light on the role of KCl at intracellular concentrations (~ 0.15M) is by observing its influence at concentrations above and below intra- cellular levels. Likewise, a means of determining specific roles for K01, is by comparing its effects with other salts. Another point that deserves some attention here is that in vivo solute concentrations are far different from those typically used for in vitro experiments. For example, one rarely sees in vitro experiments done with the protein concentration as high as 10%, as it is in vivo. It seems likely that some important basic principles about protein structure and function in vivo may be gleaned from studying proteins in concentrated protein solutions as well as in concentrated solutions of pure solutes such as neutral salts. The effect of neutral salts on macromolecules may be entirely different at high protein concentrations (10%). Some data that begin to address these points are presented in Chapter II and in reference 3 (see appendix). Apart from the physiological relevancy, there are other reasons for studying salt effects on macromolecules. As will be Shown later, salts can be used as a probe to semiquantitatively measure the extent of hydrophobic vs. electrostatic bonding in a 9 1 van macromolecular interaction. In addition, insights into the "Kilecular nature of the hydrophobic effect can be gained through the study of neutral salt effects on macromolecules. The Molecular Basis for Neutral Salt Effects on Macromolecules Previous explanations of the molecular basis for neutral salt effects (4-7) all included the concept of "water structure.“ The following description, which is based on some recent develop- ments, is intended to provide a more lucid molecular description of the effect of salt on macromolecules, without depending as much on the rather ill defined “water structure" concept (8). Most of the effects of salts on proteins can be explained in terms of direct binding by the salts to the peptide bond dipole, and, by inference, to other charged groups on the protein molecule (3). The relative influence of different salts can be inter- preted on the basis of their relative binding affinities for the dipole and the differences in their binding affinities can be explained in terms of their relative tendencies to be excluded from a hydrocarbon/water interface. Description of the Hofmeister (or Lyotropic) Series Neutral salts can be ranked with respect to their relative at>ilities to solubilize macromolecules. If the salts tend to scflubilize proteins, they are called "salting-in" type salts. If" they tend to force macromolecules out of solution, they are cal led "salting-out" type salts. This ranking, called the Hofmeister or lyotropic series, was first established by Hofmeister I" 1888 (9) in a study of the relative effectiveness of various 5&1 ts at salting-out euglobins from aqueous solution. [This series is sometimes termed, somewhat erroneously, the "chaotropic" series. The term chaotropic is commonly used in reference to the "disorganization" of water molecules in bulk solution by various solutes (7, 8). Since the chaotropic properties of different neu- tral salts (as measured by various physical methods) are not always correlated with their relative tendencies to salt-out proteins, it has been suggested that this terminology be dropped (W. P. Jencks, personal communication).] Table l is a comprehensive list of neutral salts in the order of increasing effectiveness at salting- out (i.e. decreasing lyotropicity) of the peptide analogue, acetyltetraglycine ethyl ester (10). A similar ranking has been shown for the effect of neutral salts on the collagen-gelatin phase transition, ribonuclease denaturation, DNA unfolding, the polyvinyl- methoxazolidinone cloud point (precipitation) and other phenomena (8, 11). Some aspects of the working hypothesis which will be pre- sented in the next sections are unpublished. Other aspects of the hypothesis are derived from the work of Tanford (12), von Hippel and Schleich (8), Kuntz and Kauzmann (13), Robinson and Jencks (10), Felgner and Wilson (3), Roseman and Jencks (l4) and Kauzmann (29). Basically the hypothesis represents a synthesis between the work ()f' von Hippel et a1. (6, 15, l6, l7) and Melander and Horvath (l). Table l.--Approximate Salting-gut Constants for Acetyltetraglycine Ethyl Ester at 25.o° Compound KSb .Compound KSb Lithium-3,5- diiodosalicylate -l.3 NaBr 0.00 Phenol -0.48 MgCl2 0.00 NaClO4 -0.33 (CH3)4NBr 0.018 Sodium tosylate -0.31 LiCl 0.021 C H NH Cl -0.31 NH Cl 0.035 6 5 3 4 Lil -0.28 KCl 0.046 013CC00Na -0.27 NaCl 0.046 NaSCN -O.25 CsCl 0.054 NaI -0.23 NaBrO3 0.090 KI -O.21 (CH3)3CCOONa 0.15 LiBr -0.l7 Glycine 0.16 C6H5C00Na -0.l4 NaHSO3 0.16 NH4Br -0.ll CH3COONa 0.23 BaCl2 -0.ll NaH2P04 0.36 CaCl2 -0.09 (NH4)ZSO4 0.45 NaNO3 -0.075 NaZSO4 0.48 KBr- -0.023 Na3 citrate 0.90 a From reference 10. . bLog S°/S = K M, where 5° is the solubility in water, S 18 the solubility in Other solvent, and M is the concentration of 5a] 1: moles per liter. Values of K were estimated from solubility measurements at salt concentration O-O.5 M. "Salting-in" type 531 ts have large negative values of K and "salting-out" type 53] ts have large positive values of Ks' There is an enormous amount of data regarding the effect of neutral salts on numerous solvent parameters such as water activity, viscosity, heat capacity, entropy of solution, ionic mobility, self-diffusion of water, internal pressure and spectral properties (I.R., NMR and Raman) and the literature is likewise replete with datacn1the effects of neutral salts on macromolecular structure (8, 10). All of these data must eventually be included in a coherent framework describing the effects of solvent and salts on macromolecular structure. In this chapter, however, I will concentrate on only two prominent factors that, when considered together, can explain much of the data as well as provide a molecular interpretation of the Hofmeister series. These facts are that (1) neutral salt binding affinities to the peptide bond dipole increase in the same order as the Hofmeister series (15,16) and (2) the surface tension increment (imL that amount by which the surface tension of pure water is changed due to the addition of a mole of solute) increases in the same order as the Hofmeister series (1). These two points will be dealt with separ- ately and then considered together in order to draw a molecular picture of the effects of salts on macromolecules. Neutral Salt Binding Affinities to the Peptide Bond Dipole Von Hippel and his colleagues (15, 17) have been atxle to determine relative binding constants of various O=C-NH . 2 neutral salts to polyacrylamide columns (-CH2-C - ). Comparative H 9“2¢ measurements on a polystyrene column (-CH2-C- ) indicate that the H bindings occurs only to the amide moieties and not to the hydro- phobic backbone or to the hydration shell surrounding the backbone of the polyacrylamide. Ions that bind tighter than water to the dipole, tend to destabilize macromolecular conformations, i.e. are more lyotropic. Likewise, those salts that tend to stabilize macromolecular conformations bind less tightly than water to the acrylamidecolumn. In fact, the lyotropic series as it relates to conformational stability of macromolecules is fully developed in the acrylamide binding experiments so that the ions with the greatest structure destabilizing characteristics bind tightest and those that tend to stabilize macromolecular structure, bind the weakest. From these studies then, von Hippel infers that the destabilizing characteristics of salts are the result of direct binding to the macromolecule (6, 15, 16, 17). The stabilizing characteristics probably arise because of some other effect that salts have on the solvent PrOperties of water favoring hydrophobic interactions. Insight into the nature of this structure stabilizing effect comes from the effect of salts on the surface tension of water which will be {I dealtwith in more detail later. For the present I want to emphasize that the overall effect of a particular salt on macromolecular structure depends on the relative magnitudes, for that salt, of two opposing effects: the first is a structure destabilizing com- ponent which arises from direct binding of the salt to polar moieties in the macromolecule, and the second is a structure stabilizing component which results from the tendency of the salt to stabilize hydrophobic interactions by changing the solvent properties of water. (At this point the phrase "solvent properties of water" may still seem vague. This will, hopefully, become sonewhat clearer later.) Von Hippel and his colleagues (ll, 13) have also demon- strated that the affinity of the various salts for an ion retarda- tion column varies with the solvent used to elute the salt. Elution by water was compared to the elution by various other solvents that were considered to be peptide bond analogues, such as formamide, acetamide, N-methyl acetamide, N-methyl formamide, and N,N-dimethyl formamide. The relative affinities of the salts for these solvents varied with the amount of methyl substitution around the dipole, such that the more methyl groups around the dipole, the lower the affinity of the salt for the dipole. The decrease in affinity for the dipole that results from added lnethyl groups is most marked for NaCl; then comes NaBr, NaI and NaClO4, in that order (consistent with the lyotropic series). Ill] salts bind approximately as well to formamide which represents an "ideal" peptide bond dipole with no vicinal non-polar groups. Thus it appears that the Hofmeister specificity of binding to the peptide bond dipole is dependent on the nonpolar environment in the immediate vicinity of the dipole. There are at least two ways to explain these data (8, 17): either the methyl groups around the dipole cause an inductive effect on the dipole which introduces the Hofmeister specificity, or the methyl groups perturb the water structure in the vicinity of the dipole and this leads to the Hofmeister specificity. The first hypothesis doesn't seem likely because methyl groups substituted in different positions around the dipole give approximately the same effect (i.e. salts bind about as tightly to N-methylacetamide as to N,N dimethyl formamide). If an inductive effect were operating, one would expect substitution at the carbonyl moiety to give substantially different results than substitution at the amide nitrogen. Furthermore, data on the effect of salts on the surface tension of water suggest another hypothesis that accounts for the observed Hofmeister binding specificity (see below). The Effect of Neutral Salts on the Surface Tension Increment Recently Melander and Horvath (1) reported a relationship between the extent to which a given salt influences the surface tension of pure water and its position in the lyotropic (Hofmeister) :serfies. The expression which approximately describes the surface tension of many inorganic salt solutions is 10 1=v°+om where yo is the surface tension of pure water, m is the molality of the salt, and, for a given salt, 7 is a constant called the molal surface tension increment (l, 18). In Table II is a list of inorganic salts and their surface tension increments. Notice that all salts give positive values of U which means that they all increase the surface tension of water. In Table l are listed the salting-out constants for many of the salts listed in Table 2. A comparison of these two tables indicates that, in general, the larger the salting-out constant, the larger the surface tension increment. The plot of 0 vs. KS (Figure l) emphasizes this relationship. In their paper, Melander and Horvath (l) similarly demonstrated that the con- centration dependence of salt-induced protein flocculation for different salts bears a similar relationship to the surface tension increment of the salt used. That is, salts that induce flocculation (precipitation) of proteins at lower concentrations have higher surface tension increments. Physical Description of the Surface Tension Increment According to Moore (19) and‘others (20, 21), surface tension is a reflection of the internal cohesiveness of the bulk solution. If the solution properties are such that molecules in the bulk Iflhase have a high affinity for one another, this will be reflected 'in a high value for the surface tension. If the molecules in 11 Table 2.--Molal Surface Tension Increments of Various Salts (x 103 26166?) Salt (x 103 ERIREI‘ Salt 0.45 KSCN 1.96 Kz-tartarate 0.55 100103 2.0 8a(N03)2 0.74 NH4I 2.0 LiF 0.79 LiI 2.02 Na2HPO4 0.84 KI 2.10 N1504 0.85 111141103 2.10 119304 0.86 K0103 2.10 Mnso4 1.02 NaI 2.15 00504 1.06 NaN03 2.16 (NH4)2504 1.14 NH4Br 2.27 ZnSO4 1.16 LiNO3 2.35 Naz-tartarate 1.26 LiBr 2.58 K2504 1.31 KBr 2.66 Na3P04 1.32 NaBr 2.73 Na2504 1.39 C51 2.78 1.12504 1.39 NH4C1 2.78 Fec13 1.4 K0104 2.93 8amZ 1.55 Peso4 3.12 K3-citrate 1.63 LiCl 3.16 M9012 1.64 NaCl 3.66 Cam2 1.57 051103 3.9 K4Fe(CN)6 1.82 Cuso4 4.3 K3FE(CN)6 aTaken from reference 1. 12 Figure l The Relationships Between the Solubility Coefficent and the Surface Tension Increment The values for Ks and 0 were obtained from Tables 1 and 2, respectively. 13 m6. m6 .3.— 14 solution have a low affinity for one another, the solution will Spread out easily and one will observe a low value for the surface tension. Water molecules exhibit particularly strong intermolecular interactions vig_hydrogen binding. that, in large part, leads to the high values for the surface ten- It is this special property sion of pure water relative to other liquids at comparable temper- ature (19, 20). It is misleading to think that great cohesive energy between the surface molecules is the only factor producing surface tension. The surface molecules do not hold bulk water in a ball like a rubber balloon holds air, but rather water prefers the bulk phase so that it seeks to minimize the surface area. This is reflected in high surface tension. Putting it in thermodynamic terms, the free energy of the water molecules in the bulk phase l5 1 ess than the free energy of the surface molecules. The solu- tion seeks to minimize the free energy by minimizing the surface area - Another useful way to state the point is that it takes work to move a molecule of water from the bulk phase and place it on the surface (19-22). The more work it takes to do this, the 9Fe3 ter will be the surface tension. The increase in the surface tension that results from added 53] t can be understood from the definition of surface tension. SlumFace tension is defined as the amount of work, dw, required to inchease the surface area by an amount dx (21). To do this one must remove water molecules from the bulk phase and put them on the surface. Therefore, anything that makes it more difficult to 15 extract molecules from the bulk phase will increase the surface tension. By this interpretation, salts increase the surface tension of water because they (somehow) increase the cohesiveness of bulk water. It takes more work to extract a water molecule from a salt water solution and put it on the surface, than from pure water; therefbre, one observes an increase in the surface tension. Data on ‘the enthalpy of hydration for the different anions should pre- sumably bear on the validity of this interpretation, because enthalpy is generally considered to be a reflection of bonding energy (21, 23). Thus, ions that bind water molecules tightly WOLrlcj be expected to give high values for the hydration enthalpy and would also be expected to make those water molecules less ava i‘lable to create a new interface. This decreased ability of waizeer~ to escape from the bulk phase could lead to an increase in the: ssurface tension. The enthalpies of hydration have been deter- mlrieecd for the halides and for perchlorate. For these five anions thE3 {enthalpy of hydration decreases in the expected order, i.e., F > 01’ > Br' >1' > 0104' (24). What is the Relationship Between Surface Tension, o, and the Hydrophobic Effect? Surface tension is a reflection of the tendency for a Litiuid to form spherical drops (19, 20, 21) and thereby minimize .tI‘SE lair/water interfacial surface area which, therefore, minimizes ‘tI‘EE free energy of the drop. A hydrocarbon in an aqueous solution creates a hydrocarbon/water interface, similar to the air/water 16 interface (10, 25, 26). Consequently, hydrocarbon molecules in aqueous solution tend to associate in an attempt to minimize the area of the interface. It follows then that anything one does to increase the interfacial tension will lead to an increased tendency for hydrocarbon moieties to associate, so that the interfacial area will be minimized. As mentioned earlier, the relative tendencies of different salts to salt-out proteins depends on the relative magnitudes of a salting-in component (which depends on direct binding of the salt to the macromolecule) and a salting-out component which involves the sol vent properties of the solution and the ability of the solvent to accommodate hydr0phobic surfaces. It is consistent with the prev ious interpretation that the surface tension increment is a dil‘E3<:t measure of the solvating properties of salt solutions, such that , for a larger surface tension increment, the solution will have a smaller tendency to solvate hydrophobic groups. In other WCW‘cls, salts with large surface tension increments will tend to prolhote hydrophobic interactions. The ability of the lyotropic salts to salt-in proteins is 9X91 ained by their ability to bind with greater affinities to hydrophobically shielded, charged sites as indicated by the data of I'C’VW Hippel. The reasons for these different binding affinities ‘3Elr1 also be interpreted in terms of the surface tension increment. \’S>r1 Hippel determined that all salts bind with equal affinity to il" ideal peptide bond dipole containing no hydrophobic groups around it. In line with the present hypothesis, the hydrocarbon 17 noieties around the dipole form an interface with a layer of water at the interface. As will be discussed below, this water layer acts like a barrier that interferes with the approaching salt molecules, lowering their effective concentration in the vicinity of the dipole and decreasing their apparent binding affinities. This "barrier" is rmare difficult for non-lyotropic salts to penetrate, which leads to a (decreased binding affinity for the charge site when compared to l more lyotropic salts . “'1 In thermodynamic terms the data of von Hippel can be des- cri bed in the following way. There is an intrinsic association [ constant, Keq, defined as, Ken = [1332131231]. int Which is characteristic of the interaction between any salt and an 1Peal peptide bond dipole with no vicinal hydrophobic groups. In the present situation, [salt]int specifically refers to the concen- tration of salt at the interface with the aqueous phase inmediately adjacent to the dipole; this is signified by the subscript "int." 1" 1the absence of secondary effects resulting from the presence of VicI‘inal hydrophobic groups, [salt]int is equal to the concentration i" 1the bulk solution [saltlbulk' When hydrophobic groups are placed 'i" ‘the vicinity of the dipole, the effective concentration of the 3a] 1: around the dipole is reduced by a factor, f, such that, flsalt]bu1k = [salt]int. 18 The expression for the apparent association constant, Keqapp, then becomes , = [dipole-salt] = Keqapp [dipole] 1sa1t1bulk f Keq where f is close to l for the most lyotropic salts, and approaches zero for the less lyotropic salts. Further evidence that salts are excluded from the hydrocar- bon/water interface comes from the thermodynamic treatment of the variation of surface tension with composition, as derived by J. W. Gibbs (20, 22). His model describes the change in the surface ten- sion, dy, at an interface of undefined thickness, between the bulk Phases (such as air and water or hydrocarbon and water) as, dv = -SdT - z P.dp. . (1) i=1 ' ' The quantity S, the surface excess entropy (22), is the amount by "hi ch the entropy per unit area of surface exceeds the entropy of the bulk phase. The quantity Pi’ is the moles of component i at the i"terface region (of depth x cm) in moles per cm surface area (Fri saure 2). Pi is the chemical potential of component i. For a “'0 component system at constant temperature and pressure and by aPID‘Iying a Gibbs-Duhem expression (20, 22), one gets, 'dY/d“2 = [F2'(N2/N1)r1] = I2(1)‘ Now the surface excess of component 2, i.e. 1‘2“), is the amount (1 r1 moles/cm2 surface) by which 1"2 exceeds the quantity of component 2 that would be associated with I“ of component 1 in the bulk phase. 19 Figure 2 The Relationship Between r and 6 s = solute molecules (the empty space contains solvent) x = the depth of the interface F](2) = moles of S/cm2 surface at depth x that is less than the moles of solute in a similar volume in the bulk phase (r](2) is a negative number). Likewise, r2(]) = moles of excess solvent in the interfacial phase rela- tive to the bulk phase F](2)/x = the 1nterfac1al concentrat1on of solute To get a difference in the solute concentration between a surface phase (defined at depth, x) and a bulk phase, the solute molecules may be organized in either of two ways. Either the solute molecules are distributed uniformly throughout each phase (I) or nonuniformly as in diagram II. In diagram II the value of r is the result of a surface layer, of thickness, 5, that is solute free (the ion free layer). The concentration of solute in the layer defined by the distance, y, is the same as in the bulk phase. omega 5.5 omega .3823... 21 N1 and N2 are the mole fractions of the respective components in the bulk phase. The important point to glean from this relationship is that a difference in the solute concentration at the surface of a solution relative to the bulk phase can contribute to the inter- facial tension. If component 2 is the solvent (water) and component 1 a solute (salt) then a surface excess of component 2 (positive F201) will give rise to an increase in the surface tension. Likewise, a surface decrement in component 2 (1‘2“) negative) gives rise to a decrease in the surface tension. Direct measurements of surface excess quantities and Av have verified the Gibbs equation (20). In order to get some idea what the actual concentration differences are between solutes in the bulk phase vs. solutes at an interface it is necessary to come up with some means of defining tlhe depth of the interface. One way that this can be done is by assuming that the surface excess of solvent (12(1)) at the inter- face is due to a surface layer of solvent that is completely Solute free and that the molecules of solute present in the inter- facial layer are surrounded by the same number of solvent molecules as in the bulk phase. The amount of excess solvent (r201) in the interfacial phase will then be given by the depth of the solute free layer. For an aqueous solution of salt this distance, 5, is Called the ion free layer, given by, 22 moles) 18 cm3 mole 6 cm = I" ( 2(1) cmz where 18 cm3/mole is the molar volume of water (see Figure 2). It has been demonstrated that a relationship similar to that used to define 0 can be written for 6 such that, Y = Yo 1‘ 6m VQK (4) where m is the molality of added salt, 9 is the moles of ions per nmfletof solute and 0, the osmotic coefficient of the solution (22). Plc>ts of Av vs. va yield straight lines with slope, K6 where K is a <:onstant that includes the temperature in °K, the gas constant and the molar volume of water. The order of increasing 6 for different arrions is SCN' < 0103' < Cl' < 504: < FeCN6E again consistent with the Hofmeister series (22). Since the enthalpies of hydration of the anions show an Obvious relationship to 6, (22) the inference is that the larger Values of 6 are the reflection of a larger hydration sphere around the anions. This larger hydration sphere leads to larger values of <5 because the anion is limited in its approach to the surface by 1:he depth of its hydration sphere. There are other methods for (talculating the depth of the hydration sphere around ions. The approaches have been reviewed by Marcus and Kertes (27). These Values for different anions have been tabulated in Table III. And again we see that the depth of the hydration layer follows the Table 3.—-The Hydration Radii of Several Ions , 0a A Oh A o A A hyd A hyd A re rhi sphere rhi sphere 6 F' .23 3.52 2.29 3.06 1.83 S04= .18 3.79 1.61 3.64 1.46 5.10 01’ .02 3.32 1.30 2.55 0.53 3.46 Br' .21 3.30 1.09 2.42 0.21 N03" .34 3.35 1.01 2.23 -0.11 I' .48 3.31 0.83 2.04 -0.44 C104" .64 3.38 0.74 2.43 -0.21 1.67 SCN' .57 - - 2.21 -0.36 0.80 re = electronic radius rhi = hydrated radius hyd sphere - rhi - re The two different values of rhi were obtained from two Clifferent experimental approaches (27). The values of 6 (the ion ‘Free layer) are from refernence 22. aCalculated from the partial molal ionic volume of the salt solution (27). bCalculated using ionic mobilities and Stoke's law (27). 24 lyotropic series, which is related to the surface tension increment and 6, the ion free layer. Summar The following picture then emerges for the molecular basis of the binding data of von Hippel et al.: (1) All salts bind equally well to an ideal peptide bond dipole with no hydrophobic groups in the immediate vicinity of the dipole. As hydrophobic groups are placed around the dipole, a water/hydrocarbon interface is created similar to the air/water interface. The less lyotropic salts bind less tightly to the dipole because they are excluded from the interfacial water around the hydrophobically shielded dipole. The extent of exclusion from this interface is determined by the depth of the hydration layer of the ion. (2) If the salts have a high binding affinity for the Incotein,they will tend to salt it into solution. If they do not have a high binding affinity, they will increase the interfacial tension until additional hydrophobic associations are favored and the protein will be salted-out. Or, in other words, the 'increased interfacial tension produced by added salt, results in a greater tendency for the solution to minimize the amount of ltydrocarbon/water interface and several proteins will contact each (Ither at their hydrophobic surfaces leading to large molecular \veight aggregates and salting-out. The salting-in phenomena can be described by the following equilibrium. 25 protein + salt +—»-protein-salt (insoluble) (soluble) Binding of salt to regions of the macromolecule that are not readily hydrated by water, makes that molecule more soluble by making it more readily hydrated and hence more soluble. Take, for example, an o—helical segment of a protein with its peptide bond involved in intrahelical hydrogen bonds: A 5 6+ 6+ 'C' o - — c - “I 6 ll _ c ’ 06 A‘ I + C+A- 4__ 1” I H6+ H6+ _,I _ I :6" ”6‘ 0* A B The segment of the protein molecule labeled A is expected to be less water soluble than segment B. This is because, due to ‘intrahelical hydrogen bonding, the amide group is not accessible ‘For hydrogen bonding with water. Binding of salt to the amide group (segment 3) breaks the helix and allows interaction of the amide moiety with water through hydrogen bonding, as well as allowing ionic interactions with other ions in the bulk solution. Disruption of a-helical structure by neutral salts has been demon- strated (8). This is somewhat analogous to the situation with 26 some small molecules like short chain fatty acids that are rela- tively water insoluble at low pH's (when they are uncharged) but are more water soluble at high pH's as the salt. Theoretical Treatment of the Salt Effects Data on the Hexokinase/Membrane Interaction Melander and Horvath (l) have derived, from thermodynamic principles, an equation which can numerically quantitate the rela- tive importance of electrostatic and hydrophobic forces in the interaction of small molecules or macromolecules with hydrophobic affinity columns. The equation is of the form 1n k/ko ‘ - 8 -Am + 00m (1) where k0, the capacity factor in the absence of salt, is the reten- tion volume of the solute minus the holdup volume of an unretained solute, k is the capacity factor in the presence of added salt, n1 is the molality of added salt, 0 the surface tension increment fkar a given salt, 8 an electrostatic "salting-in" component pro— FNDrtional to the dipole moments of the interacting groups and 0 'is proportional to the square angstrom contact area between the Ilydrophobic groups of a macromolecule and the hydrophobic affinity Column. As indicated by this equation, at low salt concentrations (In 5 o) the capacity factor (k) is largely determined by the Constant, B; at higher salt concentrations the other terms become Dredomi nant . 27 To convert the above equation into a form applicable to the hexokinase/membrane interaction, we note that the capacity factor, is defined (28) by the equation, 1n k = ln Kassoc - e (2) is the association constant for the equilibrium where Kassoc expression, 3 + L 5%=SL . (3) S is the concentration of free solute applied to the affinity column, L is the concentration of free hydrophobic binding sites on the affinity column, and SL the concentration of bound solute. e is a constant characteristic of each individual column. The hexokinase/membrane interaction can be similarly written E + M 5%: EM (4) kniere E, is the soluble hexokinase, M, the membrane binding site arui EM, the amount of bound enzyme. For this expression an asso- <2iation constant can be written, = [EM] (5) eq IE] IMI and equation (1) becomes K 99 ln /Ker = -B - Am + 00m (6) 28 where Ke is the equilibrium constant in the absence of added 0 salt. According to this equation plots of ln eq/Keq vs. m give straight lines with slope equal to (Oo-A) and y integcept equal to -8 as in figure 3A. This figure (as well as equation 6) illustrates that the salt dependence for solubilizationis the net result of two components, one of which, A, tends to solubilize the enzyme and the other, 00, which tends to hold the enzyme on the membrane. Since the slope, A is greater than 00, the overall effect is to solubilize the enzyme. 8 is a constant that describes the effect of low salt on the equilibrium constant. In this example (figure 3A) the positive value of 8 indicates that low salt increases the association constant of the enzyme for the membrane. The slope A is a constant characteristic of the interaction and is related to the dipole moments of the interacting species. Likewise, 0 is a constant, but the slope 06 depends on the value of 0 which is ar1 empirically determined characteristic of the salt used (Table 2) . 0 can be related by a complex equation (given in reference 1:) to the square angstrom contact area between interacting hydro- Phobic groups. The experimental data for the solubilization of hexokinase tflv salts as in reference 3 (appendix) was usually presented by FVIotting the fraction of enzyme solubilized versus the molality of 'tlie salt added. If, E, is the fraction solubilized, such that = ("Eel/E 2 eqo 0 Where E0 is the fraction soluble at zero salt. By solving equation E + EM =1.0, then Keq =0 ' 51/152. Likewise, K 29 Figure 3 Plots using Equation 6 with Assumed Values for All Constants The constants used to generate these curves were 0 = 15.7, A = 45.3, and B = -0.5. The salt used was NaCl which has a value of o = 1.64. E0 was equal to 0.05. The values were chosen to fit the experimental data (see Figure 4 and text). 30 0.5 *- l 0.1 0.3 Molal Salt 0.5 31 6 for E at different salt concentrations, plots such as figure 38 are obtained. (The values of B, A, 0 and Ker were the same as for figure 3A.) 0n the x asis is the molality of salt added to the membrane bound enzyme and on the y axis is the fraction solubilized. The addition of salt gradually solubilizes the enzyme until virtually all is solubilized. Also notice that the curve is sigmoidal indicating that low concentrations of salt have little solubilizing effect. Figure 4 shows several curves constructed in a fashion similar to Figure 3B with the same values for A, 0 and 8. Notice that the curves shift depending on the salt used. This is because each salt has a characteristic 0 which changes the slope, Oo.while A,§L and 8 stay constant (figure 3A). The constants (0, A, and 8) used to generate the curves on figures 3 and 4 were chosen to fit experimental data (3). The points plotted on Figure 4 are the experimental data points ob- tained for the hexokinase-membrane interaction. Notice that these data points fit the theoretical curves fairly well. This means that the constants A, B, and 0 are characteristic values for the hexokinase/membrane interaction. To calculate the nonpolar contact area between enzyme and membrane ME>can substitute the value for 0 (from Figure 4) into the equation 0 = 4110 - 12 32 Figure 4 Plots Using Equation 6 with Assumed Values for All Constants; the Effect of Differ- ent Salts All constants were the same as in Figure 3. The values for o, for KSCN, KI, NaNO3 and NaCl, were 0.45, 0.84, 1.06 and 1.64 respectively (Table 2). The data for KSCN (I), KI (V), NaNO3 (I) and NaCl (0) were obtained from reference 3. The lines are theoretically calculated with equation 6. 0.5 0.1 33 KSC KI NaNO, J 0.3 Molal Salt NaCl 0.5 34 where 0 is the nonpolar contact area in square angstroms (l). 0 from Figure 4 is 15.7, so that 6 is equal to 6,440 A2. Assuming a value of 0.74 gm/cm3 and a molecular weight of 100,000 for the hexokinase molecule it can be calculated that the total surface area of hexokinase (if the molecule is spherical) is about 12,000 AZ. This means that about half of the total surface area is in contact with hydrophobic groups on the membrane. In other words, half of the molecule is embedded in the hydrophobic portion of the membrane. While this calculation may be extending the implications of the theory to the extreme, it is nevertheless interesting that it gives a somewhat reasonable value for the nonpolar contact area. The value of this development is that numbers can be calculated which affix relative importance to hydrophobic vs. electrostatic effects in any given interaction where experimental data can be obtained on the effect of salts. The equations used to generate these numbers evolve out of a theoretical framework using principles of thermodynamics, so that the numbers have physical chemical meaning. It will be very interesting to see this approach applied to other systems so that the magnitudes of 0 and A can be compared. "I'm CHAPTER II SOLUBILIZATION, RECONSTITUTION AND CHARAC- TERISTICS OF THE OUTER MITOCHONDRIAL MEMBRANE BINDING SITE FOR HEXOKINASE Introduction The mitochondrial hexokinase from brain can be solubilized by low concentrations of g1ucose-6-P (30, 31). In the presence of M9012, the enzyme rebinds to the outer mitochondrial membrane. Investigations dealing with the effect of neutral salts on the association between hexokinase and the mitochondrion, have provided the basis for a hypothesis describing the interactions between the outer mitochondrial membrane and hexokinase (3, appendix). These interactions appear to be primarily electrostatic in nature, in accord with Teichgraber and Biesold (32) and include both repul- sive and attractive components. As indicated in the first chapter of this thesis, there also appears to be a substantial hydrophobic component to the inter- action. The extent of this hydrophobic component can be roughly approximated in terms of the square angstrom surface area of inter- acting hydrophobic groups. In order to reconcile the presence of a~Il.Ydrophob1‘c effect with the pr0posal that the interaction is primarily electrostatic we have suggested that the electrostatic. interactions occur in a hydrophobic milieu. When compared to "ideal" 35 36 ionic bonds (i.e. ionic bonds ngt_surrounded by hydrophobic groups), they are more difficult to dissociate because they are protected from attack by water and by salts. Lyotropic salts are more effective at disrupting the interaction because they can penetrate into a hydr0phobic milieu more readily than non-lyotropic salts. The basis for these conclusions can be found in reference 3 included in the appendix. So it develops that from the effects of neutral salts on the hexokinase membrane interaction, we were able to draw a more T." lucid picture of the hexokinase/membrane interaction. But after all this work was done, several key questions about the nature of the binding site still remained unanswered. What is the role of protein in the binding site? Is there a specific binding protein? What is the role of lipid? Is there a specific lipid or phospho- lipid ratio that is required to give the proper binding specificity and solubilization characteristics? In this Chapter I outline several approaches directed at these questions. The experiments fall into three general categories. I will now briefly summarize and give the conclusions of each approach. Unlike the studies outlined in the first chapter and in reference 3 (appendix) which utilized the rat brain enzyme and hexokinase binding sites from the rat brain particulate fraction, the studies presented in this chapter utilized rat liver outer "fitochondrial membranes (OMM) which could be obtained in a very pure form and in high yields. The OMM were found to contain 37 binding sites for hexokinase that had very similar properties to those of brain mitochondria (33). 1. Specificity of the outer mitochondrial membrane binding, §jt§:-(hrter mitochondrial membranes (OMM) from rat liver were purified and the specific activity of the binding site (in units of hexokinase bound/mg membrane protein) determined. When compared to various other membranes, the OMM had a specific activity of binding at least 40 fold greater, thus indicating that some specific factor(s) resides in the OMM that is required for binding. Some liposomes and liposome/protein mixtures could be shown to bind hexokinase but none of these showed the glucose-6-P dependent solubilization effect that is characteristic of the OMM. This result once again pointed to some specific characteristic(s) of the OMM that is (are) required to give the appropriate binding pro- perties. 2. The role of lipid in the binding site-~I discovered that treatment of the OMM with the non-ionic detergent, Emulgen 913, under the appropriate conditions would extract the phospho- lipid and most of the cholesterol from the OMM, without solubilizing the proteins. This treatment resulted in the loss of the glucose- 6-P dependent solubilization effect (3, 31 ). When chloroform/ "ethanol extracted lipids from the OMM were added back to the nembrane proteins, the glucose-6-P dependent solubilization effect was restored. (See the Methods section for an explanation of what is meant experimentally by the "glucose-6-P dependent solubilization 38 effect.") Since lipids extracted from microsomes (which do not contain a binding site for hexokinase) also reconstituted the glucose-6-P dependent solubilization effect, it was concluded that the reason for the glucose-6-P sensitive, specific binding of hexokinase did not reside in the outer mitochondrial lipids alone. 3. The role of protein in the binding site--Several experiments aimed at chemically or enzymatically inactivating a proteinacious binding site in intact OMM gave negative results and it was consequently not possible by this approach to either confirm or rule out the hypothesis that protein was involved. Another more direct approach toward determining the nature of the binding site involved solubilization, fractionation and reconstitution of a putative binding protein. The success of this approach was obviously contingent on the existence of a binding protein, for which there was no direct experimental evidence. After many fruitless attempts using different detergents and different methods of reconstitution, it finally became possible to prepare reconstituted membranous vesicles, greatly enriched in a single protein, that gave glucose-6—P-sensitve binding of hexo- kinase. Materials and Methods Chemicals Biochemicals and HEPES buffer were obtained from Sigma Chemical Company. All other chemicals were reagent grade, obtained from comercial sources . 39 Adult male and female rats (ranging from 150-500 gm) of the Sprague-Dawley type were obtained from Spartan Research (Haslett, Michigan) and maintained on a common laboratory diet and water ad libitum. Hexokinase Assay Hexokinase was assayed spectrophotometrically by the glucose-6-P dehydrogenase method as previously described (34). Preparation of Mitochondria Rat liver mitochondria were prepared by homogenizing the liver from a starved rat (15-17 hr, with water ad libitum) in 10 volumes (10 m1/gm tissue) cold 0.25 M_sucrose with a Teflon-glass homogenizer (4-6 strokes). The homogenate was centrifuged at 600 x g for 10 min and the pellet discarded. The supernatant was centrifuged at 6,500 x g for 15 min and the supernatant discarded. The 6,500 x g pellet was resuspended in 10 volumes sucrose and centrifuged at 6,500 x g for 15 min. The final pellet was resus- pended in 2 volumes (based on original liver weight) of 0.25 M sucrose and 1 ml aliquots were stored at -20°C. Crude rat brain mitochondria were prepared as described in reference 3. GT ucose-6-Phosphate Sol ubil ized Enzyme Rat brains, frozen in liquid N2, were thawed and homogenized T'1 0.25 M_sucrose (10 ml/gm). The homogenate was centrifuged at 40 1000 x g for 10 min, and the pellet discarded. The 1000 x g supernatant was centrifuged at 40,000 x g for 10 min and the resulting pellet washed once by rehomogenization in 10 volumes 0.25 M_sucrose and centrifugation. The washed pellet was resus- pended in 10 volumes 0.25 M_sucrose containing 1.2 mM_glucose-6-P and incubated for 30 min at 25°C. The solubilized enzyme was obtained in the supernatant after centrifugation at 40,000 x g for 30 min and stored frozen at -20°C in 10 ml aliquots. Just prior to use the thawed aliquots were centrifuged at 160,000 x g for 30 min to remove particulate material. Preparation of Purified Outer Mitochondrial Membranes Rat liver mitochondria were prepared from 5 or 6 rats (85-120 gms of liver) by differential centrifugation from a 10% (w/v) homogenate (3 strokes w/homogenizer) in 0.25 M sucrose. After sedimentation of the nuclear fraction at 600 x g (2,250 rpm, 9RA rotor) for 15 minutes, mitochondria were sedimented from the supernatant by centrifugation at 6500 x g (7,500 rpm, 9RA rotor) for 20 minutes. The pellet was washed twice, with one half and one fourth of the initial volume of sucrose. The fluffy layer on top of the pellet was carefully removed after the last wash by aspirating with a Pasteur pipet. This procedure was slightly different from the one used when only crude liver mitochondria were required (see above). The final pellets were suspended in 90 m1 of ice-cold 10 mM_Tris-phosphate (10 mM Tris + phosphoric acid to pH 7.5) buffer, 41 pH 7.5, by means of a Teflon pestle fitted into the centrifuge tube. After standing at 0° for 5 minutes (during which time the mitochondria underwent swelling) 30 m1 of a solution containing 1.8 M sucrose, 2 mM_ATP, and 2 mM_MgS04 was added to the suspension. A visible increase in turbidity immediately appeared due to contraction of the mitochondria. After another 5 minutes at 0°, the suspension was subjected, in aliquots of 20 ml, to sonic oscillation at 3 amperes with a Branson Sonifier for 20 seconds at 0°. The sonicated mitochondria were divided equally among freshly prepared sucrose density step gradients. Each gradient contained 5 m1 of 1.32 M, 5 ml of 1.00 fl_and 5 ml of 0.76 M_ sucrose, and 20 mls of the sonicated mitochondria. The gradients were centrifuged at 25,000 rpm in the Beckman SW 27.1 rotor for 4.5 hrs. The purest outer membrane fraction sediments to the interface between 0.76 M_and 1.0 M_sucrose. This fraction was diluted with approximately 4 volumes of water and pelleted at 100,000 x g for 90 min. The pelleted OMM were taken up in 0.25 M_sucrose to a concentration of 2-4 mg protein/ml and frozen in 0.5 ml aliquots at -20°. 1% Sodium Dodecyl Sulfate-Polyacrylamide Disc Gel Electrophoresis The method of Fairbanks et a1. (55) was somewhat modified. Samples (1-15 mg/ml) were prepared in 1% sodium dodecyl sulfate, 5-10% sucrose, 10 mM_Tris-HC1 (pH 8.0), 1 mM EDTA and 2% 42 B-mercaptoethanol. They were then heated at 100° for 15 min, 0.2% pyronin B tracking dye was added,and the samples were layered on 5.6% polyacrylamide gels (5 mm x 100 mm) prepared in tubes which had been coated with dimethyl dichlorosilane. Electrophoresis was performed at constant current of 4 ma/gel with a running time of about 4 hours. Gels were fixed and washed overnight by gently agitating each gel in a 30 ml capacity test tube with two changes of 10% TCA/33% isopropanol. The dehydrated gels were placed into 10% TCA until they regained their original size and then were stained with xylene brilliant cyanin-G (K + K Laboratories, Inc. Plainview, N.Y.) according to published procedures (35, 36). The rather extensive washing of the gels removes $05 that interferes with the staining. Mitochondrial Binding Assay For the assay, aliquots of glucose-6-P solubilized hexo- kinase (prepared as above) and either liver mitochondria, brain mitochondria, purified outer mitochondrial membranes (OMM) or reconstituted OMM were mixed in polycarbonate centrifuge tubes and 3 mMMgCl2 was added. The aliquots were incubated at 25°C for 10-15 min and spun at 40,000 x g for 10 min at 4°C or at 160,000 x g for 30 min if purified OMM or reconstituted OMM were used. Hexokinase activity in the supernatant was measured directly. Pellets were assayed after suspending them in a known volume of 0.5% Triton X-100 - 0.25 M_sucrose by vortexing in the presence of glass beads (Sargent, No. S-61740, size A-7) until 43 homogeneous. With each new membrane preparation the concentration of binding sites was determined by titrating a fixed volume of the membrane suspension with increasing amounts of hexokinase. In this way, it was always possible to determine when the binding sites were in excess for a given amount of hexokinase. Determination of Glucose-6-P Dependent SETubilization of Hexokinase from Purified Outer Mitochondrial Membranes Binding of hexokinase to purified outer membranes was done by incubating an aliquot of OMM (50 ul, 2.0-4.0 mg/ml) with enough g1ucose-6—P solubilized hexokinase to saturate all available binding sites (5 ml, 0.7 u/ml) and 3 mM_MgCl for 10 min at 25°C. 2 This suspension was centrifuged at 100,000 x g x 30 min. The pellet, containing bound hexokinase,was homogenized in a convenient volume (0.6 m1) of phosphate buffer (2 mM_Na-phosphate; 2 mM_ thioglycerol; 0.1 mM EDTA; 2 mM glucose; 0.25 M_sucrose; pH 6.6). 150 pl aliquots of the resuspended pellet were added to each of three tubes and 1.0 mM_of glucose-6-P or galactose-6-P was added to two of the tubes. After incubation at 25°C for 15 min the tubes were spun for 5 min at top speed in a Beckman Airfuge and the supernatants were assayed. The data were usually presented as the percent of the total hexokinase activity that was in the supernatant after incubation and centrifugation. The amount of enzyme solubilized in the presence of glucose-6-P divided by the amount solubilized in the presence of galactose—6-P (the G6P/Ga16P 44 ratio) was also used as An indicator of the magnitude of the glucose-6-P dependent solubilization effect (see below). Explanation of the Glucose-6-P/ Galactose-G-P Ratio and RationaTe For UsingiThis Ratio as a Means of Assessing the Intactness of the HexokinaseTBindipg Site In many of the studies presented in this chapter the amount of enzyme solubilized with glucose-6-P divided by the amount solubilized with galactose-6-P was used to reflect the degree of intactness of the hexokinase binding site (see Tables 5-8, 9, 10, and 12-16, and Figures 13 and 13). When this ratio exceeded 1.3, it was routinely concluded that some intact binding receptor was present in the preparation. Values of 1.1 or less were judged to contain negligible amounts of intact binding sites. Mere binding (adsorption) of hexokinase to a given membrane (or protein)prepara- tion alone, was not considered to adequately reflect the intactness of a native binding site (see Table 11). In thisthesis the G6P/ Ga16P ratios greater than 1.3 are often referred to as indicating "glucose-6-P dependent solubilization," "a glucose-6-P dependent solubilization effect," "g1ucose-6-P sensitive binding" or a glucose-6-P effect." In experiments where various agents were used to perturb the native binding site, decreases in the G6P/Ga16P ratios were sometimes observed. There are two possible factors that can give rise to such decreases in the G6P/Ga16P ratio, i.e. either the 45 amount of enzyme solubilized in the presence of glucose-6-P decreases or the amount solubilized in the presence of galactose- 6-P increases. When decreases in the 06P/0a16P ratio were observed, this was sometimes referred to as a "decrease in the glucose-6-P effect." This terminology may be somewhat misleading since the reason for the decreased G6P/Ga16P ratio was often due to an increase in the amount of enzyme solubilized in the presence of galactose-6-P without any marked effect on the amount solubilized in the presence of glucose-6-P. In these cases then it might have been more appropriate to describe the effect as a galactose-6-P effect. I did not, however, change the terminology in the body of the thesis, because no ambiguity arises as long as it is understood that when I refer to the "glucose-6-P dependent binding," I am referring to the G6P/Gal6P ratio. For reasons which remain as yet unclear, the percent of solubilized enzyme in the presence of glucose-6-P was found to vary from one experiment to the next as did the amount of enzyme released in the presence of galactose-6-P, sometimes without any deliberate change in the conditions of the experiment or without any appreciable change in the G6P/Gal6P ratio. Due to this variability in the percent solubilized by glucose-6-P and galactose- 6-P it was considered difficult to make firm conclusions about the precise effect that a given perturbant had on the binding site with respect to g1ucose-6-P or galactose-6-P solubilization. Despite this variability, the GGP/Gal6P ratio was considered, for the 'AT'””1 46 purposes of this thesis to be a sufficiently sensitive and quantita- tive as well as convenient indicator of the intactness of the native hexokinase binding site. Results Specificity of Hexokinase Binding to the Outer Mitochondrial Membrane The data in Table 4 indicate that the Specific activity for hexokinase binding by the purified outer mitochondrial membrane (OMM), is at least 40 fold higher than in erythrocytes, microsomes or inner mitochondrial membranes. SDS gel electrophoresis of purified outer mitochondrial membranes, with and without hexokinase, indicates that the binding is specific for hexokinase (Figure 5) since appreciable amounts of other proteins hithe crude hexokinase preparation are not adsorbed. Glucose-6-P Dependent Solubilization of Hexokinase from Purified OUter Mitochondrial Membranes The data in Table 5 indicates that glucose-6-P solubilized enzyme, rebound to the outer mitochondrial membrane, can be specifically eluted with glucose-6-P and not with galactose-6-P. This specific elution is similar to that observed in the crude particulate fraction from a brain homogenate, as reported previous- ly ( 3), except that the conditions for solubilization are somewhat different. With OMM the pH must be maintained at a relatively low value (pH 6.6) in order to get an observable effect and for reasons 47 Table 4.--Specificity of Hexokinase Binding to Outer Mitochondrial Membranes Units Hexokinase Bound Membrane mg Membrane Protein Outer Mitochondrial 4.1 Inner Mitochondrial 0.061 Microsomes 0.063 EDTA-Washed Microsomes 0.104 Erythrocytes (right side out) 0.021 Erythrocytes (inside out) 0.027 Microsomes were prepared fron11~at.liver by centrifuging a rat liver homogenate (10 volumes, 0.25 M_sucrose) at 40,000 x g for 10 min. This supernatant was centrifuged at 100,000 x g for 60 min. and the pellet was resuspended in 0.25 M_sucrose to a concentration of 30 mg/ml protein. Part of the microsomal fraction was washed with 20 mM_EDTA, pH 7.4, to remove loosely bound protein and ribosomes (37). Inner mitochondrial membranes were obtained as a pellet in the sucrose density gradient during the preparation of 011ter membranes (see Methods). Right side out and inside out human erythrocyte membranes were prepared according to Steck and Kant (38). The specific activity of hexokinase binding sites in each membrane preparation was expressed as the total number of units of hexokinase bound (in the presence of M9012 and excess hexokinase) per mg of membrane protein (see Methods for more details of hexo- kinase binding determination). 48 Figure 5 SDS Gels of Purified Outer Mitochondrial Membranes With and Without Added Hexokinase A. 30 ul of an (Miter membrane preparation containing 3 mg/ml of protein was prepared for SDS electrophoresis according to methods. B. As in A except that prior to preparation for electro- phoresis hexokinase was bound to the outer membrane, by the procedure described in Methods. The two bands at the top of the gel on scan B come from the crude glucose-6-P solubilized hexo- kinase preparation. 49 I Hexokinase absorbance top bottom 50 Table 5.--Glucose-6P Dependent Solubilization of Hexokinase from Purified Outer Mitochondrial Membranes. Units % G6P/ Solubilized Solubilized Ga16P No Additions .031 20 Galactose-6-P .031 20 4.7 Glucose-6-P .142 94 This experiment was done as described in the Methods section. The total units of hexokinase bound per 0.15 ml of the resuspended OMM was 0.152 units. 51 which remain unclear the spontaneous release from OMM is more sensitive to pH changes than that seen with intact mitochondria. At pH 8.0 all of the enzyme is released spontaneously from OMM (see below), whereas, crude particulate hexokinase remains 60-70% bound at pH 8.0 (.3, and appendix). Part of the explanation for this difference is apparently due to the removal of exogenous protein in the purified membrane preparation (see below). The Effect of Expgenous Protein and Polyethylene Glycol on the Hexo- kinase Membrane Interaction One readily apparent difference between the crude parti- culate enzyme and enzyme bound to purified outer membranes is that the amount of exogenous protein in the purified OMM is inuch lower. 1, therefore, decided to determine the influence of added protein on the binding affinity. From Figure 6 it is apparent that the addition of exogenous Protein (bovine serum albumin, BSA) results in a marked decrease 1'1 'the amount of spontaneous release from purified outer mito- chondrial membranes at pH 8.0. G1ucose-6-P, however, still causes eXtensive sol ubil ization. A possible explanation for the enhanced binding affinity Observed upon addition of BSA is that, because of the requirement f53V‘ BSA to be hydrated when it is in solution, it causes a reduction 1'1 'the free water concentration and leaving less water available to I31 “d to the sites between the enzyme and membrane. When the hexokinase molecule is not in contact with its binding site, there 52 Figure 6 The Effect of BSA on the Hexokinase/OMM Association Aliquots of OMM containing bound hexokinase were prepared as in Methods. To each of the several pellets was added phosphate buffer (see Methods), adjusted to pH 8.0, containing the indicated concentrations of bovine serum albumin (BSA). The BSA had been dialysed extensively [3 days, three changes against 15 volumes of water] and lyophilized prior to preparation of the 10% stock solution. The BSA stock solution was adjusted to pH 8.0 with NaOH.) To each resuspended pellet was added 1 mMglucose-6-P(o), 1 mM galactose-6—P(O) or 3 mM NaCl (I) and the samples were incubated at 25°C for 15 min. After centribugation at 160,000 x g for 30 min the supernatants were assayed to determine the percent of the total activity that was solubilized. 53 .m.\1 - O 5 0323 R 1.0 0.5 % BSA 54 must be some water molecules occupying the points between the enzyme and the membranes that are in contact when the enzyme is bound. Hence, it takes more water to hydrate the solubilized hexokinase in the presencecfl’the binding site, then it does to hydrate the enzyme when it is bound to its binding site. A reduction in the water availability might, therefore shift the equilibrium, between the enzyme and its binding site, more toward the bound form. To test this hypothesis I added increasing concentrations of poly- ethylene glycol (which like BSA must be extensively hydrated in solution) (Figure 7) to OMM bound hexokinase to determine its influence on the binding affinity. The effect of added polyethylene glycol-6000 (PEG) was similar to BSA, i.e. there was a marked reduc- tion in the amount of spontaneous release with a relatively smaller reduction in the amount of glucose-6-P releasable enzyme. At 15%; PEG the apparent affinity of hexokinase for the binding site was so high that glucose-6-P did not cuase any specific elution. An equally plausible explanation for the effect of BSA on the OMM/hexokinase interaction, is that the BSA is binding to the 0""Ws. thereby inducing some type of change in the enzyme binding SI'tE! to give the effects observed in Figure 6. This explanation is eSpecially tenable in light of the recent report that BSA binds SPecifically to the outer side of the outer mitochondrial membrane (39) . BSA binding to negatively charged regions of the 0m ““33! decrease the repulsive forces (3) between the negatively charged enzyme and negatively charged membrane, and thereby, 9T Ve rise to a higher binding affinity. Experiments 55 Figure 7 The Effect of Polyethylene Glycol on the Hexokinase/OMM Association The binding of hexokinase was done as in Figure 6 but the pellets were resuspended in phosphate buffer containing the indicated amounts of polyethylene glycol 6000. After addition of glucose-6-P (o) galactose-6-P(I ), or no additions (o) the samples were incubated at 25°C for 15 min and centrifuged as in Figure 6, and the supernatants assayed. 56 In 9140105 % 15 10 91 PEG 57 aimed at resolving this question have not yet been done. The effect of polyethylene glycol could conceivably be analogous to this BSA binding effect. Hexokinase Binding_to Liposomes In Table 6 are the results of several experiments designed to determine the binding properties of various liposome preparations. Liposomes with a net negative or neutral charge did not bind any enzyme even in the presence of MgClZ. Positively charged liposomes bound enzyme tightly but this enzyme could not be released with glucose-6-P. Presumably the positively charged liposomes bound the negatively charged enzyme molecule via attractive electrostatic interactions. There was no MgCl2 requirement for this effect. Negatively charged liposomes that had the positively charged protein lysozyme added to them, bound hexokinase. MgCl2 was required to facilitate this interaction, (Figure 8) as is observed with mito- chondria or purified OMM (3, 31, 33). The liposome bound enzyme could not, however, be specifically eluted with glucose-6-P. In fact, glucose-6-P caused decreased solubilization when compared with galactose-6-P. Presumably this indicates that glucose-6-P causes a conformation change in the enzyme that leads to a higher affinity of the enzyme for this type of liposome. Liposomes prepared from a chloroform/methanol extract of outer mitochondrial membranes gave results similar to those obtained with negatively charged liposomes, that is, no enzyme bound to them. These liposomes when sonicated extensively to produce 58 Table 6.-—Hexokinase Binding to Liposomes pH During Solubili- Units % Solubilized G6P zation Bound Control Ga16P G6P Ga16P Neutral liposome 6.6 ns 8.0 ns Negative liposome 6.6 ns 8.0 ns Positive liposome 6.6 0.257 ns ns ns 8.0 0.241 ns ns nS Lysozyme/Neg.lipo- 6.6 0.190 45 52 37 0.71 some 8.0 0.148 91 91 82 0.90 ns = not significant Liposomes with a net negative or a net positive charge were prepared according to Sessa and Weissman (40) in 10 11M Na-phosphate pH 7.5 at a concentration of 5 mg/ml of lecithin with the appro- priate amounts of cholesterol, and dicetylphosphate or stearyl amine. Liposomes coated with lysozyme were prepared by mixing 5 m1 of the negatively charged liposomes with 1 mg lysozyme. Hexokinase binding and solubilizationhes done similarly to the outer membranes (see Methods) using 0.25 mls of each liposome preparation. The composition of the solubilization buffer was the same as in Methods, except that half of the solubilization was done with the buffer adjusted to pH 8.0. 59 Figure 8 Magnesium Requirement for Binding to Lysozyme Coated Liposomes The indicated volume of lysozyme coated liposomes (prepared as described under Figure 8) and 3 m1 of glucose-6-P solubilized hexokinase (~0.6 units/m1) were incubated at 25°C for 10 min with (o) and without (0) 3 mM MgC12, The aliquots were centrifuged at 40,000 x g for 10 min and the supernatants assayed for hexokinase activity. The 25% precipitation of activity that is seen when MgCl2 is added to the enzyme (with zero liposomes) is normal for the crude glucose-6-P solubilized enzyme that has not been centrifuged at 160,000 x g. The high speed centrifugation removes some particulate material from the solubilized enzyme that can presumably bind hexokinase in the presence of MgClz. 60 l OOI-__.I.___l_ #1 % Soluble O L_ J 0.1 5 ml Liposome 0.3 61 unilamellar liposomes, still would not bind enzyme. Taken together all of the previous data indicate that there is (are) some specific characteristic(s) of the outer mitochondrial membrane that enables it to specifically bind hexokinase in a glucose-6-P dependent manner. The Role of Protein in the Binding Site; Attempts to Modify the BindingyPrpperties Chemically and Enzymatically Protease digestion.--Extensive protease digestion of OMM led to substantial modifications in the banding pattern as deter- mined by SDS gel electrophoresis (Figure 9). Several proteins were removed or reduced as the result of treatment with Chymotrypsin or trypsin. Only two of the major bands in intact membranes remained substantially unmodified. Despite these marked changes in the protein composition of the treated membranes, however, the effects on their binding properties when compared to control membranes were minimal (Table 7). There was an increase (50%) in the specific activity of the binding site in Chymotrypsin treated membranes and a slight decrease (15%) in the trypsin treated membranes, and the G6P/Ga1-6P ratios in both protease treated preparations were only slightly decreased (10%); the latter differences were, however somewhat variable and were not considered large enough to make any inferences about the involvement of protein in the binding site for hexokinase on the membrane. 62 Figure 9 SDS Gels of Protease Treated Membranes Membranes were treated as described in the legend to Table 7 and $03 gels were run as described in the Methods. 64 Table 7.--The Effect of Protease Treatment on the Outer Membrane Binding Site for Hexokinase Outer Membrane GGP ' Units Bound/ . Treatment Ga16P mg membrane protein Control 2.90 1.8 Chymotrypsin 2.76 3.0 Trypsin 2.65 1.47 Papain 2.90 3.1 Outer membranes at 1 mg protein/ml in buffer (50 mM_tris pH 7.5; 150 mM_NaC1; 2 mM.CaC12) were incubated with 1 mg/ml of three different proteases for 1 hr at 37°C. In this experiment the control was left on ice. The specific activity of binding and the glucose-6P, galactose-6P solubilization ratios were deter- mined for each of the protease treated outer membrane preparations, as in Methods. After proteolysis and centrifugation of the OMM to remove protease solubilized protein the mg membrane protein was determined. the bin los sta €pr [1160' hyd! tree tica indi the and lost 65 The data in Table 8 indicate (as do the data in Table 7) that Chymotrypsin and trypsin do not inactivate the hexokinase binding site. This Table also shows that more binding sites were lost as a result of the 30 minute incubation in the control than in the protease treated membrane preparation. This protease induced stabilization effect, though small, was observed in three other experiments. It also appears from Table 8 that the protease treat- ment stabilizes the OMM against decmeases in the GGP/GalGP ratio. A general approach to test for the involvement of carbo- hydrate in the binding site is to determine the effect of periodate treatment on the binding properties, since periodate characteris- tically oxidizes carbohydrate residues. The data in Table 9 indicate that this treatment did in fact have a marked affect on the binding site since about 30% of the binding sites were lost and the glucose-6-P dependent solubilization effect was completely lost (i.e. the G6P/GalGP ratio was reduced to 1.0). However, Figure 10 demonstrates that purified hexokinase, which is not a glycoprotein, is rapidly inactivated at a 40 fold lower concentration of periodate than was used to inactivate the binding site, indicating that protein, like carbohydrate, is also markedly susceptible to inactivation by periodate. This point has, in fact, already been demonstrated by Clamp and Hough (45) in a thorough study of the relative rates of oxidation of amino acids by periodate. Therefore, while it is clear from the data in Table 8 that periodate is reacting with something in the membrane that is 66 Table 8.--Proteoloysis of Intact OMM with Trypsin and Chymotrypsin % 92139.65 Solubilized GGP/Ga16P mg prot. Control 28 Control Ga16P 40 2.22 2.4 G6P 88 Control 25 Trypsin Ga16P 30 2.53 2.8 GGP 77 Control 28 Chymotrypsin Ga16P 36 2.49 3.0 GGP 91 Control 21 Unincubated Ga16P 27 2.81 3.6 Control G6P 76 To 0.15 ml of OMM (4.0 mg/ml) were added 75 ug of Chymotryp- sin or trypsin and the aliquots (including a control with no addi- tions) were incubated for 30 min. at 25 C. The proteolysis was stopped by the addition of 2 mM.phenylmenthyl-sulfonyl fluride UtBF) from a 0.3M_stock in ethanol; obtained from Sigma Chemical Co.). The aliquots were centrifuged at 160,000 x g for 30 min and the pellets resuspended in 0.25 M_sucrose (0.4 ml). Hexokinase wasadded to each aliquot and the G6P/Ga16P ratio was determined as in Methods. In this experiment the amount of membrane protein was calculated from the initial amount of membrane protein present before protease treatment. The amount of membrane protein present after proteo- lysis was not determined. 67 Figure 10 Inactivation of Purified Hexokinase by Periodate Purified hexokinase (in 50 mM_Hepes pH 7.5; 0.5 mM_EDTA) was incubated at 25°C with 1 mM.NaIO4. Aliquots were taken out at various time intervals and assayed for hexokinase activity. 8 units / ml IIK 9 d 68 J 10 . Minutes 30 69 Table 9.--Effect of Periodate on Hexokinase Binding Units % G6P/ Bound Soluble Gal-6P No Additions 21 Control 0.32 Galactose-6P 27 2.6 Glucose-6P 71 i No Additions 69 Periodate 0.20 Galactose—6P 67 1.0 Glucose-6P 67 Intact liver mitochondria in 50 mM Na-phosphate pH 7.5 and 150 mM NaCl were incubated with and without 40 mM_Na-periodate at 25° for 20 min. The mitochondria were pelleted and hexokinase bind- ing and glucose-6P solubilization determined as described in Methods. Note that the 0.25 M sucrose in the resuspension nedia will quench any residual, unreacted periodate before the binding determination is begun. Formaldehyde (40 mM) which is a potential product of periodate oxidation was added to some binding assays as a control and found to have no effect on the binding properties. 70 essential to the binding process, no inferences can be made about whether that group is a protein or a carbohydrate. Heat inactivation of the Glucose-6-P dependent solubilization gffggt,--Incubation of purified outer mitochondrial membranes at 37°C sites on the membranes as well as a decrease in the GGP/Ga16P ratio. Since a Ca++ activated phospholipase A is present in the outer membrane (41), the effect of added calcium and EDTA in the incubation medium were tested for their effects on inactivation. The data in Table 10 indicate that while EDTA partially stabilizes the binding site during incubation at 37°C (consistent with the phospholipase A hypothesis), calcium does not lead to destabilization (inconsistent with the lipase hypothesis). Table 10 also shows that thioglycerol does not stabilize the binding, suggesting that the mode of desta- bilization of this binding site is not oxidation of sulfhydryl groups. The data in Table 11 suggest that divalent metal ions tend to sta- bilize the binding properties against heat inactivation. One inter- pretation of these results might be that phospholipase A activity is the cause of the heat inactivation. The failure of calcium to cause increased destabilization of the binding site may be explained by a divalent ion stabilizing effect. Miscel1aneous.--Thiocyanate has been reported to solubilize membrane proteins (7). I, therefore, thought it would be of interest to determine the effect of this salt on the binding properties of the membrane. Extensive washing of intact mitochondria with 71 Table lO.--Heat Inactivation of the Glucose-6-P Dependent Solubili- zation Effect Total Units % Condition Bound Soluble GGP/Ga16P control 44 phosphate Ga16P .979 40 1.70 G6P 68 control 27 phos + EDTA Ga16P .933 26 2.31 GGP 60 ++ control 39 phos + Ca Ga16P .844 38 1.94 G6P 74 control 54 phos + 1.0. Ga16P .732 54 1.30 G6P 70 control 59 sucrose Ga16P .563 54 1.37 GGP 74 control 21 No Incubation Ga16P 1.367 18 3.00 GGP 54 OMM (3 mg/ml) were diluted with ten volumes of 0.25 M_ sucrose or with 10 mM_Na-phosphate (pH 7.5) containing either 5 mM EDTA, 5 mflCaCl2 or 10 mM_thioglycerol. These aliquots were incu- bated at 37°C for 1.5 hours and subsequently pelleted by centri- fugation at 160,000 g for 30 min. The G6P/Ga16P ratios were determined as described in Table 5. 72 Table 11.--Heat Inactivation of the Glucose-6-P Dependent Solubili- zation Effect Total Units % Condition Bound Soluble G6P/Gal6P Control 48 10 mM_phosphate 0.74 Ga16P 45 1.56 pH 7.5 G6P 70 control 24 5 mM_EDTA 0.707 Ga16P 26 2.27 G6P 59 control 30 5 mMIMgCl2 0.587 Ga16P 30 1.93 G6P 58 control 38 10 mM-MgCl2 0.474 Ga16P 35 1.94 GGP 68 control 65 100 mM_NaCl 0.209 Ga16P 65 1.18 G6P 77 control 34 500 mM_NaCl 0.346 Gal6P 34 1.56 G6P 53 control 42 100 mM NaCl 0.306 Gal6P 43 1.47 10 mM_MgC12 G6P 63 0.25 M_ control 49 Sucrose 0.289 Ga16P 45 1.13 (no phosphate) G6P 51 This experiment was exactly analogous to Table 10 with the indicated additions of salt. A11 tubes contained 10 mM Na-phosphate, pH 7.5, except the 0.25 M_sucrose control. 73 2 M_KSCN still left the glucose-6—P solubilization effect intact. Consequently, if a protein in the membrane is involved in the binding site, it is not removed by salt washing. Treatment with 1 M_NaCl was also found to leave the binding properties as well as the banding pattern on SDS gels, substantially intact. Another way to approach the question of whether a protein is involved in the binding site is by treating the membranes with sulfhydryl reagents and determining their effect on the binding properties. If a sulfhydryl group on a protein is important in the binding site, this treatment would be expected to modify the binding properties. ‘When intact mitochondria were exhaustively treated with DTNB, no change in the number of binding sites was observed. The effect on glucose-6-P solubilization was not checked. Neuraminidase cleaves sialic acid residues from glyco- proteins. Consequently, if membrane bound sialic acid is important in the binding site, neuraminidase treatment should modify the binding properties. When intact mitochondria were treated with neuraminidase, about 50% of the sialic acid was released but no change in the number of binding sites or in the concentration dependence of MgCl2 for binding was observed. Lectins bind to carbohydrate residues associated with membranes. If such residues are intrinsic to the binding site, then lectins would be expected to bind at the site and block 74 hexokinase binding. Pretreatment of intact mitochondria with concanavalin-A or wheat germ agglutinin did not decrease the amount of hexokinase binding on intact liver mitochondria. The G6P/Gal-6P ratio was not determined. Lipid extraction from outer membranes with Emulgen 913.-- While attempting to solubilize outer membranes with the non- ionic detergent, Emulgen 913, I noticed that the membranes seemed unusually diffiuclt to solubilize. With relatively high con- centrations of detergent (i.e. 1-2%) a large insoluble pellet remained after ultracentrifugation. A hexokinase binding site was found in this pellet but no glucose-6-P dependent solubili- zation could be detected. I suspected that the lipid had been preferentially removed from the membrane protein and subsequent assays proved that this was the case. Readdition of phospholipid back to the lipid depleted membranes, with sonication, substan- tailly restored the glucose-6-P dependent solubilization effect. The experiments reveal a role for phospholipid in the binding site. The data in Table 12 show that treatment with Emulgen 913 substantially removes the phospholipid from the outer membranes. The cholesterol level is also reduced but to a lesser extent. The thin-layer plates (Figure 11) likewise indicate a marked reduction in phospholipid upon Emulgen treatment. Figure 11 also 75 Table 12.--Lipid Extraction of OMM with Emulgen 913 % Reduction Before After Due to Emul- Extraction Extraction gen Treatment Protein 4.6 mg/ml 2.88 mg/ml Phospholipid 4110 nmol/ml - 160 nmole/ml - 94% 893 nmole/mg prot. 55 nmole/mg prot. Cholesterol 129 nmole/ml - 31 nmole/ml - 61% 28 nmole/mg prot. ll nmole/mg prot. Outer membranes were diluted with 5 volume of cold 10 mM Na-Phosphate pH 7.5 and 1% Emulgen 913 was added. The membranes were immediately centrifuged at 160,000 x g for 30 min and the pellets were resuspended in a volume of 0.25 M sucrose equivalent to the volume of membranes used initially. Assays for protein (42), chloroform/methanol extractable phosphate (43) and cholesterol were done on the membranes before and after Emulgen treatment. Choles- terol was determined by reaction with cholesterol oxidase and flourometric measurement of H202 (Pat Kelly, personal communication). A second extraction of the OMM with Emulgen reduced their cholesterol content by an additional 25%. The % reduction of lipid due to Emulgen treatment was based on the amount of protein remaining after Emulgen treatment. 1 76 Figure 11 Thin-layer Plates of the Chloroform/Methanol Extracts of OMM, Emulgen Extracted OMM, and Reconstituted Emulgen Extracted OMM Chloroform/methanol extracts of intact OMM (track #1) and Emulgen extracted OMM (track #2) (prepared as in Table 12) were spotted on silica gel plates and developed with chloroform- methanol-acetic acid-water, 25:15:422 (v/v). A portion of the Emulgen extracted material was sonicated in the presence of an excess (1 mg phospholipid/l mg protein) of liposomes prepared from microsomal lipids and the resulting "reconstituted" vesicles were pelleted. An aliquot of this material comparable to that used on track #1 and #2 was extracted with chloroform/ methanol and Spotted on track #3. 77 1 am) I:— Ible 78 demonstrates that upon sonication in the presence of phospholipid (extracted from microsomes), the.phospholipid is extensively reincorporated back into the proteins. The effect of lipid extraction on glucose-6-P dependent solubilization.--From Figure 12 it is clear that treatment with increasing concentrations of Emulgen leads to a progressive de- crease hitheglucose-G-P dependent solubilization effect (i.e., a decrease in the G6P/Ga16P ratio). In this experiment it appeared as though the reason for the reduction in the GGP/Gal6P ratio resulted primarily from an increase in the amount of release that occurred in the presence of galactose-G-P. However a rela- tively smaller decrease in the amount of glucose-6-P dependent solubilization was also observed. (These data are included in the addendum.) Restoration of glucose-6-P dependent solubilization by adding back OMM lipids to Emulgen extracted membranes.--From Figure 13 it is clear that adding back chloroform/methanol extracted lipid from OMM to Emulgen treated membranes restores the glucose-6-P dependent solubilization effect. In this experiment, the total lipid extract from a 2x aliquot of membranes was required to give optimal restoration of the glucose-6-P effect from a 1x aliquot of Emulgen extracted OMM. David Alessi has obtained data on the restoration of the glucose-6-P dependent solubilization effect from Emulgen extracted OMM using lipids extracted from microsomes. These data (Table 13) 79 Figure 12 The Effect of Emulgen Treatment of OMM on Glucose-6-P Dependent Solubilization 0.2 ml aliquots of OMM (~3 mg protein/ml) and 0.8 mls of 50% Buffer A (Buffer A: 0.3 M NaCl, 50 mM Na-Phosphate, 1 11M EDTA, pH 7.5) were incubated for 10 min at 4°C with the indicated amounts of Emulgen 913. The samples were centrifuged at 150,000 x g for 30 min (4°C) and the pellets were assayed for the G6P/Ga16P ratio as in Methods. 80 o3__ 661’ 6111- P 0.4 0.8 ‘16 Emulgen 81 on pa mm new «N.F mm om.F om oe.p so aopmw umuauwumcoumm mm me pm —osa=ou we so mop sow No.0 mm mm.o mm oo.~ mop ao_eo emumqu__mo mm mm em _osu=ou mm mm am new ~m.~ ow mm.m up m~.~ we septa —ospcou mm mp we _ogu=ou ampew umNFPPnzpom ampew uon_wn:Fom ampew emNPPenapom new a now u now a me ucmswgmaxm we pcmsmemgxm —* acmepemaxm mmsomoeowz Ease uwuoesuxm muwnws anew: .zzo emanate comp=Em cu auweem cowum~wpmn=pom ucmccmnmo aimimmouspo as» to copumsoumom pm_usmaun.mp mpnee 82 Table 13.--Continued The Emulgen extracted OMM were obtained exactly as described in Figure 12. The Emulgen treated pellet from each 0.2 ml aliquot of OMM was homogenized in 0.5 ml, 0.01 M Na-phosphate pH 7.0. To this rehomogenized pellet was added 0.05 ml microsomal lipid (pre- pared as described below) and the sample was briefly (~30 sec.) sonicated in a bath type sonicator. Hexokinase was added directly to these "reconstituted" OMM and the G6P/Ga16P ratios determined as described in Methods. The microsomal lipids were prepared by following a Folch type extraction procedure (44) on intact microsomes (30 mg microsomal protein/m1) prepared as described in the legend to Table ‘4. After drying this total lipid extract under a stream of nitrogen, the lipids were resuspended in a volume of 0.01M Na-phosphate pH 7.0 that was equivalent to the original volume of microsomes used. 83 Figure 13 Restoration of Glucose-64’Dependent Solubil- ization from Emulgen Extracted OMM by the Addition of OMM Lipids OMM were treated with Emulgen exactly as in lkible 13 using 0.8% Emulgen. The Emulgen extracted pellet was suspended in a volume of 0.01M Na phosphate (pH 7.0) equivalent to the original volume of OMM used. To aliquots of the pellet were added increas- ing amounts of liposomes prepared from the chloroform/methanol extract of OMMs. The liposomes were prepared by sonicating the chloroform/methanol extracted (44) lipids in a volume of 0.01 M Na phosphate (pH 7.0), equivalent to the original volume of OMM from which the lipids had been extracted, until they were clear. On the y-axis is the GGP/GalSP ratio, determined as in Methods. 0n the x-axis is the amount of liposomes that were added back to the Emulgen extracted pellet, expressed as the volume of the lipo- somes added per volume of the resuspended Emulgen extracted pellet. In this experiment an amount of lipid extracted from a 2 x volume of OMM was required to optimally restore the GfiP/Gal6P ratio to a l x volume of Emulgen extracted OMM. N 0.6!. —- Gal 6P t vol. Iipid/ vol. pellet 85 indicate that microsomal lipids partially restore the glucose-6-P dependent solubilization effect to Emulgen treated OMM (as do lipids extracted from OMM; seeaFigure 12). Decreases in the G6P/Ga16P ratio due to Emulgen treatment appear to result primarily from increases in the amount of enzyme solubilized in the presence of galactose-6-P; however, decreases in the amount solubilized with glucose-6-P are also sometimes observed. Restoration of the glucose-6-P dependent solubilization effect by the addition of lipid seems, reproducibly, to imply decreases in the amount of enzyme released in the presence of galactose-6-P. However, increases, as well as decreases, in the amount solubilized in the presence of glucose-6-P are observed. The amount of enzyme solubilized when no hexose-phosphate is added, is about the same as that which is solubilized with galactose-6-P. The scatter in the data on the extent of solubilization that occurs in the delipidated and reconstituted membranes by the hexose-phosphates, prohibits any firm conclusions on the exact role that lipid plays in the binding site, with respect to glucose- 6-P or galactose-6-P solubilization. It is, nevertheless, clear that the GGP/Ga16P ratio is a sensitive enough indicator of the intactness of the hexokinase binding site to support the conclu- sions presented here. These conclusions are that removal of lipid from OMM by Emulgen causes a reproducible decrease in the G6P/Ga16P ratio, to approximately a value of 1.0, so that no glucose-6-P dependent solubilization effect is apparent. Since this property 86 can be restored by the readdition of lipid from either OMM or microsomes (which do not contain a hexokinase binding site), then the lipid requirement for recovery of the binding site is not a specific property of the OMM lipids alone. There must, therefore, be some other factor(s) present in the OMM (presumably protein) that gives rise to the specificity of binding and to the glucose-6-P dependent solubilization effect. Subsequent experiments (see below) have fully supported this contention that factors in addition to lipid (i.e. a hexokinase binding protein) are required for a fully intact hexokinase binding site. I consider this experimental approach of delipidation of OMM by Emulgen an adequate one for demonstrating that lipids play some kind of role in forming a native binding site for hexokinase. I don't, however, think this approach is adequate for determining precisely what that role is since the data on the extent of solubilization by galactose-6-P and glucose-6-P were too scattered. One reason for this scatter may be that when the lipids are extrac- ted from the OMM, the remaining insoluble proteins form large, partially denatured aggregates which are, subsequent to Emulgen treatment, pelleted at high speed. The extent of recovery of.this binding site into its native cuanformation upon readdition of lipids, may depend on the method of resuspension of the pellet or on the extent of sonication, two factors which are difficult to control precisely. A better approach for solving this problem would avoid the formation of such high molecular weight aggregates as well as avoid 87 sonication. We have found that the non-ionic detergent octyl glucoside solubilizes, without irreversibly denaturing, the hexo- kinase binding protein (see below). We have also noticed that phospholipids in the presence of octyl glucoside, can be dialysed away from OMM protein (data not shown). These observations suggest that a much better approach for determining the role of lipid in the binding site would be to dialyse octyl glucoside-solubilized OMM against octyl glucoside to remove the phospholipids. Purified lipids could be added back to this lipid depleted OMM protein to determine which lipids are required to restore the binding site to its native conformation. Hopefully this approach, since it would not involve formation of large delipidated, protein aggregates and sonication, would lead to more consistent results so that the precise role of lipid, with respect to glucose-6-P or galactose-6-P solubilization, might be more clearly resolved. Comparison between outer mitochondrial membranes and micro- somes with respect to detergent solubilization.--Figure 14 indicates that, compared to microsomes, outer mitochondrial membranes are strikingly resistant to detergent solubilization. Compared to the microsomes it takes at least a 10 fold higher concentration of either cholate or Emulgen to solubilize a comparable percentage of the outer membrane protein. In the experiment presented in Figure 14 a rather high ionic strength was used (0.2M). Preliminary results suggest that, at low ionic strength, detergents are even less effective at 88 Figure 14 Solubilization of Outer Membranes and Microsomes with Emulgen 913 and Cholate Outer membranes and microsomes (approx. 0.4 mg/ml) in 50 mM Na-phosphate pH 7.5, 0.3 M NaCl and 1 mM EDTA, were incubated for 15 min on ice with increasing concentrations of detergent either Na-cholate (0) or Emulgen 913 (D) (x-axis). After centrifugation at 150,000 x g for 45 min the pellets were resuspended in an equivalent volume of buffer and assayed for protein (y-axis) (42). From this figure it is apparent that relative to microsomes, outer membranes of mitochondria are markedly resistant to solubilization by either cholate or Emulgen 913. 89 .0 01 mg / ml magib. prot. '66 .0 % Detergent 9O solubilizing the OMM. Taking these observations into consideration, the results in Figure 14 are taken to be an indication that the outer membrane proteins have an unusually high affinity for one another through ionic bonds as well as through hydr0phobic attrac- tions. Solubilization, Reconstitutioppgpg_ Purification of a Putative Hexo- kinase Binding Protein from the Outer Mitochondrial Membrane Solubilization of the OMM with octyl glucoside and reconsti- tution of the hexokinase binding properties by dialysis.--Addition of octyl glucoside (1-4%) to OMM in 50% buffer A (see figure 12) followed by centrifugation at 160,000 g x 30 min yields a clear supernatant and a small pellet. When this supernatant is dialyzed against 0.01 M Na-phosphate pH 7.0 for 4 hours to remove the detergent, membrane-like material forms which sediments out during centrifugation at 160,000 x g for 40 min. The data in Table 14 indicate that these particles are capable of binding hexo- kinase in aglucose-6-P‘sensitive manner. The G6P/Ga16P ratios approach or exceed 2. Solubilization with 2-4% octyl glucoside is required for optimal recovery of binding sites and the G6P/Gal6P ratios. When octyl glucoside is added to a cold OMM sample, the membranes clarify imediately. If this sample is kept on ice,it stays clear for at least 30 min. If, however, the sample is warmed up to 25°C, cloudiness develops within minutes. The 160,000 x g 91 Table 14.-—Solubilization and Reconstitution of the OMM Binding Site for Hexokinase; Detergent Concentration Depen- dence % Octyl- Units % G6P Glucoside Bound Soluble Ga16P control 58 1% Ga16P .023 58 1.34 G6P 78 control 26 2% Ga16P .130 27 2.07 G6P 56 control 35 4% Gal6P .101 35 1.94 G6P 68 OMM (~4 mg/ml) were mixed 1:1 with buffer A (0.3 M NaCl; 50 mM Na phos pH 7.5; EDTA 1 mM) on ice and the indicated concentra- tion of octyl glucoside was added from a 20% stock solution. The solubilized membranes were immediately centrifuged for 30 min at 160,000 x g and the supernatants were dialyzed for 4 hours against 500 volumes of 0.01 M_Na phosphate, pH 7.0. An amount of dialyzed material equivalent to 0.1 "VI of the original membranes was used in the standard binding and solubilization assay as described under Methods. 92 pellet from a warmed sample is noticably larger than the pellet from a sample that has been kept cold. The data in Table 15 indicate that warming the sample prior to centrifugation and reconstitution does not appreciably reduce the recovery of binding sites or the GGP/GalGP ratio. The procedure for solubilization and reconstitution as described in Table 15 with the 30 min incuba— tion at 25°C prior to centrifugation was adopted for all subse- quent experiments (unless otherwise indicated). A comparison of the 1% SDS gels of intact OMM (Figure 15, A) with the reconstituted membranes (Figure 15, C), indicates that the reconstitution procedure results in a substantial reduction in the number of bands. Of the 15 bands that are present in intact membranes, primarily two, a 31,000 MW band and a 61,500 MW band, remain after reconstitution. SDS gels of the octyl glucoside insoluble material indicates that most of the membrane proteins are insoluble in the detergent and pellet out in the 160,000 x g centrifugation step (Figure 15, B). When the reconstituted membranes are solubilized a second time, centrifuged, and dialysed, the reconstituted membranes obtained give the SDS gel banding pattern observed in Figure 16, C. From this figure, it is clear that 2x reconstituted membranes contain primarily the 31,000 M.W. protein. The data in Table 16 indicate that the 2x reconstituted membranes retain a substan- tial number of binding sites for hexokinase that are sensitive to glucose-6-P solubilization, although there are decreases in both the binding sites and the GGP/Gal6P ratio. 93 Table 15.--Solubilization and Reconstitution of the OMM Binding Site; the Effect of Preincubation of Solubilized Membranes at 25°C Prior to Centrifugation % Recovery of G6P/Ga16P Binding Sites Untreated OMM 1.99 E 100% Control Reconstituted 1.85 52% 10 min; 25°C 2.15 40% 30 min; 25°C 1.75 42% The membranes were treated identically to Table 14 with 2.6% octyl glucoside except that the solubilized membranes were either kepton ice (control reconstituted) or incubated at 25°C for 10 minutes or 30 minutes, prior to centrifugation. The re- covery of binding sites in 1x reconstituted OMM is consistently around 50%. The most likely reason for this persistent result is that half of the binding sites in the reconstituted membranes are located on the inner side of the reformed vesicle and are, there- fore, made inaccessible for hexokinase binding. While much prece- dence for this type of randomization of membrane sidedness during reconstitution exists (57), there could be other reasons for the 50% loss in binding sites observed in Table 15. 94 Figure 15 SDS Gel Electrophoresis of Reconstituted OMM OMM were solubilized and reconstituted exactly as in Table 15 with the 30 min preincubation at 25°C. The octyl glucoside insoluble material, the reconstituted membranes and intact OMM were applied to $05 gels exactly as described under Figure 5 and in the Methods section. ABSORBANCE control pellet C reconstituted OMM top 95 "RP \ HBP \ HBP \ bottom 96 Table l6.--Hexokinase Binding Properties of 1X and 2X reconsti- tuted OMM % Recovery of G6P/Ga16P Binding Sites Intact OMM 2.22 E 100% 1X reconstituted 3.35 46% 2X reconstituted 1.51 22% 1X reconstituted OMM were prepared exactly as described under Table 15 and a portion of these membranes were used to determine the binding capacity and the G6P/Gal6P ratio. Another portion of the 1X reconstituted membranes were pelleted, and resuspended in a volume of 0.25 M sucrose equivalent to the original volume of the intact membranes. These membranes were resolubilized, incubated, centrifuged and dialyzed exactly as in Table 15 to give the 2X reconstituted membranes. The increase in the G6P/Ga16P ratio observed in this experiment for the 1X recon- stituted membranes is not always seen. One usually sees a decrease in the G6P/Gal6P ratio after reconstitution. 97 Figure 16 SDS Gels of 1X and 2X Reconstituted OMM; Comparison of Gels With and Without Bound Hexokinase Aliquots of intact OMM (A,B) 1X reconstituted OMM (0,0), and 2X reconstituted OMM (E,F), were prepared as in Table 16 and run on $05 gels according to the procedure described under Figure 5. For comparison, the same membrane preparations with hexokinase bound to them (as in Figure 5) were also run. 99 Figure 16 also gives a comparison of the SDS gels of intact membranes, 1x reconstituted and 2x reconstituted membranes, with and without bound hexokinase. One finds that the amount of hexo- kinase bound is in each case somewhat less (based on the staining intensity of the bands) than the amount of the 31,000 M.W. protein. Electron microscopy of the reconstituted vesicles.--Negative staining of 1x reconstituted OMM reveals the obvious formation of large vesicles (Figure 17, b) some of them comparable in size to the intact OMM (Figure 17, a). The vesicle diameter for both intact and reconstituted membranes ranges between 500 and 7000 A but in the reconstituted preparation a larger fraction of the vesicles are of the small variety. The membrane thickness in both preparations appears to be about 70-140 A, consistent with the view that the vesicles are unilamellar. Thin-sections of OMM and reconstituted membranes after pelleting and fixation again revealed a similarity in the mem- brane thickness and vesicle diameter of the two preparations (Figure 18, 19). However, the reconstituted vesicles apparently collapse around one another during the pelleting leading to figures that often appear multilamellar. Close examination of the photo- graphs, keeping in mind the differences in plane of sectioning through various types of collapsed vesicles, leads to the conclu- sion that the vesicles are primarily unilamellar not multilamellar. The differences in the electron micrographs of the two preparations is apparently a consequence of differences in the flexibility 100 Figure 17 Negative Staining, Electron Microscopy of Intact and Reconstituted OMM Intact (A) and reconstituted (B) OMM were prepared and stained briefly with 2% phosphotungstic acid on a Formvar coated electron microscopy grid. 102 Figure 18 Thin Sectioning, Electron Microscopy of Intact and Reconstituted OMM Intact (A) and reconstituted (B) OMM preparations were fixed in cold 2% glutaraldehyde in 0.01 M_phosphate buffer (pH 7.0) for 2 hr. The pellets were then washed in three changes of the same buffer, and postfixed in 1% buffered osmium tetraoxide for 30 min. They were dehydrated in an ascending ethanol series and prwipylene oxide, and embedded in an Epon-Araldite mixture. Ultrathin sections, 500-600 A in thickness, were then cut and examined with a Philips EM 201 electron microscope. , .. . M. .. «to he). ..eCLWe . ..- . ‘z .. r. .b a. m. l .0 . J . w... v)... a ’6‘ .. k ._ h th 6!. u . .0 s 1 l. I. . m5» ,1‘. ‘a 104 Figure 19 Thin Sectioning, Electron Microscopy of Intact and Reconstituted OMM The intact (A) and reconstituted (B) OMM were processed in the same manner as in Figure 18. 106 and/or the osmotic sensitivity of the two preparations, the intact OMM being more rigid. Protease treatment of OMM followed by reconstitution.--The data in Table 17 indicate that prior treatment of OMM with either Chymotrypsin or trypsin does not destroy the ability of the protease treated OMM to be reconstituted. The % recovery of binding sites as well as the G6P/Ga16P ratio is very good in all preparations. Since [arotease treatment alone apparently led to a purification of the binding protein (Figure 9) it seemed that combined protease treatment and reconstitution would lead to a greater purification. The gel scans (Figure 20), which compare protease treated/ reconstituted membranes to the ordinary reconstituted membranes, indicate that a somewhat more purified preparation was obtained from the protease treated membranes, but the differences weren't considered substantial enough to adopt protease treatment (with its possible artifacts) as a step in the standard purification procedure. The data in Tables 7 and 8 which give the binding characteristics of the various preparations may indicate that Chymotrypsin treatment actually improves the recovery of intact binding sites, although the differences are small. The data in Table 18 indicate that trypsin treatment of octyl glucoside-solubilized OMM prior to reconstitution destroys the binding site. The reconstituted membrane material obtained from the trypsin treated, sol ubilized membranes lost about 70% of the binding sites and the binding that remained was practically 107 Table l7.--Chymotrypsin and Trypsin Treatment of OMM Prior to Reconstitution % Recovery of G6P/Ga1-6P Binding Sites Untreated OMM 3.73 100% Reconstituted OMM 3.40 49% Chymotrypsin/Reconstituted 4.08 54% Trypsin/Reconstituted 2.41 47% Aliquots of OMM were diluted 1:1 with buffer A containing 2 mM CaCl2 and 1 mg/ml Chymotrypsin or trypsin were added. The samples were incubated at 25°C for 45 min, 1 mM_PMSF in ethanol was added and the samples were centrifuged at 160,000 x g for 45 min. The resulting pellets were resuspended in 0.25 M sucrose to their original volume and carried through the ordinary reconstitu- tion procedure (Table 14 and 15) without the 30 min preincubation. The resulting membranes were checked for hexokinase binding ability and the G6P/Gal6P ratio was determined in Methods. 108 Figure 20 SDS Gels of Protease Treated/ Reconstituted OMM Aliquots of the reconstituted membranes from Table 17 were applied to $05 gels as in Figure 5. mozgzcmm< 110 Table 18.--Trypsin Treatment of Octyl Glucoside Solubilized OMM, Fo owed by Reconstitution Relative Recovery of GGP/GalGP Binding Sites Control Reconstituted 1.73 1.0 Trypsin Treated Reconstituted 1.09 0.29 Octyl glucoside solubilized OMM (Table 15) were centrifuged at 160,000 g for 30 min and the supernatant was treated at 25°C with 0.5 mg/ml trypsin for 30 min. 1 mM PMSF was added and the sample was dialyzed against 0.01 M phosphate pH 7.0 (as in Table 15). Binding assays on the trypsi 11 treated and control samples were done as in Methods. 111 insensitive to glucose-6-P. The gels of the trypsin treated/recon- stituted membranes (Figure 21) indicate that the reconstituted membranes were substantially modified and the binding protein was greatly reduced. To summarize the current status of our understanding of the hexokinase binding site, the evidence iinplicating the involvement of a protein in the binding site is: l. Lipids alone do not give a glucose-6-P sensitive hexokinase binding site. 2. Other isolated membrane preparations, including some with a lipid composition similar to OMM, Show no significant ability to bind hexokinase. 3. The reconstitutability of octyl glucoside solubilized OMM is protease sensitive. The evidence favoring the 31,000 M.W. protein as the hexokinase binding site is that: l. Reconstituted OMM which retain substantial amounts of the 31,000 M.W. protein also retain 25-50% of their original hexo- kinase binding capacity. Other proteins in some of these reconsti- tuted membrane preparations have been reduced by 95% or more without comparable losses in the binding capacity. . 2. Hexokinase associates with intact or reconstituted OMM in amounts (gm/gm) that are roughly comparable to the 31,000 M.W. pro- tein. The amount of hexokinase binding does not correlate with the amounts of any other protein present in intact or reconstituted OMM. 3. The protease treatment of octyl glucoside solubilized OMM results in the loss of the 31,000 M.W. protein as well as the loss in the reconstitutability of the hexokinase binding site. 112 Figure 21 SDS Gels of Trypsin Treated Reconstituted OMM SDS gels were run according to Figure 5 of the control reconstituted and trypsin treated-reconstituted membranes in Table 18. ABSORBANCE A CONTROL 3 TRYPSIN top 113 ..{1 IIBP ‘A bottom 114 The experiments contained in this thesis do not definitively prove that the 31,000 M.W. protein is the one and only protein required for binding of hexokinase to the native OMM. Chemical crosslinking studies which are currently in progress may eventually resolve this point. If hexokinase can be shown to specifically crosslink to the 31,000 M.W. protein in intact OMM, then this certainly will be consistent with our contention that the 31,000 M.W. protein is the native binding protein. The preparation of antibodies against the hexokinase binding protein will be attempted in the near future. If this antibody preparation specifically modifies the binding properties of hexokinase to OMM, this will be further proof that the 31,000 M.W. protein is the native binding protein. The antibody prepar- ation should also prove very useful for screening different tissues of the rat (including brain) to determine whether all rat mitochondria contain the binding protein. Other mammalian mitochondria (e.g. cow, human) as well as non-mammalian (e.g. yeast, plant) could be checked for the presence of the binding protein to determine whether it is ubiquitous among all organisms. Subunit molecular weights and tentative identification of some of the protein bands on SDS_gels of 0MM.--Figure 22 lists the molecular weights of the 15 bands that we routinely see on polyacry- lamide gels of OMM. The putative hexokinase binding protein is band 11 and has a molecular weight of 31,000. Monoamine oxidase has been recently purified and shown to have a molecular weight on $05 115 of about 65,000 (46), approximately equivalent to band 6. An excellent cytochrome 05 spectra is easily obtainable from the purified outer membrane preparations. Since the cytochrome oxidase spectrum is nearly absent (and sometimes completely undetectable) in these preparations the outer mitochondrial membrane preparation as reported herein is judged to be virtually free from inner mitochondrial membrane contaminents. Cytochrome b5 has been reported to have a molecular weight of approximately 16,000 corresponding to band 14. There are several other activities that have been reported to be associated with the outer mitochondrial membrane (41). Among these are a "rotenone insensitive" NADH cytochrome 05 reductase and a phospholipase A2. The cytochrome bs in the 0111 has been reported to be different from the one found in microsomes (47) but the reductase has recently been shown to be the same (48). The phospholipase A2 may be involved in the inactivation of the hexokinase binding protein (Tables 10, 11) and may also be involved in the uncoupling of isolated rat liver mitochondria (49). Bovine serum albumin (BSA) may bind to the lipase thus inactivating it and giving rise to enhanced stability of oxidative phosphorylation in isolated mitochondria (39). Te BSA binding to the phospholipase A2 may also change the properties of the hexokinase binding site producing the effects observed on Figure 6 of this chapter. (The ability of BSA to bind free fatty acids might also give rise to some of these observed effects.) 116 Figure 22 Tentative Identification and Molecular Weight Determination of the Proteins in Native OMM SDS gels of intact OMM and of the Dalton Mark VI standard mixture (Sigma) were run as in Figure 5. The molecular weights of the OMM proteins were assigned by comparison with a standard curve generated from the Dalton Mark VI mixture. 117 Band Molecular Tentative Number Weight Comments Identification I 135 000 Very --- 2 125. 000 Faint Pair --- 3 85 000 Medium --- 4 79 000 Dark Pair --- 5 70000 Very Faint --- 6 61 500 Broad. Dark Monoamine Oxidase 7 55.000 Faint to --- 8 50 000 Medium Dark --- 9 46 000 Group --- 10 35 000 Medium Dark . ll 31 000 Broad Dark Hexokinase Binding Protein 12 29 000 Dark --- 13 21 500 Faint --- I4 16 000 Dark Cytochrome b5 15 I 500 - 3 500 Variable --- l l 118 Recently S. J. Singer and his colleagues have presented data indicating that mitochondria in cultured cells are linked to microtubules (50). This suggests that the OMM contain a protein that binds to microtubules. It has long been known that the OMM is markedly insensitive to changes in osmolarity when compared to the inner mitochondrial membrane, both in vitro and in vivo (51). This suggests that the OMM contains a pore that is permeable to small molecules such as sucrose. This pore is most likely formed by a protein or a complex of proteins. The outer membrane contains saturable, high affinity bindings sites for ribosomes (52). This suggests that there is a specific binding protein(s) similar to the ribophorins I and II that have been demonstrated to be responsible for the binding of ribosomes to the rough endoplasmic reticulum (53). If this could be demonstrated it would suggest an attractive hypothesis for the mode of insertion of 'intramitochondrial proteins that are synthesized from extramitochondrial RNA. This hypothesis would be similar to the "signal hypothesis“ proposed by G. Blobel (54). Ultimately all of the proteins responsible for the above mentioned effects will be identified and assigned molecular weights. But at the present time it is clear from Figure 22 that the identifies of most of the proteins in the OMM are unknown. It may be possible to identify some of the remaining proteins in the OMM using techniques similar to those that were used to identify the hexokinase binding protein. ADDENDUM 119 Addendum Data for Figure 12 Vol. Lipid % G6P Vol. Protein Units Bound Condition Soluble ESTER condition 45 0.0 0.24 Ga16P 53 0.98 G6P 52 control 32 0.5 0.28 Ga16P 32 1.5 G6P 47 control 29 1.5 0.25 Ga16P 29 1.7 G6P 49 control 21 2.5 0.33 Gal6P 20 2.4 G6P 48 control 26 3.5 0.25 GalGP 25 2.3 G6P 58 control 21 4.5 0.27 Gal6P 27 1.9 GGP 52 120 Addendum Data for Figure 13 % GGP % Emulgen Units Bound Condition Soluble Ga16P control 20 0.0 0.61 Gal6P 20 4.7 GGP 94 control 35 0.2 0.29 Gal6P 38 2.6 G6P 97 control 51 0.4 0.31 Gal6P 51 1.5 G6P 79 control 58 0.6 0.34 Ga16P 70 1.1 G6P 78 control 56 0.8 0.35 Ga16P 58 1.1 G6P 63 121 REFERENCES 122 10. 11. 12. 13. 14. 15. REFERENCES Melander, W., and Horvath, C. Arch. Biochem. Biophys. 1’92. 200 (1977). Lehninger, A. L. Biochemistry, 2nd ed. N.Y., N.Y.: Worth Publishers, 1975. Felgner, P. L. and Wilson, J. E. Arch. Biochem. Biophys. 1&2, 282 (1977). Hanstein, W. 8., Davis, K. A., and Hatefi, Y. Arch. Biochem. Biophys. 152; 534-544 (1971). Jencks, W. P. Catalysis in Chemistry Enzymology. New York: McGraw-Hill, 1969} Hamabata, A., and Von Hippel, P. H. Biochemistry. 12, 1271- 1282 (1972). Hatefi, Y., and Hanstein, W. 0. Proc. Nat. Acad. Sci. 62, USA, 1129 (1969). Von Hippel, P. H., and Schleich, T. in Structure and Stability of Biological Macromolecules (S. N. Timasheff and G. D. Fasman, eds.) New York: Marcel Dekker, Inc., 417 (1969). Hofmeister, F. Arch. Exptl. Pathol. Pharmakol. 24, 247 (1888). Robinson, 0. R., and Jencks, W. P., JACS. 81, 2470-2479 (1965). Von Hippel, P. H., and Wong, K. Y. Science. 145, 577 (1964). Tanford, C. Advances in Protein Chemistry Vol. 24 (C. B. Anfinsen Jr., J. T. Edsall and F. M. Richards, eds.) New York and London: Academic Press, 1 (1970). Kuntz, I. D., and Kauzmann, W. Adv. In Prot. Chem. Vol. 28 (C. B. Anfinsen, Jr., J. T. Edsall, and F. M. Richards, eds.) New York and London: Academic Press, 239 (1974). Roseman, M., and Jencks, W. P. JACS. 91, 631 (1975). Von Hippel, P. H., Peticolas, V., Schack, L., and Karlson, L. Biochemistry. 12, 1256-1264 (1973). 123 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 124 Hamabata, A., and Von Hippel, P. H. Biochemistry. lg, 1264- 1271 (1973). Von Hippel, P. H., and Hamabata, A. J. Mechanochem. Cell Motil. p, 127-138 (1973). Heydweiller, G. Ann. Phys. 13,145 (1910). Moore, M. J. 4th Ed. Englewood Cliffs, New Jersey: Prentice- Hall, Inc. (1972). Adamson, A. W. Physical Chemistry of Surfaces 3rd ed. New York: John Wiley and Sons, 1976. Barrow, G. M. Physical Chemistry 2nd ed. New York: McGraw- Hill Book Co., 1966. Randles, J. E. 8. Adv. in Electrochem. and Electrochemical Egg. Vol. 3 (P. Delahany ed.) New York: Interscience Pub- ishers, 1963. Cotton, F. A. and Wilkinson, G. Advanced Inppganic Chemistry 3rd ed. New York: Interscience Publishers, 1972. Halliwell and Nyburg, S. C. Trans Farady Soc. §9, 1126 1963 . Tanford, C. The Hydrophobic Effect. New York: John Wiley and Sons, 1973. Frank, H. S., and Evans, M. W. J. Chem. Phyg, 1;, 507. Marcus, Y. and Kertes, A. S. Ion Exchange and Solvent Extraction of Metal Complexes. New York, New York: John Wiley and Sons, 1969} Horvath, C., Helander, H., and Molvar, I. Analyt. Chem. 49, 142 (1977). Kauzmann, W. Advances in Protein Chem. 14, 1 (1959). Rose, I. A., and Warms, J. V. B. J. Biol. Chem. g4g, 1635- 1645 (1967) Wilson. J. E. J. Biol. Chem. ggg, 3640-3647 (1968). Teich raber, P., and Biesold, D. J. Neurochem. 1§, 979-989, (19581 —— KropP. E. S., and Wilson, J. E. Biochem. Bipphys. Res. Commun. pp, 74-79 (1970). 34. 35. 36. 37. 38. 39. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 125 Chou, A. C., and Wilson, J. E. Arch. Biochem. Biophys. 99, 74. Diezel, N. et al. Anal. Biochem. 49, 617 (1972). Molick, N., and Erzie, A. Anal. Biochem. 49, 173 (1973). Picciano, 0., and Anderson, W. F. in Methods in Enzymology {01. KXX (Moldave and Groesman, eds.). Academic Press, 171 1974 . Steck, T. L., and Kant, K. A. Methods in Enzymology. 91, 172-180 (1974). Barbour, R. L., and Chau, S. H.P. J. Biol. Chem. 245, 3295-3301. Racker, E. Membranes of Mitochondria and Chloroplasts. Van Nostrand Reinhold Company, 1970. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. and Randall, R. J. J. Biol. Chem. 199, 265-275 (1951). Bartlett, G. R. J. 8161. Chem. ggg, 466 (1959). Radin, N. S. in Methods in Enzymology Vol. XIV (Lowenstein, J. M., ed.). Academic Press, 245 (1969). Clamp, J. R.,and Hough, L. Biochem. J. 94, 17-24 (1965). McCauley, R. Arch. Biochem. Biophys. 199, 8-13 (1978). Fukushima, K. and Sato, R. J. Biochem. 24, 161-173 (1973). Kowahara, S., Okada, Y., and Omura, T. J. Biochem. 99, Japan, 1049 (1978). Seppala, A. J., Saris, N. E. L., and Gauffin, M. L. 919- chemical Pharmacology. 99, 305-313 (1971). Haggeness, M. H., Simon, M., and Singer, S. J. PNAS. 79_, 3863-3866 (1978). Tedeschi, H. in Current Topics in Membranes and Transport (F. Bronner and A. Kleinzeller, eds.). Vol. 2, Aeademic Press, 1971. Schatz, G., and Mason, 1. L. in Annual Review of Biochemistpy (E. E. Snell, ed.) Vol. 43. Palo Alto, Califlz Annual Reviews Inc., 1974. 53. 54. 55. 56. 57. 126 Czako-Graham, M., Sabatini, D. D., Algranati, 1., Bard, E., Morimoto, T., and Kreibich. Federation Proceedings. 91, no. 6, 1568 (1978). Jackson, R. C., and Blobel, G. PNAS. Z4, 5598-5602 (1977). Fairbanks, G., Steck, T. L., and Wallach, D. F. H. 949- chemistry. 19, 2607 (1971). Olmsted, J. B., and Borisy, G. G., in Annual Review of Bio- chemistr (E. E. Snell, ed.) vol. 42. P510 ATto, CElif.: nnua Reviews Inc., 507-540 (1973). Racker, E. A New Look at Mechanisms in Bioenergetics. Acadmic Press (1976). APPENDICES 127 APPENDIX A 128 129 Vol. 68, No. 2, 1976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS PURIFICATION OF NONBINDABLE AND MEMBRANE-BINDABLE MITOCHONDRIAL HEXOKINASE FROM RAT BRAIN Philip L. Felgner and John E. Wilson Biochemistry Department, Michigan State University East Lansing, Michigan 48824 Received November 24,1975 SUMMARY: MgClz-induced binding of glucose-6-P solubilized rat brain hexo- kinase to rat liver mitochondria has been found to be markedly diminished by increasing ionic strength. Using a modified assay of binding ability, it has now been possible to demonstrate that purified preparations of brain hexokinase do retain appreciable ability to bind to mitochondria. A slight modification of the previous DEAE-cellulose chromatography pro- cedure (4), permits resolution of the hexokinase into two major components designated as Type Ib and Type In based on their ability to bind and not bind, respectively, to mitochondria. 1b and appear to be identical in molecular size and subunit composition, but di fer slightly in net charge. It has been demonstrated (1-3) that the Type I isozyme of hexokinase, as found in brain, selectively binds to the outer mitochondrial membrane, presumably indicating some specific component (or components) of this membrane which selectively interacts with the enzyme. In order to better understand the nature of the interaction between hexokinase and the outer mitochondrial membrane it would be advantageous to purify the enzyme and the requisite component(s) of the membrane. Although the enzyme has been purified to homogeneity (4), it was not previously possible to demonstrate that the purified enzyme retained the ability to interact with the mito- chondrial membrane. It is the purpose of this communication to describe a method for resolution of purified bindable and nonbindable forms of hexokinase and to indicate some of the reasons why this was not observed earlier. MATERIALS AND METHODS Chemicals and rats were obtained from commercial sources given earlier (4,5). Hexokinase was assayed as described previously (4) except that the NAB? and g1ucose-6-P dehydrogenase concentrations were doubled. Rat liver mitochondria were prepared according to the method of Sottocasa g£_§l, (6). Copyright C5. 1976 by Academic Press, Inc. 592 All rights of reproduction in any form reserved. I30 Vol. 68, No. 2, 1976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Three times washed rat brain particles (5) and glucose-6-P solubilized hexokinase (7) were obtained as previously described. Rat brain hexokinase was purified according to Chou and Wilson (4) except for slight modification in the DEAE-cellulose column procedure (see Fig. 3). Binding ability was assayed by incubation of the enzyme with excess mitochondrial binding sites. Except where noted (Fig. l), the conditions were as described previously (5) with the following modification: a) hexo- kinase preparations were diluted with 0.25 M sucrose such that the total ionic strength in the incubation medium did not exceed 0.005 M, b) rat brain particles were used in place of rat liver mitochondria, and c) incubation was at 25° rather than 0°. These modifications resulted from observations that rat brain mitochondria bound an appreciably greater proportion of the hexokinase activity than did liver mitochondria, and the binding was less susceptible to inhibition by increasing ionic strength (see Fig. l); with brain particles (but not with liver mitochondria), binding was also enhanced by increasing the temperature to 25°. Binding assays were done l l O (105 0J0 Ionic «moth Fig. 1. Inhibition of Binding by Increasing Ionic Strength. Binding ability of glucose-6-P solubilized hexokinase was assayed as described previously (5), using rat liver mitochondria. The indicated salt concentrations were added to the tubes before initiating binding with MgCl . Open symbols (o,A) represent KCl and closed symbols (0,5) potassium phosphate (pH 7.0). 593 131 Vol. 68, No. 2, I976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS in polypropylene microcentrifuge tubes (Brinkmann Instruments) which had been dipped in 22 bovine serum albumin (BSA) solution then dried at approx. 60°; coating with BSA was found to prevent artifacts due to adsorption of the hexokinase to polypropylene (unpublished observation). RESULTS From Fig. 1, it is clear that low ionic strength severely inhibits the Mg012 - induced binding of g1ucose-6-P solubilized hexokinase to liver mitochondria, which have previously been routinely used for the binding assay (5). Fifty percent inhibition occurs at 0.02-0.03 M ionic strength. Using a modified binding assay (see Methods), it has been possible to show for the first time that purified hexokinase can bind to rat brain particles (Fig. 2). In this experiment more than 502 of the enzyme was bound when excess binding sites were present. When the conditions for DEAR-cellulose column chromatography used by Chou and Wilson (4) are slightly modified by using a shallower gradient of KCl, elution patterns such as those shown in Fig. 3 are obtained. Under these conditions, two peaks of hexokinase, one at about 0.065 3 KCl and the other at about 0.075 §_KC1, have been reproducibly resolved (eight experiments). é Ill/POVMSII'I‘I \ pollen wcontmrocomy O 1 0 325 650 pmpftaig:hondrisl Fig. 2. Binding of Pure Hexokinase. Hexokinase (0.065 units, purified according to Chou and Wilson (4)) was incubated with brain particles as described in Methods. Total volume was 0.25 ml. 594 132 Vol. 68, No. 2, 1976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS The relative amounts of activity in the two peaks have, however, varied from about 752:252 to 251:752; the reasons for this remain under investigation. Fig. 3 also indicates that the enzyme eluted at higher KCl concentrations is at least partly bindable to brain particles, while hexokinase eluted at lower salt concentrations is completely nonbindable. For this reason we refer to these two forms of the enzyme as Type I for the bindable enzyme, b and Type In for nonbindable. The asymmetry of the eluted peaks, together with preliminary analytical isoelectric focusing experiments, suggest that 20 100+ ‘32 'o c a O O Q a E c 13» 350~ #01 5 CL 2, E 5:? - D L) X 50* 25 Percent baund 60 80 100 Fraction number Fig. 3. DEAR-Cellulose Column Chromatography of Rat Brain Hexokinase. Chromatography was performed as previously described (4) except that shallower salt gradients were used. In the upper elution pattern a 600 m1 linear gradient from 0.0 to 0.20 §_KC1 in column buffer was used, collecting 3.8 m1 fractions. The lower elution pattern was done exactly the same except the gradient went from 0.0 to 0.15 §_KC1. The KCl concentration (0) was determined by conductivity measurements on the fractions. Hexokinase activity (0) and percent bound (A) were measured as described in the Methods section. In both experiments bindable enzyme was found only in the high salt peak (0.075 g KCl). The proportion of the enzyme in the second peak which could be bound was, however, found to vary in different experiments e.g. in the lower profile, only 351 of the activity in the second peak was bindable, whereas in the experiment shown in the upper profile, 702 was bindable. 595 133 Vol.68, No. 2, 1976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Type In and Type I are in turn composed of more subforms (at least six b total). Variation in the percentage bindability of Type Ib (see Fig. 3) also suggests that some of these subforms are nonbindable. DISCUSSION It had been previously observed that high ionic strength could cause solubilization of mitochondrial hexokinase (l, 8, 9). In contrast, low ionic strengths actually appeared to slightly enhance the strength of association of the enzyme with the membrane (8). In view of this obser- vation, it was presumed that low to moderate ionic strengths would have little or no effect on the MgCl -induced rebinding of the enzyme to 2 mitochondria. The results presented here show that this presumption was incorrect. Furthermore, we have also found that (for reasons which remain under investigation) binding of hexokinase by brain mitochondria is less susceptible to the ionic strength effect than is binding by liver mitochondria which have, for reasons of practicality (5), previously been routinely used in assays of binding ability. Using this new information, and making appropriate modifications of the binding assay, it has now been possible to demonstrate that purified brain hexokinase can bind to mitochondria. Furthermore, slight modification of the previously described DEAE- cellulose chromatography procedure (4) has permitted the demonstration of a previously undetected heterogeneity of the enzyme. Two major forms, designated Type In and Type I have been resolved. Although In and Ib are b readily distinguished by the difference in their ability to interact with the mitochondrial membrane, the physical or chemical basis for this distinct- ion remains under investigation. Since mitochondrial enzyme (presumably all Type Ib) serves as the starting point for the purification, it seems quite probable that Type In represents, at least to some extent, an artifactual form produced in variable amounts (see Fig. 3) during purification. The enzyme does not contain detectable carbohydrate (10, and J. E. Wilson unpublished observations) or lipid (5), so the difference between I and In b 596 '134 Vol.68, No. 2, 1976 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS is not likely to be due to such factors. Partial proteolysis during purification of the enzyme has not been entirely ruled out, but seems unlikely in that I and In have not been found to differ in N-terminal b amino acid (by the dansylation technique) or apparent molecular weight (98,000 by SDS-gel electrophoresis). 1 The latter results suggest that the principle difference between I and In lies in a modification affecting b their net charge, with I being more negatively charged (based on its b greater affinity for DEAR-cellulose) than is In' Such modification might include phosphorylated and dephosphorylated forms e.g. perhaps Ib is a phosphorylated form which is (enzymatically?) converted to the dephosphorylated (and thus less negative) In during purification with resulting loss of binding ability. ACKNOWLEDGEMENT Financial support for this work was provided by Grant NS-09910 from the National Institutes of Health. REFERENCES 1. Rose, I. A., and Warms, J. V. B. (1967) J. Biol. Chem” fl, 1635-1645. 2. Craven, P. A., Goldblatt, P. J., and Basford, R. E. (1969) Biochendstry, g. 3525-3532. 3. Kropp, E. S., and Wilson, J. E. (1970) Biochem. Biophys. Res. Commm., gg, 74-79. 4. Chou, A. C. and Wilson, J. E. (1972) Arch. Biochem. Biophys., 121, 48-55. 5. Wilson, J. E. (1973) Arch. Biochem. Biophys. 124, 332-340. 6. Sottocasa, C. L., Kuylenstierna. B., Ernster, L., and Bergstrand, A. J. (1967) J. Cell. Biol. 22, 415-438. ' 7. Wilson, J. E. (1973) Arch. Biochem. Biophys., 152, 543-549. 8. Wilson, J. a. (1968) J. Biol. Chem., gig, 3640-3647. 9. Teichgraber, P., and Biesold, D. (1968) J. Neurochem. 15, 979-989. 10. Craven, P. A., and Basford, R. E. (1974) Biochim. Biophys. Acta 338, 619-631. 1 These results are consistent with earlier studies using the enzyme prepared according to the original Chou and Wilson (4) procedure. These enzyme preparations probably were, in visvof the present observations, mixture of I and I , but were found to be homogeneous with regard to molecular welght (e?g., SDS-gels or centrifugal methods). 597 APPENDIX B 135 136 Hexokinase Binding to Polypropylene Test Tubes Artifactual Activity Losses from Protein Binding to Disposable Plastics In the course of our studies with rat brain hexokinase (ATon—hexose 6ophosphotransferase, EC 2.7.1.1), we have frequently used disposable polypropylene centrifuge tubes. As long as relatively crude systems with substantial amounts of extraneous protein were used, no difficulties were encountered. However, with more purified preparations, substantial apparent activity losses were incurred. Investigation of the situation disclosed that the reason for these apparent losses was adsorption of the hexokinase to polypropylene vessels, an unusual adsorption that appeared to be particularly effective at the air-water interface. Subsequently, we found similar effects with other enzymes. Although binding of some proteins, such as ribonuclease (1,2), to glassware has been reported. we do not know of a similar phenomenon being described for plastic ware. The increased use of disposable plastic laboratory ware necessitates an awareness of this potential problem. We present here the results of our study of the adsorption of hexokinase to polypropylene. MATERIALS AND METHODS Polypropylene microcentrifuge tubes were obtained from Brinkman Instruments. In some experiments these tubes were coated by dipping them in a 2% bovine serum albumin (BSA) solution, then drying in a 60°C oven. Glucose-6—P-solubilized (3) and purified hexokinase (4) were prepared as previously described. Hexokinase activity was assayed according to Chou and Wilson (4). RESULTS AND DISCUSSION As shown in Fig. l, successive transfers of hexokinase between poly- propylene tubes caused a progressive loss of activity. Coating the tubes with BSA before enzyme addition diminished the observed losses; washing uncoated tubes with ordinary dishwashing detergent, 1:1 chloro- form:methanol, or 6 N BC] or treatment with 1% dichlorodimethyl silane in benzene had no effect. That the tubes were binding (rather than totally inactivating) enzyme was demonstrated by the presence of residual glucose phosphorylating activity in tubes which had been ex- 631 Copyright © l97b by Academic Press. Inc. All rights of reproduction In any form reserved. 137 632 SHORT COMMUNICATIONS 100 ISA eoeted e mooted percent activity remaining ta 0 I I I J O 2 4 6 tube number FIG. 1. Activity losses resulting from multiple transfers of hexokinase between poly- propylene tubes. Glucose-SP-solubilized hexokinase (l ml. 0.52 unit) was added at 0°C to polypropylene tubes which had (0) or had not (0) been pretreated with BSA as described in the text; these initial tubes are designated as "tube 1" in the above figure. After removal of an aliquot for assay. the solution was transferred to a second tube (also with or without BSA) using a Pasteur pipet. This procedure was repeated for a total of six transfers. posed to hexokinase solutions and subsequently washed with 0.25 M sucrose (e.g., see Fig. 3). As shown in Fig. 2, successive aliquots taken from the same tube showed decreasing activity. This somewhat puzzling result could be ex- plained if the enzyme were preferentially adsorbed at the air-water interface. This interpretation is supported by the experiment described in Fig. 3, which demonstrated the preferential adsorption of the enzyme to regions where a meniscus (i.e., an air—water interface) had passed. When formation of an interface between air and the enzyme solution was prevented by overlay of the enzyme solution with water, less adsorption occurred (Table 1). Our initial reaction to these observations was that this preferential adsorption at the air-water interface seemed so unusual that it was likely to be a property restricted to hexokinase and perhaps a few other proteins. We were somewhat surprised, therefore, to find that nearly 40 years ago a strikingly similar phenomenon was studied by Langmuir er al. (5), who demonstrated the adsorption of proteins to a “hydrocarbon surface“ (a brass plate coated with barium stearate) which was passed through a sur- 138 SHORT COMMUNICATIONS 633 '8' fi ISA coated to 0 t uncoated percent activity remaining J l 20 0 I00 200 pl removed FIG. 2. Activity losses resulting from removal of successive aliquots from a single poly- propylene tube. Glucose-bP-solubilized hexokinase (0.2 ml. 0.43 U/ml) was placed in polypmpylene centrifuge tubes at 0°C which had (0) or had not (0) been pretreated with BSA. Successive 50ml aliquots were removed for assay. face layer of protein formed at an air-water interface. We believe our results are analogous to those of Langmuir er a1. (5), with the highly hydrophobic polypropylene being the equivalent of the “hydrocarbon surface” used by Langmuir er 01. Although the earlier workers (5) produced the surface layer of protein by experimental manipulation, such surface films are reported to form spontaneously (2) in all protein solutions. Thus, in analogy with the experiments of Langmuir et a]. (5), we would interpret results such as those shown in Fig. 2 in the following way: As the aliquot of solution is removed, the meniscus passes down the hydrophobic polypropylene surface with resulting adsorption from the surface layer of protein to the polypropylene. Subsequent spontaneous reformation of the surface layer (2) results in net removal of enzyme from the bulk solution. Repetition of this process would account for the progres- sive depletion of enzyme activity in the bulk solution, accompanied by adsorption of enzyme to hydrophobic surfaces which had been traversed by the air-water interface. Furthermore, we have observed that a tube containing protein differs markedly from one that did not, since after withdrawal of the protein solution the sides of the tube are observably wet, whereas a tube that never contained protein sheds water. Langmuir er al. similarly observed that the “hydrocarbon surface" (in our case. polypropylene) shed water unless it was coated by a protein film, while coating with the hydrOphilic protein layer allowed the otherwise hydro- phobic surface to bind water molecules. Since protein surface films (2) and “monolayer adsorption" (S) are 139 634 SHORT COMMUNICATIONS v mole oer/25 min A . . ones a 0.043 c cots Fro. 3. Preferential binding of hexokinase at the air-water—polypropylene interface. A 0.5-ml aliquot of 0.25 M sucrose-0.1M glucose was added, at 0°C. to each of three polypropylene tubes. Pure hexokinase (0.05 ml, 11.0 U/ml) was then injected directly into the sucrose-glucose, followed by gentle stirring. In tube A, the enzyme was slowly drawn out (over a period of 5 see) with a Pasteur pipet. and the tube was then washed twice with 0.25 M sucrose-0.1 M glucose. Assay mix (3) (0.5 ml) minus the glucose-6P dehydrogenase was added. and the tube was incubated at 25°C for 25 min; subsequently. glucose-6P formation was measured spectrophotometrically after addition of glucose- 6-P dehydrogenase. Tube B was like A except that the enzyme was raised from the bottom of the tube by under-laying it with 0.5 ml of 0.9M sucrose. After careful removal of the enzyme solution and the 0.9 M sucrose with a Pasteur pipet. the tube was washed as above. Sucrose (0.9M, 0.5 ml) was then placed in the tube. overlaid with 0.5 ml of assay mix, and assay for hexokinase activity done as described above. Tube C was exactly like 8 except that the 0.5 ml of assay mix was placed directly into the tube. In the above figure. the solid lines indicate the regions of the tubes assayed for adsorbed hexokinase. and the dotted lines indicate the regions of the tube through which an air— water interface has passed. The results of the assays are given in the column to the right. generalized phenomena that pertain to a wide variety of proteins, it seemed reasonable that other proteins might adsorb to polypropylene tubes. Therefore, we performed experiments similar to that shown in Fig. 3 using yeast hexokinase, beef heart lactate dehydrogenase, and glucose- TABLE I INJECTION or ENZYME UNDER A Wares LAYER" Micromoles 06? produced! 10 minutes Control .030 Experimental .016 ° In the control, 0.5 ml of glucose-6-P-solubilized enzyme in 0.25 M sucrose was added to a polypropylene tube. then drawn back out. The experimental sample was treated exactly the same way except that the enzyme was underlaid beneath a 0.3-ml water layer and drawn back out from the bottom of the tube. The tubes were then washed and assayed for adsorbed hexokinase as described in the legend to Fig. 3. 140 SHORT COMMUNICATIONS 635 6-P dehydrogenase. Each of these enzymes was shown to be adsorbed (with retention of catalytic activity) in amounts comparable to those seen with rat brain hexokinase. Furthermore, since BSA diminishes hexo- kinase adsorption, presumably by contributing to the surface film and competing for hydrophobic binding sites, it too must be adsorbed. In addi- tion, we have observed that plastics other than polypropylene also give adsorption artifacts with hexokinase. Referring to protein adsorption on glass, Sobotka and Trumit (2) men- tion that “many proteins are adsorbed onto solid surfaces from solution to form complete, well-adhering monolayers of undenatured molecules. This phenomenon has caused—and is still causing—emoneous results in work with highly diluted protein solution.“ The observations in this paper, coupled with the increasing use of plastic laboratory products, suggest that an awareness of this problem is necessary whenever a dilute protein solution is used with polypropylene (or similarly hydro- phobic) vessels. REFERENCES l. Godson. G. N. (1967) in Methods in Enzymology (Grossman, L., and Moldave. K.. eds.). Vol. X11, Part A. p. 504, Academic Press, New York. 2. Sobotka. H.. and Trurm't, H. J., (1961) in Analytical Methods in Protein Chemistry (Alexander. P.. and Block. R. J., eds.). Vol. 3. pp. 212-243. Pergarnon. New York. . Wilson. .1 . E. (1973) Arch. Biochem. Biophys. 159, 543-549. . Chou. A. C., and Wilson. J. E. (1972) Arch. Biochem. Biophys. 151, 48-55. 5. Langmuir. 1.. Schaefer, V. 1.. and Wrinch. D. M., (1937) Science 85, 76-80. #1.» PHILIP L. FELGNER JOHN E. WILSON Department of Biochemistry Michigan State University East Lansing. Michigan 48824 Received March 11. I 976; accepted April 12. I976 APPENDIX C 141 142 Dansylation Of Tyrosine: Hindrance by N-Ethylmorpholine and Photodegradation of O-Dansylated Derivatives PHILIP L. FELGNER AND JOHN E. WILSON Department of Biochemistry. Michigan State University East Lansing. Michigan 48824 Received December 30, 1976; accepted March 8. 1977 Dansylation of free tyrosine or of rat brain hexokinase (A’I'P:o-hexose 6-phosphotransferase: EC 2.7.1.1). which contains an N -terminal tyrosine residue, yields both the didansyl and N-rnonodansyl derivatives if N-ethylmorpholine is used in the dansylation procedure [W. R. Gray (1972) in Methods in Enzymology (Hirs, C. H. W., and Timasheff, S. N., eds.). Vol. 25, pp. 121-138. Academic Press. New York]. If the N-ethylmorpholine is replaced by NaHCO, buffer (pH 9.5). the didansyl derivative is formed almost exclusively, and the ambiguity resulting from the formation of two derivatives from a single N-terminal residue is thereby eliminated. Therefore. a slightly modified dansylation procedure. using NaHCO, buffer is recommended: the validity of the modified procedure was demonstrated by its successful application to six different proteins having previously known N-terminal amino acids. The didansyl and O-monodansyl derivatives of tyrosine are remarkably photolabile as compared to the N-dansyl derivatives. Unless specific precautions against unnecessary irradiation are observed. photolytic degradation Of the didansyl tyrosine derivatives could occur during experimental manipulations: loss of the didansylated compound and formation of photolysis products complicates the interpretation of experiments in which a single didansyl derivative (of the N-terminal residue) is expected. Dansylation’ is one of the most frequently employed methods for determining the N-terminal amino acid of proteins or peptides. The popularity of the method derives from its relative simplicity. sensi- tivity, and general reliability. While working with a protein (rat brain hexokinase: ATon-hexose 6-phosphotransferase, EC 2.7.1.1) found to have an N-terminal tyrosine (see below), we encountered some technical problems with the dansylation procedure which led to equivocal results. Additional studies disclosed that these technical problems were not unique to the N-terminal tyrosine of hexokinase, but were, in fact, generally ' Abbreviations used: Dansyl chloride, S-dimethylarnino-l-naphthalene sulfonyl chloride; SDS, sodium dodecyl sulfate: PPO. 2,5-diphenyloxazole: POPOP. 1.4-bis[2(5- phenyloxazoylllbenzene. Dansylation will be used to include both the reaction with dansyl chloride. or. in reference to the determination of N-terminal amino acid residues. in the more inclusive sense of reaction with dansyl chloride followed by hydrolysis and subsequent identification Of dansyl derivatives. 601 Copyright i' 1977 by Academic Press. Inc. All rights 0t reproduction If) any lorm rescued. ISSN 0003-2697 143 602 FELGNER AND WILSON encountered when working with dansyl derivatives Of tyrosine. Briefly. these problems are due to (a) a marked photosensitivity of the didansylated2 tyrosine derivative: unless special precautions are taken to prevent photolysis. a multiplicity of fluorescent photolysis products are formed which hinders the straightforward interpretation Of chromato- graphic identification procedures; (b) a formation of appreciable (or even predominant) amounts of the N-monodansylated derivative under condi- tions recommended (1) for the dansylation reaction. This result contrasts with the frequent assumption that the reactivity Of the amino and phenolic groups are sufi‘iciently similar that the didansyl derivative will be the predominant (or exclusive) form resulting from N-terminal tyrosine residues. The present work reveals that this is not the case and that the extent of didansylation depends markedly on the reaction conditions. Conditions have been devised which yield almost exclusively the didansyl derivative and thus avoid the potential uncertainty which can result from formation of both monOdansyl and didansyl derivatives Of N-terminal tyrosine residues. MATERIALS AND METHODS Materials. Dansyl chloride, L-tyrosine, and standard dansyl amino acids were obtained from Sigma (St. Louis, MO.). Sodium dodecyl sulfate (sequenation grade), N-ethylmorpholine (sequanal grade), and dimethyl- formamide (silylation grade) were purchased from Pierce Chemical Company (Rockford. 111.). L-[U-“C]Tyrosine and [G-3Hldansyl chloride were products of Amersham-Searle (Arlington Heights, Ill.) and l-nitroso—Z-naphthol was from Mallinckrodt (St. Louis, MO.). Polyamide thin-layer sheets were Obtained from Gallard-Schlesinger (Carle Place, N.Y.). All other chemicals were reagent grade and were obtained from commercial sources. Thin-layer chromatography of dansyl derivatives. Thin-layer chromatography was performed as described by Weiner et al. (2). Samples were spotted on 5 x S-cm polyamide plates, and chromatography was carried out in the first dimension with 1.5% formic acid in water (Solvent 1). After drying. chromatography was performed in the second dimension using. successively. benzene:acetic acid, 9:1 (Solvent II) and the ethyl acetatezacetic acidzmethanol, 20: 1 :1 (Solvent III). [The further chromatog- raphy in the second dimension employed by Weiner et al. (2) was not required for the present studies] Dansylation. Tyrosine (10 pl of a 1 mM solution, raised to pH 10 with " Tyrosine contains two groups which can react with dansyl chloride. the aoamino and the phenolic hydroxyl. which would give the N-dansyl and O-dansyl derivatives. reSpectively. The didansyl derivative refers to the compound produced by reaction Of both the amino and phenolic functions. 144 DANSYLATION OF TYROSINE 603 NaOH to promote solubilization) was added to 20 pl of 0.1 M sodium bicarbonate, pH 9.5. The reaction was started by adding 30 pl of 10 mM dansyl chloride in acetone. After 1 hr at room temperature, 50 pl of 0.1 N NaOH was added to promote hydrolysis of any remaining dansyl chloride (1), and then the solution was acidified with 20 pl of 6 N HCl. After removal of acetone under a stream of nitrogen, the samples were completely dried under vacuum at 35°C on a rotary evaporator, dissolved in 10 p1 of acetic acid—acetone (40:60, v/v), and then chromatographed. This procedure yields the didansyl tyrosine derivative almost exclusively; however, in some experiments, other buffers at various pHs replaced the bicarbonate in order to generate significant amounts of the N -monodansyl derivative (see below). For dansylation of proteins, the method outlined by Gray (1) was routinely followed with three modifications: N-ethylmorpholine was replaced by 0.5 M sodium bicarbonate, pH 9.5; protein samples were not performic acid oxidized; and the dansylated protein sample was washed more extensively. The procedure was as follows. Lyophilized protein. 50250 pg, was placed in a 6 x SO-mm culture tube and was dissolved in 50 pl of 1% SDS by heating in a boiling water bath for about 5 min; if the protein sample contained buffer or salts, it was dialyzed overnight against distilled water before lyophilization. After cooling, 50 pl of 0.5 M sodium bicarbonate, pH 9.5, followed by 75 pl of a 25-mg/ml dansyl chloride solution in dimethyl formamide. was added. The sample was mixed with a Vortex mixer, covered with Parafilm, and incubated at room temperature for at least 1 hr. Since the dansyl chloride was not entirely soluble in this aqueous solution, the reaction mixture had a cloudy appearance that cleared as the reaction went to completion. One volume Of20% TCA (w/v) was added to stop the reaction and precipitate the protein. The precipitated protein was washed once with 1 N HCl and once with 80% acetone:water. To the dried precipitate was added 50 pl of constant-boiling HCl; the culture tube was sealed in vacuo, and the sample was hydrolyzed at 110°C for 18 hr. The HCl was removed on a rotary evaporator, and the residue was dissolved in 10 pl of 40% acetic acid:acetone and was spotted on polyamide-layer sheets for chromatography. Photolysis of didansyl tyrosine on thin-layer plates. Radioactive didansyl tyrosine was prepared from tyrosine and [3H]dansyl chloride as described above. The dansylated tyrosine was chromatographed in Solvents l and 11 (see Materials and Methods), and the chromatograms were then irradiated with an ultraviolet light source (Sargent-Welch Catalog NO. 844260: long wavelength: ~366 nm maximum emission, 100 W) for various time intervals. The plates were placed exactly 20 cm from the lamp and were centered in the most intense portion of the beam. At this distance from the source, the temperature was 35°C. After irradiation, further chromatography using Solvent III was done to resolve the 145 604 FELGNER AND WILSON remaining didansyl tyrosine from photolysis products. The relevant areas were scraped from the plates. and radioactivity was determined by scintillation counting. The scintillation solution used contained 4 g of PPO and 0.05 g of POPOP per liter of toluene. RESULTS Characterization of Dansylated Tyrosine Derivatives Both commercially obtained (Fig. 1) samples of “dansyl tyrosine" or samples prepared under various conditions in the laboratory (see below) can contain several fluorescent species. We are primarily concerned here FIG. 1. Fluorescent species in commercially Obtained sample of dansyl tyrosine. Approximately 0.5 pl of a solution of commercially obtained dansyl tyrosine (0.25 mg/ml in 40% acetic acid:acetone) was spotted on a polyamide sheet and was chromatographed according to Weiner er al. (2). Several fluorescent species (some of which may be photolytic degradation products: see text) are present in trace amounts. but the principal components observed are: (I) didansyl tyrosine; (II) N-dansyl tyrosine: (III) O-dansyl tyrosine: (IV) unknown: (Vidansylsulfonic acid (the hydrolysis product ofdansyl chloride). Neither 111 nor IV was seen in dansyl tyrosine samples prepared in our laboratory by the procedure given in Materials and Methods. I46 DANSYLATION OF TYROSINE 605 with the species labeled 1, II, and III in Fig. 1. Compound III was present in the commercially obtained sample of dansyl tyrosine, and, based on its bright yellow fluorescence and chromatographic properties, it is presumed to be the O-dansyl derivative of tyrosine (1-3) such as would be formed by dansylation of non-N-terminal tyrosine residues in proteins or peptides: dansylation of a tripeptide (gly-leu-tyr) with a non-N-terminal tyrosine, which is expected to yield O-dansyl tyrosine, did give a fluorescent derivative corresponding to Compound 11] (data not shown). Compound III was not produced in appreciable amounts when we directly dansylated tyrosine except at high pH (see below), and we have not attempted to characterize this species in the present work. In contrast, Compounds I and II were readily produced from both free tyrosine and N-terminal tyrosine. These species have been reported to be the didansyl (1-3) and N -dansyl (3) derivatives, respectively, but the experimental basis for this assignment of identities has not, to our knowledge, been published. Therefore, we further characterized these compounds in the present study. Dansyl chloride reacts with unprotonated amines and phenols (l ,4), and, therefore, the relative reactivity of these groups is markedly dependent on the pH of the reaction mixture. Since the pK of the phenolic hydroxyl of tyrosine is 10.1 and the pK of the amino group is 9.1 (S), at lower pH, the amino group should be more reactive than the phenolic hydroxyl. The results shown in Table 1 are consistent with this expectation: At pH ~7-8, substantial amounts of Compound 11, proposed to be the N-dansyl derivative, are formed. As the pH is raised, the phenolic function becomes more reactive, and the formation of the didansyl derivative (1) is favored. At pH 10.5 in NaI-ICOa buffer. the phenolic hydroxyl apparently reacts sufficiently fast so that the O-dansyl derivative comprises an appreciable percentage of the reaction mixture. l-Nitroso-Z-naphthol, in the presence of nitric acid, has been shown to react specifically with phenolic groups to give a red-colored product (6). When sprayed with this reagent, Compound I] reacted to give a red color while Compound I did not. Thus, Compound 11 contained a free phenolic group (consistent with its identification as the N -dansyl derivative). while Compound I did not (as expected for the didansyl derivative). Finally. Table 2 shows the results of an experiment in which [”C]tyrosine was dansylated with [3H]dansyl chloride, and the 3H/ “C ratio was determined in various reaction products after their separation by thin-layer chromatography. If Compound I is the didansyl derivative and II is the N-dansyl derivative of tyrosine, the 3H/“C ratio of Compound I should be twice that found for I]; the observed value was O.72/O.42 = 1.7: essentially identical results were obtained in other experiments [mean ratio of 1.7 t 0.1 (SD) in four separate experiments]. We cannot presently explain the departure of this ratio from the expected value of 2.0, but, nevertheless, it is clear that Compound 1 contains approximately twice as I47 606 FELGNER AND WILSON TABLE I RELATIVE AMOUNTS or DANSYL TvnOSINE DERIVATIVES PRODUCED AT VAanus rHs‘ Didansyl N-Dansyl O-Dansyl tyrosine tyrosine tyrosine Bufl'er pH (‘7c) (‘72) (%) Phosphate 7.2 50 50 nd’ Phosphate 8.0 62 38 nd Bicarbonate 8.3 70 30 nd Bicarbonate 9.5 99 1 nd Bicarbonate 10.5 91 1 8 Borate 8.5 99 1 nd Borate 9.5 98 1 l ' Dansylation was carried out as described in Materials and Methods. except that 1 uCi of [’Hldansyl chloride was mixed with the unlabeled dansyl chloride before addition to buffered tyrosine. Buffers used were all 0.1 M. had the indicated pH. and were prepared with the sodium salts. Samples were chromatographed. the appropriate spots were scraped off, and radioactivity was determined. 0 nd. Not detected. many dansyl groups per tyrosine residue as does II, which would support their identification as the didansyl and monodansyl derivatives, respec- tively. In summary, all of these experiments are consistent with the previous identification (1-3) of Compounds I and II as the didansyl and monodansyl derivatives of tyrosine. We have also confirmed the identification of Compound I as the didansyl derivative by direct-probe mass spectrometry TABLE 2 DETERMINATION or THE DANSYLITYROSINE RATIO 1N COMPOUNDS 1 AND ll" DP 3” 10C SHICN Compound” (dpm) (dpm) (dpm) (l) Didansyl tyrosine 2566 3569 0.72 (ll) N-Dansyl tyrosine 1964 4672 0.42 (V) Dansyl sulfonic acid 16032 390 41 " Dansylation was carried out as described in Materials and Methods using [3H]dansyl chloride and [“Cltyrosine: the buffer was 0.1 M sodium phosphate. pH 8.0. After chromatography. the indicated spots were scraped off. and the radioactivity was determined. In this experiment only. the scintillant used contained 200 g of naphthalene. 20 g of PPO. 1.6 g of dimethyl POPOP, 2 liter of xylene. and 1.1 liter of Triton X-114: standard curves for calculating the disintegrations per minute of 3H and "C were provided by Dr. A. J. Morris (Biochemistry Department. Michigan State University. East Lansing. Mich.). " Compounds 1, II. and V are as designated in Fig. l. I48 DANSYLATION OF TYROSINE 607 TABLE 3 EFFECT OF N-ETHYLMORPHOLINE ON DANSYLATION or TYROSINE” Didansyl N-Dansyl O-Dansyl N-Ethylmorpholine tyrosine tyrosine tyrosine (M) (‘70 (‘72) (7:) 0. 1 44 56 nd" 0.4 42 58 nd 1.6 29 71 nd 7.8 (neat)‘ I 99 nd ‘ Tyrosine was dansylated as described in Materials and Methods using the indicated concentration of N-ethylmorpholine in place of the 0.1 M bicarbonate: the pH of the diluted N-ethylmorpholine solutions was 10.3. [3H]Dansyl chloride was used to permit quantitation by radioactivity determination. ’ nd. Not detected. ‘ The final concentration of N-ethylmorpholine in the reaction mixture was 2.6 M (see Materials and Methods): this is comparable to the concentration (2.2 M) used in the dansylation procedure Of Gray (1). Of the compound eluted from thin-layer plates: The mass spectrum was in excellent agreement with that previously reported by Seiler et al. (7). Unfortunately, for unknown reasons, we have not been able to obtain a mass spectrum Of the similarly isolated Compound 11, and, hence, mass spectrometry has not yet been useful in confirming the identity of this species as the N -dansyl compound. Use of N-Ethylmorpholine as Buffer in Dansylation Reactions Gray (1) has recommended N-ethylmorpholine for dansylation of proteins, stating that this compound not only serves as a base but also has a detergent-like action which facilitates reaction with proteins. When we utilized N-ethylmorpholine for dansylation Of rat brain hexokinase (8) using the procedure of Gray (1), both didansyl and N -monodansyl tyrosine derivatives were obtained, the ratio of didansylzN-monodansyl derivatives being approximately 2:3. In contrast, when 0.5 M sodium bicarbonate, pH 9.5, was used as buffer, only the didansyl tyrosine (>99%) derivative was formed.3 These results clearly suggested that N-ethylmorpholine affects the reactivity of the tyrosine residues, and this was confirmed by experiments with free tyrosine (Table 3): Increasing the N-ethylmorpholine content of the reaction mixture drastically reduced the reactivity of the phenolic hydroxyl. resulting in an increasing yield of the N-monodansyl derivative. Conceivably, this may result from an interaction of the deter- gent-like N-ethylmorpholine (l) with the phenolic ring, thereby hindering its dansylation. 3 We are grateful to Dr. Dean Ersfeld for confirming the presence of N-terminal tyrosine by the Edman method. I49 608 FELGNER AND WILSON Dansylation of Proteins in Bicarbonate Buffer In contrast to the results with N-ethylmorpholine, dansylation of rat brain hexokinase in bicarbonate buffer gave a single didansylated derivative of the N-terminal tyrosine. To check on the general applicability of the method, other proteins with known N-terminal amino acids were examined by this same procedure. In each case, the expected N-terminal amino acid was unambiguously identified. The proteins used and their N-terminal residues were: P. putida 2-ketO3-deoxy-6-phosphogluconate aldolase, threonine (9); rabbit muscle aldolase, proline: egg white lysozyme, lysine: bovine pancreatic ribonuclease, lysine; horse heart cytochrome c, glycine; bovine fetal hemoglobin, a-chain valine, B-chain methionine [see Ref. (10) for references related to the last five proteins]. We therefore conclude that this method is an acceptable alternative to that of Gray (1) and is, in fact, an improvement in cases in which an N-terminal tyrosine is involved. Photolability of O-Dansyl Tyrosine Derivatives As shown by the results in Fig. 2, the didansyl derivative Of tyrosine is remarkably labile when irradiated with ultraviolet light. A 20-min photolysis under the conditions described in Materials and Methods resulted in virtually complete conversion of the didansyl derivative to unidentified photolysis products (compare Figs. 1 and 2). Photodegrada- tion Of the O-dansyl derivative (Compound 11]) was also evident, while, in contrast, no obvious photolytic degradation of N-dansyl tyrosine (again compare Figs. 1 and 2) or dansyl glycine (not shown) was observed under these conditions. Thus, it would appear that the O-dansyl group is much more susceptible to photolytic degradation as compared to N-dansyl derivatives. The rapidity of this photodegradation is emphasized by the results in Fig. 3, which demonstrate that over 50% of the didansyl tyrosine is converted to photolytic degradation products within 3-4 min under the conditions employed. It should be noted here that photodegradation was even more rapid when the distance from the light source was decreased (as would usually be the case when thin-layer plates were being examined for fluorescent compounds), or when a short-wavelength (mineralogical) ultraviolet source (e.g., Sargent-Welch Catalog No. S-44240, 254 nm maximum emission) was used. DISCUSSION The present results demonstrate two potential sources of difficulty that may be encountered when utilizing the dansylation procedure with proteins containing N-terminal tyrosine residues. (a) One of these is formation of two derivatives (didansyl and N -dansyl) when dansylation is done by the procedure of Gray (1), employing 150 DANSYLATI ON OF TYROSINE 609 FIG. 2. Separation of didansyl tyrosine from photodegradation products. Commercially Obtained dansyl tyrosine was chromatographed as in Fig. l (with which this figure should be compared) except that. prior to running in Solvent III (see Materials and Methods). the plate was irradiated for 20 min with ultraviolet light as described in Materials and Methods. Subsequentchromatographyin Solvent III resolved ‘ " ")frnm photolytic degradation products (X). Also note the marked loss of 0-dansyl tyrosine (III Fig. l) as a result of irradiation while no appreciable destruction of the N-dansyl derivative (II) was noted; these results clearly illustrate the much greater photolability of O-dansyl derivatives as compared to N-dansyl compounds. N-ethylmorpholine in the solvent; the formation of two derivatives of a single N-terminal amino acid obviously is not an overwhelming problem, but, nevertheless, it does make the interpretation of thin-layer chromato- grams (Or analogous identification procedures) less straightforward and represents an unnecessary complication. (b) Another source of difficulty is the remarkable photolability of the O-dansyl group: unless precautions against photolysis are taken, substantial amounts of photolytic degradation products may be formed, markedly diminishing the yield of the expected didansyl derivative and complicating the interpretation of experiments in which a single derivative (of the N-terminal residue) is expected. Although the photolability of 151 610 FELGNER AND WILSON l‘ m ‘V Pore-M Didansyl Tyrosine 3, . H 710 r. '2 i Mm FIG. 3. Rate of photolysis of didansyl tyrosine. The experiment was conducted essentially as described in the legend to Fig. 2. except that the dansyl tyrosine was prepared by the procedure given in Materials and Methods using [‘Hldansyl chloride. The duration of irradiation was varied as indicated on the abscissa. After chromatography in Solvent III. the areas corresponding to didansyl tyrosine and its photolysis products (areas I and X, respectively, shown in Fig. 2) were scraped off, and their radioactivity was determined. The results are expressed as the percentage of total counts (T + X) found as undegraded didansyl derivative (1 ). We attribute the Observation of substantial (~20%) photodegradation products at zero time of irradiation to their formation during the preceding preparative and chromatographic procedures: no special precautions were taken against light exposure, i.e. . the Operations were conducted under ordinary laboratory lighting‘ conditions. various N-dansyl derivatives has been previously studied (11,12), we are not aware Of any results which emphasized the much greater photolability of O-dansyl derivatives. The present results further provide the basis for avoiding these two potential problems. (A) Dansylation should be done as described in Materials and Methods, using sodium bicarbonate as buffer. (B) Obviously, prevention of photolysis requires the avoidance of significant exposure to light, especially ultraviolet, during the manipula— tions. Previously, it was not uncommon in our laboratory (and, we suspect, in Others) to examine briefly thin-layer plates under ultraviolet light before chromatography to verify that sufficient sample was spotted to provide readily observable fluorescence; in the light of the present results, such “sneak previews" are obviously to be avoided. It is likely that the technical problems described here contributed to the previous erroneous identification of glycine as the N-terminal amino acid of rat brain hexokinase (8). Weiner et al. (2) have summarized some technical problems which may complicate N-terminal analysis by the dansylation procedure. In addition, the oxidative capability of dansyl chloride (13) might also cause problems with N-terminal cysteine residues. The present paper draws attention to the difficulties possible when proteins with Noterminal tyrosine are encountered. The effective use of the dansylation procedure demands an awareness of these potential sources of difficulty. 152 DANSYLATION OF TYROSINE 61 1 ACKNOWLEDGMENTS We are grateful to Dr. W. A. Wood for providing a sample Of purified 2- ketO-3-deoxy-6-phosphogluconate aldolase from P. putida. The financial support provided by NIH Grant No. NS 09910 is gratefully acknowledged. 01:.pr 1‘) 13. REFERENCES . Gray. W. R. (1972) in Methods in Enzymology (Hirs. C. H. W., and Timasheff, S. N., eds.). Vol. 25. pp. 121—138, Academic Press. New York. . Weiner. A. M., Platt, T., and Weber. K. (1972)]. Biol. Chem. 247, 3242-3251. . Hartley. B. S. (l970)Biochem. J. 119, 805-822. Gros. C., and Labouesse. B. (l969) Eur. J. Biochem. 7, 463-470. . Lehninger. A. L. (1975) Biochemistry. p. 79. Worth, New York. . Greenstein. J. P.. and Winitz. M. (1961) Chemistry of the Amino Acids. Vol. 3. p. 2353. John Wiley. New York. . Seiler, N., Schneider. H. H., and Sonnenberg. K. -D. (1971)Anal. Biochem. 44,451-457. . Chou, A. C., and Wilson. J. E. (l972)Arch. Biochem. Biophys. 151, 48-55. . Robertson. D. C., Hammerstedt. R. H.. and Wood. W. A. (1971).]. Biol. Chem. N6. 2075—2083. . Dayhoff. M. O. (1972) Atlas of Protein Sequence and Structure, Vol. 5. National Biomedical Research Foundation. Silver Spring. Md. . D'Souza. L., Bhatt. R., Madaiah. M.,and Day. R. A. (l970)Arch.Biochem. Biophys 141, 690-693. Pouchan. M. l.. and Passeron. E. J. (1975)Anal. Biochem. 63, 585-591. Schulze, E.. and Neuhoff. V. (1976) Hoppe Seyler's Z. Physio]. Chem. 357, 225—231. APPENDIX D 153 154 Effect of Neutral Salts on the Interaction of Rat Brain Hexokinase with the Outer Mitochondrial Membrane PHILIP L. FELGNER AND JOHN E. WILSON Biochemistry Department, Michigan State University, East Lansing, Michigan 48824 Received February 24, 1977 The glucose 6—phosphate (Glc-6-P)oinduced solubilization of mitochondrial hexokinase (ATPzn-hexose 6-phosphotransferase, EC 2.7.1.1) from rat brain can be reversed by low concentrations (ionic strength <~0.02 M) of neutral salts. When compared to the original particulate enzyme (i.e. . enzyme found on the particles prior to solubilization by Glc-G-P), the rebound enzyme is similar in distribution on sucrose gradients, K... for ATP, inhibition by antiserum to purified brain hexokinase. and resistance to removal by exhaustive washing of the particles. The effectiveness of chloride salts at promoting rebinding increases in the order Cs+ < Rb+ < K+ 5 Na" < Li’ 4 Mg". This salt-induced rebinding is attributed to the screening of negative charges on the enzyme and/or membrane by cations. thereby decreasing repulsive forces and enhancing attractive interactions between enzyme and membrane. Solubilization of the enzyme, both in the presence and absence of Glc-G—P, is increased at alkaline pH, as would be expected due to increasing repulsive interactions between negative charges on membrane and enzyme as the pH is increased beyond the pI of the enzyme (pI s 6.3). In contrast to previous interpretations, P. displayed no special efficacy at reversing Glc-6-P-induced solubiliza- tion, being comparable to other neutral salts on an ionic strength basis. However, P, and its structural analog, arsenate, were shown to counteract specifically the Glc-6-P- induced inhibition and conformational change in the enzyme. At higher concentrations (ionic strength >~0.02 at) neutral salts themselves lead to reversible dissociation of the enzyme from the mitochondria. The efficacy of the salts depends primarily on the pH and on the position of the anion in the Hofmeister series, with salts of chaotropic anions (SCN’, 1', Br') being most effective. At pH 6, both chaotropic and nonchaotropic salts solubilize the enzyme, while at pH 8.5, only the chaotropes retain this ability. Neutral salts also have a reversible effect on the conformation of the enzyme, as reflected by enzymatic activity, with chaotropic salts again being most effective; there is no pro- nounced influence of pH (in the range of pH 6-8.5) on the ability of the salts to cause conformational change in the enzyme. Based on a lack of correlation between salt- induced solubilization and conformational changes affecting activity, it is concluded that the latter are not directly responsible for release of the enzyme from the membrane. In the presence of KSCN, the extent of solubilization decreased with increase in tempera- ture, indicating a negative enthalpy for solubilization. In contrast, in the absence of salt, the enthalpy for solubilization was positive. These temperature effects and the effects of neutral salts on the hexokinase-membrane interaction are interpreted in terms of a model in which electrostatic forces are considered to be Of major importance. At low ionic strength. repulsive forces between negative charges on enzyme and membrane predomi- nate; screening of these charges by cations diminishes the repulsion, effectively enhanc- ing attractive electrostatic forces between enzyme and membrane and thus promoting their interaction. At higher ionic strengths, the attractive electrostatic forces are themselves disrupted, resulting in dissociation of the enzyme from the membrane. It is proposed that the greater effectiveness of chaotropic salts at disrupting these attractive forces is due to their increased ability to penetrate through hydrophobic regions of enzyme and membrane to relatively inaccessible sites of electrostatic interaction. 282 Copyright c 1977 by Academic Press, Inc. All rights of reproduction in any form reserved. ISSN 0003-9861 155 NEUTRAL SALTS AND HEXOKINASE BINDING The mitochondrial hexokinase (ATP:D- hexose 6-phosphotransferase, EC 2.7.1.1) from brain can be solubilized by low con- centrations (~1 mM) of Glc-6—P‘ (1, 2). In the presence of MgClz, the enzyme rebinds to the outer mitochondrial membrane but not to inner membranes or microsomes (1, 3) nor to myelin and other heterogeneous membrane fragments derived from brain (this paper). This presumably indicates that some specific component (or compo- nents) of the outer mitochondrial mem- brane selectively interacts with the en- zyme. Pure bindable enzyme has recently been obtained in this laboratory (4, 5), and an investigation of the membrane compo- nent(s) necessary for binding is currently in progress. The Objective of these studies is to gain an understanding Of the molecu- lar basis for the selectivity Of binding and the forces involved in the hexokinase—mi- tochondrial membrane interaction. The present investigation, dealing with the effect of neutral salts on the associa- tion between hexokinase and the mito- chondrion, has provided the basis for a hypothesis describing the interactions be- tween the outer mitochondrial membrane and hexokinase. These interactions appear to be primarily electrostatic in nature, in accord with the suggestion of Teichgraber and Biesold (6) and include both repulsive and attractive components. MATERIALS AND METHODS Adult Sprague-Dawley rats were obtained from Spartan Research (Haslett, Michigan) and main- tained on a common laboratory diet and water ad libitum. Brain tissue from these animals was stored under liquid N,. Tris, ATP, NADH, phosphoenol- pyruvate, lactate dehydrogenase, and pyruvate ki- nase were products of the Sigma Chemical Co. Particulate hexokinase was prepared by the fol- lowing procedure. Brain tissue, which had been stored frozen in liquid N,, was thawed at room tem- perature in 0.25 M sucrose and, after thawing, the supernatant was decanted. The tissue was homoge- nized in 10 volumes of 0.25 M sucrose at 0°C using a Teflon-glass homogenizer (size C, A. H. Thomas Co.). The homogenate was centrifuged at 1000g x 10 min at 4°C and the pellet discarded. The superna. ‘ Abbreviation used: GIc-6-P, glucose 6-phoso phate. 283 tent was centrifuged at 40,000g x 10 min and the supernatant discarded. The pellet was washed by resuspending in 10 volumes of 0.25 M sucrose (0°C) followed by centrifugation at 40,0003 x 10 min, and the particulate enzyme then resuspended in 10 vol- umes of 0.25 M sucrose. Total hexokinase activity was determined after treatment of the particles with 0.5% (v/v) Triton X-100 (2). Unless noted otherwise, solubilization was stud- ied at pH 6.6 t 0.1 in unbufi‘ered 0.25 M sucrose containing the indicated salt concentrations. Ad- justment of pH was done with 0.02 N KOH or 0.02 N HCl, as necessary. Except where indicated, hexokinase was assayed by the Glc-G-P dehydrogenase method as previously described (4). For studying inhibition by Glc-6-P. a coupled assay employing pyruvate kinase plus lac- tate dehydrogenase was used. This assay solution contained, in a total volume of 1.0 ml. 2 mar glucose. 6 mil ATP, 10 mar MgCl,, 50 mm Tris-Cl buffer, pH 7.4, 0.8 IU of pyruvate kinase, 1 IU of lactate dehy- drogenase, 1 mM phosphoenolpyruvate, 0.15 mM NADH, and 50 mM potassium chloride. The reaction was initiated by the addition of hexokinase, and the progress of the reaction was recorded as the decrease in absorbance at 340 nm. RESULTS Sucrose Density Gradient Fractionation of MgClg-Rebound Hexokinase Brain hexokinase has been shown to be associated with mitochondria, being bound to the outer membrane of that organelle (1, 3, 7). Solubilization of the enzyme with Glc-6-P and subsequent addition of MgClg leads to rebinding of the enzyme to the mitochondrial fraction (1, 2). Rebinding does not occur with microsomes or inner mitochondrial membranes (1, 3), indicat- ing a specific interaction with the outer mitochondrial membrane rather than an indiscriminate adsorption of the enzyme to membranes in general. Additional support for the specificity of this interaction is pro- vided by the results shown in Fig. 1. Frac- tionation of the crude mitochondrial frac- tion on discontinuous sucrose density gra- dients, according to De Robertis et al. (8), shows the hexokinase to be primarily lo— cated in fraction D (a heterogeneous frac- tion containing predominantly nerve end- ings and mitochondria) (Fig. 1, panel 1). These results are in excellent agreement with those previously obtained in this lab- 156 284 3n- —- so If : 2 -( s C ' s F 8 1'1 ,7 2’! L ,\o .E :7 so 3 ...., - 5 2 ' "» z = O a a . ~ II “ g O a are... h i j E . LP . ‘50 U 4 n . s e o' r FIG. 1. Distribution of native and rebound par- ticulate hexokinase on sucrose gradients. Particu- late hexokinase was prepared as described in the text. Aliquots (5 ml) containing 5.9 units total activ- ity were treated in the following manner: Panel 1, maintained at 0°C; panel 2, incubated at 25°C for 15 min with 3 ml MgCl.; panel 3, incubated at 25°C for 30 min with 1 mar Glc-6-P; panel 4, incubated at 25°C for 30 min with 1 mar Glc-6—P followed by addition of 3 ml MgCl, and a further l5-min incubation. After the indicated treatments, all samples were pelleted at 40,000g x 10 min, washed once with 0.25 at su- crose, then fractionated on discontinuous sucrose gradients as previously described (9). Fractions A-E are as defined by De Robertis et al. (8) and are enriched in: (A) myelin fragments; (B) heteroge- neous membrane fragments; (C) nerve endings; (D) nerve endings and mitochondria; and (E) mitochono dria. Hexokinase activity (solid lines) is expressed as a percentage of the original 5.9 units; the milli- grams of protein in each fraction are shown by the broken lines. A oratory (9) which is perhaps somewhat surprising since the present studies were done with frozen brain while in the pre- vious work (9) fresh tissue was used; evi- dently freezing of the intact brain does not lead to appreciable alteration of the sedi- mentation properties of subcellular parti- FELGNER AND WILSON eles derived therefrom. The significance of the Observed distribution has been previ- ously discussed (9) and will not be reiter- ated here. Treatment of the crude mitochondrial fraction with 3 mM MgCl, does not mark- edly affect the distribution of hexokinase on the gradient (Fig. 1, panel 2). Incuba- tion of the crude mitochondrial fraction with Glc-6-P results in extensive solubili- zation of the enzyme (1, 2), with a corre- sponding reduction in the amount of hexo- kinase found in the particulate fractions (Fig. 1, panel 3). Solubilization with Glc-6- P, followed by rebinding in the presence of MgClz, restores the hexokinase distribu- tion (Fig. 1, panel 4) to one essentially identical to that seen with the original particles (Fig. 1, panels 1 and 2). Of partic- ular note is that, although there are sub- stantial amounts Of membranous struc- tures [primarily myelin and heteroge- neous membranous fragments (8)] in frac- tions A and B, there is virtually no observ- able interaction Of the hexokinase with the membranes represented in these fractions. These results, therefore, support the view that MgClz-induced rebinding of solubi- lized hexokinase restores the enzyme to its original binding site on the outer mito- chondrial membrane and does not result in significant nonspecific adsorption of the enzyme to other membranous components of the crude mitochondrial fraction. Additional criteria indicating the simi- larity of the rebound enzyme and the origi- nal particulate hexokinase are (a) both forms have an apparent K ,,, for ATP that is substantially less than'that of the solubi- lized enzyme (Fig. 2), and (b) both forms are inhibited to exactly the same extent in titrations with antiserum prepared against the purified (4) enzyme, and both are equally resistant to elution by washing of the particles with 0.25 M sucrose or 2 mM EDTA, pH 7.4 (G. P. Wilkin and D. K. Young, personal communication). I Salt-Induced Reversal of Solubilization Glc-6—P is a potent inhibitor of brain hexokinase and P, is effective at reversing this inhibition (ll-13). Binding of GIc-G-P results in marked conformational changes, 157 NEUTRAL SALTS AND HEXOKINASE BINDING I I I -s.o Vim! «3.9 '09 FIG. 2. Apparent K. for ATP of the native particulate, Glco6-P-solubilized, and rebound hexokinase. Particulate hexokinase was prepared as described in the text. Solubilized enzyme was prepared by incubation of the particulate enzyme with 1 mar Glc-6-P for 30 min at 25°C, followed by centrifugation at 40,000g x 10 min. Rebound enzyme was prepared by solubilizing the enzyme with Glc-6-P as just described, but then adding either 50 mu NaCl or 3 mu MgCl, and incubating a further 30 min at 25°C before centrifuging; the pellets, containing the rebound enzyme, were resuspended in 0.25 M sucrose. Initial velocities were determined using the Glc-6- P dehydrogenase coupled assay with the indicated ATP concentrations and other conditions as described in the text. The data were analyzed using the method of Wilkinson (10). An aliquot of the original particulate enzyme, incubated at 25°C for 60 min without addition of Gle-6-P or salts was used in this comparison to detect any effect that incubation under these conditions might have on the apparent K... but no such effect was Observed, i.e., both freshly prepared particulate enzyme and particulate hexokinase subjected to these incubation conditions showed equivalent apparent K .. values. Similarly, controls in which the original particulate enzyme was incubated with 3 mil MgCI, or 50 ml NaCl also displayed apparent K... values identical to the unincubated particulate enzyme. Thus the slightly increased K. seen with "rebound” enzyme must reflect a real, albeit rather small, difference between the original and rebound particulate enzyme which is not attributable to the incubation per se. The K ., values (ATP) are: (I) particulate enzyme, 0.12 z 0.01 ml; (A) Glc-6-P-solubilized enzyme, 0.27 z 0.01 mas; (D) MgCl,-rebound enzyme, 0.17 r. 0.01 mil; (A) N aCl-rebound enzyme, 0.16 z 0.01 mm. A4— 15.0 285 and P, counteracts the Glc-G-P-induced changes (14, 15). A similar antagonism be- tween solubilization by Glc-6-P and rever- sal by P, was previously interpreted in terms of ligand-induced conformational changes (14), i.e., the conformation in- duced by Glc-6-P was considered to have low affinity for the outer mitochondrial membrane while that induced by P, was preferentially bound. However, as shown in Fig. 3, the effectiveness Of P, at reducing the solubilization by Glc-6-P does not ap- pear to be attributable to any specific ac- tion of P, since other salts are also effective at bringing about reversal. There are, in fact, two antagonistic effects of increasing ionic strength: Lower ionic strengths (<~0.02 M) bring about reversal of Glc-6- P—induced solubilization while further in- crease in ionic strength leads to solubiliza- tion. A similar but less marked effect of ionic strength has been seen in the ab— sence of Glc-6-P (2). The effectiveness of various salts at reversing Glc-6-P-induced solubilization and at directly solubilizing the enzyme is related to the position of the salt in the Hofmeister series (16); "salting in”-type salts (KSCN, KI, KN03), are 158 286 0. .3. .3. ionic mm FIG. 3. Reversal of solubilization by various _salts. Particulate hexokinase (0.97 unit/ml) was in- cubated with 1 mu Glc-6-P at 25’C for 30 min to solubilize the hexokinase. Appropriate additions of 1 is salt solutions were added to 0.34 ml aliquots and the volumes adjusted with 0.25 M sucrose so that the final volume of each sample was 0.5 ml. These sam- ples were incubated for an additional 30 min at 25°C and the tubes then spun at 40,000g x 10 min. Super- natants were assayed for hexokinase and results expressed as a percentage of the enzyme solubilized (0.47 unit/ml) when Glc-6-P and no added salt were used. It could be shown that the remaining activity was indeed present in the pellet, i.e.. the activity in supernatant plus pellet was >901: of the initial ac- tivity. more effective at solubilizing the enzyme than are "salting out”-type salts (KH,PO,, Na2804). The solubilization of hexokinase by various salts is explored in greater de— tail in experiments described below. The relative effectiveness of the various salts at causing rebinding is inversely related to their effectiveness at bringing about solu- bilization, as would be expected since re- binding and solubilization represent com- peting processes. Although the rebinding observed with P, does not appear to be a specific effect re- sulting from interaction of P, with the en- zyme as previously thought, P, does have FELGNER AND WILSON direct effects on the activity and conforma- tion of hexokinase which cannot be attrib- uted to ionic strength effects per se. Of all the salts tried, only P, or its structural analog, arsenate, cause reversal of the in- hibition by Glc-6-P (Fig. 4) and the protec- tion against chymotryptic digestion (Fig. 5) which results from Glc-6-P binding (14). These results clearly demonstrate that the specific effects of P, on the kinetic proper- ties and conformation of the enzyme are not directly connected to P,-induced rever- sal of solubilization. Kosow et al. (17) reached a similar conclusion in their work with the Type II isozyme of sarcoma 37 ascites tumor, but these authors also ap- parently did not recognize the generality of the salt effect on solubilization, i.e. , that reversal was not specific to P,. Effectiveness of Different Cations at Re- versing Glc-6-P-Induced Solubilization The effectiveness of monovalent cations in causing rebinding varies inversely with their crystal radius with the largest cation (08*) being least effective (Fig. 6). Al- though the differences are relatively small, they are consistently observed PO; AIO‘ 81 °‘\. i: hexokinase (”m-ll 25 50 MM F16. 4. Release of- Ole-6P inhibition by phos- phate and arsenate. All assays were performed with pure hexokinase (4) (6.1 units/ml in the absence of Glc-6—P) and 0.1 mm Glc-6-P, using the pyruvate kinase/lactate dehydrogenase coupled assay system with the indicated additions (sodium salts were used): phosphate (0), arsenate (O). acetate or sul- fate (which gave identical results) (I), and chloride (A). The inhibition by elevated levels of phosphate has been previously reported (11). NEUTRAL SALTS AND HEXOKINASE BINDING A H.. O \‘n‘. ....... .szM :E fa \‘M‘OOP .- .g. ‘\Efi . 3 ‘. .. E ‘. ‘-QOS- \ ' ‘6 \ ' 5 “, 097mm x; \ 06' \ \ a \ \ \ 0.02 '- u L 1 ”minutes 30 FIG. 5. Phosphate reversal of Glc-6-P protection against chymotryptic digestion. The conditions for this experiment were as previously described (14). The ordinate shows (on a logarithmic scale) the activity remaining at various times after chymOo trypsin addition. Two different Glc6P concentra- tions were used as indicated, either without arse- nate (O) or with 5 mm arsenate (I) or 5 mm phos- phate (A). Five millimolar concentrations of all other salts tried (NaCl, Na,SO,, NaN03) gave ex- actly the same results as with Glc-6-P alone. when saturating levels, e.g. , 25—50 mM, are used. Magnesium, as well as other divalent cations (Ca2+, Be“, Mn“; data not shown), are much more effective than monovalent cations when compared on an ionic strength basis (Fig. 6). The Effect of pH on Solubilization It had previously been found (18) that various commonly used buffers have a marked influence on the extent of solubili- zation at a given pH, i.e. , the proportion of hexokinase solubilized by Glc-6-P depends not only on pH but on the buffer used to maintain that pH. To avoid this complica- tion, we have determined the pH depend- ence of solubilization using a pH-Stat rather than using buffers to maintain pH. The results (Fig. 7) are consistent with the existence of repulsive electrostatic interac- tions between membrane and enzyme. Sol- ubilization in the absence of Glc-6-P in- creases at alkaline pH when negative 159 287 charges on both membrane and enzyme [pI = 6.3 (4)] would be increased; Glc-6-P in- creases the extent of solubilization, pre- sumably due to ligand-induced changes in conformation (14) which result in still less favorable interactions with the mem- brane. At pH values below the isoelectric point of the enzyme, negligible solubiliza- tion occurs either in the absence or in the presence of Glc-6-P. Solubilization of Mitochondrial Hexoki- nase by Neutral Salts Solubilization of particulate brain hexo- kinase at elevated concentrations of var- ious salts has been reported previously (1, 2), but there has been no extensive com- parison of the efficacy of different salts. Such a comparison is shown in Fig. 8. The cations appeared of minimal importance in this process, e.g., KCl and NaCl gave es- sentially identical results. In contrast it is quite apparent that different anions are Sfifififi Ezfifiir . . 0| j §§§ a o u in as FIG. 6. Reversal of solubilization by various chloride salts. The conditions for this experiment were as described in the legend to Fig. 3. There was 0.43 unit/ml solubilized in the absence of added salts. The inset shows the results of a similar but separate experiment in which the enzyme was solu- bilized with Glc-6—P. then rebound in the presence of the chloride salts of the indicated cations. Each sam- ple was run in triplicate. The values given are the mean activities : SD (units/milliliter) remaining in the supernatant; in the absence of added salt, 0.59 unit/ml was solubilized. 160 288 IOO)- ’66:, ./°\ 3 . in» -66 at /: .~ . I ..m-e/c L 4 1 ‘0 CD 7.0 u ‘0 pH FIG. 7. The pH dependence of solubilization by Glc-6-P. Particulate hexokinase (0.95 unit/ml) was adjusted to the indicated pH in a pH Stat with either 0.02 N HCl or 0.02 N NaOH, and solubilization was initiated by addition of 1 mac Glc-6-P. After a further incubation for 25 min at 25°C, the samples were centrifuged at 40,000g x 10 min and supernatants assayed. Results are expressed as a percentage of the original activity which was found in the super- natant; the slight increase in total activity seen in the presence of Glc-6oP has been previously observed (2). markedly different in their solubilization abilities. Salts of chaotropic anions (16, 19), e.g., SCN', 1‘, Br’, are highly effec- tive at solubilization of the enzyme, while anions at the opposite end of the Hofmeis- ter series, e.g., chloride, sulfate, are rela- tively ineffective. Solubilization by salts is readily reversi- ble. The results of an experiment using ° KSCN are shown in Table I. Treatment of particulate hexokinase with 80 mM KSCN leads to extensive solubilization; subse- quent dilution to a concentration of 40 mM KSCN results in rebinding, with the final soluble:particulate distribution being com- parable to that seen after directly solubi- lizing the enzyme with 40 mM KSCN . Sim- ilar reversal is seen after dilution of the KSCN from an initial concentration of 40 mM to one of 20 mM. The slight increase in solubilization which results from dilution alone (i.e., with maintenance of the origi- nal KSCN concentration) is expected for a dissociative process such as the release of hexokinase from the membrane (1). FELGNER AND WILSON 7 ”gt/é a; “‘2"... FIG. 8. Solubilization of particulate hexokinase by neutral salts. Particulate hexokinase (0.80 unit/ ml) was incubated at 25°C for 30 min in the presence of the indicated salts; the samples were centrifuged at 40,000g x 10 min (4°C) and the supernatants assayed for solubilized hexokinase. The final con- centration of the salts in the assay medium was always <10 m; thus the direct effects of the salts on enzyme activity (see Fig. 9 and text) were negligi- ble. A similar experiment. with comparable results, was done with enzyme that had first been solubi- lized with 1 mu Glc-6oP and then rebound by addi- tion of 3 m MgCl,. Thus, there is no obvious differ- ence between native particulate enzyme and re- bound hexokinase in their susceptibility to solubili- zation by neutral salts; additional similarities be- tween the native and rebound enzymes are de- scribed in the text. /_—J— .l 0. TABLE I Rsvsasraxurv or KSCN-Innvcsn Sowsnszarion or Haxoxmass" Initial Diluent Solubilized [KSCN] (9%) (mM) 80 None 61 0.25 is Sucrose 34 80 mm KSCN 68 40 None 19 0.25 n Sucrose 13 40 mm KSCN 28 0 None 7 0.25 n Sucrose 7 ' Particulate hexokinase (0.95 unit/ml) in 0.25 M sucrose was incubated for 30 min at 25°C with the indicated concentrations of KSCN. Subsequently, aliquots (1 ml) were diluted with an equal volume of KSCN or 0.25 M sucrose or left undiluted. After an additional 30 min at 25°C, the samples were centri- fuged at 40,000g x 10 min, and the solubilized hexo- kinase was determined by assaying the superna- tants. 161 NEUTRAL SALTS AND HEXOKINASE BINDING The extent of KSCN-induced solubiliza- tion was decreased at increasing tempera- tures (Table 11), indicating a negative en- thalpy for the dissociation process. The Effect of Salts on Hexokinase Activity The solubilization of hexokinase by salts could conceivably be due to an effect of these salts on the enzyme, on the mem- brane, or on both. In an attempt to assess directly the effects of salts on hexokinase, the experiments shown in Fig. 9 were per- formed. At concentrations below ~0.1 M, all salts tested activated the enzyme to a comparable extent, the maximal increase in activity being about 20% over that seen in the absence of added salt. At higher salt concentrations, inhibition was observed. Analogous salt effects have previously been found to occur with pig heart fumar- ase (20). As was the case with salt-induced solu- bilization, the cation had little effect; thus, the inhibitory action of various salts de- pended primarily on the nature of the an- ion, with highly chaotropic anions (SCN’, I') being much more effective than less chaotropic anions (Cl', SO42“). The inactivation was rapid and reversi- ble. Thus, if active enzyme was added to the assay mix containing inactivating con- centrations, e.g., ~0.4 M, of KSCN, the TABLE II Tzursaaruna Dsrsnnsncs or KSCN-INDUCED SowsmzA-non or stoxmxss‘ Temperature Solubilized (°C) (9? )° 10 64 20 59 30 53 40 50 ‘ Particulate hexokinase (0.88 unit/ml) was incu- bated with 80 mm KSCN for 30 min at the indicated temperatures; a preliminary experiment demon- strated that 30 min was adequate to attain maximal solubilization at each of these temperatures and that no detectable loss of activity occurred, even at 40°C. After centrifugation (40,000g x 10 min, at a temperature equal to the incubation temperature), hexokinase activity was determined in the superna- tant. ‘ Expressed as a percentage of the original activ- ity. / .. ... /i——~~—~r //:/ ((7 / .” c/ FIG. 9. Effect of neutral salts on hexokinase ac- tivity. Hexokinase was assayed as described in' the methods section with addition of the indicated salts to the assay medium; the results shown were ob. tained at the usual assay pH [40 mm N-2-hydroxy- ethylpiperazine-N'-2-ethanesulfonic acid (HEPES) buffer, pH 7.5]. but identical effects were also ob- served at pH 8.5 (40 mm HEPES buffer) and at pH 6.0 [40 mm 2(N-morpholino)ethanesulfonic acid (MES) as buffer]. The reaction was initiated by ad- dition of hexokinase. In the experiment shown, crude Glc-6-P-solubilized enzyme was used, but the same results were observed in experiments with pure (4) or particulate hexokinase. The coupling enzyme in the assay. Glc-G-P dehydrogenase, was also inhibited at elevated salt concentrations, but was still in excess with respect to the hexokinase; increasing the amount of coupling enzyme had no effect on the observed reaction rate. It is also con- ceivable that added anions might compete with ATP for available Mg“ and thereby reduce the availabil- ity of the ATP-Mg“ complex required as substrate; however, this was not an important factor in the observed inhibition since doubling the MgCl, con- centration had no effect on activity. indicated (Fig. 9) inhibition of' the activity was seen from the first observable moment in the spectrophotometric assay. Con- versely, if the enzyme was pretreated with inactivating concentrations of KSCN and then diluted into assay mix with no KSCN, the activity expected from the final KSCN concentration in the assay was ob- served immediately, with no evidence of concavity in the reaction progress curve which would be indicative of a slow rena- turation process. Effect of pH on Salt-Induced Rebinding and Solubilization As shown in Fig. 3, solubilization of mi- tochondrial hexokinase with Glc-6P fol- lowed by subsequent addition of neutral l62 290 salts leads first to reversal of the Glc-6-P- induced solubilization, followed by resolu- bilization of the enzyme with an efficacy depending largely on the chaotropicity of the anion. The effect of pH on this process when the relatively nonchaotropic NaCl is the salt involved is shown in Fig. 10A. At pH 6.0, there is very little solubilization due to Glc-6P and, hence, no reversal of Glc-6-P-induced solubilization to be ob- served, but moderate concentrations of NaCl are shown to be quite effective at extensively solubilizing the enzyme. At pH 8.5, Glc-6oP is very effective at solubi- lizing the enzyme, with reversal by low concentrations of NaCl being dramatically evident. However, at pH 8.5, NaCl is nota- bly ineffective at bringing about resolubili- zation of the enzyme. At an intermediate 11.5"“ as If: a“ u “I I H! FIG. 10. Effect of pH on solubilization by a chao- tropic (KSCN) and nonchaotropic (KCl) salt. (A). Particulate hexokinase (0.73 unit/ml) was incubated with 1 mM Glc-6~P for 30 min at 25°C. Portions of this mixture were adjusted to the indicated pH, and 0.8-ml aliquots were diluted to a final volume of 1.0 ml with appropriate amounts of 2 M NaCl or 0.25 M sucrose to yield the indicated final NaCl concentra- tion. After 30 min at 25°C, the samples were centri- fuged at 40,000g x 10 min, and hexokinase activity in the supernatant was determined. Final salt con- centrations in the assay medium were such as to have negligible effect on enzyme activity. (B), The procedure was essentially as for the experiment in (A), except that KCl and KSCN were used at the indicated pH values. In both (A) and (B), the activ~ ity in the supernatant is expressed as a percentage of the total original activity. Control experiments confirmed that recovery (in supernatant and pellet) of initial activity was >90%. FELGNER AND WILSON pH (pH 7.1), both activities of NaCl, i.e., reversal of Glc-6-P-induced solubilization, and resolubilization, can be observed. In contrast to the relatively nonchao- tropic KCl, KSCN was effective at solubi- lizing the enzyme at both pH 6.0 and 8.5 (Fig. 103). Although both salts solubilized the hexokinase at pH 6.0, KSCN was ap— preciably more effective than was a com- parable concentration of KCl, clearly indi- cating that this was not simply an ionic strength effect. DISCUSSION Clearly, neutral salts have two distinct effects on the interaction of hexokinase with the mitochondrial membrane. At low concentrations (<~0.02 M), salts serve to enhance this interaction while elevated concentrations have a disruptive effect. In the present discussion, we will first con- sider these effects separately and then sug- gest a model for the hexokinase-mem- brane interaction which accounts for these antagonistic effects of neutral salts. We propose that the enhancement of hexokinase—membrane interactions ob- served at low salt concentrations is due to the shielding of negative charges on the membrane and/or enzyme, thereby de- creasing repulsive electrostatic forces be- tween the two. This enhancement cannot be explained simply as electrostatic screening in the Debye-Hiickel sense since the effectiveness of various salts is not solely a function of ionic strength but, rather, varies with the salt used. Clearly this effectiveness depends directly on the charge density of the unhydrated cation (Fig. 6). This kind of relationship has also been observed for the binding of cations to pyrophosphate, AMP, ADP, and ATP (21). Ross and Scruggs (22, 23) found a similar order of effectiveness of these cations in promoting the double-helix formation in DNA and attributed their results to a shielding of negatively charged phosphate groups on the DNA by the cations, thereby diminishing repulsive interaction between the DNA strands. Ross and Scruggs also suggest that water of hydration must be released from the cations during binding. Otherwise Na+ and K“( would be expected to be more effective than Lit which has a 163 NEUTRAL SALTS AND HEXOKINASE BINDING hydrated radius larger than that of hy- drated Na+ and K+ ions. By analogous reasoning, it would appear that the ionic interactions leading to rebinding of hexo- kinase involve the unhydrated cation. Rot- tem et al. (24) demonstrated that mem- brane reconstitution is facilitated by low pH and the presence of cations (monova- lent as well as divalent), while Okada et al. (25) found that divalent cations pro- moted adhesion of cells to protein-coated surfaces. All of these investigators offer a similar explanation for these cation ef- fects, nanmly, that the cation serves to diminish repulsive electrostatic interac- tions and allows association to occur. Thus, there is ample precedent for the role that we have proposed for cations in en- hancing the interaction of hexokinase with the mitochondrial membrane. Elevated concentrations of neutral salts serve to solubilize the enzyme. Ifthis were purely an ionic strength effect, one would expect comparable effectiveness of various salts compared on an ionic strength basis. Clearly this is not the case, since the pres- ent results show the solubilization to de- pend greatly on the nature of the anion, with salts of chaotropic anions being much more effective than salts of relatively non- chaotropic anions. The net effect of a salt on the hexokinase-membrane interactions is thus a composite one, with charge screening effects (primarily due to the cat- ion) resulting in enhanced interactions while counteracting chaotropic effects seek to weaken the interaction. The charge- shielding action (observed at low ionic strengths) of the cations apparently out- weighs their chaotropic effect: The effec- tiveness of the chloride salts at promoting membrane-hexokinase interactions (Fig. 6) increases in the order Cs+ < Rbt < K+ 2 Na+ < Lit, which is also the order of in- creasing chaotropicity of the ions (16). Thus, if the chaotropic effects of the cat- ions were predominant, their effectiveness in promoting hexokinase—membrane in- teractions should be the reverse of that observed. We interpret these results to mean that, at low ionic strengths, chao- tropic properties of the salts are of mini- mal influence and the major effect of the salt is to enhance hexokinase-membrane 291 interactions by the chargevscreening ef- fects of the cations. At higher ionic strengths, chaotropic effects primarily due to the anions became predominant, result- ing in disruption of the association of en- zyme with membrane. Neutral salts also have a marked effect on the activity of the enzyme (Fig. 9), pre- sumably reflecting alterations in the ter- tiary structure of the protein (20, 26). Is this alteration per se necessary for solubil- ization? We believe the answer to be "No,” based on the following observations. (a) K01 and NaCl are relatively ineffec- tive at inactivating the enzyme at any pH in the range of 6.0-8.5 (Fig. 9) but, at comparable concentrations, they are quite effective in solubilizing the enzyme, at least in the pH range of ~6-7 (Figs. 8 and 10). (b) At appropriate concentrations, neu- tral salts virtually instantaneously inacti- vate the enzyme; in contrast, we have found solubilization to be noticeably time dependent, being complete within 15-20 min at 25°C under the conditions used in the present study. (c) Although the concentration depend- ence for solubilization (Fig. 8) and inacti- vation (Fig. 9) have some similarity, there are also notable differences; e.g., compare the relative effectiveness of NaCl at solu- bilizing (Fig. 8) or inhibiting (Fig. 9) the enzyme. We conclude that a direct effect of neutral salts on the conformation of the enzyme, as reflected by catalytic activity, is not sufficient to explain the observed solubilization. ' At pH 6, both relatively nonchaotropic salts such as KCl as well as chaotropes such as KSCN are highly effective at solu- bilizing the enzyme (Fig. 10). Further- more, this solubilizing action is apparent at salt concentrations well below those typically required to have pronounced ef- fects on water structure and to bring about extensive disruption of hydrophobic inter- actions (16, 19). These observations lead us to suggest that, at pH 6, the solubilization is primarily due to the effectiveness of these salts at disrupting nonhydrophobic interactions. Certainly the most obvious possibility would be attractive electro- static interactions between oppositely l 64 292 charged groups on enzyme and membrane, as already suggested by Teichgraber and Biesold (6). But it was also observed that, even at pH 6.0, noticeable differences could be observed between chaotropic (KSCN) and relatively nonchaotropic (KCl) salts (Fig. 10). Von Hippel and his colleagues (16, 27—30) have suggested that the effectiveness of various salts in bring- ing about conformational change in pro- teins may be due to interaction of the salts with the peptide bond dipole. This effec- tiveness varies with the position of the salt in the Hofmeister series, i.e. , its chaotrop- icity, reflecting the differences with which the various salts can penetrate hydropho- bic regions of the protein and thereby ex- tend their range of action. Hexokinase must exist in rather intimate contact with the outer mitochondrial membrane; e.g. , it is not readily removed by washing with 0.25 M sucrose or dilute salt solution, and it is largely protected from inhibition by antihexokinase serum (31); given the high lipid content of the outer mitochondrial membrane (32), it seems likely that the enzyme must be semi-embedded in a rather hydrophobic milieu. Extrapolating from the thoughts of Von Hippel and his co-workers (16, 27—30), we suggest that the relatively greater efficacy of chaotropic salts reflects their increased proficiency in penetrating through hydrophobic regions to the interacting charges on membrane and enzyme. At higher pH, e.g., 8.5, relatively non- chaotropic salts, e.g., KCl, are ineffective at bringing about solubilization whereas chaotropic salts retain their effectiveness. It is, of course, possible that hydrophobic interactions (which would not be disrupted by nonchaotropes) could become dominant at elevated pH. However, we feel that this is unlikely. First, there is no inherent rea- son to expect hydrophobic interactions to be markedly pH dependent; since they did not appear to play a prominent role at pH 6, there is no a priori reason to expect them to predominate at higher pH. Sec- ond, as at pH 6, the concentrations of chao- tropic salts that bring about solubilization are considerably lower than those gener- ally found to be effective at disrupting hy- FELGNER AND WILSON drophobic interactions (16, 19). As an al- ternative, we suggest that electrostatic forces remain the prime attractive interac- tion, but at higher pH, the interacting ionic groups become less accessible to dis- ruptive exogenous ions. Decreased accessi- bility could be a consequence of more inti- mate contact between interacting groups on enzyme and membrane resulting from pH-induced changes in membrane and/or enzyme structure. By virtue of their greater ability to penetrate through hydro- phobic regions and then exert disruptive influences on electrostatic interactions, chaotropes would retain their capability for bringing about solubilization whereas the effectiveness of nonchaotropes would be greatly diminished. The thoughts ex- pressed in the above discussion are pre- sented diagrammatically in Fig. 11. Finally, we would like to comment briefly on the negative enthalpy seen for solubilization in the presence of KSCN (Table II). In the absence of salts, solubili- zation increases with increase in tempera- ture (2, and unpublished observations), in- dicating a positive enthalpy. In the ab- sence of salt, the interactions between en- zyme and membrane are pictured as being primarily electrostatic in nature and oc- curring in a generally hydrophobic envi- ronment; under these conditions, we would anticipate a negative enthalpy for formation of the ionic interactions (27, 33) and conversely, a positive enthalpy for their disruption. Thus, in the absence of salts, A{Isolubilization = ”AHME > 0, where AH M; is the enthalpy for formation of the attractive electrostatic interactions ‘ between enzyme and membrane. Solubili- zation in the presence of salts, however, is considered to be the result of interactions of exogenous salts with the charged groups on enzyme and membrane, with conse- quent rupture of the attractive interaction between the latter species. Thus, in the presence of salts: AHsolubtllzation = ’AHME + AHsssu where AH s; 3,, is the enthalpy for forma- tion of new electrostatic interactions be- 165 NEUTRAL SALTS AND HEXOKINASE BINDING pit 6.0 '0! 9.5 F10. 11. A model for interaction of hexokinase with the outer mitochondrial membrane. The forces between enzyme and membrane are considered to be primarily electrostatic in nature. Repulsive interac- tions, which are more marked at pH 8.5, can be masked by cations. Neutral salts can disrupt the attractive interactions if the salts can penetrate to the region between enzyme and membrane. At pH 6.0, the association between the interactants is con- sidered relatively "loose” in the sense that both chaotropic and nonchaotropic salts can readily enter the region of attractive electrostatic interaction be- tween the membrane and enzyme; thus, chaotropes would possess only slight advantage in gaining ac- cess to this region via hydrophobic routes. At pH 8.5, the region of electrostatic interaction is considered to be much less accessible to nonchaotropes, while chaotropes can still gain access by penetration of hydrophobic regions. The representation of the binding site as an invagination into the membrane is diagrammatic and should not necessarily be taken to indicate a penetration of the hexokinase molecule (or portion thereof) into the hydrophobic interior of the membrane. Nor should the emphasis on electro- static interactions be taken to exclude the possible involvement of hydrophobic interactions; the pres- ent results do not provide any direct evidence for the latter, however, and suggest that electrostatic inter- actions are the predominant if not the exclusive forces involved. tween the added salts and groups on the enzyme and membrane; AHSESM is ex- pected to be negative (27, 33), and thus the AH for solubilization in the presence of salt is predicted to be more negative than that for solubilization in the absence of salt, as observed. ACKNOWLEDGMENTS We are pleased to acknowledge the technical as- sistance of Suzanne Murrmann and the financial support of NIH Grant NS 09910. REFERENCES 1. Ross, 1. A., AND WAruas, J. V. B. (1967)J. Biol. Chem. 242. 1635-1645. ‘0 10. ll. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 26. 293 WnsON, J. E. (1968) J. Biol. Chem. 243, 3640- 3647. Kxorr, E. 8., AND WDsON J. E. (1970) Biochem. Biophys. Res. Commun. 38. 74—79. Cnov, A. C., AND WnsON, J. E. (1972) Arch. Biochem. Biophys. 151, 48-55. FxDGNn, P. L., AND WnsON, J. E. (1976) Bio- chem. Biophys. Res. Commun. 68, 592-597. 'I‘xrcuaaAsxa, P., AND Brasom, D. (1968) J. Neurochem. 15, 979-989. CxAva, P. A., GOLDBLA‘I‘T, P. J., AND BAsronD, R. E. (1969) Biochemistry 8, 3525-3532. . Dc Roasarrs, E., Pstuaamo Dc IaAun, A., Romucvzz DI Loam AaNArz, G., AND SAL- cANrcorr, L. (1962) J. Neurochem., 9. 23-35. . WILSON, J. E. (1972) Arch. Biochem. Biophys. 150, 96-104. WILersON, G. ‘N. (1961) Biochem. J. 80, 324- 332. EmsON, W. R., Luscx, J. D., AND Fnouu, H. J. (1975).]. Biol. Chem. 250. 1864-1871. TIstMANN, H., AND BORN, J. (1959) Z. Natur- forsch. MB, 477. ELLISON, W. R., Lvscx, J. D.. AND Faoma, H. J. (1974) Biochem. Biophys. Res. Commun. 57, 1214-1220. WILSON, J. E. (1973) Arch. Biochem. Biophys. 159, 543-549. Rsnm, V. D., AND KsNans, U. W. (1975) Biochemistry 14. 4704-4712. VON HIPPEL, P. H., AND Scnuncn, T. (1969) in Structure and Stability of Biological and Mac- romolecules (Timasheff, S. N., and Fasman, G. D., eds.). pp. 417-251, Marcel Dekker, New York. Kosow, D. P., Osxr, F. A., WAnus, J. V. B., AND Ross. I. A. (1973) Arch. Biochem. Biophys. 157. 114-124. WILSON, J. E. (1968) Technical Report, EATR 4204, US. Army Medical Research Labora- tory, Edgewood Arsenal, Md. HArsn, Y., AND HANsrer, W. G. (1969) Proc. Nat. Acad. Sci. USA 62, 1129-1136. WAnasN, J. C., SrowruNc, L., AND Mamas, M. F. (1966) J. Biol Chem. 241. 309-316. Sustraa, C. H. (1974)in Metal Ions in Biological Systems (Sigel, H., ed.). Vol. 3, pp. 201-251, Marcel Dekker, New York. Ross, P. D., AND Scavccs, R. L. (1964)Biopoly- mers 2, 79-89. . Ross, P. D., AND Scnvacs, R. L. (1964)Biopoly- mers 2, 231-236. . Roman, S., Stein, 0., AND WIN, S. (1968) Arch. Biochem. Biophys. 125. 46-56. . OxADA, T. S., TAxsicm, M., YAsvDA, R., AND VIDA. M. J. (1974) Advan. Biophys. 6, 157- 181. WAanaN, J. C., AND CxxArUu, S. G. (1966)Bio- chemistry 5. 1702-1707. 166 294 FELGNER AND WILSON 27. VON I-Irrrst,,P. H., Prncous, V., ScrrAcx, L., Mechanochem. Cell Motil. 2, 127-138. AN” K4330“ L- (1973) 13““me 12. 31. CaAva, P. A., an sum», 11. s. (1969) Bio- 1255'12“ chemistry 8, 3520—3525. ~ 28' ahmtrsg'gizgngiwh P. H. (1973) 32. PAasONs, D. E., AND YANO, Y. (1967)Biochim. 29. HAMABATA, A., CHANG, S., AND VON HML, P. Emphy" Am 135' 362-364: H. (1973) Biochemistry 12. 1271-1282. 33. KAuznANN, W. (1959)Advan.Protein Chem. 14. 30 VON HIPPEL, P. H., AND HAWATA, A. (1973).]. 1-63. "'Tll'liifluijjiifillflilflifliifliylil”