IIIIIIIIIIII IIIIIIIIIIIIIII IIIIIII II III 3 1293 01094 8747 ‘ ""‘“'“ "‘ j»: 1e, . i ‘ \./ ; “v 1' ‘I‘ .{l [I JAI M. IV; I i 1 I ..., - - ‘I I Ursa--1“ I I ‘ ‘3' I I' ",1 . I \. 2123.4" * - ,e $75,395” This is to certify that the dissertation entitled PHYSICAL PROPERTIES or GRAN NEGATIVE MEMBRANES AND LIPOPOLYSACCHARIDE presented by RICHARD THOMAS COUGHLIN has been accepted towards fulfillment of the requirements for PA. D degree in B ; OIPA/YS IQS Date 67//:) L/g7// 0-12771 MS U is an Affirmatiw Action/Equal Opportunity Institution MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will Hbe charged if book is i returned after the date stamped beIow. PHYSICAL PROPERTIES OF GRAN NEGATIVE MEMBRANES AND LIPOPOLYSACCHARIDE By Richard Thomas Coughlin A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Biophysics 1982 ‘1‘. i.':l .In A r5: 4“ ":n J t ‘II 1‘ I‘ New! . ‘9 i Is II; 5 P 6.. H51." D IIA 3 II" " ha I Q U ‘14 v “ | A‘i I‘IICF “~:i. . 1" II . ‘ a 'd.‘;: vu" (ii/fjfijYD ABSTRACT PHYSICAL PROPERTIES OF GRAM NEGATIVE MEMBRANES AND LIPOPOLYSACCHARIDES By Richard Thomas Coughlin The isolated membranes and lipopolysaccharide (LPS) of Escherichia 3211 were analyzed using the electron spin probes 4-(dodecyl-dimethyl ammoniunl-l-oxyl-2,2,6,6-tetramethyl piperidine bromide, CAle, and S-doxyl stearate. The position of the head group probe CAT12 in the outer membrane was determined to be on the outer monolayer in LPS-protein domains. The results indicate that both the inner and outer monolayer of the outer membrane experience the same cooperative membrane phase transition despite strong transmembrane asymmetry. Using inductively coupled plasma emmission spectroscopy, the outer membrane was found to be enriched in divalent cations when compared to the cytoplasmic membrane. The higher cation affinity was attributed to the presence of anionic LPS in the outer membrane. Elemental analysis of LPS revealed the presence of not only high levels of Mg and Ca but significant levels of Fe, Al, and Zn as well. A purified sodium salt of LPS isolated from both a rough and deep P0"9",§-‘gg11 was prepared. It was relatively free of contaminating cations and had significantly greater head group and acyl chain mobilities when compared to the starting material. Elemental analysis suggests that the sodium salt of the rough LPS was not CONPIEPEIY charge neutralized. The binding affinity and capacity of LPS from a variety of "(w Iui~VI .' '4 5.2.": 4 cations was shown to be quite high yet variable depending upon the cation used. The information obtained with LPS from Escherichia coli was then applied to Yersinia pggtlg, Chromatia vinosum, and Thiocapsa roseopersicina LPS with attention to the potential biological significance of the observed physical phenomenon. ’Pv‘ ”I ‘P ACKNOWLEDGEMENTS I would like to express my sincere appreciation to my adviser, Dr. Estelle McGroarty, and to the members of my dissertation committee, Drs. Alfred Haug, Ashraf El-Bayoumi, and Harold Sadoff. Their helpful (iirection and friendship was invaluable in the progress and review of this work. I acknowledge the expert technical assistance of Steven Tonsonger in Use of operating the inductively coupled plasma emission spectrometer. ‘tiiis instrument was obtained through Dr. H. Braselton of the Department of Pharmacology and Toxicology in cooperation with the Animal Health Diagnostic Laboratory of Michigan State University. I would also like to acknowledge Dr. Karen Baker, coordinator of the Center for Electron Optics at Michigan State University, for her assistance in the preparation of the electron micrographs. Finally, I would like to express my thanks to my mother, father, and fiance" , Carol, for their encouragement. TABLE OF List of Tables . . . . . . . . . LiSt Of F1 gures. O O O O O O O 0 Chapter I. II. III. 12V. BACKGROUND....... Gram Negative Membranes. Biochemistry of LPS. . . References . . . . . . . A PROGRAM FOR TRANSITION DATA 0 O O O O 0 O O 0 Abstract . . . Introduction . Program. . . . Discussion . Summary. . . References . A CATIONIC ELECTRON SPIN CATION INTERACTIONS WITH Summary. . . . . . . . . Introduction . . . . . . Materials and Methods. . Results and Discussion . References . . . . . . . CONTENTS POINT ANALYSIS OF RESONANCE PROBE USED TO EXPERIMENTAL LIPOPOLYSACCHARIDE. QUANTITATION OF METAL CATIONS BOUND TO MEMBRANES EXTRACTED LIPOPOLYSACCHARIDE OF ESCHERICHIA COLI Abstract . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . Materials and Methods. . . . . . . . . . . . . . Growth Conditions and Procedures for Membrane LPS Isolation. . . . . . . . . . . . . . . Assays. . . . . . . . . . . . . . . . . . . . Sample Ashing Procedures for Elemental Analysis . E] mentaI AnaIYSiSO O O O O O O O O O O O O O O 0 RESUItSo O O O O O O O O O O O O O O O O 0 O 0 O O O Membrane Ionic Composition. . . . . . . . . . . . LPS Ionic Composition . . . . . . . . . . . . . . Effects of Growth Media on LPS Ionic Composition. iii ANALYZE AND and m. 31' S: Refi ELE' DOM Sun: Int Mat Res Dis Ref ' IQ“ Chapter IV. V. VI. VII. VIII. LPS Salt Preparation and Characterization . . . Ionic Content of LPS from a Heptoseless Mutant. DISCUSSIO" O O O O O O O O O O O O O O O O O O O O Raferences O O O O O O O O O O O O O O O I O O O 0 PHYSICAL PROPERTIES OF DEFINED LIPOPOLYSACCHARIDE SALTS. O O O O O O I O O O O O O O O O O O O O I 0 Abstract 0 O O O O O O O O O O O O O O O O O O O 0 Introduction . . . . . . . . . . . . . . . . . . Materials and Methods. . . . . . . . . . . RESU] ts. O O O O O O O O O O O O O O O O O O O ESR Probing of DZI Defined LPS Salts. . Scatchard Analysis of CAT1 Binding to NaLPS. . Magnesium Titration of NaL S from Strain DZI . Electron Micrographs of D21 LPS . . . . . . . . ESR Probing of a Heptoseless Mutant of Strain DZ D1 SCUSSION O O O O O O O O O O 0 0 O O O O O O O O 0 References 0 O O O O O O O O O O O O O O O O O O O O i ELECTRON SPIN RESONANCE PROBING OF LIPOPOLYSACCHARIDE DOMAINS IN THE OUTER MEMBRANE OF ESCHERICHIA COLI. sunmary. O O O O O O I O O O O O O O O O O O O 0 Introduction . . . . . . . . . . . . . Materials and Methods. . . . . . . . . Cell Growth and Membrane Isolation. Lipid Isolation . . . . . . . . . Porin-LPS-Peptidoglycan Isolation Assays. O O O O O O O O O O O O 0 Spin Labeling . . . . . . . . . ReSUI ts. O O O O O O O O O O 0 0 O Membrane Spin Labeling. . . . . Lipid Spin Labeling . . . . . . Porin-LPS-Mucopeptide Labeling. Discussion . . . . . . . . . . . . References . . . . . . . . . . . . IONIC BRIDGING HITHIN LIPOPOLYSACCHARIDE AGGREGATES. AN ELECTRON SPIN RESONANCE, pH TITRATION STUDY . . . . AbStraCt O I O O O O O O O O O 0 O O O O O O O O O O O IHtrOdUCtion O O O O O O O O O O I O O 0 O 0 O I O 0 0 Results and Discussion . . . . . . . . . . . . . . . . References 0 O 0 O O O O O O O O O O O O O O O O O O 0 PHYSICAL PROPERTIES OF YERSINIA PESTIS LIPOPOLYSACCHARIDE MAY DETERMINE VIRULENCE. . . . Introduction . . . . . . Materials and Methods. . Results and Discussion . References . . . . . . . iv 105 105 106 107 107 107 108 108 108 109 109 110 111 111 114 122 122 123 127 132 138 138 139 140 143 turn . - * fer IX. CHARAC CHRCV} 155E Intr0< Nater‘ Resul' Refer: inedix 1. USE OF HESAT F Intrcd: Hateri Discus' Refere THE NC; BY HEN STUDY Abstra Introd Nateri Pre Ele Oat Result Con Eff Str Discus Conclu RBIQre LIGANB NaI’K+ BRAIN Summer Introd HEIDOQ RESult ESR ESR SR DISCug RefEre Chapter IX. CHARACTERIZATION OF THE PHYSICAL PROPERTIES OF CHROMATIACEAE LIPOPOLYSACCHARIDE . . . . . . . fitraCt O O O O O O 0 Introduction . . . . . Materials and Methods. Results and Discussion References . . . . . . Appendix USE OF ELECTRON SPIN RESONANCE TO STUDY BACILLUS ,A. B. C. Smmw........ Introduction. . . . . Materials and Methods Discussion. . . . . . References. . . . . . MEGATERIUM SPORE MEMBRANES. . . . . . . . . . . . THE MODIFICATION OF HUMAN BY MEMBRANE STABILIZERS: STUDY 0 O O O 0 O O O 0 Abstract. . . . . . . . Introduction. . . . . . Materials and Methods . Preparation of Cells Electron-Spin Resonance Spectroscopy Data Analysis. . . . . . . . . . . . Results . . . . . . . . . . . . . . . . Control Erythrocytes . . . . . . . . . Effects of Membrane Stabilizing Agents Structure. . . . . . . . . . . . . . . . . DISCUSSIOH. O O O 0 O O O O O 0 O O O O O O 0 Conclusions . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . ERYTHROCYTE MEMBRANE STRUCTURE AN ELECTRON-SPIN RESONANCE O O O O O C O O O O O O C O O O O O O O O O O O O O O O O C O O O C O n e O 3 3 00.00.0000... 0.00%.000000. 0.003000000.0 0000 LIGAND EFFECTS ON MEMBRANE LIPIDS ASSOCIATED WITH Na+,K+-ATPase: COMPARATIVE SPIN PROBE STUDIES NITH RAT BRAIN AND HEART ENZYME PREPARATIONS . . . . . smary O O O O O O O O O O O O O O O O O IntrOdUCtion. O O O O O O O O O O O O O O Methads O O O O O O O O O I O O O O O O 0 RESUItS O O O O O O O O O I O O O I O O O ESR Spectra in the Absence of Ligands. ESR Spectra in the Presence of Ligands ESR Spectra in the Presence of Ouabain Discussion. . . . . . . . . . . . . . . . RaferenceSO O O O O O O O O O 0 O O O 0 O Page 149 149 150 152 152 156 162 162 162 163 164 166 171 171 172 174 174 174 176 176 176 177 178 181 181 188 189 191 193 193 196 197 198 203 1. 2. 1. 2. 3. 4. 5. 1. 1. 2. 3. 1. LIST OF TABLES Chapter III Elemental Analysis of LPS Samples Expressed as Molar Ratio 0 O O O O O O C O O O O O O O O O O O O O O O O O O O 0 Displacement of CAT12 from edLPS by 50 uM Salt. . . . . . . . Chapter IV Elemental Composition of Membranes from E. coli Strain H1485F' Grown in Either Nutrient or Minimal Media . . . . . . Elemental Composition of LPS from E. coli Strain N1485F' Grown in Either Nutrient or Minimal Medium. . . . . . Elemental Composition of LPS and of Porin-LPS-Peptidoglycan Complexes from E. coli Strain DZI Grown in Nutrient Broth . . Elemental Composition of Specific Salts of LPS from-E. coli Strain D21 Grown in Nutrient Media. . . . . . . . . . . . . . Elemental Composition of LPS from E. coli Strain D21f2 Grown 1n "Utrient Media 0 O I O O O O O O O O O O O O O O O 0 O O 0 Chapter VII Hyperfine Splitting of CAT12 in LPS Salts at 37°C . . . . . . Chapter VIII Elemental Analysis of Yersinia pggplp Native LPS. . . . . . . Elemental Analysis of Yersinia pggplg Sodium LPS. . . . . . . Low Field Line Width of CAle in NaLPS of Yersinia pggpig . . Chapter IX Elemental Analysis of Native Lipopolysaccharide from Several Gram Negative Bacteria. . . . . . . . . . . . . . . . . . . . vi 46 69 7O 71 72 73 133 145 146 142 158 Table Page Appendix C I. Background Level of Ions in Samples . . . . . . . . . . . . . 194 vii TranS‘ (m IranS‘ inter: Transf Partit temper Scatci DISpla functi concer Struct rough) Thete Darame LPS fr Thete Pardme from E TemPEr (a), . and TR. SCatghé LPS (de AI 37°: Figure 1. 2. 3. 1. 2. 3. 1. 2. 3. 4. LIST OF FIGURES Chapter II Transition point analysis of data generated from two lines (0's 0.25) intersecting at x a 27.0 . . . . . . . . . . . . Transition point analysis of six lines (0'3 0.25) intersecting at x 8 11.0, 20,0, 25.0, 34.0 and 43.0 . . . . Transition point analysis of human erythroycte membranes. . Chapter III Partitioning of CAT12 into nLPS as a function of increasing temerature O O O O O O O O O O O I O I O O O O O O O I O O Scatchard analysis of CAle binding to edLPS. . . . . . . . Displacement o; CAT1 fro% edLPS was measured at 37°C as a function of CaL' 01,MgI +(A), and spermidine (I) concentrations. 0 O O O O O O O O O O O O O O O O O I O O 0 Chapter V Structure of E. coli K12 strain 021 (rough) and 021f2 (deep rough) LPS. 0 O O O O O O O O O O O O O O O O O O O O O I O The temperature dependence of the hyperfine splitting parameter, 2T", of CAT12 bound to define salts of LPS from E. coli strain 021 . . . . . . . . . . . . . . . . The temperature dependence of the hyperfine splitting parameter, 2T". of SDS bound to defined salts of LPS from E. coli strain 021 . . . . . . . . . . . . . . . . . . Tem erature dependence of SDS order parameter in native (.I, electrodialyzed (I), sodiun (A), magnesiun (A), and TRIS (a) LPS 0 O O O O O O O O O O I O O O O O O O O O Scatchard analysis of CAT binding to electrodialyzed LPS (dashed line) and NaLPS (o) from E. co_l_i K12 grown at 37°C 0 O O O O O O O O O O O O O O O O O O O O O O O O 0 viii 48 49 50 90 91 92 93 94 Fieure '1 1‘0 The Dr: 1: Out l. The Dar 9"01 Figure 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 1. 2. 3. Chapter V Titration of NaLPS from E. coli strain 021 with MgClg . . . 95 Electron micrograph of the sodium salt of LPS isolated from E. coli strain 021, stained with sodium phosphotung- State, pii 7.0 O O O O O O O O O O O O O O O O O O O O 0 O O 96 Electron micrograph of the TRIS salt of LPS isolated from E. coli strain 021, stained with sodium phosphotungstate, 3H 0 O O O O O O O O O O O O O O O O O C O O O O O O O O O 9 7 Electron micrograph of NaLPS with CaClz added (Ca/LPS, 1) isolated from E. coli strain 021. . . . . . . . . . . . . . 98 Electron micrograph of NaLPS with MgClz added isolated from E. coli strain 021 (Mg/LPS, 4) stained with sodium phosphotungstate, pH 7.0. . . . . . . . . . . . . . . . . . 99 Electron micrograph of NaLPS with FeCl3 added (Fe/LPS, 1) isolated from E. coli strain 021, stained with sodium phosphatunQState, p” 2 .0. O O O O O O O O O O O O O O O O O 100 Temperature dependence of the hyperfine splitting parameter, 2T... of CAle in the sodium LPS salts of 021 (O) and 021f2 (.)O O O O O O O O O O O O O O O O O O O O O O O 101 Temperature dependence of the hyperfine splitting parameter, 2T... of 505 in the sodiun LPS salt of 021 (O) and 021f2 (.)O O O O O O O O O O O O O O O O O O O 0 O O O O O 102 Temperature dependence of SDS order parameter, S, in the sodium LPS salts of 021 (O) and 021f2 (O) . . . . . . . . 103 Electron micrograph of nLPS from E. coli strain 021, stained with sodium phosphotungstate, pH 7.0. . . . . . . . . . . . 104 Electron micrograph of edLPS from E. coli strain 021, stained with sodium phosphotungstate, pH 7.0. . . . . . . . . . . . 104 Chapter VI The chemical structure of the two electron spin resonance prObes CATIZ and SDSO O O C O O O O O O O O O O O O O O O O 116 Electron spin resonance spectra of CAT; bound to intact outer membranes from cells grown at 37 g. . . . . . . . . . 117 The temperature dependence of the hyperfine splitting parameter, 2T", of CAT12 bound to outer membranes (0) and cytoplasmic membranes (0) from E. coli N1485F‘ grown at 37°C 0 O O O O O O O O O O O O O O O O O O O O 0 O 118 ix 4. Sea ele the par of par PEP gr01 1. Chen obta T 1" Egg; caic SUDQ cal' and calc Figure 4. 5. 6. 1. 2. 3. 4. 1. 2. 1. 2. Chapter VI Scatchard analysis of CAT12 binding to electrodialyzed LPS (I) and to phospholipid (0) from the outer membrane of.§. coli H1485F‘ grown at 37°C . . . . The temperature dependence of the hyperfine splitting parameter, 2T", and the order parameter, 5, (Insert) of CAT12 (0) and 505 (I) bound to electrodialyzed LPS . . The temperature dependence of the hyperfine splitting parameter, 2T", of CAle bound to porin-LPS- peptidoglycan complexes isolated from E. coli N1485F' grown at 37°C 0 O O O O O O O O O O O O O O O O O O O O O 0 Chapter VII The pH dependence of the rotational correlation time (I!) of CAle bound to NaLPS from E, coli strain 021 . . . . . . The pH dependence of the hyperfine splitting parameter, 2T... (0); and low field line width,f,' , (O), of CAT12 bound to NaLPS from E, coli strain 021f2. . . . . . . 31P NMR spectra of the NaLPS of E. coli strain 021 taken at 81 MHZ . O . . C . . C C O C C . O O O O C C . C O 31P NMR spectra of the NaLPS of.§. coli strain 021f2 taken at 81 MHZ . . . O . . C . . . . . . . . . . . C . . . Chapter VIII Chemical structure of Yersinia pestis LPS based on data obtained from reference 8 . . . . . . . . . . . . . . . . . Temperature dependence of the hyperfine splitting parameter, 2T», of CAT12 in NaLPS from 37°C grown Yersinia estis cal' with (A) and without (A) supplemented ca ciun; and cal+ cells with ac?) and without (it) supplemented calcium; and 26°C grown Yersinia estis cal‘ with (:3) and without (II) supplemented caiciun; and cal“ cells with (o) and without (0) supplemented ca1c1m. O O O O O O O O O O C O O O O O 0 O I O O O O O 0 Chapter IX Pyran ring containing the gnggzgx sequence . . . . . . . . Temperature dependence of the hyperfine splitting parameter, 2T", of CAT1g in Chromatia vinosum native LPS (C?) S and sadim L (.5 O O O O O O O O O O O O O O O O O O O O 119 120 121 134 135 136 137 147 148 153 159 5-dc in i The ine abse Pres Figure 3. 4. 1. 2. 1. 2. 3. 1. 2. 3. Chapter IX Temperature dependence of the hyperfine splitting parameter, 2T", of CAT 2 in Thiocapsa roseopersicina native LPS (0) ané SOdim‘ LPS .)0 O O O O O O O O O O O O O O 0 Temperature dependence of the hyperfine splitting parameter, 2T», of 505 in Chromatia vinosum ((0) and Thiocapsa roseoper51C1 "a to S "LPS. 0 O O O O O O O O O O O O O O O 0 Appendix A Temperature dependence on 2T“ in dormant spore mranes O O O O O O O O O O O O O O O O O O O 0 O O O O 0 Effect Of L-pr011ne O O O I O O O O O O O O O O O O O O O 0 Appendix B Electron-spin resonance spectra of human erythrocytes labeled with 5-doxyl stearate . . . . . . . . . . . . . . . Hyperfine splitting parameter, 2T“ (Gauss), as a function of temperature in erythrocytes labeled with 5-doxyl stearate in the absence of perturbants (A), and in the presence of 5 x 10‘ M propranolol (B) . . . . . . . The temperature dependence of the order parameter, S, in erythrocytes labeled with 5-doxyl stearate in the absence of perturbants (closed circles) and in the presence of 3 x 10"4 M diazepam (open circles). . . . . . . Appendix C Typical ESR spectrum of rat brain (Na+,K+)-ATPase preparations at 37°C. . . . . . . . . . . . . . . . . . . . Temperature dependence of the order parameter of SDS bound to A) rat brain enzyme in the presence (0), and absence (I) of 15 mM KCl; B) rat heart enzmne in the presence (0) and absence (-) Of 15 M KC]. 0 O O O O O O O O O O O O O O O 0 Temperature dependence of the order parameter of SDS bound to A) rat brain (I) or heart (0) enzyme in 5 mM MgCl2, 100 mM NaCl, 5 mM Tris-ATP and either 1 uM (rat brain) or 100 (IN (rat heart) ouabain; B) rat brain (I) or heart (0) enzyme in 5 mM MgClz, 100 mM NaCl, 5 mM Tris-ATP, 15 mM KCl and either 1 uM (rat brain) or 100 uM (rat heart) ouabain. . . . . . . . . . . . . . . . . . . . . . . xi 160 161 168 169 184 185 186 206 208 210 figure 4. Temper to A) 100 mi 100 ul enzyme KCl ar ouabai Temper to A) 100 mi (0) er and 1C ESL"? P_ag_e_ Appendix C 4. Temperature dependence of the order parameter of SDS bound to A) rat brain (I) or heart (0) enzyme in 5 nM MgClz, 100 mM NaCl, 5 mM Tris-ATP and either 1 uM (rat brain) or 100 uM (rat heart) ouabain; B) rat brain GI) or heart (0) enzyme in 5 mM MgClp, 100 mM NaCl, 5 mM Tris-ATP, 15 mM KCl and either 1 uM (rat brain) or 100 "M (rat heart) ouabain O O O O O O O O O O O O O O O O O O O O O O O O O O 212 5. Temperature dependence of the order parameter of SDS bound to A) dog brain (I) or heart (0) enzyme in 5 11M MgClz, 100 nM NaCl, and 5 11M Tris-ATP; B) dog brain (I) or heart (0) enzyme in 5 mM MgClp, 100 mM NaCl, 5 mM Tris-ATP, aTIdIOOHMOlJabaIn.....................214 xii Gran Neea Elec we disti enzymes f tranSport dimension Covalent) lenbrane SIIe foy- - InIEQr-ity in )0»; H; Mptidog‘,‘ The . "31h othe :mOSIIll IOU? pTIns gratEl'ns I the Out er 5991') refer CHAPTER I BACKGROUND Gram Negative Membranes Electron micrographs of thin sections of gram-negative bacteria show two distinct membranes. The inner or cytoplasmic membrane contains the enzymes for oxidative phosphorylation, electron transport, and active transport. The characteristic shape of the cell is maintained by a two dimensional sheet of peptidoglycan distal to the cytoplasmic membrane. Covalently linked to the peptidoglycan and extending into the outer membrane is the Braun lipoprotein. This protein serves as an attachment site for the outer membrane and is crucial for outer membrane structural integrity. In lipoprotein deficient mutants of Escherichia p211, grown in low Mg media, the association of tfiZ outer membrane for the peptidoglycan is weakened (I). The outer membrane is a highly specialized structure. In comparison with other membranes (e.g., the cytoplasmic membrane) its protein composition is relatively simple. In most enteric bacteria there are four principle proteins, excluding the lipoprotein. Two of these proteins 1a and 1b (Schnaitman nomenclature) (2) form trimeric units in the outer membrane and act as pores (3). The pore proteins (porins) have been referred to as "molecular sieves“ because they allow passive nonspecif hydrophl) Set) distribut monolayer bacteria membrane exception be carri e detergent Up to thr 59 predom reconstit however, 1 Wins. '35 fon‘ne: CDHSIdnt 1 CWDIEX. HDEr ”‘99 diff. Quite Imp hidrophob. eff)“Cy . bacteria. mane ( Q'D‘GIySaCC My “ch nonspecific diffusion of low molecular weight (generally <600 daltons) hydrophilic solutes through the outer membrane. Both proteins and lipids of the outer membrane are asymmetrically distributed. Lipopolysaccharide (LPS) is found exclusively on the outer monolayer of the outer membrane while phospholipids in most gram negative bacteria are found predominantly on the inner monolayer of the outer membrane (4). The association of LPS with outer membrane proteins can be exceptionally strong. In fact, the separation of porin and LPS can only be carried out at high ionic strength (5). In the absence of high salt, detergent solubilization of porin results in a porin complex containing up to three moles of LPS. The interaction of LPS with porin appears to be predominantly hydrophobic in nature since porin can be effectively reconstituted with only the lipid A portion of LPS (6). It is apparent, however, that the polysaccharide region of LPS does interact with the porins. In reconstituted lipid A-porin complexes, a hexagonal lattice was formed as was also detected with LPS-porin complexes, but the lattice constant was significantly less than that of the intact LPS-porin complex. whereas most biological membranes are only weakly resistant to the free diffusion of small hydrophobic compounds (7), the outer membrane is quite impermeable to these molecules. Nikaido (8) first noticed that the hydrophobicity of antibodies correlated exceptionally well with their efficacy in inhibiting deep rough (short LPS polysaccharide chain) mutant bacteria. The velocity of diffusion of such molecules across the outer membrane of LPS mutant cells increased as the length of the O-polysaccharide chains decreased. Even deep rough mutants which have only lipid A and the 3-deoxy-D-manno-octulosonic acid (KDO) sugars had one tenth hydropho: sdective The recent) y :9). Ba: to mai nt.‘ the temp! temperat- native L oithe t pheSphol tamerat Chains ( tTaHSitt activitj When tht phase I! one tenth the diffusion velocity of the cytoplasmic membrane for small hydrophobic molecules. Thus the porins and LPS form a passive yet selective diffusion barrier to low molecular weight solutes. The fluidity of the hydrocarbon interior of the outer membrane has recently been shown to be a determinant of the growth limits of.§..ppl_ (9). Bacteria adapt their outer membrane lipid and protein composition to maintain a mixture of gel and liquid crystalline acyl chain domains at the temperatures of growth. When adaptation is unsuccessful at either temperature extreme and membrane acyl chains are exclusively either gel or liquid crystalline, growth ceases. Nakayama pg 31. (10) have shown in an.§..pp11 B smooth strain using X-ray diffraction analysis that isolated native LPS undergoes a phase transition which is relatively independent of the temperature at which the cells had been grown. Moreover, the phospholipids of the outer membrane do adapt to changes in growth temperature by changing the ratio of saturated to unsaturated acyl chains (10). In a fatty acid auxotroph of g. ppll K12 the outer membrane transition temperature as determined by fluorescent probing and enzyme activity shifted in response to the exogenously supplied fatty acid (11). When the rigid trans A C18:1 fatty acid was supplied, the membrane phase transition occurred at higher temperatures than with the more fluid cis A C13:1. Since there are concomitant changes in the outer membrane protein profile as a function of temperature, no complete model exists as yet which explain the adaptation of the membrane physical state. The interaction of LPS with phospholipid domains is complex. Onji and Liu (12) have shown that LPS will bind to neutral phospholipid (phosphatidylcholine) vesicles and reduce their electrophoretic mobility. ; :a ‘I': :: 314' A " u " 02‘ "ll: '0." This effec phosphati'c interact i N 11M E2 {see Chap' Converse). interacti undoubted Hiropho: Head grou 0f LPS wi The iron est: phOSDhat- onb’ SH! ”hoSphat mobility TakEuchi that LPS ph°SDhol “as Dre: This effect was potentiated by prior incubation of the phosphatidylcholine vesicles with a cationic detergent. LPS did not interact as well with phosphatidylserine vesicles, however, the addition of 1 mM EDTA which presumably increased the net negative charge of LPS (see Chapter VI) eliminated any LPS-phosphatidylserine interaction. Conversely, 1 mM calcium addition promoted LPS-phosphatidylserine interaction almost to that of phosphatithcholine vesicles. Thus, LPS undoubtedly can associate with phospholipid domains through strong hydrophobic interaction between lipid A and the phospholipid acyl chains. Head group charge repulsion, however, may prevent the initial association of LPS with the acceptor complex. The assumption that LPS and phospholipids freely intersperse is far from established. Schindler st 21. (13) showed that LPS and phosphatidylethanolamine mixtures have diffusion coefficients which are only slightly different from their pure lipid diffusion coefficients. Phosphatidylethanolamine was thought to have slightly decreased lateral mobility because its motion was hindered by slower moving LPS monomers. Takeuchi and Nikaido (14) challenged this point and suggested instead that LPS and lipids are essentially immiscible. They found that LPS and phospholipids do mix very slowly (Tl/2 several hours) when magnesium was present and mixing was maximum when sodium, magnesium, and porin were present. The mobility under these conditions was relatively independent of the lipid head group charge since phosphatidylethanolamine and phosphatidylglycerol were roughly equivalent. An important observation of this work was that very low phospholipid-to—LPS ratios produced an apparently randomly mixed aggregate. An explanation was advanced whereby LPS and phospholipid mixtures excess 0‘ begins t Any mixi* equilibr The also ill that LPS phosphat and Ma surface i phosphat‘ PF835ure Phosphat- indicate annulus ‘ 1lght of WDOIayp but 350v afipear H mixtures behave much like water and phenol. When either is in great excess of the other, a single phase develops. As the ratio of the two begins to reach unity, the binary nature of the mixture becomes apparent. Any mixing which occurs is governed by the obligatory thermodynamic equilibrium which exists between immiscible liquid phases. The facile insertion of rough LPS into phospholipid monolayers is also illustrative of this point. Fried and Rothfield (15) have shown that LPS from Salmonella typhimurium G3OA readily inserts into phosphatidylethanolamine monolayers. These monolayers were quite stable and invariant with pressure changes (non compressible) until the LPS surface area fell below 45% of the total area. As the phosphatidylethanolamine content of the monolayer increased, the surface pressure was a function of the surface concentration of phosphatidylethanolamine. The authors interpreted those results to indicate that LPS at low phosphatidylethanolamine/LPS is surrounded by an annulus of “boundary“ phospholipids. This explanation may be reworded in light of Takeuchi's theory of LPS miscibility in phospholipids. In monolayers phosphatidylethanolamine miscibility in LPS domains is high but above a critical molar ratio pure phosphatidylethanolamine domains appear which are highly compressible and independent of the presence of LPS in the monolayers. The miscibility of LPS in phospholipid domains may also be dependent upon differences in the LPS core or polysaccharide sugars. Since deep rough mutants of.§..ppll are thought to have large amounts of phospholipids in the outer monolayer of the outer membrane (16). Biochemistgy of LPS In the following section and throughout the remainder of this dissertation it will become increasingly clear that LPS is a name attributed to an extraordinarily broad class of lipids. No generalized skeletal diagram can encompass all of the overlapping structural features of LPS nor is any component of its structure common to all bacteria. The recent biochemical characterization by Nexler and Oppenheim (17) of the Listeria monocytogenes endotoxin-like component has even broken the exclusive tie between LPS and grampnegative bacteria. Even the cell wall of Cyanophyta (blue-green alga) are reported to contain LPS (18). The most completely characterized LPS is that of the enteric bacteria. Conceptually, it has become the point of reference for LPS structural and physiological comparison. The classic lipid A of enteric bacteria consists of a a 1,6-linked D-glucosamine dissacharide (19) where the amine groups serve as attachment sites for p-hydroxy fatty acids. The hydroxyl group of one of these amide-linked B-hydroxy fatty acids is often esterified with another fatty acid. The reducing and non-reducing termini of the diglucosamine are usually phosphorylated. They may be pyrophosphorylated, ethanolamine phosphate substituted or even arabinosamine phosphate substituted (e.g. Salmonella, 20) depending on strain and culture conditions. An early supposition that LPS is cross-linked via phosphodiester bridges has proven incorrect (21). Nevertheless the strength of LPS self-association is so great that attempts to obtain free monomeric LPS have failed (22). Lipid A is linked to the core polysaccharide through an acid labile glycosidic bond to KDO. In many LPS molecules there are three molecules of K305 and etha sugars ( Alt well con confound oetails CONNOR 5 Sequence: bacteria. Overwhelr neutral 5 have a h' Percenta. The (25). T idditio” independ of KDOs in the inner core which acts as cation binding sites. Phosphorus and ethanolamine phosphate substitution is also reported to occur on KDO sugars (23). Although the sugars present in the LPS core region are reasonably well conserved from species to species, details of the sugar linkages are confounded by a host of variations and substitutions. The relevant details will be discussed in appropriate sections of the text. The most common sugars of the core are D-glucose, D-galactose and L-glycero-D-manno-heptose (referred to thoughout the text as heptose) and, of course, KDO. The O-polysaccharide chain of LPS consists of repeating sugar sequences which are the immunological determinants of gram-negative bacteria. Their variability in sequence and sugar composition is overwhelming. While most enteric O-polysaccharides are composed of neutral sugars, some species (e.g. Chromatia and Rhodgpseudomonas, (24)) have a high percentage of cationic sugars while others (e.g. Bordetella, Klebsiella, Proteus, Brucella, and Citrobacter, (25) have a high percentage of anionic sugars. / The assembly of LPS takes place on the cytoplasmic membrane (26). The core is attached to the completed lipid A by sequential addition of sugars. The synthesis of the O-polysaccharide chain occurs independently on the C55 polyisoprenoid lipid carrier. The polymerization of the repeating sugar segments that make up the O-polysaccharide is not uniform. Rather, the distribution of repeating units can range from 19 to 34 units (77 mole percent) in'S. typhimurium (27). The completed polysaccharide is then attached to the lipid A core complexes. This too occurs on the cytoplasmic membrane but is rapidly followec nenbrane conpl ex 97,28) the cute :ytool as: irreverS' smooth st iurlbert smooth 5: trawl ca enterics Hunf SLJpernata ltlease m. nem'llsls D i‘fn'pw.¥te followed by translocation into the outer monolayer of the outer membrane. The attachment of the O-polysaccharide to the lipid A-core complex is not always complete and in two recent studies of Salmonella (27,28) it was found that two thirds of the LPS which is translocated to the outer membrane has no O-polysaccharide attached at all. Since the cytoplasmic to outer membrane translocation step is essentially irreversible, it appears that in enteric bacteria the majority of LPS in smooth strains is actually rough LPS. This is not true of all LPS. Hurlbert (29) has shown that all of the LPS in Chromatia and Thiocapsa smooth strains has an O-polysaccharide chain attached. The utility of translocating both smooth and rough LPS to the outer membrane in the enterics has not been explained. Munford (30) and coworkers have compared the size heterogeneity of .§- typhimurium LPS isolated from outer membrane and from media supernatant membrane fragments (blebs). As Enterobacteriaceae grow they release membrane fragments which contain protein, phospholipid, and LPS (31). The protein profile is characteristic of the outer membrane and contains few periplasmic or cytoplasmic membrane proteins. The LPS in the membrane fragments is enriched in long chain polysaccharide LPS. Although the enrichment is only slight, it may reflect a greater tendency for more hydrophilic LPS patches of outer membrane to bleb. Since these . LPS containing membrane fragments are endotoxins, elucidating the forces responsible for their release into the serun will undoubtedly expand our practical understanding of endotoxin pathology. Host responses to endotoxins include fever induction, bone marrow necrosis, leukopenia, leukocytosis, depressed blood pressure, mitogenic lymphocyte stimulation, macrophage activation, complement activation, Hageaann productio infection features beming A st induce sh from _S_al_m Pyrogenic associati appeared activity these unu re‘iations One and L. B 1(liked by radUCing ardbanSE but )5 p) An 6 PUT-pl E ”C haCkbone 1t ls dis With C, v Hagemann factor activation, prostaglandin synthesis, interferon production, induction of endotoxin tolerance, nonSpecific resistance to infection, tumor necrosis and death (32). Attempts to link structural features of LPS with either endotoxicity or outer membrane function are becoming increasingly successful. A strong link has been made between the ability of endotoxin to induce shock and the structure of the lipid A moiety of LPS. Lipid A from Salmonella and other enterics has been amply documented to be pyrogenic, toxic, and reactive with complement (32,33). This association, however, is not absolute and numerous exceptions have appeared in the literature as the chemical composition and biological activity of new bacterial LPS are documented. A brief review of some of these unusual lipids may illustrate the complex structure-function relationship. One of the more structurally complicated lipid A groups is that of Chromobacterium violaceum (34). The diglucosamine_backbone has both 0— and L- a hydroxy fatty acids attached. The reducing end glucosamine is linked by a phosphodiester to another D-glucosamine while the non- reducing glucosamine is linked by yet another phosphodiester to an arabinosamine. This lipid A will not react with the complement system but is pyrogenic and quite lethal. An even more complex lipid A is that of Rhodospirillum.tgggg, a purple non sulfur bacterium (35). The lipid A has a diglucosamine backbone which is like that of enteric LPS with three notable exceptions. The reducing end glucosamine is phosphodiester linked to a D-arabinose; it is also linked to a 4-amino-L-arabinose via the C4 hydroxyl group. As with.§. violaceum the non-reducing end glucosamine is phosphodiester linked 1 complemi W the int; affect ‘ mild aC' cleavage molecule that are detennir Pev that of LPS may of LPS. lipid. reCOgm'z diglucos QVOUps a esterjfi Meme only m“ ability Of 3‘hyn‘ SQr . Q1 10 linked to a D-glucosamine. This lipid A will not react with the complement system, but is pyrogenic and toxic. Curiously, lipid A of g. ‘tgggg is over 100 times more toxic than the intact LPS. Hydrolysis of enteric LPS to lipid A does not usually affect lethality (36). Upon closer examination, it was found that the mild acid treatment used to obtain lipid A from 3..tggug resulted in the cleavage of the 4-aminoarabinose and arabinose-l-phosphate from the molecule. The lipid A obtained also became immuno—crossreactive with Salmonella lipid A. Thus reactive groups on the classic enteric lipid A that are blocked in LPS from 3. Egggg are apparently critical determinants of lipid A toxicity. Perhaps the least structurally complex lipid A yet characterized is that of Rhodopsudomonas viridis (37). Classification of this lipid as a LPS may exceed both the chemical and functional limits of the definition of LPS. There is neither an O-polysaccharide nor any core sugar in this lipid. It is considered a LPS by virtue of its similarity to more recognizable LPS of closely related bacteria (24,36). Instead of diglucosamine, this LPS has a 2,3-diamino-D-glucose backbone. The amino groups are partially substituted by D-3-hydroxymyristic acid. No esterified or nonhydroxylated fatty acids are present. Furthermore, the molecule contains no phosphorus. Although this LPS is nonpyrogenic and only mildly toxic, it compares well with LPS from a Salmonella in its ability to interact with complement. The importance of distinct chemical groups in lipid A are becoming apparent. Lugowski and Ramanowska (38) have shown that the hydroxamate of 3-hydroxymyristic acid alone is a potent inhibitor of the Shigella ‘gggggi lipid Alanti lipid A system. A link between the amount of phosphorus {39). LPS and Rb fr: LPS, low 1 C. vinosum. in 2‘ animal is occurs aln is delayec interactic m anti those LPS' induce the activityl' size. Th circulatin after more HF’Ollt‘otei “ting as 11 phosphorus in lipid A and anticomplement activity has also been made (39). LPS which is rich in phosphorus such as LPS from mutant strain Ra and Rb from.§. minnesota, are not able to inactivate complement, while LPS, low in phosphorus, from such sources as Anabaena variabilis, g. vinosum, and g; viridis are able to inactivate complement. .19.!119 anticomplement activity measured after injecting LPS into an animal is biphasic (40). An early depression in serum complement level occurs almost instantaneously upon LPS injection while a second decrease is delayed 6 to 9 hours. The early phase is thought to be due to the interaction of large particle size LPS aggregates which also exhibit in ‘11359 anticomplement activity. The delayed interaction occurs only with those LPS's which are also toxic. Thus, LPS from 3. viridis does not induce the second phase of inactivation despite its anticomplement activity'lnglt 9 (41). The delayed phase is not dependent upon particle size. The recent work by Ulevitch gt 31. (42) on the fate of circulating LPS makes it quite clear that LPS remaining in the serum after more than an hour is almost entirely bound to high density lipoproteins. The possible significance of high density lipoproteins acting as mediators in the LPS complement interaction has not as yet attracted attention in the literature. The serologically diverse family Chromatiaceae produces several other classes of LPS which can provide new insights into endotoxaemia. These purple sulfur bacteria are not recognized human pathogens yet they do produce LPS which retains the capacity to induce several classic endotoxic shock symptoms. Most notably g, vinosum has been assayed for its toxicity and anticomplement activity (43). Its lipid A moiety lacks the classic diglucosamine backbone and instead has a glucosamine-mannose ”iv" "l' group ' these i .8: :22 that o‘; neither was was salts o (vide 1' of eith anticom lie in sDecies been sh 12 group with typically esterified and‘amide linked fatty acids. Despite these differences 9. vinosum is 25% as toxic to mice on a weight basis as S, abortus LPS. Considering the higher average molecular weight of Q. vinosum LPS the actual molar toxicity of its LPS is nearly comparable to that of S. abortus. In marked contrast with enteric LPS, however, neither toxicity nor anticomplement activity of this LPS changed after it was washed with NaEDTA. Galanos.gtigl. (44) have shown that the sodium salts of enteric LPS typically have greatly enhanced endotoxic activity (vide infra). Even more startling was the observation that small amounts of either calcium or magnesium actually enhance Q. vinosum anticomplementary activity. The explanation for these discrepancies may lie in the differences in the charge character of enteric and Chromatiaceae LPS. Approximately 90% of the total flora in adult human consists of the species of Bacteriodes (45). The LPS of several Bacteriode species have been shown to have low endotoxin activity (46). Biochemical analysis of the LPS from Bacteriodes fragilis has revealed the complete absence of both K00 and L-glycero-D-mannoheptose (47). More recently, Nollenweber (48) and coworkers have attempted to explain the low endotoxicity of Bacteriodes by their unusual fatty acid composition. Both amide and ester linked fatty acids were found; however, the composition was quite unusual. B-hydroxyl (14:0, 15:0, 16:0, 15-Methyl—16:0, and 17:0) acyl chains were observed as well as significant amounts of iso and anti-iso myristic acid acid. Teleologically, Bacteriodes may have adapted their LPS so that they are less hazardous to their hosts. The acyl chain pattern which they have developed may represent a method of detoxifying their LPS without loss of its function within the outer membrane. establish l. (49). phosphoru toxic in enteric L only mild B-cells i I‘th bovi activity carrier 5 of lipid lfictive Sub: bacteria‘. that [My Elevated backbone. phasDhodi 0f no” re “term-o withstand 5996:1th mutant Ce .Etained : liSlStanc. 3f the 0m 13 Haemophilus influenzae provides yet another complication in establishing a clear structure-endotoxin activity relationship with lipid A. (49). Although lipid A from 5. influenzae contains glucosamine, phosphorus, B-hydroxy fatty acids as well as other fatty acids, it is not toxic in mice at doses up to 50 mg/Kg. The L050 against mice for enteric LPS is usually about 1 mg/Kg (50). Lipid A from g. influenzae is only mildly pyrogenic but can induce a mitogenic response in polyclonal B-cells in cell culture. Furthermore, when this lipid A was complexed with bovine serum albunin to enhance water solubility, all immunological activity was lost. Thus the proposal by Galanos (51) that a protein carrier such as bovine serum albunin may be essential for full expression of lipid A toxicity, must be weighed against the possibility that reactive groups on Lipid A may be masked in such a complex. Subtle lipid A modifications are also apparently critical in bacterial resistance to antibiotics. Vaara and coworkers (52) have shown that polymyxin resistant mutants (pmrA) of Salmonella typhimurium have elevated substitution at the non-reducing terminus of the diglucosamine backbone. Sixty to seventy percent of the C4 hydroxyls are phosphodiester linked with 4-amino-4 deoxy-L-arabinose. Only 10 to 15% of non resistant wild type LPS has this substitution. Moreover, this alteration enhances the ability of the pmrA mutant outer membranes to withstand tris-EDTA-lysozyme treatment which is commonly used in the separation of outer membrane and cyt0plasmic membrane (53). The pmrA mutant cells were also less sensitive to protamine and polylysine but remained sensitive to several other cationic antibiotics. Thus, the resistance generated by this mutation influences the barrier properties of the outer membrane. It was proposed that the mutation simply lowers the hind att chine Alt toxiphor properti activity from a w' nitogeni< develooe< divide in cells cox including polysaccr could ind The abili arm thoUth l D‘Gi curc lOlvsaccl sulfate. The ei‘fec DOlysaCCh telluioSe 14 the binding affinity of LPS for polymyxin by eliminating an anionic attachment site and substituting a repulsive cationic charge. Although lipid A has been endowed with a reputation as the LPS toxiphore, it is clear that its activity is modulated by the physical properties of the core and O-polysaccharide sugars. The mitogenic activity of LPS demonstrates this principle quite clearly. LPS isolated from a wide variety of gram negative bacteria has been shown to be mitogenic for B-lymphocytes of normal mice. A mouse strain C3H/HeJ was developed which produced defective spleen cells that were unable to divide in response to LPS (54). It was later found that these spleen cells could be induced to divide in response to LPS from several sources including Brucella, 359532;, and Bordetella (55). Further, the polysaccharides but not the lipid A moiety of Brucella pertussis LPS could induce mitogenicity in these cells as well as normal spleen cells. The ability of §.pertussis 0-polysaccharide but not common Salmonella or Escherichia O-polysaccharides to elicit the mitogenic response is thought to lie in the anionic character of the Brucella O-polysaccharide. D-Glucuronic acid is present in the repeating 0-antigen. Other anionic polysaccharides such as the 0-polysaccharide from E, mirabilis, dextran sulfate, and S. 31251 Vi polysaccharides have similar mitogenic behavior. . The effect is not, however, a general property of anionic polysaccharides. Sulfate, phosphate, and carboxymethyl derivatives of cellulose and heparin sulfate are ineffective mitogens against C3H/HeJ spleen cells. Thus, the enhanced mitogenicity of Brucella, firgtgus, and Bordetella LPS can not be explained solely by the presence of repeating negative charges although this property distinguishes them from the neutral or cationic O-polysaccharides of most bacterial LPS. The cotibinati trisaccha doualy Sl acid (56) 2-amino-2 relativel existence Neve coordinat In g. E esotectio A su of L519 Virulent devoid of Fucose wh aiso abse 3”“ glucc aviruiem “id cont The vimIEflCe vascular oi rd "1 L93 a 15 The polysaccharide of Bordetella pertussis represents an interesting combination of acidic, basic, and neutral sugars. Recently a trisaccharide was identified into 0-antigen which consists of a D-glucose doubly substituted with D-glucosamine and 2-amino-2-deoxy-D-galacturonic acid (56). LPS from Shigella.§gngg1 has also been reported to contain 2-amino-2-deoxy-L-altruronic acid (57). Aminohexuronic acids are relatively uncommon sugars in gram negative bacteria and reports their existence in LPS are quite rare. Nevertheless, the potential exists for these unusual sugars to coordinate cations, and provide inter and intra molecular cross bridging. In §,pertussis, the unusual trisaccharide confers considerable protections upon its LPS against chemical degradation. A survey of the sugar and fatty acid content of LPS from 38 strains of Neisseria gonorrhoeae LPS has revealed major differences between virulent and avirulent strains (58). Virulent strains were totally devoid of rhamnose whereas this sugar was common among avirulent strains. Fucose which is found in non-pathogenic.fl..§1gga and fl, lactamica was also absent in virulent fl. gonorrhoeae. The levels of mannose, galactose and glucose compared to K00 in virulent strains were greater than in the avirulent strains. More subtle variations were also observed in fatty acid content. The polysaccharides of LPS also contribute significantly to the virulence of plant pathogens. In both temperate and tropical regions, vascular wilt caused by Pseudomonas solanacearum is one of the most important bacterial diseases affecting crops. Nhatley gt g1. (59) have demonstrated that a correlation exists between the presence of 0-antigen on LPS and the ability of the bacteria to avoid the hypersensitive 4 In“ ""h m— response 4 host cell bacteria. in interce and potatc alone can challenge O-polysacc core and t The c expressior attEmpted Electrodia PrOGUCt wi Galanos ar does not l‘ result in Draper-ties LPS fr0m§ sadlmenta: liahiy m DYPQQEMC “Must, apparent p ma” the n. '362). 16 response (HR). The HR occurs when invading bacteria become attached to host cell wall resulting in a rapid collapse of host cells trapping the bacteria. Virulent cells do not attach to cell walls and multiply freely in intercellular fluid. Leach gt 31. (50) has found lectins on tobacco and potato mesophyll membranes which bind bacterial LPS. Although LPS alone can not induce the HR, it can block the HR during subsequent challenge by the intact bacterium. It has been proposed that the 0-polysaccharide in smooth strains masks the antigenic sites in the LPS core and thus avoids triggering the HR. The cations associated with LPS have a profound effect on the expression of endotoxic symptoms. Galanos and Luderitz (61) first attempted to generate uniform salts of LPS. Their procedure consisted of electrodialysing crudely isolated native LPS and neutralizing the acidic product with an appropriate base. As we will show in Chapter IV and as Galanos and Luderitz pointed out in their original work, this procedure does not result in a completely uniform salt. It does, nevertheless, result in dramatically different physiochemical and biological properties of the resulting LPS preparation. The triethylamine salt of LPS from Salmonella smooth strains was found to have a very low sedimentation coefficient suggesting a low particle size. It was also highly water soluble. ,Triethylamine LPS was also highly toxic and pyrogenic in rabbits but did not interact with the complement system. In contrast, the divalent salts of LPS were poorly soluble and had high apparent particle size. They were also less toxic and less pyrogenic than the native LPS but did interact strongly with the complement system (62). The by us to and MM upon the retained contrast, and R595 The quite hig POiymyxin (Kd = 0.3 flu0resce high affi The Kd f0 Values we- Strain) 0 Site respl trisaccha IIIl Conf binding a Only been Dr. SChln 330 and S: cm the l “thin the The i 17 The principle cation removed during electrodialysis has been shown by us to be magnesium. The capacity of the LPS salts prepared by Galanos and Luderitz to retain the substituted cation was apparently dependent upon the length of the polysaccharide chain. Salts of smooth LPS retained their altered physical properties in the presence of serum. In contrast, triethylamine salts of rough strains such as S. minnesota R3 and R595 became indistinguishable from their native LPS in serum (62). The binding affinity of LPS anionic sites for polyvalent cations is quite high. Storm gt g1. (63) have demonstrated that LPS has Kd for polymyxin B of between 2.5 and 5 uM. An even higher binding affinity (Kd - 0.3 to 0.5 pH) was suggested by Schindler and Osborn using a fluorescently labeled rough Salmonella LPS (64). They also identified a high affinity binding site for divalent cations using their labeled LPS. The Kd for Mg and Ca were 15 and 6 uM, respectively. Nearly identical values were obtained using LPS from either S. typhimurium G30 (rough strain) or G30A (heptoseless strain). They suggested that the anionic site responsible for this high affinity cation binding is within the K00 trisaccharide complex in the LPS core. We have independently (Chapter III) confirmed that this site does have a very high divalent cation binding affinity. Furthermore, the LPS used by Schindler and Osborn had only been electrodialysed. In studies carried out in cooperation with Dr. Schindler, we have found that electrodialyzed LPS from both strains 630 and G3OA retain a considerable nunber of divalent cations leaving open the possibility that even higher affinity cation binding sites exist within the molecule. The high affinity cation binding of LPS is not specific for divalent cations. As we will show in Chapter IV, a wide variety of trace elements including affinity. been deno As a conalounii bacteria I it was dis inhibitim The l nutrients acluire ar ll‘Oll (less variety 0f evolutiona Neila 73-80 K da below 1 u.” 18 including iron, aluminum, zinc and lanthanides will bind to LPS with high affinity. The potential significance of this observation has already been demonstrated in the literature. As early as 1944, A. Schade and L. Caroline (65) reported that conalbumin could inhibit the in 11339 growth of several gram-negative bacteria by virtue of its profound affinity for ferric iron. Later (66), it was discovered that supplemental iron could offset the serum inhibition of nonpathogenic Pasteurella septica strains. The keen competition between host and pathogen for essential nutrients has caused both organisms to develop multiple mechanisms to acquire and store minerals. The extraordinarily low solubility of ferric iron (less than 10'33M in water) and its irreplaceability in a variety of enzymes has made it the object of particularly fierce evolutionary adaptation. Neilands (67) has pointed out that Escherichia synthesizes several 70-80 K dalton outer membrane proteins when iron levels in the media drop below 1 uM. These proteins are receptors for bacterial enterochelin which in turn have association constants for the ferric ion of up to 1050. Thus in hindsight, it should not have been surprising when it was discovered that serum free iron levels drop dramatically during gram-negative infections (68). The startling observation was that the induced hypoferraemia could be mimicked completely by an intravenous injection of LPS (69). The association of iron with LPS not only stresses the host but may also be critical for the pathogens survival. Kochan gt.g1. (70) have shown that when virulent g..ggli strains were grown on mammalian sera their growth was inhibited and they eventually died. The bacteriocidal activity I exogenous grovn in ‘ inhibited Althi parent bar LPSis a l dependent LPS is 00' from endot (72) have obviated t Ei‘ecti ve 1- Osbor 369-4 2° 86351 (197: R0Ser Huhln 1397. 19 activity of sera could be overcome in these strains by the addition of exogenous iron or enterochelin. Interestingly, LPS isolated from cells grown in the presence of iron could also act as an iron source for serum inhibited cells. Althouh LPS can bind trace quantities of iron to the benefit of the parent bacteria, iron binding to LPS can benefit the host. Although free LPS is a potent toxin, the lethality of LPS has been shown to be dependent upon its solubility (44). The solubility of the iron salt of LPS is quite low as Sourek.gt.gl. have reported (71). Host protection from endotoxins by cations is not limited to iron. Snyder and Walker (72) have shown that injection of zinc chloride prior to challenge by LPS obviated the endotoxin symptoms. Chromium and manganese were even more effective than iron or zinc in detoxifying LPS. REFERENCES 1. Osborn, M.J. and Hu, H.C.P. (1980) Ann. Rev. Microbiol..§4, 369-422. 2. Bassford, P.J., Diedrich, D.L., Schnaitman, C.L., and Reeves, P. (1977) J. Bacteriol. 131, 608-622. 3. Steven, A.C., Ten Heggeler, 8., Muller, R., Kistler, J., and Rosenbusch, J.P. (1977) J. Cell Biol. 12, 292-301. 4. Muhlradt, P. and Gelecki, J. (1975) Eur. J. Biochem..§;, 343-352. 5. Yu, F. and Mizushima, S. (1977) Biochem. Biophys. Res. Commun. 14, 1397-1403. 6. Yamada, H. and Mizushima, S. (1980) Eur. J. Biochem.‘lg§, 209-218. 23. Dunelli,i habranes" . Nikaido, F 337-420. . Janoff, A. Acta.§§§, . Nakayama, Biochim. . . Overath, . Onji, T. - Schindler 346-350. . Takeuchi, - Fried, V. . Gmeiner, - Hexler, l - Heckessel J° Bactev ° Gilleiner, 370. ' Vaara, M. RlEtschel ° "Miro. . 22. Shands , J‘ 1221-1226. "liar, H. ’ 357-368. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 20 Danielli, J. and Davson, H. (1952) "Permeability of Natural Membranes“ Chpt. 8, Univ. Press, Cambridge. Nikaido, H. (1979) Angewandte Chemie Intermat. Edition, 18 (5), 337-420. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta 555, 56-66. Nakayama, H., Mitsui, T., Nishihara, M., and Kito, M. (1980) Biochim. Biophys. Acta 691, 1-10. Overath, P. and Trauble (1973) Biochemistry, 12 (14), 2625-2634. Onji, T. and Liu, M.-S. (1979) Biochim. Biophys. Acta §§§ 320-324. Schindler, M., Osborn, M.J., and Koppel, D.E. (1980) Nature 283, 346-350. Takeuchi, Y. and Nikaido, H. (1981) Biochemistry, 29, 523-529. Fried, V.A. and Rothfield (1978) Biochim. Biophys. Acta 514, 69-82. Gmeiner, J. and Schlecht, S. (1980) Arch. Microbiol.'lgz, 81-86. Hexler, H. and Oppenheim, J.D. (1979) Infect. Immun. 23, 845-858. Heckesser, J., Katz, A., Orews, 6., Mayer, H., and Fromne, I. (1974) J. Bacter. 120 (2), 672-678. Gmeiner, J., Luderitz, 0., Hestphal, O. (1969) Eur. J. Biochem. '1, 370. Vaara, M., Vaara, T., Jensen, M., Helander, 1., Nurminen, M., Rietschel, E.T., and Makela, P.H. (1981) FEBS Lett. 122, 145-149. Nowotny, A. (1961) J. Am. Chem. Soc.‘§§, 501-503. Shands, J.H., Jr. and Chum, P.H. (1980) J. Biol. Chem. 255 (3), 1221-1226. Mayer, H., Rapin, A.M.C., Schmidt, 6. (1976) Eur. J. Biochem.‘§§, 357-368. 2h lb 27. 2h 29 n. 3h Heckesse g, 215- . Girard, 31 (1). Robbins: Heinbaur Palva, l Goldman, Hurlbert 983-994. Munford 630-640. . Russell . Hestpha' Biochem‘ - Hestpha' Hase, S. o Tharanai Biochem. Galanos, ”- (1977 ' Keilich, ' Lueovski and Hecke Galanos, 1 0°C”: All}. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 21 Heckesser, J., Orews, 6., and Mayer, H. (1979) Ann. Rev. Microbiol. 33, 215-239. Girard, R., Chaby, R., and Bordenave, G. (1981) Infect. and Immun. 31 (1), 122-128. Robbins, P. and Wright, A. (1971) in Microbiol Toxins, Vol. 4, ed. Heinbaum, S. and Ajl, 5., Academic Press, London, 351-368. Palva, E.T. and Makela, P.H. (1980) Eur. J. Biochem. 191, 137-143. Goldman, R.C. and Leive, L. (1980) Eur. J. Biochem. 191, 145-153. Hurlbert, R.E. and Hurlbert, I.M. (1977) Infect. and Immun. 16 (3), 983-994. Munford, R.S., Hall, C.L., and Rick, P.D. (1980) J. Bact. 144 (2), 630-640. Russell, R.R.B. (1976) Microbios. Lett. g, 125-135. Hestphal, 0., Luderitz, 0., Rietschel, E.T., and Galanos, C. (1980) Biochemical Reviews 3, 191-195. Hestphal, 0. and Luderitz, 0. (1954) Angew. Chem. 66, 407-417. Hase, S. and Rietschel, E.T. (1977) Eur. J. Biochem. 15, 23. Tharanathan, R.N., Heckesser, J., and Mayer, H. (1978) Eur. J. Bi ochem. _8_4;_, 385. Galanos, C., Roppel, J., Heckesser, J., Rietschel, E.T., and Mayer, H. (1977) Infect. and Immun. 16 (2) 407-412. Keilich, 6., Roppel, J., Mayer, H. (1976) Carbohydr. Res. 51, 129. Lugowski, C. and Romanowska (1974) Eur. J. Biochem. 48, 81. Luderitz,-0., Galanos, C., Lehmann, V., Mayer, H., Rietschel, E.T., and Heckesser, J. (1978) Naturwissen. 65, 578-585. Galanos, C. (1977) Microbiology, Schlessinger, 0., ed., Washington, 0.0., Am. Soc. Microbiol. 269. 41. Freudenbi az. Ulevitch. Clinical (3. Hurlbert J. Bioch . Galanos, :— .f. . . Hawker, form, an 46. Sveen, K Hicrobic 47. Kasper, 48. Hollenwe Lindberg. “ so . Raichvarg Infect. a 50. Galanos, i (1977) m 51. Galanos, c Eur. J. 31‘ 32. yam, H., Rietschel, ( 53' Vaara, H. (1 54' Rudbach, J.A. 513-517. 55. Moreno, E. m n. We“, H., Ch 27.35. 4‘ 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 22 Freudenberg, M.A. and Galanos, C. (1977) Infect. and Immun. Ulevitch, R.J., Johnston, A.R., and Heinstein, D.B. (1981) J. Clinical Invest. 61, 827. Hurlbert, R.E., Heckesser, J., Mayer, H., and Fromme, I. (1976) Eur. J. Biochem. 68, 365-371. Galanos, C. (1975) Z. Immun-Forsch. Bd.,112, 214-229. Hawker, L.E. and Linton, A.H. (1979) Micro-organisms: FunctionL form, and environment, 2nd ed., Univ. Park Press, 302. Sveen, K., Hofstad, T., and Milner, K. (1977) Acta. Pathol. Microbiol. Scand. Sect. B. 85, 358-396. Kasper, D.L. (1976) J. Infect. Dis. 111, 59-66. Hollenweber, H.-H., Rietschel, E.T., Hofstad, T., Heintraub, A., and Lindberg, A.A. (1980) J. Bact. 144 (3), 898-903. Raichvarg, 0., Guenounov, M., Brossard, C., and Agneray, J. (1981) Infect. and Immun. 33 (1), 49-53. Galanos, C., Freudenberg, M., Hase, 5., Jay, F., and Ruschmann (1977) Microbiology, ibid, 269-276. Galanos, C., Rietschel, E.T., Luderitz, 0., and Hestphal, O. (1972) Eur. J. Biochem. 31, 230-233. Vaara, M., Vaara, T., Jensen, M., Helander, 1., Nurminen, M., Rietschel, E.T., and Makela, P.H. (1981) FEBS Lett. 112, 145-149. Vaara, M. (1981) J. Bacter..11§, 426-434. Rudbach, J.A. and Reed, N.D. (1977) Infect. and Immun. 16 (2), 513-517. Moreno, E. and Berman, D.T. (1979) J. Immunol..1g§, 2915-2919. Moreau, M., Chaby, R., and Szabo, L. (1982) J. Bacter. 150 (1), 27-35. 57. 58. 59. 63. 51. 620 63. 998.1 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 23 Kontrohr, T. (1977) Carbohydr. Res. 58, 498-500. Hiseman, G.M. and Caird, J.D. (1977) Infect. and Immun. 16, 550-556. Hhatley, M.H., Hunter, N., Cantrell, M.A., Hendrick, C., Keegstra, K., and Sequeira, L. (1980) Plant Physiol. 65, 557-559. Leach, J., Cantrell, M.A., and Sequeria, L. (1978) Phytopathol. News 1;, 197. Galanos, C. and Luderitz, 0. (1975) Eur. J. Biochem. s5, 603-610. Galanos, C. and Luderitz, 0. (1976) Eur. J. Biochem. 65, 403-408. Storm, D.R., Rosenthal, K.S., and Swanson, P.E. (1977) Ann. Rev. Biochem. 16, 723-763. Schindler, M. and Osborn, M.J. (1979) Biochem. 18 (20), 4425-4430. Schade, A.L. and Caroline, L. (1944) Science 199, 14-15. Bullen, J.J., Hilson, A.B., Cushnic, G.H., and Rogers, H.H. (1968) Immunology 15, 889-898. Neiland, J.B. (1979) TIBS, 115-118. Kampschmidt, R.F. and Schultz, G.A. (1961) Proc. Soc. EXp. Biol. Med. 1116, 870. Butler, E.J., Curtis, M.J., and Hatford, M. (1973) Res. Vet. Sci. 15, 267-269. Kochan, I., Kvath, J.T., and Wiles, T.I. (1977) J. of Infect. Dis. 135 (4), 623-632. Sourek, J., Tichy, M., and Levin, J. (1978) Infect. and Immun. g1, 648-654. Snyder, S.L. and Halkner, R.I. (1976) Infect. and Immun., 13 (3), 998-1000. A FOE for data t lines. Ir the data i °fC°mPuta iterative Exam; Presented transltlor It 15 of com,Ect .Ublished CHAPTER II A PROGRAM FOR TRANSITION POINT ANALYSIS OF EXPERIMENTAL DATA* ABSTRACT A FORTRAN IV program is presented which determines transition points for data that can be represented by a set of intersecting regression lines. Initial estimates are not required, and no specific spacing of the data is needed. A spline function is used to reduce the large number of computational steps arising from complex linear models while an iterative least squares procedure is used to obtain the final solution. Examples of experimental and artificially generated data are presented along with an F-statistic for differences in slope at the transition point. INTRODUCTION It is often convenient to represent experimental results with a set of connected regression lines. An example can be seen in the sharp *Published in Computers in Biology and Medicine (1981) 11, 9-15. 24 change in 5' temperature transitions determine a vhere 1) tr the locatic this anal y! paper (1). analysis 0‘ A nut: Points (4, has pointe in detenni PCSsible 1 We!“ of ”“779" of Griffiths ' mtlphase UUr m continua“ "TIA this m 25 change in slope of the logarithm of enzyme activity vs. reciprocal temperature. Considerable physical significance is associated with such transitions while the points of transition are frequently difficult to determine accurately. In this paper we are concerned with the cases where 1) the data before and after the transition behave linearly, and 2) the location of the transition is not known. Detailed applications of this analysis to actual experimental data will be given in a separate paper (1). The program has already been used for transition point analysis of membrane fluidity data (2,3). A number of methods have been developed to estimate transition points (4,5,6). While all of these methods provide line sets, Hudson (6) has pointed out the computational obstacles that this search represents. In determining such transitions there are 5‘: 25 (O ("'E'Z) S'O possible line sets, where r is the number of transitions and n is the number of data points. Clearly, if either n or r becomes large, the number of solutions to be tested becomes prohibitive. This has led Griffiths and Miller (7) to suggest the use of a Spline function to model multiphase relationships. Our model assumes that suitable data can be represented by a set of continuous non-overlapping regression lines. While these lines are not constrained to meet, discontinuous line sets are rejected. In accordance with this model, the variance is assumed to be constant and normally distributed over the entire data range. It is e be correlate functions ar (8). Aspli snoothl y at extremely se excellent en normalized i the number a Spline funci estimate of if: iterative points. Th, the curve f- An F-5« indication e The Dr °9519ned to to ”V9 tra {95¢ "“99 by a Spline rout 3.x. .. Mmilnei 26 It is expected that the behavior of the data in one phase will not be correlated with that in any other phase. As mentioned above, spline functions are particularly useful approximations under this condition (8). A spline function is defined as a set of polynomial pieces joined smoothly at points called knots. Their piecewise character makes than extremely sensitive to the local prOperties of the data and they are excellent empirical functions. He have adapted an algorithm using normalized B-spines developed by Dierckx (9) which automatically selects the number and position of the knots. Information obtained from the Spline function and its derivatives can then be used to obtain an initial estimate of the transition points. The final solution is determined by an iterative linear regression analysis about the estimated transition points. The efficiency of this method over others stems directly from the curve fitting used to obtain initial estimates of the transitions. An F-statistic is calculated for each transition which gives an indication of the difference in slope of the two regression lines at the transition point. PROGRAM The program package consists of four FORTRAN IV subroutines. It is designed to accept up to 200 weighted data pairs and will search for up to five transition points (6 lines). In the initial steps, data are fitted by a nonmalized B-spline utilizing the program of Dierckx (9); the spline routines have been slightly modified by the addition of a subroutine from de Boor (10) to-speed calculations. The UT approximate are the km data point: repeated va value are a weight and aspect of t is detennir residuals 5 user are th overestimat FOT‘ Du TUOCtlon. 1' estimates 0 3f the 5p“ ! Smooth fu Su9935180 5 0f the thir Smcthlng f decrEfiSes t littor fro“ transition t'dhsitmn ‘ the “00th“ lemes 0f t) Fittegwe or 27 The user determines the degree of the polynomial pieces used to approxinate the data. A few advantages of this particular spline program are the knots need not be specified by the user and that equally spaced data points are not required. Since the spline program does not allow repeated values of the independent variable, data points with the same x value are analyzed as a single point; this point is given increased weight and represents the mean of the combined data pairs. An important aspect of the spline package in our analyses is that the closeness of fit is determined by a smoothing factor which is related to the sum of the residuals squared. The spline parameters which must be assigned by the user are the degree (K), initial smoothing factor (S), and an overestimate of the number of knots (NK). For our analyses the zeroes of the third derivative of the Spline function, i.e., the intervals with the greatest slope change, are used as estimates of the transition points. Note that this requires the degree of the spline to be at least four in order for the third derivative to be a smooth function. Also, polynomials of degree greater than five are not suggested Since they may result in overfitting (8). The number of zeroes of the third derivative of the spline function are dependent upon the smoothing factor of the Spline calculation; as the smoothing factor decreases the fit becomes tighter. The program varies the smoothing factor from an initial estimate in order to obtain the desired number of transition points; if iterations occur in which the assigned number of transition points is bracketed, a bisection method is used to determine the smoothing factor. Newton-Raphson procedure is utilized to locate zeroes of the third derivative. Starting points for the Newton-Raphson procedure are obtained by inspection of the third derivative of the spline. Thi: thus approxil The app portions. A combinations estimated tr. user. Regre sets of line The best sol i’inSition p residuals Sq Mon transi (WPlEtely d DClnts in th TO aid flit: “H ler‘erlCeS h’lll hypotheg million o “C ' its Corre 28 spline. This procedure is completed when the proper number of zeroes and thus approximated transition points are found. The approximated transition points allow grouping data into linear portions. An interative procedure is then used to consider all possible combinations of data groups which adjoin in an interval about the estinated transition; the size of this interval is determined by the user. Regression lines are calculated for all of these data groups as sets of lines: L1(x) ‘ a1 x1 + b1 + e1 Ldm'askibai% where X; represents X values on line i and e1 represents the random variable accounting for error of each data point on line i. The best solution is selected based on the following criteria: (1) the transition point lies between the data groups; and (2) lowest sum of the residuals squared. The intersection of these regression lines are the final transition points and are returned to the user in two arrays which completely describe the lines fitted to the data; thus the first and last points in the array represent the first and last data points. To aid in the interpretation of the results, an F-statistic for differences in slope of each adjacent pair of lines is calculated; the null hypothesis for this test is that the lines are parallel, i.e. no transition occurred (11). For each transition point analyzed, an F value and its corresponding degrees of freedom are returned. The call anon (X,Y,l~' A description given below. INPUT X-tl Y - ti COEFF XERK 29 The calling sequence for the set of programs described is: CALL BRKANL (X,Y,H,N,NK,KNOT,KN1,COEFF,KD,S,NB,INT,XBRK,YBRK,F,DF,ERRMIN,IER). A description of each parameter and its suggested or restricted values is given below- 11301 X - the array of the X (independent) values. Y - the array of the Y (measured property) values. H - the array of the weights for each data point. N - the number of data points (N<200). NK - the maximum number of knots (3 ghu An .mcovupmcacu amuse as mcsuu>esu o: mzbsm an o» a o_ao>uev mpasuwmmc as» mo uo—a < as .uom.mm wen m.- an umecauuo mesuaerEwu msmco> gpu gaumEaeea Pneuumam use mo aucwuemnou cemcpp as» c. mcopuwmcae» .m:.~a§ogneo—gu z aoH x m mo mucmmwca as» :. mueemoum pxxocam mecca mocmcomme c—am cognac—m us» say: um—wnu— mew: abscesses .mmcacnsws muxuocguaem cuss; mo mvmapacm Haven copupmemch .m meampm 8L 2232.th No mVVVOW Mmmtmlow RN CW. 0.. w... m ¢ 0 cu ’ .vn .;wm .¢fiu .mwm .9 .mfiu JFN java ,va Sway mum r 8 o s o e o: e e we. 0 o o 0099 o 009 e 0 fl P no a Be 9.» Po .0 o co m ofi 40%.”. 41 “0“ 4? hp! 4 P} ‘ ph o . o O > D ‘1 04 9 ‘ O .66 ‘ O .0. O . O 0 Pbbrbbbbb - bDbbbbPbbbbbhhhIPbPNlPDbthbbIbhbbbthbhhbbbbb CHAPTER III A CATION ELECTRON SPIN RESONANCE PROBE USED TO ANALYZE CATION INTERACTIONS WITH LIPOPOLYSACCHARIDE* SUMMARY The partitioning of a cationic electron Spin resonance probe, 4-(dodecyl dimethyl anrnoniun)-1-oxy-2,2,6,6,-tetramethyl piperidine bromide, into lipopolysaccharide from Escherichia coli H1485 was shown to increase markedly above approximately 15°C, presumably reflecting a thenmal transition. Partitioning was also highly dependent on probe and lipopolysaccharide concentrations, and Scatchard analysis of electrodialyzed lipopolysaccharide revealed a single non-interactive binding site for the probe. Several cations were able to displace probe bound to this site. At concentrations above 30 pM, Ca2+ and M92+ displaced probe bound to electrodialyzed lipopolysaccharide while various polyamines and other cations were less effective. Since this probe is very sensitive to the environment of the lipopolysaccharide, it Should prove to be a valuable tool in analyzing lipopolysaccharide structure and interactions with other molecules. *Published in Biochem. Biophys. Res. Commun. (1979).§2_(2), 565-570. 40 41 INTRODUCTION The outer membrane of gram negative bacteria consists of a highly asymmetric bilayer. The outer monolayer is comprised of a relatively impermeable lipopolysaccharide (LPS)-protein complex, while the inner monolayer contains phospholipid and protein. The role of LPS or the LPS-protein complex in excluding hydrophobic and amphipathic molecules such as fatty acids, bile salts and certain antibiotics from the intact cell has been studied extensively (1). It has been shown that divalent cations are important in stabilizing the LPS structure. In the intact outer membrane of Escherichia coli two populations of LPS have been described which respond differently to EDTA extraction (2). The LPS population which is readily extracted with EDTA apparently consists of clusters stabilized predominantly by divalent cations. To further analyze cation-LPS interactions we have utilized the cationic electron spin resonance probe CAle described by Quintanilha and Packer (3) which partitions into lipid bilayers according to surface charge. We report here that partitioning of CAle into LPS complexes is sensitive to the LPS thermal transition and to cation-LPS interactions. MATERIALS AND METHODS Growth of E. coli strain H1485 was as previously described (4). Cells grown at 37°C were washed three times in distilled water. Hot aqueous phenol was used to extract LPS (5). Extensive dialysis against distilled water yielded the nLPS product. Electrodialysis of the nLPS 42 resulted in an acidic deionized sample which was immediately neutralized to pH 7.0 with NaOH and stored at 4°C. Ketodeoxyoctanoic acid was assayed to quantitate LPS recovery (6). LPS samples were prepared for elemental analysis by wet ashing with concentrated nitric acid (instra-analyzed, Baker Chemical Co.) at 70-75°C for 24 h. Elemental analysis was performed in a Jarrell-Ash 955 Plasma Atomcom plasma emission spectrometer. The CAT12 probe was synthesized according to the procedures of Hubbell, gt 21. (7). Stock solutions of CAT12 were made up at 1 mM concentration in distilled water and were added to LPS samples at room temperature with mild vortexing. All electron spin resonance experiments were carried out with a Varian Century line ESR spectrometer, model E112. An external calibrated thermistor probe monitored the temperature of the sample. The free probe concentration, [F], was determined by the peak to peak height of the free probe high field Signal, corrected with a free probe standard curve. The bound probe concentration, [8], was calculated by subtracting the free probe from the total probe added. The temperature dependent partitioning parameterfl’t, was determined as follows: [Bl/[F1 W t = - Tlogm where T is in degrees Kelvin and Ct is the value of [Bl/[F] at 310°K. The ionic strength dependent partitioning parameter,‘P1, was calculated 43 according to the following equation: [BI/[F] Ci ‘l’x " - 10910 where C; is the value of [BJ/[F] in the absence of added cations. The temperature dependence of CAle partitioning was analyzed using a linear regression program developed for transition point analysis (8). RESULTS AND DISCUSSION Since LPS from E. coli undergoes a broad thermal transition with a midpoint at 22°C (9), it was important to determine whether binding of CAT12 to LPS was sensitive to thermally induced structural changes. Figure 1 shows that partitioning of CAle into nLPS increased dramatically with increasing temperature, apparently reflecting a thermal transition at 15.5 1 1.5°C. Furthermore, this temperature dependent binding was completely reversible. CAT12 binding to edLPS showed a similar temperature dependence but the effect was not as well resolved due to the tighter binding of CAT12 to edLPS. Thus, the partitioning of CAle into LPS is highly dependent on the physical structure of the LPS and is maximal at temperatures above the thermal transition. These results suggest that the barrier function of LPS, i.e., the exclusion of detergent type compounds like the CAle molecule itself, is greater at temperatures below the thermal transition of the LPS. To further characterize the CATIZ binding site in LPS and to determine the stoichiometry of binding, partitioning was analyzed at 37°C 44 as a function of probe concentration. The Scatchard plot of CAT12 binding to edLPS (Figure 2) indicates a single non-cooperative binding site. Saturation occurred at a molar ration of 1:1.5, LPS to CAle. It has been shown that LPS from E. coli grown under similar conditions as used here contains three negatively charged phosphates attached to the diglucosamine groups in lipid A (10). In addition, the core region of LPS polysaccaride chains contains carboxyl and phosphate groups. The lack of charge neutralization upon binding CAle may be precluded by the rigidity of the LPS, the presence of tightly bound cations (Table 1), or by other physical constraints. Table 1. Elemental Analysis of LPS Samples Expressed as Molar Ratios LPS Ca/P Mg/P Ca + Mg/LPSa Preparation nLPS 0.117 0.434 3.31 edLPS 0.090 0.318 2.45 aAssuming 6 moles P/mole LPS. It has been reported that a high affinity divalent cation binding site exists in edLPS from Salmonella typhimurium (11). This Site is reported to occur in equimolar ratios with LPS and is thought to involve carboxyl groups in the LPS core and a phosphate in the lipid A diglucosamine backbone. Table 1 indicates that in E. coli approximately one divalent cation was removed from each LPS molecule upon electrodialysis, consistent with the number of high affinity binding sites reported in edLPS from S. typhimurium. To determine whether Cillz is b and H92+ w Elanental levels of reports (I vas added cation Spe concentrat 50 oh con: ionic radi 332+, ioni cations we d1“Splacing consistent the lrival. '35 as eff. WAN-lys- r—‘ '— .4 ‘hlstjdu e 5 ”Ethiopia bound to ed 45 cnrlz is binding near the high affinity cation binding site, Ca2+ and MgZ+ were titrated at 37°C against edLPS bound with CAT12. Elemental analysis indicated that edLPS initially contained significant levels of these divalent cations (Table 1) consistent with previous reports (12). Figure 3 shows that CAle was displaced as either ion was added at concentrations above 30 pM. Furthermore, the trivalent cation Spermidine was effective in displacing CAle but only at concentrations above 500 uM. When small cations were added to edLPS at 50 uM concentration, CAT12 displacement appeared to depend on the ionic radius (Table 2). Since Yb3+ was almost as effective as Ca2+, ionic charge does not appear as critical for binding. Hhen cations were added at high concentrations (1 mM) their efficacy in displacing the probe was Lanthanides3+ > Sr2+ > Ca2+ > Mng. consistent with their order in ability to screen charge (13). Neither the trivalent ion spermidine nor the tetravalent ion, spermine, at 1 mM was as effective as Ca2+ in displacing probe, and poly-I-arginine and poly-I-lysine each at 1 mM monomeric concentrations were ineffective (data not shown). None of the other cationic molecules tested at 1 mM (I-histidine, 1-lysine, I-arginine, 1-ornithine, choline and triethanolamine) was able to displace significant levels of CAT12 bound to edLPS. The one exception was Tris-HCl, pH 7.5, which, at 1 mM, displaced significant amounts of probe (data not shown). Control SrCiz LaCi3 H9212 YbCl3 C3512 \ *t‘ifferent f rc The resul am through Site may invoi Furthermore, t “d“ied by the interactiOns. ”mm that CA1 it appears the Lps‘pY‘OtE‘in lr 361‘491. u 46 Table 2. Displacement of CAT12 from edLPS by 50 pH Salt an. 9, 325132 A Control 0.0 -- SrClz -0.043 1.13 LaCl3 0.004 1.15 MgClz 0.056* 0.55 YbCl3 0.101* 0.94 oa012 ‘0.111* 0.99 *Different from control. p<0.05. The results reported here indicate that CAle interacts with edLPS through a single type of non-cooperative binding site and that this site may involve the high affinity divalent cation site of interaction. Furthermore, binding of the probe is sensitive to the structural change induced by the LPS thermal transition and to competitive ion interactions. In preliminary studies (manuscript in preparation) we have shown that CAT12 can partition into outer membranes of E. coli, and it appears that this probe will be useful in further analyzing LPS-protein interactions. REFERENCES 1. Nikaido, H. (1979) in Bacterial Outer Membranes, ed. Inouye, M., pp. 361-401, Wiley Interscience Publications, N.Y. Levy, S-B 1435-1439 Quintanil Janoff, A Acta §§§§, Nakamura, 371-393. Droge, E. J. Bioche Hubbell, (1970) Bi Brunder, Biol. Mac Emmerling Biochem. ° Rosner, p Bio], Che SChlndlei Gaianos’ McLaughii 3. 667.5 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 47 Levy, 8.8. and Leive, L. (1968) Proc.Natl. Acad. Sci. USA.§1, 1435-1439. Quintanilha, A.T. and Packer, I. (1977) FEBS Lett.,.1§, 161-165. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta §§§, 56-66. Nakamura, K. and Mizushima, S. (1977) Biochim. Biophys. Acta 51;, 371-393. Droge, E., Lehmann, V., Luderitz, 0., and Nestphal, 0. (1970) Eur. J. Biochem. 14, 175-184. Hubbell, N.L., Metcalfe, J.C., Metcalfe, S.M., and McConnell, H.H. (1970) Biochim. Bi0phys. Acta 212, 415-427. Brunder, D.G., Coughlin, R.T., and McGroarty, E.J. (1981) Comput. Biol. Med., 1;, 9-15. Emmerling, G., Henning, U., and Gulik-Krzywicki (1977) Eur. J. Biochem. 1g, 503-509. Rosner, M.R., Tang, J., Barzilay, 1., and Khorana, G. (1979) J. Biol. Chem. ggg, 5906-5917. Schindler, M. and Osborn, M.J. (1979) Biochemistry 19, 4425-4430. Galanos, C. and Luderitz, 0. (1975) Eur. J. Biochem. g5, 603-610. McLaughlin, S.G.A., Szabo, 6., and Eisenman, G. (1971) J. Gen. Phys. _5_§, 667-687. 90-“ so- 70- h 60 I 30- 20~ Figure 1, p MSW t 48 IO“ 00 5 66202530354045 TEMPERATURE (°C) Figure 1. Partitioning of CAT 4 into nLPS as a function of increasing temperature. The CA 12 to nLPS molar ratio was 0.1. 40¢ Figure 2. partltioni “nCthH o 49 1 1 l 1’ 1 ICC 200 300 400 500 600 [314M Figure 2. Scatchard analysis of CAT12 binding to edLPS. CAT12 partitioning was measured at 37°C on samples of 500 uM edLPS as a function of probe concentration. 20. to #1 Qure 3. '“ncti on tonce'll: ra 50 2.0 . .. [.5 " .J-II-""" .. 4’: LO - .. 05 - A K , j l IO IOO IOOO 10000 ION CONCENTRATION (,uM) Figure 3. Displacement 0; CAle from edLPS was measured at 37°C as a function of Ca + (0), Mg +, (A ), and spermidine (I) concentrations. The CAT12 to edLPS molar ratio was 0.1. C'Jr'Ji'TITA' Inductiv' quantitate th to extracted strains. The Qinesium re) $1'ini'iicant l content of H Morena. Li contained hig b“: the “Ediu; “-9 as we” as .n ContraSt, . CHAPTER IV QUANTITATION 0F METAL CATIONS BOUND T0 MEMBRANES AND EXTRACTED LIPOPOLYSACCHARIDE 0F ESCHERICHIA COLI ABSTRACT Inductively coupled plasma emission spectroscopy was used to quantitate the metal cations bound to outer and cytoplasmic membranes and to extracted lipopolysaccharide from several Escherichia coli k12 strains. The outer membrane was found to be enriched in both calcium and magnesium relative to the cytoplasmic membrane. Both membranes contained significant levels of iron, aluminum, and zinc. The multivalent cation content of lipopolysaccharide-resembled that of the intact outer membrane. Lipopolysaccharide extracted from wild type k12 strains contained higher levels of Mg than Ca regardless of the growth medium, but the medium used for growth did effect the relative amounts of bound Mg as well as the levels of the minor cations, iron, aluminum and zinc. In contrast, lipopolysaccharide isolated from a deep rough mutant strain, 021f2, contained more Ca than Mg. Electrodialysis of lipopolysaccharide from wild type k12 strains removed one mole of Mg per mole of ‘lipopolysaccharide but did not significantly affect the level of other bound metal ions. Dialysis of lipopolysaccharide against sodium (ethylenedinitril0)tetraacetate removed most of the Mg and Ca resulting 51 ina sodium sini'iar to i :at'lon neuti The sodium : oethane°HCl ‘ioopoiysac was shown t but not aii of the othe Tsuitlng 1': isolated fr: into a sodii natralize i ”POPOTysaCI S'OUps in t) Strain resuf cations and Divaien Otter mammal ”$111510“ bei divalent Catj LPS "1th memo Emma PFOte 52 in a sodium salt. The molar content of Na bound to the sodium salt was similar to the level of divalent cations removed, but the monovalent cation neutralized less of the anionic charge on the lipopolysaccharide. The sodium salt was dialyzed against either Tris(hydroxymethyl)amino- methane'HCl, CaClz, MgClz, FeClZ, or TbCl3, and the resulting lipopolysaccharide salts were analyzed for their ionic composition. It was shown that Tris(hydroxymethyl)aminomethane and Ca can replace some but not all of the Na bound to the sodium salt, but it appears that all of the other multivalent cations can nearly completely replace Na, resulting in uniform lipopolysaccharide salts. Lipopolysaccharide isolated from the deep rough mutant strain, 021f2, was also converted into a sodium salt. Relative to the wild type LPS, Na was able to neutralize the anionic charge to a greater extent in the mutant lipopolysaccharide. These results suggest that the loss of specific groups in the core region of the lipopolysaccharide from the mutant strain results in a more Open structure that allows the binding of larger cations and of more monovalent cations. INTRODUCTION Divalent cations play an important role in the stabilization of the outer membrane of gram negative bacteria. In addition to reducing charge repulsion between the highly anionic lip0polysaccharide (LPS) molecules, divalent cations are thought to bridge adjacent LPS molecules and to link LPS with membrane proteins (1). The formation of complexes of outer membrane protein and LPS requires divalent cations such as magnesium :2). Chelati effect the re The phys are aiso 1'an L95 aggregrat resoit, the a aggregated tr which have be resistant to 515913)) soiu The tend 13% Even af Citations. mad by el :ations "emai cflietal cati 33th t0 be 53 (2). Chelating agents such as (ethylenedinitrilo)tetraacetic acid (EDTA) effect the release of up to 50% of the LPS from whole cells (3). The physiochemical and pathophysiological properties of isolated LPS are also influenced by the ions bound. The solubility and morphology of LPS aggregrates vary greatly depending on the salt form (4). As a result, the anticomplement activity of the soluble sodium salt of LPS from Salmonella abortus is much greater than that of the highly aggregated triethylamine salt (5). It has also been observed that mice which have been injected with iron salts prior to LPS administration are resistant to endotoxin stress (6). Iron salts of LPS generally are only slightly soluble and have lower toxicity (7). The tenacity with which LPS binds cations has been previously noted (8). Even after extensive dialysis LPS remains complexed with a variety of cations. Although monovalent cations and several polyamines may be removed by electrodialysis, most of the tightly associated divalent cations remain bound (4,9). In this study we have quantitated the levels of metal cations bound to isolated membranes and extracted LPS in an attempt to better characterize cation-LPS interactions. MATERIALS AND METHODS Growth conditions and procedures for membrane and_LPS isolation . Escherichia coli k12 strains H1485 F', 021, and 021f2 were grown at 37°C in either M9 minimal medium plus 0.4% glucose or in nutrient broth (1% tnyptone, 0.2% yeast extract, 0.4% NaCl). Outer and cytoplasmic membranes were isolated (10), and the purity of the isolates equalled or exceeded that described earlier. LPS was isolated using either hot a0. extraction or yielded the I resoited in . (edLPS). Th edLPS agains T'ciiowed by sait forms l TRIS'HCI, Ci diaiysis ag Porin- Ntrient by fiShEd in : 39‘315r1'bori iEii'i'iii‘aneg Porin-LPs. iEIDi‘ane f ehtensi ve' 54 either hot aqueous phenol (11) or chloroform-petroleum ether (12) extraction procedures. Extensive dialysis against double distilled water yielded the native LPS (nLPS) product. Electrodialysis of the nLPS resulted in an acidic product which was analyzed in the acidic form (edLPS). The sodium salt of LPS, NaLPS, was obtained by dialysis of edLPS against three to five changes of 10 mM NaEDTA, pH 7.0 at 4°C, followed by extensive dialysis against double distilled water. Other salt forms were obtained by dialysis of the sodium salt against 10 mM TRIS'HCl, CaClz, MgClz, TbCl3. or FeCl3, followed by extensive dialysis against double distilled water. Porin-LPS-peptidoglycan complexes were isolated from cells grown in nutrient broth and harvested as previously described (10). Cells were washed in 10 mM HEPES, pH 7.5, and lysed using a French pressure cell. Deoxyribonuclease I (Sigma Chemical Co.) was added at 20 ug/ml, and total membranes were pelleted and washed twice with distilled water. The porin-LPS-peptidoglycan complex was isolated by extracting the total membrane fraction with 1% SDS (w/v) twice and washing the complex extensively with distilled water. Assays 3-deoxy-D-manno-octulosonic acid (K00) was assayed using the method of Droge, gt 31. (13) to quantitate LPS recovery. Protein concentrations were determined using the procedure of Lowry _et _al. (14). Membrane proteins were characterized on SDS polyacrylamide slab gels as previously described (10). Peptidoglycan was quantitated by assaying the levels of muramic acid (15). Bet asi nitric acid cap vial (T) to 75°C for log/ml, an: soiution. Dry as :ruci‘ole, l sanpies wer saoples wer cobalt as a Sample both wet an analyses we ashed mater Elmer; % Quanti SPECtrOSCOp "I 1 Va ria for the We 214'9; Na, h. 350.9, The area dr and the bag by “Wei [3 55 Sample ashing_procedures for elemental analysis Het ashing was accomplished by combining equal volumes of sample and nitric acid (Ultra R, Atomergic Chemicals Corp.) in a 15 ml teflon screw cap vial (Tuf-tainer, Pierce Chemical 00.). Samples were incubated at 70 to 75°C for 24 hours. Cobalt was added to give a final concentration of 1 ug/ml, and ashed samples were diluted to give a 15% (v/v) acid solution. Dry ashing was accomplished by transferring the sample to a quartz crucible, lyopbylizing, and then heating in a muffle furnace. The samples were ashed at 500°C for 24 hours. After dilution with water, the samples were filtered through acid washed glass wool and analyzed with cobalt as an internal reference standard. Samples of nLPS and edLPS were prepared for elemental analysis by both wet and dry ashing procedures. For convenience, most elemental analyses were carried out on wet ashed material, since analyses of dry ashed material yielded virtually identical results. Elemental analysis Quantitative elemental analysis was accomplished by plasma emission spectroscopy using a Jarrell-Ash Model 955 Atomcomp Spectrometer with an N + 1 variable wavelength accessory. The observation wavelengths (nm) for the specific elements were set as follows: Ca, 370.6; Mg, 279.0; P, 214.9; Na, 330.3; K, 766.5; Fe, 259.9; Al, 308.2; Zn, 213.8; Co, 228.6; Tb, 350.9. Potassium and terbium were quantitated on the N + 1 channel. The area around each emission line was examined for spectral interference and the background effects by dynamic profiling. Profiles were obtained by manual micrometer adjustment of the entrance slit position. To corral: for in Viscosit' standard' gl/nin by a are the ave unless WI ratios 0f t Since the structu 0f ions bou characteriz levels of d cytoplasmic to contain high levels result from the underl y cmhosition {Table 2). ”OH. 56 correct for changes in nebulizer spray characteristics due to variations in viscosity and/or dissolved solids, added cobalt served as an internal standard. Samples were introduced into the nebulizer at a rate of 1.5 ml/min by a peristaltic pump. Elemental analysis results presented here are the average of measurements on three or more individual isolates unless indicated otherwise. Metal cation levels are expressed as molar ratios of the cation to phosphorus content. RESULTS Membrane ionic composition Since multivalent cations are reported to be critical in stabilizing the structure of the outer membrane in gram negative bacteria, the levels of ions bound to outer and cytoplasmic membranes of g. c_ol_i_ were characterized and compared. The outer membrane contained three times the levels of divalent cations on a per phosphorus basis compared to the cytoplasmic membrane (Table 1). Furthermore, both membranes were shown to contain small but significant levels of iron, aluminum and zinc. The high levels of divalent cations bound to outer membranes presumably result from the presence of anionic LPS in the outer monolayer, and from the underlying anionic peptidoglycan. Note that the multivalent ion composition of nLPS was similar to that of the intact outer membrane (Table 2). A porin-LPS-peptidoglycan complex was purified and characterized to determine whether certain cations are involved in the interaction of outer membrane proteins with other membrane structures. This complex was shown to cm pep-110091310 n. the come and 1b with )5 indicate mnovalent mp) ex HES intact oute oata whethe anionic muc cmplex. L93 ionic C LPS is phenol or c ireparatior nLPS isolat Phosphorus Jansson g phosPilate c nLPS fro,“ c total catic eii Sher ca]c 57 shown to contain approximately 70% protein, 16% LPS and between 7 and 15% peptidoglycan by weight. SDS polyacrylamide gels showed that the protein in the complex was comprised of approximately equal amounts of porins 1a and 1b with only small amounts of contaminating protein (data not shown). As indicated in Table 3, this complex was enriched in divalent and monovalent cations compared to nLPS. Specifically the Mg content of the complex was nearly twice the amount found in the purified LPS and in the intact outer membrane. However, it could not be determined from these data whether the excess cationic charge was bound to protein, to the anionic mucopeptide layer or between the anionic groups within the complex. LPS ionic composition LPS isolated from g. 2911 strain H1485 F' using either hot aqueous phenol or chloroform-petroleum ether methods resulted in nLPS preparations which had essentially identical ionic compositions. The nLPS isolated under these conditions contained between 7 and 8 moles phosphorus per 3 moles KDO (Table 2) consistent with the findings of Jansson gt 31. (16). Figure 1 indicates the likely position of the phosphate groups in the LPS structure. The ionic composition analysis of nLPS from cells grown in nutrient broth indicated that approximately five metal cations were bound to each LPS molecule with four of these being either calcium or magnesium. In an attempt to remove weakly bound cations including organic amines, the LPS was electrodialyzed, and the ionic composition of the acidic form (pH 3.5 to 4.0) is given in Table 2. Electrodialysis consistently removed one mole of Mg per mole LPS while the levels of all other metal cations were unaffected. The results suggest that I mtrient medii ‘38, and one ( Forthermore, 1 per three molt electrodialys E“ects of gr: To asses: 1’3, cells we and the LPS w. present in th in Table 2, 'efiECt t0 va content of th '45 found to nutrient brot SM)‘ Thus higher levels and PEr Dhc WITIEnt brot the difffirent 58 suggest that the native form of LPS from g, cglj_cultures grown in nutrient medium contains 1 mole calcium and 3 moles magnesium per mole LPS, and one of the magnesium ions is removed with electrodialysis. Furthermore, this LPS contains one mole total of iron, aluminum, and zinc per three moles of LPS, and these ions are not removed with electrodialysis. Effects of;growth media on LPS ionic composition To assess whether the growth medium influences the ionic content of LPS, cells were also grown in M9 minimal medium containing 0.4% glucose, and the LPS was extracted and analyzed. The levels of metal cations present in the extracted LPS from cells grown in the two media are shown in Table 2. The differences in the ionic content of the two LPS isolates reflect to varying extents the differences in the ionic and phosphorus content of the two media. From elemental analysis, the minimal medium was found to be higher in M9, Na. K and P content compared to the nutrient broth, whereas the Ca concentrations were similar (data not shown). Thus LPS isolated from cells grown in minimal medium contained higher levels of Mg and is higher in the level of total divalent cations bound per phosphorus compared to LPS isolated from cells grown in nutrient broth. This difference in divalent cation content may reflect the different phosphate content in the two LPS isolates. The phosphorus levels in the two media were dramatically different, and it has been reported that the level of phosphorylation of LPS is sensitive to the level of phosphate in the growth media (17). 59 LPS salt preparation and characterization Dialysis of electrodialyzed LPS against NaEDTA, pH 7.0, resulted in replacement of essentially all of the divalent cations with sodium (Table 4). Extensive dialysis of the sodium salt with 10 mM TRIS'HCl, pH 7.0, decreased the level of bound Na from 5 Na/LPS to approximately 1 Na/LPS (mole/mole). Thus it appears that the TRIS cation was unable to replace all of the Na. The Na and TRIS salts contained similar levels of Fe, Al and Zn. They both formed clear pellets upon centrifugation, and the resu5pended samples were very soluble and almost water clear. The calcium and magnesium LPS salts were slightly opalescent in solution and upon sedimentation formed clear to opalescent pellets. Elemental analysis of the CaLPS indicated that although the calcium bound was equivalent in bound cationic charge to that in the nLPS, the CaLPS still contained one to two sodium per LPS even after extensive dialysis (Table 4). In contrast, Mg completely replaced sodium, and the total cationic charge bound as Mg was greater than that seen in the nLPS. The TbLPS appeared as a white gel upon sedimentation but readily resuspended in water. As shown in Table 4, Tb replaced nearly all the Na as well as a significant amount of the Fe. The amount of cationic charge bound in the terbium salt was somewhat less than that bound to the MgLPS but similar to that detected in the nLPS. The ferric salt of LPS formed an orange gel which was not as soluble as the other salts but which could be dispersed by sonication. Elemental analysis of FeLPS indicated that Fe replaced all of the sodium as well as most of the other trace cations (Table 4). The level of Fe recovered in this salt, however, was exceedingly high, probably due to the formation of insoluble iron , . Fm?! N ‘I 60 hydroxylates that co-sedimented with the LPS and which could not be removed by dialysis. Ionic content of LPS from a heptoseless mutant LPS was isolated from E. 9911 strain 021f2. LPS from this mutant strain lacks the carbohydrate groups distal to the K00 units (see Figure 1).‘ Extraction of the LPS with chloroform-petroleum ether, followed by electrodialysis of the sample resulted in an acidic edLPS which was analyzed for the levels of metal cations bound. As shown in Table 5, the mutant LPS contained more Ca than Mg in the electrodialyzed form. Treatment of this heptoseless LPS with 10 mM NaEDTA, pH 7.0, resulted in nearly total replacement of the divalent cations with Na. Analysis of the levels of KDO/P in the LPS of this mutant indicated that this molecule contained approximately 4 phosphate groups. These results imply that the sodium salt of the mutant LPS has bound approximately 8 metal cationic charges, approximately 1.5 more cationic charges than were detected bound to the acidic electrodialyzed form. NaLPS from the mutant strain 021f2 also retained significant levels of iron, aluminum, and zinc. DISCUSSION The outer and cytoplasmic membranes of gram negative bacteria differ significantly in their composition and function. The results presented here indicate that within the envelope of the gram negative bacteria, the outer membrane is enriched in divalent cations (Table 1). The anionic LPS on the outer monolayer of this membrane and the underlying anionic Hi H l: “I A; r" 1‘ c1- 61 peptidoglycan layer provide multiple sites for cation interaction. Previous evidence suggests that porin proteins in the outer membrane associate with the peptidoglycan through divalent cation bridges (19). We found that the porin-LPS-peptidoglycan complex is enriched in Mg compared to LPS or to the intact outer membrane (Tables 1 and 2). Presumably either porin-LPS or porin-peptidoglycan complexes or the porin proteins themselves contain Mg binding sites. Divalent cations also appear to be critical in stabilizing the structure of pure LPS domains within the outer membrane. The major metal cations detected in nLPS were Mg and Ca (Table 2). The level of ions recovered in extracted LPS was dependent on the medium in which the cells were grown. The different levels of magnesium recovered in nLPS isolated from cells grown in the two media used in this study could be the result of differences in the LPS structure. Conversely, the varying levels of bound metal cations may result from differences in the levels of organic polyamines bound to LPS. Polyamines such as spermidine and spermine may be binding to and neutralizing part of the anionic charge on nLPS, and the levels of polyamines present may be dependent on the growth medium. Regardless of the growth medium used, electrodialysis of the extracted nLPS resulted in the removal of approximately one Mg per LPS (Table 2). This Mg binding site in the LPS is likely to reside within the K00 groups in the core polysaccharide region (see Figure 1). Electrodialysis results in a drop of the pH of the sample to approximately 4. The partial neutralization of the carboxyl group of the KDO would then allow release of the bound cation. Previous studies have shown that either Ca or Mg can readily bind to this site after neutralizing the edLPS to pH 7.0 with NaOH (19). Electrodialysis nay 62 also remove spermine and spermidine bound to nLPS (4) although the removal of polyamines cannot be detected directly by elemental analysis. Our elemental analysis of metal cations associated with LPS differs from that reported by Galanos and Luderitz (4) for LPS isolated from several Salmonella strains. First, we detected relatively low levels of monovalent cations in nLPS from g. ggll_whereas Galanos and Luderitz detected relatively high levels. This discrepancy may reflect differences in the structure of the cation binding sites in the LPS of these two organisms. Otherwise, the low monovalent ion content in our nLPS may be the result of extensive dialysis against double distilled water prior to elemental analysis. It should be noted that Galanos and Luderitz described uniform salts prepared from edLPS which was neutralized with an appropriately chosen base. Although LPS prepared in this manner was shown to have dramatically different endotoxin activity (5), it is clear from our results that these LPS samples were not unifonn salts but were heterogeneous with respect to their associated cations. Secondly, we found significant levels of iron, aluminum and zinc tightly bound to nLPS and edLPS. The presence of these ions represents a major unreported component in LPS. We believe the presence of these ions is not the result of contamination of LPS during isolation since the intact outer membrane isolated using entirely different procedures contained similar levels. Considering the profound influence iron has on the pathology of LPS, the detection of iron bound to LPS may be of considerable importance. It has been proposed that for some gram negative organisms, the ability of cells to c0mpete for iron within a host correlates with the virulence of the strain (20). In addition, it has been shown that LPS when injected into animal iron the 52 Since bacteria SUI chelators ca iron-sideroc senbrane rec affinities f constant for uPtake systes bound to LPS. cultures of E medium, growt added LPS (24 Memoir acc. The Press fortUltous. s Present in the Ciliplexes have surfaces ( 25 ) (25) sh ° Aluminu: 0"” to Inhib‘ ATP' Billdl'ng ( Sequestgr the The Log b natnaiogy of t' 63 into animals can induce hypoferremia as if the LPS were removing iron from the serum (21,22). Since the ferric ion has an exceedingly low solubility in water, bacteria such as g, 2211 produce and excrete several low molecular weight chelators called siderophores which can complex with Fe (III) ions. The iron-siderophore complex is then taken up by the cell utilizing specific membrane receptor proteins (23). These siderophores have very high affinities for ferric ions; enterochelin, for instance, has an affinity constant for Fe+3 of 1052M'1. With such potent ferric iron uptake systems, it is somewhat surprising to find high levels of iron bound to LPS. It has been observed, however, that when iron starved cultures of g. 9911 were given LPS isolated from cells grown in complete medium, growth resumed, suggesting that iron was associated with the added LPS (24). The ferric iron bound to LPS may thus serve as an iron reservoir accessible to the siderophore-receptor systen. The presence of aluminum in LPS and outer membranes may be fortuitous. However, at neutral pH, aluminum hydroxylate complexes present in the growth medium would contain a net positive charge. Such complexes have been shown to bind tenaciously to negatively charged surfaces (25) and can dramatically alter membrane physical pr0perties (26). Aluminum is potentially very toxic for the cell (27) and has been shown to inhibit enzymes such as Na,K-ATPase by binding to the substrate, ATP. Binding of aluminum to LPS on the cell surface may serve to sequester the toxic ion and prevent it from entering the cell. The LPS binding capacity for zinc may also be critical in the pathology of this endotoxin. Zinc has been shown to reduce the lethality of endotoxin-challenged mice (28). Protection was maximized when mice are intra lager Tl-‘JS, l D‘. ' (i he 6 uni 4 . 1.1 Ila: nhf; 64 were pretreated with zinc. It has also been observed that intraperitoneally administered zinc chloride prevented endotoxin-induced hyperaninoaridemia and the elevation of plasma transaminase levels (29). Thus, the ability of LPS to bind these transition metals may be important in pathogen-host reactions as well as for the bacterial cell physiology. The only other metal reported to be bound to LPS is copper. Sourek (30) found that Shigella dysenteriae LPS isolated using the hot aqueous phenol method was contaminated with copper. We have found no significant levels of copper or of manganese, molybdenum, arsenic, cadmium, chromium, mercury, lead, selenium, thallium or cobalt in any of our LPS isolates. We have shown that the divalent cations bound to LPS can be removed and defined salts can be formed. The NaLPS contained lower cationic charges bound per phosphorus than were detected in nLPS (Tables 2 and 4). The lack of equivalent charge neutralization with Na likely results in a LPS complex with a higher net negative charge compared to nLPS. This conclusion is supported by studies of the LPS head group mobility, as discussed in the following paper. Upon formation of the TRIS salt, all but approximately one Na per LPS molecule was displaced. The number of TRIS cations bound could not be determined in this study, but the overall anionic charge of the TRIS salt of LPS is likely to be equal to or greater than that of the sodium salt. Presumably the larger TRIS cation cannot associate with more anionic sites than can Na. The calcium salt also contained between one and two Na. The reason that Ca was unable to completely displace Na in the LPS is not clear. The ionic radii of Ca2+ and Na+ are nearly the same. Perhaps dialysis was not extensive enough for complete exchange. However, Mg which is substantially smaller, was able to replace all of the Na following dial nagnesium may ability of for :alciun salts charge bound t nay result frc ”BPCVEO upon 1 0f the ti band cationic divalent salt Iota2+ and ma hrsnall ions essentially al igpg CODPIEX insolubie fer: exceedingly hi ”tilll‘etatio: Thus. the tat‘IOns of SW bi”ding Sites ta ”9“le ize mirage, LPS in ore "open“ Cilia apWTTlatell or. - ms mutant ii the n was See 65 following dialysis of NaLPS with MgClz (Table 4). Calcium and magnesium may interact with the anionic sites differently since their ability of form chelation complexes is different. Both the magnesium and calcium salts of LPS from the k12 strain had higher levels of cationic charge bound than was detected in nLPS (Table 2 and 4). This difference may result from the presence of undetected polyamines in nLPS which were renoved upon formation of the defined salts. 0f the trivalent salts which we formed, the terbium salt contained bound cationic charge at levels significantly lower than that in the divalent salt forms (Table 4). The Tb3+ ion is close in ionic radius to Ca2+ and may be excluded from one or more sites which are specific for small ions. In contrast, the smaller ferric cation replaced essentially all of the other detectable cations bound to NaLPS. The FeLPS complex was relatively insoluble, due perhaps to the presence of insoluble ferric hydroxylates trapped within the FeLPS aggregates. The exceedingly high levels of iron detected in FeLPS is consistent with this interpretation. Thus, the nLPS from k12 wild type strains preferentially binds cations of small ionic radius. Furthermore, at least some of the cation binding sites in this LPS are unable to accomodate two monovalent cations to neutralize the charge originally balanced by one divalent ion. In contrast, LPS from the heptoseless mutant strain, 021f2, appeared to have a more "Open" structure which could accommodate the binding of larger cations. We have shown that the edLPS from this strain contains approximately two Ca and only one Mg per LPS. Furthermore, the Na salt of this mutant LPS appeared to neutralize more of its anionic charge with Na than was seen in the NaLPS from the parent 021 strain (Tables 4 and 5} Such I Lesa two I grasp 17001. One 5‘ he bindinl ofethanol mines in ‘ groups wit: other cati 2' Nakan‘n 371. 3' Leive Galam & Gelanr Snyge, Immun. P”gal 30c. .1 Schinc 4425.4 CheDte 66 5). Such differences in charge neutralization in the sodium salts of these two LPS molecules was also detected by differences in the head group mobility as described in the accompanying paper. One structural difference in the mutant LPS which may account for the binding of larger cations and more monovalent cations is the absence 'of ethanolamine groups (Figure 1). The presence of covalently bound amines in the LPS of the k12 strain could allow crossbridging to anionic groups within and between LPS molecules, rigidifying and restricting other cationic interactions. REFERENCES 1. van Alphen, L., Verkleij, A., Leunissen-Bijvelt, J., and Lugtenberg, B. (1978) J. Bacteriol. 134, 1089-1098. 2. Nakamura, K. and Mizushima, S. (1975) Biochim. Biophys. Acta 413, 371. 3. Leive, L. (1974) Ann. N.Y. Acad. Sci. 235, 109-129. 4. Galanos, C. and Luderitz, O. (1975) Eur. J. Biochem. 54, 603-610. 5. Galanos, C. and Luderitz, O. (1976) Eur. J. Biochem. gg, 403-408. 6. Snyder, S.L., Walker, R.I., and Monoit, J.V. (1977) Infect. and Immun. 15 (1), 337-339. 7. Prigal, S.J., Herp, A., and Gerstein (1973) J. Reticuloendothel. Soc. 14, 250-257. 8. Schindler, M. and Osborn, M.J. (1979) Biochemistry 18 (20), 4425-4430. 9. Chapter III. ._4 (3 I . o b, . l\_-, C-) a IO. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 67 Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta 555, 56-66. Westphal, 0., Luderitz, 0., and Bister, F. (1952) Z. Naturforsch. 16, 148-155. Galanos, C., Luderitz, 0., and Westphal, O. (1969) Eur. J. Biochem. 2, 245-249. Droge, W., Lehmann, V., Luderitz, 0., and Westphal, O. (1970) Eur. J. Biochem. 14, 175-184. Lowry, O.H., Roberts, N.R., Leiner, K.Y., Wu, M.L., and Farr, A.L. (1954) J. Biol. Chem. 201, 1-17. Huramic acid assay. Jansson, P.-E., Lindberg, B., Lindberg, A.A., and Wollin, R. (1979) Carbohyd. Res. 68, 385-389. Rosner, M.R., Khorana, H.G., and Satterthwait (1979) J. Biol. Chem. 254 (13), 5918-5925. Sonntag, I., Schwarz, H., Hirota, Y., and Henning, U. (1978) J. Bacteriol. 136, 280-285. Chapter III. Mickelson, P. and Sparling, P.F. (1981) Infect. and Immun. 33 (2), 555-564. Horvath, Z. (1971) In Trace Element Metabolism in Animals, Ed. C.F. Mills and S. Livingstone, Edinburgh, Scotland, 328-336. Butler, I.J., Curtis, M.J., and Watford, M. (1973) Res. Vet. Sci. 15, 267-269. Neilands, J.B. (1976) In Deve10pment of Iron Chelates for Clinical ‘Qgg, Eds. Anderson, W.F. and Hiller, M.C., NIH Publ. 76-994, 5-44. 24. 26. 27. 29. 24. 25. 26. 27. 28. 29. 30. 68 Kochan, I., Kvach, J.T., and Wiles, T.I. (1977) J. Infect. Dis. 135 (4), 623-632. Matijevic, E. (1973) J. Colloid and Interf. Sci. 43 (2), 217-245. Viersta, R. and Haug, A. (1978) Biochem. Biophys. Res. Commun. 84 (1), 138-143. Viola, R.E, Morrison, J.F., and Cleland, W.W. (1980) Biochemistry 12, 3131-3137. Snyder, S.L. and Walker, R.I. (1976) Infect. and Immun. 13 (3), 998-1000. Sobocinski, P.Z., Powanda, M.C., Canterbury, W.J., Machotka, S.V., Walker, R.I., and Snyder, S.L. (1977) Infect. and Immun. 15 (3), 950-957. Sourek, J. and Tichy, M. (1975) Zbl. Bakt. Hyg. I. Abt. Orig. A. 231, 259-268. Table —# :iéiiEilt/PT {Volar Ca/P hng h+HyP Fe/P Al/P Zn/P \ 69 Table 1 Elemental Composition of Membranes from E, coli Strain W1485F Grown in Minimal Media Element/Phosphorous (Molar Ratio) Ca/P Mg/P Ca + Mg/P Fe/P Al/P Zn/P Cytoplasmic Membrane (4 Isolates) 0.03 1 0.01 0.15 1 0.04 0.18 1 0.05 0.006 1 0.003 0.006 1 0.004 0.0014 1 0.0005 Outer Membrane (4 Isolates) 0.09 1 0.03 0.45 1 0.04 0.54 1 0.07 0.009 1 0.005 0.009 1 0.002 0.0006 1 0.0002 Ca + H Fl 3". 69 Table 1 Elemental Composition of Membranes from E, coli Strain W1485F Grown in Minimal Media Element/Phosphorous Cytoplasmic Membrane Outer Membrane (Molar Ratio) (4 Isolates) (4 Isolates) Ca/P 0.03 1 0.01 0.09 1 0.03 Mg/P 0.15 1 0.04 0.45 1 0.04 Ca + Mg/P 0.18 1 0.05 0.54 1 0.07 Fe/P 0.006 1 0.003 0.009 1 0.005 Al/P 0.006 1 0.004 0.009 1 0.002 Zn/P 0.0014 1 0.0005 0.0006 1 0.0002 EDEN: #55:“: Lo ace—Luz). Lazar; a; CZCLQ umwc~3 c—MLHW p—co .u SOL.» mm..- $O COpuanQEOU ~Mu=®=5~m N mlcos 70 .o.z oo.~ Hoe.o H woo.c ~o.o H fio.o Ho.o H No.o No.0 H wo.o eo.o H ~¢.o No.o H om.o No.o H -.o .a.z ¢.o H ~.~ mm.“ mm.o Noo.o Hoo.o H Nfio.o No.o Hoo.o H Ngo.c Ho.o Ho.o H No.0 mwo.o ¢o.o H vo.o mm.o mo.o H mm.o m¢.o no.9 H -.o NH.o coo.c H mmH.o m.o H m.~ H~.H oo.o H Ho.o Ho.o H Ho.o Ho.o H eo.o mo.o H mo.o Ho.o H m¢.o ~o.o H mm.c mco.o H o-.o max m\¢ a} a\=~ ¢\~< g\mm ¢\mz ¢\mz+mu a\mz g\ou ma; eonapa_eocboo_u mad eseeoz _ae.=_z oases—u a: o>_uaz mad eo~a_a_eaceoo.m mag m>Fumz e=_eoz oeo.cu=z «o1oam ca_o1H maogogamo;¢\a:meo—m Ezwuoz wwsv:_z Lo acopeazz emguwu :_ czogw mmmegz cpagum F—ou .m 551% mm; 90 co_upmoqsoo poucwEa—m N «pack 71 Table 3 Elemental Composition of LPS and of Porin-LPS-Peptidoglycan Complexes from g, coli Strain 021 Grown in Nutrient Broth Element/Phosphorous Porin-LPS-Peptidoglycan Native LPS ,(Molar Ratio) Complex Ca/P 0.12 1 0.01 0.116 1 0.003 Mg/P 0.61 1 0.04 0.33 1 0.01 Ca+Mg/P 0.73 1 0.05 0.45 1 0.01 Na/P 0.65 1 0.08 0.09 1 0.05 Fe/P 0.07 1 0.05 0.04 1 0.01 Al/P 0.02 1 0.01 0.01 1 0.01 Zn/P 0.010 1 0.004 0.01 1 0.00 72 .9.z .9.2 9.9 H ~.9 .9.z .9.z ~._ H 9.9 99x 9\m .9.z v~._ 99.— 99.~ «9.9 ¢~.9 ax + —auch .9.z ~9.9 H 99.9 .9.z .9.2 .9.2 .9.z 9\mw ~999.9 H -99.9 ~99.9 H "99.9 9 999.9 H 999.9 ~99.9 H ~99.9 9999.9 H 9999.9 mch 999.9 H ~99.9 99.9 H 99.9 999.9 H ~99.9 999.9 H 9~9.9 ~9.9 H ~9.9 999.9 H "—9.9 m\—< 9.9 H 9.9 999.9 H 999.9 ~99.9 H 999.9 999.9 H ~99.9 99.9 H 99.9 999.9 H 9g9.9 9\mm 9 «9.9 H 99.9 «9.9 H No.9 “.9 H 9.9 99.9 H -.9 99.9 H 99.9 axe: ~99.9 H 999.9 999.9 H 9~9.9 “9.9 H 95.9 99.9 H ~9.9 «9.9 H 99.9 999.9 H 999.9 9\9z+u9 9 999.9 H 9~9.9 ~9.9 H 95.9 999.9 H 999.9 -9.9 H 99.9 999.9 H 999.9 m\9t ~99.9 H 999.9 ~999.9 H 9999.9 ~99.9 H ~99.9 99.9 H 99.9 ~9.9 H 99.9 9999.9 H 9999.9 9\a9 Amman—cm. 99 magic; idmouupom_ my mmainh «hobo—om. n9 mad-a: «mega—cm~ 99 medium AmuuapoH_ 99 madam—sh Amman—cm. 99 magic: ~a_oa¢ co_eaH «nogogamogmxucmeo_u o_eo: ueo.ca=z e. ezocu _No =.acum :8.mStmfice33m3183eosznasu3282“eozz 73 Table 5 Elemental Composition of LPS from g, coli Strain 021f2 Grown in Nutrient Media Element/Phosphorous Electrodialyzed-LPS Na-LPS (Molar Ratio pH 4.0 (3 Isolates) (3 Isolates)_) Ca/P 0.46 1 0.05 0.019 1 0.004 Mg/P 0.18 1 0.02 0.014 1 0.001 Ca+Mg/P 0.64 1 0.07 0.032 1 0.005 Na/P 0.2 1 0.1 1.5 1 0.1 Fe/P ' 0.032 1 0.003 0.021 1 0.005 Al/P 0.03 1 0.02 0.08 1 0.02 Zn/P 0.02 1 0.01 0.0007 1 0.0002_ Total + /P 1.51 1.84 P/3 KOO N.D. 4.3 1 0.2 Th dimethy '10 char 0c:Spect l-LPS). 51 elec the mid In cont Electra Sodium p016? a salt 0f Cations 315909511 The shag; bath hea phosphOr hLPs. A CHAPTER V PHYSICAL PROPERTIES OF DEFINED LIPOPOLYSACCHARIDE SALTS ABSTRACT The electron spin resonance probes 5-doxyl stearate and 4-(dodecyl dimethyl ammonium)-1-oxyl-2,2,6,6-tetramethyl piperdine bromide were used to characterize the fluidity of the acyl chain and head group regions, respectively, of defined salts of lipopolysaccharide from 1, £911 K12 (LPS). The removal of the weakly bound divalent cations from native LPS by electrodialysis and their replacement by sodium had little effect on the mid-point of the lipid phase transition or on head group mobility. In contrast, lipopolysaccharide acyl chain mobility increased following electrodialysis. The replacement of most of the remaining cations with sodium resulted in a further dramatic increase in mobility in both the polar and nonpolar regions of LPS. Head group mobility of the sodium salt of LPS was shown to be reduced with the addition of divalent cations. Furthermore, evidence is presented which suggests that low magnesium concentrations may actually induce phase separations in NaLPS. The magnesium salt of LPS closely resembled the native form of LPS in both head group and acyl chain mobility although the cation charge to phosphorous ratio in the magnesium salt was greater than that detected in nLPS. Additional analysis of other LPS salts support our hypothesis that .74 many of propert‘ charge T'hi impact 1 the pari inactivi‘ (e.g. ti "19h par conplemc Thi and sodi RCtivat: Anc Ehdotoxi ability re.DOrteC Simply 1 this (303 filitOEleni Att CoEffl'ci 30070010 75 many of the observed differences in the physical and pathological properties of LPS salts may simply be explained by the degree of LPS charge neutralization. INTRODUCTION The physical state of LPS has been reported to have a profound impact on its endotoxicity. A strong correlation seems to exist between the particle size of enteric LPS aggregates and their ability to inactivate serum complement (1). LPS salts which have low particle size (e.g. triethylamine) do not interact with the complement system, whereas high particle size salts (e.g. sodium) are potent inactivators of complement. This pattern may not be common to all LPS. For example, the native and sodium salts of Chromatia vinosum LPS are equally effective activators of complement (2). Another indication of the importance of the physical state of LPS on endotoxicity is the effect that mild alkaline treatment has on the ability of LPS to act as a mitogen (3). Although high pH has been reported to cleave esterified fatty chains from LPS, mild treatment simply lowers particle size without apparent chemical denaturation. In this case, alkaline treated LPS of low particle size had enhanced B cell mitogenicity. Attempts to characterize LPS aggregates by sedimentation coefficients or particle sizes can be complicated by gross changes in the morphology of LPS as it is purified and transformed into defined salts. Furthermore, correlations between LPS aggregate size or shape and endotoxi internol Ne L?S salt in the h understa lipid, a Cel carried . taperat 5980 fT'O. CORSeque and resu 76 endotoxicity do not give attention to subtle differences in the intermolecular interactions between LPS within these aggregates. We have attempted to characterize the differences between various LPS salts in their head group and acyl chain mobilities. This was done in the hope that such differences may point toward a better molecular understanding of the interaction of cations with this complex anionic lipid, and its involvement in bacterial pathology. MATERIALS AND METHODS Cell growth and LPS isolation, computer analysis, and ESR were carried out as in Chapter III. All ESR results were reversible in the temperature range indicated. We observed that LPS solutions which had been frozen in solution and then thawed did not resuspend well. Consequently, all ESR experiments were performed on fresh or lypholized and resuspended LPS samples stored in solution under nitrogen at 4°C. RESULTS ESR probingiof 021 defined LPS salts Defined salts of 021 LPS were probed with either CAle or 505 to measure head group and acyl chain mobility, respectively. Spectral parameters were measured as a function of temperature. We found that nLPS was only slightly more restricted in head group mobility than edLPS (Figure 2). Even this small difference diminished with increasing temperature. In contrast, acyl chain mobility of these two samples probed with 503 was significantly different as determined when either 21. (iii functior depender at 29 1 Mg- chain Ol between The to be rn< the difi In Cont: dramatic Examiner it 15 i; Propert' Th1 anoman‘. than th. ”lit of “Diesel T'Eporte‘ S a w The hating me P1593 by binding 77 2T. (Figure 3) or order parameter S, (Figure 4) was plotted as a function of temperature. Both probes indicated that a broad temperature dependent transition occurs in native and electrodialyzed LPS with Tm at 29 1 9°C and 24 1 5°C, respectively. Mg-LPS closely resembled nLPS particularly with respect to acyl chain order (Figure 4). Head group mobility appeared to be intermediate between native and electrodialyzed LPS (Figure 2). The hydrophobic acyl chain region of NaLPS probed with 505 was shown to be more fluid than that of edLPS up to 30°C (Figure 3). Above 30°C the difference in acyl chain mobility between NaLPS and edLPS was slight. In contrast, head group mobility detected with CAT12 in NaLPS was dramatically greater than that of edLPS over the entire temperature range examined. Again a broad acyl chain transition was observed with a Tm at 15 1 12°C. Tris-LPS appeared to be similarly to NaLPS in its physical properties (Figures 3 and 4). The terbium salt of LPS from strain 021 represented something of an anomally. Acyl chain mobility of this salt was more severely restricted than that of nLPS. The spectra at all temperatures examined was like that of a rigid glass (4). In contrast, the head group mobility of TbLPS represented the most fluid of the LPS salts. Even at -5°C CAle reported an unusually fluid environment. Scatchard analysis of CAT12)binding to NaLPS The above results suggested that the interaction of CAT12 with NaLPS may be complex. Thus we examined the affinity of CAT12 for NaLPS by Scatchard analysis. Figure 5 indicates that CAT12 has 4.4 binding sites on NaLPS isolates from strain 021. In addition some of these 5 to our to a si neight of prob Madnesi 3 °C, w (Figure the far salt is AS the drainati t"‘0 TTIEC increas bIowan haLPg d which 5 log Tie that of width 0 not Cha populat aeals’ur‘e 78 these sites show strong positive cooperativity (5). this is in contrast to our previous results with edLPS which indicates that CAT12 bound to a single site in a noncooperative fashion. The CAle to LPS weight ratio used throughout was 0.074. This is equivalent to one mole of probe per 50 moles of LPS or 300 moles of LPS fatty acids. Magnesium titration of NaLPS from strain 021 The sodium salt of LPS from strain 021 probed with CAT12 at 37°C, was mixed with increasing concentrations of magnesium chloride (Figure 6). The low field line width, 1: , was used as an indicator of the formation of multiple spin probe domains. At 37°C, I: the sodiun salt is already rather broad, presumably due to lifetime broadening (6). As the magnesium concentration approached that of the NaLPS, I; increased dramatically. The observed broadening could be explained by either of two mechanisms 1) as the level of Mg increased CAT12 was forced into increasingly smaller “pure" NaLPS domains resulting in spin-spin exchange broadening; or 2) CATIZ partitioned randomly between both "pure" NaLPS domains and domains bound with Mg. Figure 6 provides evidence which strongly supports the latter interpretation. At least two distinct low field peaks could be resolved when the concentration of Mg equalled that of NaLPS. As the M9 to NaLPS molar ratio approached 2, the line width of the rigid population reached a value of about 4 Gauss and did not change even at very high Mg concentrations. In the rigid spin probe populatidn, the hyperfine splitting parameter, 2T" could be measured. CAT12 mobility was greatly reduced in this population and was nea infra). Th an incri of NaLP: explain. been sh: The nicroscc 7-0. Na aPproxin twice tl Shaped V Structur The The com; be SEEn loop bac oinimum 79 was nearly equal to that of CAT12 in the defined salt of MgLPS (vide infra). The observed decrease in I; of the rigid papulation accompanied by an increase in 2Tu is taken as evidence that the head group mobility of NaLPS has decreased upon addition of Mg. These changes can not be explained by altered polarity of the spin probe environment since.[; has been shown to be insensitive to such changes (7). Electron micrographs of 021 LPS The structure of various LPS salts was characterized in the electron microscope with samples negatively stained by sodium phosphotungstate, pH 7.0. NaLPS of 021 was characterized as long tubular structures approximately 9 nm in diameter. These tubes are frequently swollen to twice this diameter (Figure 7). There were also a number of irregularly shaped vesicles present. The Tris salt had a nearly identical structure (Figure 8). The addition of one mole of calcium per mole of NaLPS resulted in the complete conversion of the tubes to vesicles (Figure 9). Tubes can be seen at several places on most of the vesicles. The tubes appear to loop back onto the vesicles. The diameter of these tubes is equal to the minimum tube diameter seen in NaLPS before the addition of calcium. Although most vesicles were approximately 60 nm in diameter, several very large vesicles were also present. Magnesium was less effective than calcium in inducing the vesicularization of NaLPS. In fact, four moles of magnesium per mole of NaLPS were required to completely convert tubes into vesicles (Figure 10). Again, in virtually every case, there were one or more looping 80 tubes present on each vesicle. In contrast to the round shape of calciumpinduced vesicles, the magnesium induced vesicles appeared irregular. Both terbium (data not shown) and ferric chloride (Figure 11) when added to NaLPS produce similar results. When either is added at molar ratio of one, the tubular structure of NaLPS appeared to stabilize. Furthermore, there appeared to be fewer regions of swelling within tubes of either the terbium or iron treated NaLPS. The minimum tube diameter remained 9 nm. ESR probing of a heptoseless mutant of strain 021 The sodium salt of LPS from strain 021f2, a deep rough mutant of strain 021, was probed with CAle and 505. Figure 12 indicates that the heptoseless mutant produced LPS which in the sodium salt was substantially more restricted in head group mobility than the head group of NaLPS from the parent strain. Above 45°C, however, head group mobility of NaLPS from strain 021f2 increased dramatically. In contrast, SDS probing showed the acyl chain mobility and order to be virtually identical in the sodium salts of 021 and 021f2 (Figure 13). Furthermore, both sodium salts showed an abrupt change in the temperature dependence of order parameter around 35°C (Figure 14). DISCUSSION In the preceding paper we have shown that E. coli LPS extracted and purified according to standard protocol remains heavily contaminated with a variety of cations (8). With further treatment we were able to obtain asodiu NaLPS w He repo physica fact th LPS. Hh differe complern activit‘ differ-e that 0u with lo HaOH wh Th in head TESerVa DOlyam- binding could i ml‘lc‘ir r not alt 81 a sodium salt of LPS which was relatively free of these contaminants. NaLPS was then used to obtain various chemically defined salts of LPS. We report here that these specific salts are very distinct in their physical properties. These differences are interesting in light of the fact that certain cations are able to modulate the endotoxic activity of LPS. Nhen acidic edLPS is neutralized with an appropriate base, the different salts had greatly altered ability to inactivate serum complement (9). A strong correlation has been observed between endotoxin activity and the apparent aggregate weight of these salts. One important difference between our defined LPS salts and those of Galanos 33 31. is that our preparations were converted to a relatively pure sodium salt with low levels of contaminating ions. Thus, our edLPS neutralized with NaOH which may be analogous to the sodium salt of Salmonella (10). The ESR Spin probes CAT12 and 505 were used to monitor changes in head group and acyl chain mobilities of the LPS salts. Our initial reservations concerning cationic “detergents" as a spin probe of a polyanionic lipid system were confirmed. NaLPS has multiple CAT12 binding sites; some of which display strong cooperativity. This probe could induce vesicularization of NaLPS tubes at high CAT12- to NaLPS- molar ratios (data not shown). Nevertheless, CAT12 apparently did not alter either the local LPS environment or the morphology of LPS aggregates when applied at or below a probe-to-LPS weight ratio of 0.074. The relatively high degree of head group mobility in the NaLPS compared to edLPS, pH 7, may simply result from the low cation charge-to- phosphorus ratio as noted in the preceding paper (8). Incomplete charge neutralization may also be responsible for the tube-like structures seen in the ele curvature packed cor group mobi The i partially carboxyl . thought i: with amin dimension are large detected The ”Edtrall'; Ilithin it high app in mm his sin. techniq WeCul bEthieer ”Won. anal/Sc that O carbox 32 in the electron micrographs of NaLPS. The high degree of surface curvature afforded by this conformation would result in less densely packed core sugars. This is consistent with the high degree of head group mobility observed with CATIZ. The head group structure of the NaLPS from strain 021 may be partially stabilized by proton transfer complexes between phosphate or carboxyl groups and ethanolamines. Proton transfer complexes are now thought to exist at low ionic strength in virtually all phospholipids with amine containing head groups. These non covalent bonds act as a two dimensional polymeric network in phosphatidylethanolamine monolayers and are largely responsible for the observed rigidity of the head group detected with deuterium NMR (11). The observation that NaLPS of the 021 strain may not be fully charge neutralized leaves open the potential for hydrogen bonding between and within NaLPS molecules. An early model which attempted to explain the high apparent molecular weights of LPS proposed that the phosphate groups in Lipid A were diester linked (12). The existence of such LPS multimers has since been disproven using both biochemical and 31P NMR techniques (13). Nevertheless, no explanation for the high apparent molecular weight of LPS is apparent. Divalent cation cross bridging between adjacent LPS is certainly reasonable in native salts. Recently though, Schands gt_gl. (14) have reported analytical centrifugation analyses of NaLPS which indicate that its minimum aggregate weight is that of a dimer. We suggest that hydrogen bonding between ethanolamine groups and the carboxyl groups of the K00 units or the phosphate groups in NaLPS may 83 contribute to the stability of NaLPS aggregates. The presence of such bonds in the presence of divalent cations is, however, uncertain. Systematic studies of the association of cations with anionic lipids have only recently been reported (15,16,17). Mixtures of acidic and neutral phospholipids can form multilamellar tubes which are visible under a light microscope (18). Upon exposure to calcium, these tubes can twist back upon themselves into helices. We found that equimolar calcium converted NaLPS tubes into vesicles as did magnesium, albeit at higher magnesium concentration. Calcium has been reported to converted normally bilayer vesicles of sodiun phosphatidic acid into H11 phase (19,20). Furthermore, phosphatidic acid can form a calciun ionophore across lipid bilayers by forming an inverted micelle in the presence of calcium (21). In the pure sodium salt of LPS, the tube structure is thought to be H; whereby the polysaccharide chains are extended radially into the aqueous phase. The possibility remains that at low divalent cation concentrations, restricted region of inverted H11 phase exist. This may be what we have identified as loop on vesicles (Figures 9 and 10). In fact, both nLPS and NaOH neutralized edLPS consist of bilayer sheets which are regularly and heavily pitted (Figure 15 and 16). This sort of structure has also been recognized by Burnell and coworkers (22). In nLPS the pits are arranged in both random, cubic, and hexagonal arrays. The hexagonal packing of the pits was most commonly observed. The repeat distance for the hexagonal array was 144 A. Comparison of Figure 15 and 16 indicates that the loss of one magnesium through electrodialysis and its replacement with sodium results in the formation of tubes similar to what had been seen in the sodium salt of LPS. This morphological change 84 as mentioned earlier is accompanied by a dramatic increase in LPS toxicity (1). Surprisingly, head group mobility of nLPS and edLPS were very similar (Figure 2). Unfortunately, electron micrographs are often misleading and can lead to erroneous interpretations with regard to molecular organization. These may be particularly true when analyzing what are thought to be lipidic particles and cubic phase lipids (23). CAT12 proved to be very sensitive to the binding of cations to NaLPS. The addition of Mg to NaLPS produced a dramatic decrease in the mobility of bound CATIZ. This decrease in head group mobility was not the result of a change in the surface curvature of LPS aggregates. Electron micrographs showed that Mg could not induce complete vesicularization of NaLPS tubes at molar ratios below 4 (Mg:LPS). Furthermore, examination of the low field line width provided strong evidence that magnesium can promote the formation of separate domains within NaLPS aggregates when the molar ratio of Mg to NaLPS is below one. Apparently, not all cations altered the mobility of the LPS head group in the same way. The similarity between native and electrodialyzed LPS of strain 021 detected with CAT12 suggested that the removal of one weakly bound divalent cation and its replacement by sodium does not significantly alter the packing of the core sugars. Experiments described in the preceding paper suggest that the most weakly bound divalent cation in nLPS from strain 021 is magnesium which is presumably located in the K00 core. Curiously, the divalent cations induced vesicles of NaLPS almost without exception but the vesicles retained regions of tubular LPS which 85 loop back onto the vesicles. The minimum diameter of these tubular regions was about 9 nm. The bilayer thickness of LPS is 10 nm detected by X-ray diffraction analysis (25). For this reason, we believe that the tubes and loops are regions of LPS that are in a typical H1 conformation (24). Ferric and terbium chloride seemed to relax the twisted tubes of NaLPS into long smooth tubes. This may partially explain the exceedingly high head group mobility detected by CAle and the very low acyl chain mobility detected with 505 in TbLPS. Conversely, the tight packing of the acyl chains of TbLPS may not allow CAle to completely partition into TbLPS. The broad phase transition observed in nLPS with both CAle and 505 was shifted down 5°C after electrodialysis and the temperature over which the transition occurred, narrowed substantially. As evidenced by fluorescent probing and X-ray diffraction analysis, a broad order-disorder transition occurred at 22 t 5°C in nLPS of the‘g..ggli strain B/r (25). Neither the midpoint nor width of the observed transition in this smooth strain was altered by electrodialysis. X-ray diffraction experiments demonstrated that the phase transition temperature of nLPS from g. 3911 B grown at 37°C to be 25°C (26). Below 25°C the diffraction data suggested that the LPS acyl chains were still more disordered than acyl chains of phospholipids. In contrast, a comparison of the order parameter calculated for 505 in extracted phospholipid, (Figure 4) and nLPS showed just the opposite results. nLPS acyl chains are reported to be substantially more ordered than phOSpholipids. This apparent discrepany may be explained by the 86 difference in time scales of detection for the two techniques rely. X-ray diffraction analysis may be considered a static measurement of the hydrocarbon chain order. Order determined by electron spin resonance probes is also a static parameter; however, the technique measures the distribution of acyl chain conformations in the time range of 10'9 to 10‘7 sec. Thus, our order parameter measurements indicate that the motion of the hydrocarbon interior of nLPS aggregates is greatly decreased relative to that of phospholipids. In this context, it is interesting to note that lateral diffusion of LPS determined by the technique of flourescence recovery after photobleaching (27) indicated that phospholipids diffuse much faster relative to one another than does LPS (DLp5/DPL . 0.6). Thus, LPS domains in the outer membrane are most accurately thought of as more restricted than phospholipid domains regardless of the conformation of the acyl chains. Our laboratory has previously shown that outer membranes extracted from g, 5911 grown at different temperatures have altered physical properties (28). Most notably the beginning and end of the membrane lipid transition seen with SDS correlates well with the limits of the growth temperature. The acyl chains of both phospholipids (26) and LPS (29) have been reported to change in response to growth temperature. We have found no detectable changes in either acyl chain mobility or transition temperature (data not shown) in response to changes in growth temperature in the range of 12 to 43°C. The similarity between the midpoints of the phase transitions of nLPS and outer membrane from g. 2211 grown at 37°C, therefore, may be coincidental. Emmerling 33 31. (25) reached a similar conclusion in their smooth E. coli strain. 87 Finally, a comparison of NaLPS of strain 021 and the heptoseless mutant strain 021f2 shows that the two have virtually identical acyl chain mobility. The Lipid A moiety of the two LPS isolated from bacteria grown under the same condition has been reported to be the same (30). Consistent with the elemental analysis presented in the preceding paper, NaLPS from strain 021f2 has less head group mobility than the NaLPS from the parent strain. Furthermore, electron micrographs (data not shown) indicate that NaLPS from strain 021f2 is vesicular. 1. 2. 3. 4. 5. 6. REFERENCES Galanos, C. (1975) Z. Immun.-Forsch. Bd. 152, 214-229. Hurlbert, R.E. and Hurlbert, I.M. (I977) Infect. and Immun. 16 (3), 983-994. Goodman, G.w. and Sultzer, B.M. (I977) Infect. and Immun. 17 (1), 205-214. ' Schreier, S., Polnaszek, C.F., and Smith, I.C.P. (1978) Biochim. Biophys. Acta 515, 375-436. Cantor, C.R. and Schimmel, P.R. (1980) The Behavior of Biological Molecules, Part II, H.H. Freeman, and Comp., 850-861. Mason, R.F., Giavedoni, E.B., and Dalmasso, A.P. (1977) Biochemistry 16 (6), 1196-1200. Mason, R.F. and Freed, J.H. (1974) J. Phys. Chem. lg, 1321. Chapter IV. . Galanos, 0. and Luderitz, 0. (1976) Eur. J. Biochem. §§, 403-408. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 88 Galanos, 0. and Luderitz, 0. (1975) Eur. J. Biochem. 54, 603-610. Browning, J.L. (1981) Biochemistry 20, 7144-7151. Nowotny, A. (1961) J. Am. Chem. Soc. §§J 501-503. Rosner, M.R., Khorana, H.G., and Satterthwait, A.C. (1979) J. Biol. Chem. 254 (13), 5918-5925. Shands, J.w., Jr. and Chun, P.H. (1980) J. Biol. Chem. 255 (3), 1221-1226. Eisenberg, M., Gresalfi, T., Riccio, T., and McLaughlin, S. (1979) Biochemistry 18 (23), 5213-5223. Lau, A., McLaughlin, A., and McLaughlin, S. (1981) Biochim. Biophys. Acta 645, 279-292. Mauser, H., Darke, A., and Phillips, M. (1976) Eur. J. Biochem. 62, 335-344. Lin, K.-C., Heis, R.N., and McConnell, H.M. (1982) Nature 226, 164-166. Verkleij, A.J., DeMaagd, R., Leunissen-Bijvelt, J., and DeKruijff, B. (1982) Biochim. Biophys. Acta 684, 255-262. Vasilenka, 1., DeKruijff, B., and Verkleij, A.J. (1982) Biochim. Biophys. Acta 684, 282-286. Serhan, C., Anderson, P., Goodman, E., Durham, P., and Neissmann, G. (1981) J. Biol. Chem. 256, 2736-2741. Burnell, E., VanAlphen, L., Verkleij, A., Kruijff, B., and Lugtenberg, B. (1980) Biochim. Biophys. Acta 521, 518-532. Hui, S. and Boni, L. (1982) Nature 226, 175. Cullis, P.R. and DeKruijff, B. (1979) Biochim. Biophys. Acta §§2, 399-420. 25. 26. 27. 28. 29. 30. 89 Emmerling, 6., Henning, V., and Gulik-Krzywicki, T. (1977) Eur. J. Biochem. 18, 503-509. Nakayama, H., Mitswi, T., Nishihara, M., and Kito, M. (1980) Biochim. Biophys. Acta 601, 1-10. Schindler, M., Osborn, M.J., and Koppel, D.E. (1980) Nature 283, 346-350. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta 555, 56-66. . Kasai, N. (1966) Ann. N.Y. Acad. Sci. 133, 486-507. Gmeiner, J. and Schlecht, S. (1980) Arch. Microbiol.1121, 81-86. 90 0 I‘ T.I.-101 " Ii :10] - l‘ [cm] is“ 0 n W m E." n, l.‘ 9 I u E ‘.‘ w n) in (C‘ u‘ u- v 4. 7w "" v . VA " u H -T 3 e in E 'b‘ ' v “ 0- VA ‘1 ‘ a g . 7'. 74 I h- e. E: [n alt-i I} 36'.“ 3’10'4! Fir-lib n .1- in il‘ ‘ 1 V: i , :7. L A- _I 02le DZI coli K12 strains 021 (rough) and 021f2 o I-I-Ii ‘4- O Q) L 3 4.) o UM =0. 8...! 4.9 (”A .2 0'0 0: u—cO L 0 LG. 30 m0 w-‘U th 91 52- 50*- l J ono I I I I I e 20 30 4O 50 60 Temperature,°C ‘:i gure 2. The temperature dependence of the hyperfine splitting Parameter, 2T", of CAT1 bound to defined salts of LPS from _E_. coli strain 021. The LBS salts included native LPS ( O ), e'ectrodialyzed LPS neutralized with NaOH ( I ), NaLPS ( A ), TRIS-LPS (in ), MgLPS ( A ), and TbLPS ( o ). All LPS salts were resuspended n double distilled water at an approximate concentration of 0.5 mM and Mixed with approximately 10 pM CAT g (final concentration). The hyperfine splitting parameter of CA 12 is inversely proportional to the head group mobility. 92 0 ' A 0 I0 20 3O 4O 50 60 Temperature, °C Figure 3. The temperature dependence of the hyperfine splitting Parameter 2T.., of 505 bound to defined salts of LPS from _g. coli . Strai n 021. The LPS salts studies were native LPS (0 ), electrodialyzed LPS neutralized with NaOH (I), NaLPS (A), TRIS-LPS (U), MgLPS (A), and TbLPS (o ). All LPS salts were suspended in double distilled water at an approximate concentration of 0.5 M and mixed with approximately 5 PM 50$ (final concentration). The hyperfine splitting parameter of SDS is inversely preportional to the lipid acyl chain fluidity. 93 o #6 36 4'6 Temperature, °C F'i gure 4. Temperature dependence of SOS order paramater in native (O ), e1 ectrodialyzed (I), sodium (A), magnesium (A), and TRIS (D) LPS. LPS to 50$ molar ratio was 50. 94 we co_aucam a mm s: c we: m=v=o_a_ueoa my . .copumeucmocou waogq wo co_pmgacmocou ma; :- mu_3.mwpa5mm co uosm an umeamoos :3 8.3 be 253 c. :8 .w 8.: :3 $32 2; 3:: umcmmuv ma; um~a_e_uoguuopm op m:_n:_n N p_emee on cue mxeee or» we; ecu +mz Leweswaeo x_eueswxeeaae p< .eeeee mw upon: me acescegw>=e eeegq «gm we xuwecemegeueg ocwmeeeecw meueeweew xeea e—eww sew use we spew: mew— mew .prm: we meaneEe mcwmeececw saw: eeueguwu we: Acewueeuceeeee peewwv Hw) and 021f2 (ID). The Probe to weight ratio was 0.074. 103 (IT .6 2'0 GT Temperdiue, °C Figure 14. Temperature dependence of 505 order - parameter, 5, in the :gcsiignolizs salts of 021 (O) and 021f2 (O). The probe to weight ratio F‘i gure 15. Electron micrograph of nLPS from E. coli strain 021, stained W1th sodium phosphotungstate, pH 7.0. i i i l i l l Figgre 16. Electron micrographs of edLPS from E. coli strain 021, StaIned with sodium phosphotungstate, pH 7.0. CHAPTER VI ELECTRON SPIN RESONANCE PROBING OF LIPOPOLYSACCHARIDE DOMAINS IN THE OUTER MEMBRANE OF ESCHERICHIA COLI SUMMARY Two electron spin resonance probes, an anionic probe 5-doxyl stearate and a cationic probe 4-(dodecyl dimethyl ammonium)-I-oxyl- 2.2,6,6-tetranethyl piperdine bromide, were used to analyze the physical PPOperties of the phospholipid and lipopolysaccharide domains in the Outer membrane of Escherichia coli. Scatchard analysis of the binding of the cationic probe to lipopolysaccharide and to phospholipid indicated that this probe has a 5 fold greater affinity for anionic HPopolysaccharide than for phospholipid. In the intact outer membrane this cationic probe likely associates with the lipopolysaccharide cOntaining outer monolayer. The temperature dependence of the nobility 01" the cationic probe in the outer membrane indicates that a structural tY‘ansition occurs at 9°C in the outer monolayer. A similar 9°C transition was detected in the outer membrane using 5-doxyl stearate. This anionic probe has been shown to partition into the phospholipid em"iched domains. A porin-lipopolysaccharide-peptidoglycan complex pr°bed with the cationic probe also was shown to undergo a thermally ind“(Zed structural change at 9°C. In contrast, purified 105 106 lipopolysaccharide was shown to have structural transitions at 20°C and 40°C. It is proposed that a structural rearrangement of the intact outer membrane occurs at approximately 9°C in both the lipopolysaccharide and phospholipid domains of the outer membrane. Furthermore, this structural transition appears to be highly dependent on lipid-protein interactions. A second thermotropic transition that occurs in the outer membrane at approximately 40°C to 42°C appears to result mainly from changes in the lipopolysaccharide domain structure. INTRODUCTION The outer membrane of gram negative bacteria is a highly asymmetric structure. The outer monolayer is composed of anionic lipopolysaccharide (LPS) and protein, whereas the inner monolayer contains phospholipid and protein. Moreover, phospholipids and LPS isolated from Escherichia coli upon reconstitution are reportedly unable to coexist in a single phase (1). Within the intact outer membrane of Salmonella, spin labeled fatty acids have been shown to preferentially partition into the phospholipid danains of the inner monolayer (2) perhaps as a result of the greater fluidity of these domains compared to that of the LPS domains, and to charge repulsion between the anionic head groups of the fatty acid probe and LPS. Previous analysis of extracted phospholipid from the outer unmbrane of 37°C grown E. £911 indicated that the phospholipid undergoes a structural transition at approximately 4° and 24°C (3). Results reported here demonstrate that the structural transition of purified LPS begins at 20°C and ends at 40°C. He also found that the cationic electron Spin resonance (ESR) probe 4-(dodecyl dimethyl amnonium)-1-oxy- 107 2,2,6,6-tetramethyl piperdine bromide (CAle) preferentially partitions into LPS compared to phospholipid domains. Applying this cationic probe and the acyl chain probe 5-doxyl stearate (508) we have evidence to suggest that both lipid domains located on either side of the outer membrane undergo a cooperative tenperature dependent rearrangement. The onset of this rearrangement at 9°C in both lipid domains of the intact membrane appears to be determined mainly by protein-lipid interactions. In contrast, the end of the thermotropic phase change in the intact outer membrane at around 42°C may be determined by rearrangements within the LPS domains. MATERIALS AND METHODS Cell:growth and membrane isolation Cultures of E. 3911 strain w14es F' were grown at 37°C in M9 minimal medium supplemented with 0.4% glucose (final concentration). Cultures were harvested, and the cytoplasmic and outer membranes were isolated as described previously (4). The degree of purity of the separated membranes equalled or exceeded that previously described (5). .Lipid isolation LPS was extracted from whole cells with aqueous phenol (5). The extracted LPS was sedimented at 78,000 g for 60 min and washed twice with double distilled water. The resuspended sample was dialyzed extensively against double distilled water yielding the native LPS (nLPS) product. Electrodialysis of nLPS resulted in an acidic sample (pH approximately 4) Which was neutralized with NaOH to pH 7 (edLPS). Phospholipid was 108 extracted from isolated outer membranes with chloroform/methanol (2/1) as previously indicated (6). Porin-LPS-peptidoglycan isolation A porin-LPS-peptidoglycan complex was isolated from cells grown and harvested as described above. Cells were washed in 10 mM N-2-hydroxyethylpiperiazine-N'-2-ethanesulfonic acid (HEPES), pH 7.5, and lysed using a French pressure cell at 12,000 psi. Total membranes were treated with 20 ug/ml deoxyribonuclease I (Sigma Chemical Co.), pelleted and washed twice with double distilled water. The porin-LPS- peptidoglycan complex was extracted with 1% sodium dodecyl sulfate (SDS). The protein composition of the complex was characterized on SDS polyacrylamide slab gels (7). Assays Succinate dehydrogenase activity was determined by the method of Osborn gt_gl, (8) to quantitate cytoplasmic membrane levels. Lipopolysaccharide levels were quantitated by analysis of 3-deoxy-D- manno-octulosonic acid (K00) content (9). Protein concentration was assayed using the procedure of Lowry g§_gl. (10). Peptidoglycan levels were quantitated by measuring the muramic acid content of acid hydrolyzed peptidoglycan (11). Spin labeling The fatty acid spin probe 505 was dissolved as a 30 mM solution in absolute ethanol. An aliquot of the probe was dried onto the bottom of a clean test tube under a stream of N2, and the sample was added at room 109 temperature with mild vortexing. The spin label comprised less than 0.1% of the lipid weight. The CAle was always less than 0.3% of that of the lipid acyl chains present in the sample. The partitioning of CAT12 into lipid was calculated as previously described (12). All electron spin resonance experiments were carried out as described earlier using a Varian Century Line electron spin resonance spectrometerm X-band, model E112 (13). The temperature dependence of the order parameter, S, and the hyperfine splitting parameter, 2T", was analyzed in terms of linear components using a linear regression program developed for transition point analysis (14). Discontinuities in plots of the temperature dependence of 2T" and S are interpreted to reflect structural rearrangements of the lipid domains in which the probe is bound (15). RESULTS Membrane spin labeling Two electron spin resonance probes, CAle and 50$ (Figure 1), were used to monitor intact outer membranes, isolated LPS and porin-LPS-peptidoglycan complexes. Negatively charged 50$ likely partitions into those lipid domains which contain lower levels of anionic groups. The nitroxide group probes acyl chain mobility. In contrast, the positively charged CAle is thought to partition into more anionic lipid domains and probe the motion within the lipid head group region. ESR spectra indicate that the spin label movement in the head group region is sensitive to temperature (Figure 2). At higher temperatures broadening of the low field peak suggests that the probe 110 probe partitions into more than one environment. Figure 3 shows the temperature dependence of 2T" reported by CATIZ and by 505 (Figure 4) bound to intact outer and cytoplasmic membranes. The 2T" parameter can be calculated from spectra of CAT12 bound to outer membranes only at temperatures below 30°C. At low temperatures, the CAle in the outer membrane appears to experience a slightly more rigid environment than does 505. In contrast, the CAle probe in the cytoplasmic membrane appear to reside in a similar or slightly more fluid environment. Lipid spin labeling The partitioning of CAle into purified phospholipid and LPS was analyzed at 37°C as a function of probe concentration in order to assess the preferential association of CAT12 with specific domains in the outer membrane. From Scatchard analysis of CAle binding (Figure 4) it can be concluded that this cationic probe has approximately 5 fold greater binding affinity for LPS than for phospholipid. The binding of this probe to both types of lipids appears to be non-cooperative and at a single type of site. Similar Scatchard analysis of SOS binding is not possible since micelles form at high probe concentrations. Plots of the temperature dependence of the spectral parameters, 2Tn and S, for 505 and CAT12 bound to purified LPS, indicate discontinuities at 20°C and 40°C (Figure 5). These results suggest that structural transitions occur at these temperatures which can be detected within both the head group and the acyl chain regions. Comparison of ESR spectra of CAT12 bound to isolated LPS and intact outer membrane measured at 37°C indicates that the head group region in the pure LPS 0 111 appears to be more rigid than that of LPS in the intact membrane (data not shown). Porin-LPS-mucopeptide labeling Isolated porin-LPS-mucopeptide complexes were shown to contain approximately 70% protein, 15% LPS and between 5 and 15% peptidoglycan by weight. SDS polyacrylamide gels indicated that the protein was comprised of approximately equal amounts of porins 1a and 1b with only minor contamination from other membrane proteins (data not shown). These complexes were probed with CAT12, and the plots of the temperature dependence of 2T" indicated a discontinuity at 9°C (Figure 6). The mobility of the probe within this complex was similar to that of the probe bound to intact outer membranes when compared at temperatures above 9°C. At temperatures below 9°C, the LPS head group appeared to be more rigid in the isolated porin-LPS-mucopeptide complex than LPS within the intact outer membrane. DISCUSSION Thermotropic structural rearrangements have been shown to occur in the lipid domains of biological membranes, and these transitions usually occur over a broad temperature range (15). These changes often affect membrane enzyme activities and other membrane functions (16). It is also known that the presence of membrane proteins and the composition of the lipid's acyl chains dramatically affect the temperature of the membrane's thermotropic rearrangement of the lipid components and the lipid acyl chain mobility. In this study we set out to determine whether specific 112 nembrane components modulate the lipid domain structure in the outer membrane of E, 5911, Previously we had shown that SOS labeled outer membranes undergo thermotropic structural transitions at temperatures different from those of the extracted phospholipids (3). Outer membranes from cells grown at 37°C labeled with 505 appear to undergo transitions at 9° and 42°C whereas the phospholipid extracted from the same membrane undergoes structural transitions at 4° and 24°C (3). Since 503 is reported to probe the phospholipid containing domains of the outer membrane (2), we believe that other components within the outer membrane are modulating the temperature of this transition. We report here the use of a cationic electron spin resonance probe, CATIZ. to analyze the structure of the outer membrane. We have shown that CAle preferentially partitions into LPS more readily than into phospholipids, and in the intact outer membrane this probe likely partitions with the same specificity. Using this probe to analyze the structure of LPS in the intact membrane we have found that LPS associated with protein in the intact membrane and in the isolated porin-LPS-peptidoglycan complex has a thermotropic transition beginning at 9°C. In contrast extracted LPS probed with either CAle or 503 undergoes a thermotrOpic transition starting at 20°C and ending at 40°C. Since the LPS domains in the intact outer membrane, probed with CAle, appear more fluid compared with pure LPS, it is proposed that proteins when associated with LPS disorder the head group region. Such disordering of the LPS may also cause the decrease in the temperature of the beginning of the thermal transition phospholipids have been reported when the lipid is mixed with membrane proteins (17). The end of the thermal transaction in the intact outer membrane can only be detected 113 using the anionic probe 505 which presumably probes the phospholipid domains (2). The temperature of the end of the transition of the intact membrane detected with SDS is 42°C, very similar to the end of the transition of the isolated LPS (approximately 40°C). Perhaps temperature dependent structural rearrangements which occur in the LPS domains of the intact membrane affect the structure of the phospholipid domain structure as well. Since in the intact outer membrane both SDS and CAle detect a 9°C structural transition, both lipid domains may be coupled in their structural rearrangements, perhaps through the structure or packing of transmembrane proteins. ,In the porin-LPS-peptidoglycan complexes, CAT12 detected a more rigid environment below 9°C than was detected with the same probe in the intact outer membrane. The domains being probed in these two structures are likely to be similar in composition. However, in the porin-LPS-peptidoglycan complex the LPS present is probably tightly bound to the protein. We have shown that this porin-LPS complex is enriched in divalent cations compared to the intact outer membrane (Chapter IV), and these cations may be involved in tightly binding LPS to the porin proteins. The cationic probe CAle detected a slightly more rigid environment in the outer membrane than did SOS (Figure 3). These differences are more difficult to interpret. First, the nitroxide free radical in the CAle probe is located in the head group region of the probe and probes the mobility of the head group of the lipid into which it partitions. In contrast, the nitroxide free radical in 505 is located five carbons into the acyl chain of the probe and is likely probing the mobility of the lipid acyl chains at this level. Secondly, as previously 114 suggested, these two probes are probably probing different lipid domains within the outer membrane which may have dramatically different head group and/or acyl chain mobility. 1. 3. 4. 6. 7. 8. 10. 11. 12. REFERENCES Takeuchi, Y. and Nikaido, H. (1981) Biochemistry gg, 523-529. Nikaido, H., Takeuchi, Y., Ohnishi, S.-I., and Nakae, T. (1977) Biochim. Biophys. Acta 46E, 152-I64. Janoff, A.S. gtflal,, manuscript in preparation. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta.1§§§, 56-66. Galanos, C., Luderitz, 0., and Westphal, O. (1969) Eur. J. Biochem. 2, 245-249. Folch, J., Lees, M., and Stanley, G.H.S. (1957) J. Biol.Chem. ggg, 497-509. Ames, C.F.-L. (1974) J. Biol. Chem. ggg, 634-644. Osborn, M.H., Gander, J.E., Parisi, E., and Carson, J. (1972) J. Biochem. E41, 3962-3972. Droge, H., Lehmann, V., Luderitz, 0., and Westphal, 0. (1970) Eur. J. Biochem. 14, 175-184. Lowry, 0.H., Rosebrough, M.J., Farr, A.L., and Randal, R.J. (1951) J. Biol. Chem. 193, 265-275. Hadzija, 0. (1974) Anal. Biochem. g9, 512-517. Melhorn, R.J. and Packer, L. (1979) Methods in Enzymology, Vol. LVI, 515-526. 13. 14. 15. 16. 17. 115 Janoff, A.S., Coughlin, R.T., Racine, F.M., McGroarty, E.J., and Vary, J..C (1979) Biochem. Biophys. Res. Commun 89 (2), 565-570. Chapter II. Steim, J.M., Edner, 0.J., and Bargoot, F.G. (1968) Science.1§2, 909. Linden, C.O., Wright, K.L., McConnell, H.M., and Fox, C.F. (1973) Proc. Natl. Acad. Sci. USA 70 (8), 2271-2275. Boggs, J.M., Stamp, 0., and Mascarello (1981) Biochemistry gg, 6066-6072. 116 CAT}; 505 Figure 1. The chemical structure of the two electron spin resonance Probes CAle and 505. 117 10 lComte“ Figure 2. Electron spin resonance spectra of CATlfi bound to intact outer membranes from cells grown at 37°C in 10 mM epes, pH 7.5. The membrane concentration was approximately 20 mg/ml protein and probe was added at a 1:300 (probe; lipid acyl chain) dilution. The spectra were recorded at 5° and 18°C. 118 65 ' 60_ . 2T" 55 _ - 50- \ ‘ 45 . . . . . . . . . 1__,_: O 5 IO I5 20 25 30 35 4O 45 50 TEMPERATURE ('C) Figure 3. The temperature dependence of the hyperfine splitting parameter, 2T11, of CAle bound to outer membranes (0) and cytoplasmic membranes (0) from E. coli WI485F grown at 37°C. 505 results (solid line) are for outer membrane (top) and cytoplasmic membrane (bottom) (ref. 4). 119 400 CAle BINDING K‘LPS/ KP‘. = 4.7 [B] [F] 200 130 100 200 300 400 500 600 [3] uM Figure 4. Scatchard analysis of CAT 2 binding to electrodialyzed LPS (I) and to phospholipid (I) from t e outer membrane of E. coli W1485F grown at 37°C. CAle partitioning was measured at 37°C on samples of LPS and phospholipid at 500 M concentration as a functibn of probe concentration. Figure 5. The temperature dependence of the hyperfine splitting 120 )- TEMPERATURE (“CI parameter, ZTII’ and the order parameter, 5, (Insert) of CAle (0) and 50S I) bound to electrodialyzed LPS. L‘L''''‘'""''''"'_'""''“"""""""""""o 5 l0l5 20253035404555 121 sop . . ‘ 0 IO 20 Temperoiure, °C Figure 6. The temperature dependence of the hyperfine splitting Pararneter,.2T}1 of CAT12 bound to porin-LPS-peptidoglycan ”le exes iso ated from E..coli W1485F grown at 37°C. CHAPTER VII IONIC BRIDGING WITHIN LIPOPOLYSACCHARIDE AGGREGATES. AN ELECTRON SPIN RESONANCE, pH TITRATION STUDY ABSTRACT The head group packing of lipopolysaccharide (LPS) from Escherichia £211 021 and 021f2 was examined under conditions of varying hydroxide ion concentration. It was concluded that fixed charges in the core region contribute significantly to the surface potential of LPS aggregates, and that the polysaccharide core is a region of high polarity where the pK of ionizable groups may be altered substantially. Using the electron spin resonance probe CAT12 as a reporter of the head group mobility, we analyzed the structure of sodium salts of LPS from both strain 021 and its heptoseless mutant strain 021f2. At neutral pH, the head group mobility of the magnesium salt of LPS from strain 021 was substantially less than that of the sodiun salt. However, at high pH (pH 12) the two salts were approximately equally fluid. The changes at high pH were completely reversible precluding the possibility that alkaline hydrolysis of LPS had occurred. We have interpreted these results as an indication that ionic bridging stabilizes pure LPS aggregates and LPS domains in membranes. 122 123 INTRODUCTION The charged nature of enteric LPS has already been amply discussed here (1) and in the literature (2). Ionizable groups such as phospho mono and diesters, carboxyl and amine groups as well as polar sugars all contribute to produce a complex three dimensional zone of high charge density. The capacity of this zone to bind counterions (e.g., metals, antibiotics) and form ionic bridges for long range cooperative interactions is unquestionably defined by the mutual interaction of charges within the zone. In general, the pK of an ionizable group is strongly influenced by the polarity of its environment. When the ionizable group is fixed to a surface, it alters the environment of its neighbors by generating a surface potential ‘1 . When the charges on a surface are uniformly and closely packed in a low ionic strength solution, they change the surface pH, pHs (3). The relationship described by equation 1 assumes that protons behave as a diffuse double layer of counterions. equation 1 pHs = prulk + ¢ "P. 2.3 kT The constant e is the electronic charge (-1.61 x 1049 coul.), PHbulk is the bulk pH, K is Boltzmann's constant (1.38 x 10'23 Joules/°K) and T is the temperature in degrees Kelvin (310 for our purposes). Although this simple relationship is not fully applicable to a three dimensional zone of charge as is expected in the core region of LPS, the 124 usefulness of more complex models are limited by the uncertain position of charges. Overall, the relationship described by equation 1 does provide a useful insight into the balance of electrostatic forces in the LPS core. Romeo and coworkers (4) have given the cross sectional area of Salmonella typhimurium LPS in a monomolecular film at an air-water interface to be 232 e 5041 At the time it was thought that LPS was a polymer having 13.5 acyl chains. The corrected cross sectional area (assuming 6 acyl chains per LPS) is 103A2. From this we can compute the surface potential contribution from the three phosphorus of lipid A i and the three carboxyl groups of the inner core. The surface potential is: equation 2 M103 - 0.0506 volts ln (272 0') Z where Z = 1 when we assign a single charge to each group and 0' is the surface charge density (6/103A2). The contribution of just these inner core groups to the surface potential is -140 millivolts. From this we can calculate the change in pHs. Substituting “S into equation 1 yields an increase of about 2.3 pH units at the hydrophobic-hydrophilic interface of LPS. Thus, all ionizable groups distal to this interface may be expected to have their pK's shifted up by as much as 2.3 units. If the interface is reasonably planar, the surface potential influence on ionizable groups should fall off as 1/r. Although this effect may seem large, it is actually very typical of the field influence of fixed charges on monolayers. The pKa of acetic 125 acid in aqueous solution is 4.75, while the apparent pKa of arachidic acid micelles is greater than 9 (5). In this case, the close packing of fatty acids produced a ‘E of more than -200 mV. Similarly, the phosphodiester group of phosphatidylinositol has a higher pKa in monolayers than a phosphodiester free in bulk solution (6). The simple relationship described above can be greatly complicated by the presence of neighboring polar groups and counterions. The influence of the carboxyl group of phosphatidylserine on its amine group is quite strong. Although the pKa of the phosphodiester group of both phosphatidylethanolamine and ph05phatidylserine are the same (pKa-4.2), the pKa of the amine group in phosphatidylserine is 1.6 units higher than that of phOSphatidylethanolamine [pKNH2 7.8 and 9.4, respectively, (7)]- The influence of counterions depends strongly on the nature of their binding. Neither sodium nor potassium alter the sz of most ionizable groups on phospholipids by much more than half a pH unit (6). In contrast, monovalent detergents such as sodium deoxycholate can greatly influence surface pH by virtue of their ability to intercalate between lipids and thus reduce their overall surface charge density. More important in the case of LPS are the effects of tightly bound polyvalent cations. McLaughlin and coworkers (3) found that when bimolecular lipid membranes were formed out of acidic lipids (e.g., phosphatidylserine) the observed decrease in the surface potential due to the binding of Sr2+ and Ba2+ was well described by the diffuse double layer model. When either Ca2+ or Mg2+ were used, however, it was necessary to consider some specific lipid-ligand interactions of the Stern type. Both Ca2+ and Mg2+ were able to effect a decrease in membrane 126 potential at a much lower concentration than either Sr2+ or Ba2+ by coordinating with anionic sites rather than simply screening those sites. Emmerling 33 El- (8) observed a single pH titratable group on the native LPS of a smooth strain of E. 9911 B/r. The pH of the neutral product was approximately 8.0 and the lone titratable group in the alkaline range had an estimated pK of about 8.5. They assumed that the group they were titrating was an ethanolamine of the inner core of lipid A. The use of native LPS which undoubtedly contained a variety of polyvalent organic (amine containing) and inorganic cations certainly complicates the interpretation of their results. Furthermore, most published accounts of the chemical composition of enteric LPS agree that either phosphomono esters or diesters are the most prevalent ionizable groups. In addition three carboxyl groups are also present in the core of enteric LPS (9). Thus, it would be difficult to rationalize the pH of a neutral LPS suspension being just below the pK of an amine residue. Olins and Warner (10) working with Azotobacter vinelandii have reported the pH titration of extracted LPS. Unfortunately, during isolation they treated their LPS with 0.1 M HCl (pH 2.2) which likely hydrolysed their LPS to lipid A and the polysaccharide chains were probably removed in subsequent purification. Interestingly though, they observed three titratable groups in their NaEOTA washed sample (pKl = 1.3; pKz = 2.5; pK3 = 6.5). With what is known now of the chemical structure of lipid A, the first two ionizable groups might well have been phosphomono or phosphodiester and the last ionizable group a phosphomonoester. 127 The cation binding results presented in chapter III suggest that divalent cations may lower the surface potential of LPS from E. ggflj, This, in turn, may dramatically influence the electrostatic forces governing LPS-LPS self association. We report here evidence for the existence of multiple pH titratable groups above pH 7 in a suspension of NaLPS from E. ggli_021 and of a single titratable group in the NaLPS from the heptoseless mutant 021f2. RESULTS AND DISCUSSION Our assumption is that at neutral pH the NaLPS from both strains 021 and 021f2 is close to but below the second pK of phosphomonoester groups. Although an accurate measurement of the pH of a highly viscous sample such as NaLPS is not possible, we have estimated by dilution that the pH of both 021 and 021f2 sodium LPS is close to 7. Figure 1 shows the change in the rotational correlation time"?: of the ESR spin probe CAle bound to NaLPS from E. 9911 021 LPS as a function of increasing hydroxide ion concentration. It is apparent that even below the actual LPS concentration, the addition of hydroxide ions has measurably increased head group mobility. We interpret this as an indication that at pH 7 021 NaLPS is very close to the pK of one of its ionizable groups. As the added hydroxide ion concentration exceeded the LPS concentration, the rotational correlation time did not increase with further hydroxide ion addition. Apparently, no change in the net charge of the LPS core occurred until pH 11 (calculated) where head group mobility again increased dramatically. Further increases in head group mobility are evident at pH 12 and 12.5 (calculated). 128 The reported pK of ethanolamine in solution is 9.4 (11). Taking the surface potential contribution into account, amine group titrations in the pH range 11 to 12 are quite reasonable. The more distal the amine group is from the interface, the lower its apparent pKa. The existence of up to three different ethanolamine groups on 021 LPS is supported by biochemical characterization performed by Prehm g; 31. (9) and our own 31P NMR analysis (Figure 3). Our belief that at pH 7 the NaLPS is below the pK of the second ionization of a phosphomonoester is supported by the work of Abramson 33. ‘31. (12). They found that the pK of the second ionizable group of phosphatidic acid is about 8.6. In the presence of sodium, this pK is shifted down to 8.0. Furthermore, Rosner g; 31. (13) have determined that the pKa's for phosphomonoesters of LPS from a heptoseless mutant of an E. 2911 K12 strain using 31P NMR ranged from 6.75 to 8.10. The pKa of the monoester of the lipid A linked pyrophosphate group was 7.27. If the pH of our NaLPS is close to 7, it would seem quite reasonable that the first observed increase in head group mobility with increasing pH is the result of the double ionization of phosphomonoesters on LPS. Although the influence of the surface potential within the core is likely to be unequally felt even by chemically equivalent groups, some simple assumptions about the charged state of NaLPS of 021 should be possible. At pH 7.0: 1) All ethanolamine groups are fully charged. ' 2) All KDO carboxyl groups are fully charged. 3) All phOSphodiester and most phosphomonoesters are singly charged. 129 Analysis of the effect of pH in the magnesium salt of LPS from strain 021 was complicated by its low head group mobility at neutral pH. Thus the parameter 1/2 W (the low to mid field peak splitting) was used to compare the different salts. Calculation of the rotational correlation time for comparison with the sodium salt was not possible except at very high pH. As Table 1 indicates, at high pH 1/2 W is nearly identical in both the sodium and magnesium 021 LPS salts. The effect of high pH on the MgLPS was completely reversible upon neutralization with HCl, indicating that the increased head group mobility was not due to hydrolysis of acyl chains. The slight increase in head group mobility of Mg°LPS after neutralization can be accounted for by the presence of sodium ions in solution. Thus the quintessential factor governing LPS self-association at high pH is charge repulsion within the LPS core regardless of the counterions present. Our results with the NaLPS of 021f2, a heptoseless deep rough mutant, were even more obvious. We found that head group packing of NaLPS from strain 021f2 increased abruptly above its isoelectric point and did not increase further. Since 021f2 LPS lacks ethanolamine groups (Figure 4) (2), we suggest that the lone titratable group seen in Figure 2 is that of the second pK of phosphomonoester groups. The relatively broad (4.3 Gauss) low field linewidth of CAT12 in 021f2 NaLPS decreased to a minimum of 3.8 Gauss with added hydroxide ion concentration. Once the hydroxide ion concentration exceeded the LPS concentration, the low field linewidth increased dramatically. At least two models may explain this behavior. I) At neutrality not all phosphomonoesters are singly charged. The substoichiometric addition of hydroxide ions reduced the heterogeneity of spin probe environments thus 130 reducingf; . The subsequent increase in linewidth is the result of our inability to determine linewidths accurately on such fluid samples. 2) The dramatic increase in linewidth at the same hydrogen ion concentration that produces an equally dramatic increase in head group mobility are distinct events which both describe the partial double ionization of ph05phomonoesters. Accordingly, the 6.0 Gauss linewidth just above the apparent pKa of the phosphomonoester would be the result of spectral addition of rigid and fluid spin probe populations. The preliminary nature of these experiments has prevented us from supporting either model. We were, however, encouraged by the precipitous drop in 2T" occuring at three times the actual LPS concentration. This is completely consistent with our expectation that the three phosphomonoesters of 021f2 carry a single charge at neutral pH. Browning (14) has recently presented deuterium NMR results which substantiates the existence of proton transfer complexes between adjacent lipids having both anionic and cationic groups. The hydrogen-bonded ion pair suggested was of the type: equation 3 -P02 ° ° - H - NH2 - although other donor and acceptor combinations are possible in phospholipid dispersions. The existence of such a complex in phosphatidylethanolamine and phosphatidylcholine has previously been suggested on the basis of infrared spectroscopy measurements (15). The emerging model is that phospholipids with amine containing head groups are free to form "polymeric" two dimensional complexes with other adjacent amine containing lipids. Moreover, there is an additional 131 possibility for hydrogen bond formation with the carboxyl group of phosphatidylserine with its amine. Browning (14) has shown that the methyl ester of a phosphatidylserine may also form a proton transfer complex between its amine and an adjacent carbonyl group. Increasing the intermolecular spacing of either phosphatidylcholine or phosphatidylethanolamine by addition of cholesterol, increased the rate of head group motion as detected by deuterium NMR (14). Since the addition of phosphatidylcholine to phosphatidylethanolamine did not increase head group mobility, it appears that proton transfer complexes between these lipids are sterically possible. Phosphatidylserine, however, was shown to be unaffected by either cholesterol or phosphatidylcholine addition. Thus, intramolecular hydrogen bond formation may also exist. The functional role of a network of noncovalent, proton transfer complexes in LPS is only speculative. It is important to remember that native LPS typically contains many polyvalent cations which undoubtedly are more effective in competing for anionic groups than are the ethanolamine moieties of LPS. Furthermore, it has long been recognized that the addition of EDTA to gram negative membranes promotes the release of a large percentage of the LPS (16). Although the importance of divalent cations in stabilizing LPS aggregates is not in doubt, proton transfer complexes in LPS may have considerable biological importance. Recent biochemical characterization of polymyxin (a pentavalent amine containing antibiotic) resistant mutants of gram negative bacteria indicate that resistance may occur via alteration in the extent of amine substitution (sugar amines or ethanolamines) on LPS (17) modulating the amount of intermolecular complex stability. 1. 2. 3. I.CL 1.1. 1J2. 123. :14, 155. 1(5. 117. 132 REFERENCES Chapter IV. Gmeiner, J. and Schlecht, S. (1980) Arch. Microbiol. 1g], 81-86. McLaughlin, S., Szabo, G., and Eisenman, G. (1971) J. of Gen. Physiol. EE, 667-687. Romeo, 0., Girard, A., and Rothfield, L. (1970) J. Mol. Biol. EE, 475-490. Goddard, E. (1974) Adv. Colloid Interface Sci. 4, 45. Abramson, M. and Katzman, R. (1968) Science Egg, 576. Seimiya, T. and Ohki, S. (1972) Nature New Bio. ggg, 26. Emmerling, G., Henning, V., and Gulik-Krzywicki, T. (1977) Eur. J. Biochem. 1g, 503-509. Prehm, P., Schmidt, G., Jann, B., and Jann, K. (1976) Eur. J. Biochem. 29, 171-177. Olins, A. and Warner, C. (1967) J. Biol. Chem. g5; (21), 4994-5001. The Merck Index, 8th edition (1968) ed. P.G. Stecher, Merck and Co. Abramson, M., Katzman, R., Wilson, C., and Gregor, H. (1964) J. Biol. Chem. g§g_(12), 4066-4072. Rosner, M., Khorana, H., and Satterthwait, A. (1979) J. Biol. Chem. ggg (13), 5918-5925. Browning, J. (1980) Biochem. E9, 7144-7151. Akutsu, H., Kyogoku, Y., Nakahara, H., and Fukuda, F. (1975) Chem. Phys. Lipids lg, 222-242. Leive, L. (1974) Ann. N.Y. Acad. Sci. gag, 109-129. Vaara, M. (1981) J. Bacter. E48 (2), 426-434. 133 Table 1 Hyperfine Splitting of CAT12 in LPS Salts at 37°C Preparation 1/2 W, Low to Mid Field Peak Splitting (Gauss) NaLPS, pH 7 18.90 NaLPS, pH 12 18.65 MgLPS, pH 7 25.12 MgLPS, pH 12 18.95 MgLPS, pH 7 pH 12 pH 7 24.70 MgLPS, pH 7 + 58 M NaCl 24.85 .mae on» we :owmme aaeem ewe; on» we Auwwwees on» e» peceweeeeeea xwueecwe mw N~ww:= ewcezeeez .zgumeegu we peegem .eceseez .eo he coxeu we: eeueeem egw .Aaomv seemewe ecu “sowv eces seen zewgz mezeem manganesaeexa Eeew wee eEeLLxe Ie—eww sow; oz» esw .o. m me we euepegexxeee esweem we cexeu Hue :weeum _Pee .m we maeez ecu we eeueeam as: can .m eezmwu Eng 3i «7. OF... oi 0.. VI an o N v — _ . . p . F . h . 3e :22 a: 2...: we 137 .ewpeeum=< ..=.m.z .eexg :ueez .auwmce>w== eweesceez .xeumeesu we _eezem .eceeeez .eo we cexou we: eeeeeam mew .eeuzuwumeamca xpeuepeeee ego managemecaeexn mmw .o.w :e mum—ezexxeee saweem he eemceemwe mm: meeez .sz am we case» Nwflma :wecum w—ee .w we maeez one we ecuueam mzz man .a eeaaww Eng v.7 NT. D.T. at 0... at m... .o N i b p b '— - n ‘V <9 «:3 :22 9.5.: B CHAPTER VIII PHYSICAL PROPERTIES OF YERSINIA PESTIS LIPOPOLYSACCHARIDE MAY DETERMINE VIRULENCE INTRODUCTION It is difficult to begin a discussion of the physical properties of Yersinia pestis lipopolysaccharide (LPS) without at least touching upon the historical importance of this pathogen. The initial shock, 1346-50, was severe. Die-offs varied widely. Some small communities experienced total extinction; others, e.g., Milan, seem to have escaped entirely. The lethal effect of the plague may have been enhanced by the fact that it was propagated not solely by flea bites, but also person to person, as a result of inhaling droplets carrying bacilli that had been put into circulation by coughing or sneezing on the part of an infected individual. Infections of the lungs contracted in this fashion were 100 per cent lethal in Maudiaria in 1921, and since this is the only time that modern medical men have been able to observe plague communicated in this manner, it is tempting to assume a similar mortality for pneumonic plague in fourteenth-century Eur0pe. Whether or not pneumonic plague affected Europeans in the fourteenth century, die-off remained very high. In recent times, mortality rates for sufferers from bubonic infection transmitted by flea bites has varied between 30 and 90 per cent. Before antibiotics reduced the disease to triviality in 1943, it is sobering to realize that in spite of all that modern hospital care could accomplish, the average mortality remained between 60 and 70 per cent of those affected (1). The role that LPS plays in the pathogenicity of 1. pg§31§_has only recently been addressed. Virulence has been linked to a 42 md plasmid which confers calcium dependence and the ability for cells to autoagglutinate (2). Virulent and avirulent strains grow normally at 25°C with or without calcium in their growth media. At 37°C, however, ~138 139 virulent strains are unable to grow in the absence of added calcium (2.5 mM). Changes in the outer membrane protein profile have been reported to occur as a function of growth temperature in a virulent strain of 1. pggtis (3). Nonetheless, these changes were not affected by the presence or absence of calcium. More recently, temperature and calcium dependent changes in outer nembrane proteins have been found in cal+ (virulent) and cal’ (avirulent) strains of 1..pg§gig (Dr. W. T. Charnetzky, personnel communication). The expression of these proteins on SOS-PAGE gels was found to be dependent upon the bound LPS. That is, LPS from calcium deprived cal+ cells grown at 37°C did not influence protein migration whereas LPS from 37°C grown calcium supplemented cal+ cells or cal' cells with or without calcium did alter the migration of the proteins. The new positions of the proteins and LPS cofactors were identical with the position of the proteins isolated from cells grown under the same conditions as the LPS cofactors themselves. Thus, the apparent differences in migration of these outer membrane proteins was a consequence of altered LPS and not altered proteins. MATERIALS AND METHODS 1} p§§£1§_nLPS was a gift of Dr. Willard T. Charnetzky of the Department of Bacteriology and Public Health, Washington State University, Pullman, Washington 99164. The bacteria were grown in broth consisting of: N-Z amine (3%), lactic acid (10 mM), MgCl2 (2 mM), KZHP04 (25 mM), citric acid (10 mM), MnClz (0.01 mM), and FeClz 140 (1 mM). After autoclaving, xylose (10 mM), Na23203 (25 mM), and potassium gluconate (10 mM) were added. Calcium was removed from N-Z amine by treatment with sodium oxalate. Where specified, CaClz (2.5 mM) was added. The medium was adjusted to pH 7.0 with 5.5 N NaOH. Cells were grown aerobically at either 26 or 37°C as specified in shaken water baths. Cells were harvested at mid exponential phase. The hot aqueous phenol extraction procedure was used to isolate LPS. Electrodialysed and sodium salts of LPS were obtained as described in Chapter III. ICP analysis of LPS isolated was carried out as indicated in Chapter III. ESR spectra of probed samples were analyzed as in Chapter III with a probe to LPS weight ratio of 0.074. RESULTS AND DISCUSSION Although the endotoxin activity of Yersinia LPS is not thought to be of critical importance in the pathology of the bacteria, LPS may be a critical determinant of virulence. We considered LPS to be a logical candidate for the cell component responsible for the strong calcium requirement in virulent strains. Indeed, the calcium levels in native LPS from cells supposedly grown in low calcium media are comparable to that of cells grown in calcium supplemented media (Table 1). This was true of LPS from cal+ and cal‘ cells. In fact, LPS from cal+ cells generally bound calcium at levels equal to or greater than magnesium regardless of growth conditions. nLPS from cal‘ cells, like nLPS of E. £911 021, consistently bound more magnesium than calcium. Calcium starved, 37°C grown cal+ cells produced nLPS which was conspicuously low in its 141 cation charge to phosphorous ratio. This was the direct result of exceptionally low sodium content. The results of the elemental analysis of 1. pg§§1§_Na-LPS are summarized in Table 2. Unfortunately, the ashed samples of Na-LPS from cal+ 37°C grown cells with added calcium and cal’ 37°C grown cells without added calcium were heavily contaminated with sodium and aluminum fran an unknown source. The data do not indicate any major differences metals bound to in the Na-LPS correlated with the strain or growth conditions. The Na/P ratio of NaLPS from I, pgggjg is, however, four times that from E, 5211' 021. .1..pggE1§ LPS has been reported to have only three phosphorus per LPS whereas our analysis indicates E. gglj_021 LPS has between seven and nine (4). Figure 1 clearly shows that Na-LPS of 1. pggglg grown at 37° and 25°C are distinctly different in their head group mobility. This type of temperature adaptation is opposite to that observed in the hydrocarbon interior of both prokaryotic and eukaryotic membranes (5). Typically, when cells are grown at lower temperatures, membrane lipids are altered so that their acyl chains become more fluid (6). Homeoviscous adaptation is still a controversial topic and the argument has never been proposed to extend beyond the membrane hydrocarbon interior into the polar head group region. The significance of head group mobility adaptation in 1. pggglg LPS nay lie in the single departure from the pattern. Na-LPS from cal+.1. pestis grown at 37°C in the absence of calcium appears to have head group mobility intermediate between that of Na-LPS from 37 and 25°C grown cells. It is apparent that the critical adaptation of LPS characterized 142 by increased head group mobility is incomplete in cal+ cells shifted up from 26 to 37°C in the absence of calcium. The nature of the differences in the chenical modification(s) between LPS from X: pggglg grown at 26°C and at 37°C may be quite subtle. We could not discern a pattern to the phosphorus to KDO ratio which could account for the observed pattern of LPS head group mobility. There were no perceptible differences in the migration pattern of any of the Na-LPS on 12% SDS PAGE gels (data not shown). There was, however, one additional ESR spectral parameter which correlated extremely well with growth temperature and may provide some insight as to the nature of the LPS modification. Table 3 Low Field Line Width of CAT12 in Na-LPS of Yersinia pestis + 25° + 37° al al’ Cal Cal‘ magi +ca2+ tag: +Ca2“ -Ca2+ +63“ 4382" +5623”... Gauss 4.15 4.00 3.90 4.06 4.63 4.50 4.40 4.85 Mean 4.02 Gauss Mean 4.60 Gauss SD t 0.10 Gauss SO x 0.19 Gauss var i 0.01 var 1 0.03 The low field linewidth,I:, of CAT12 bound to LPS, was measured for each NaLPS sample at 37°C. The results are summarized in Table 3. Simply, E of CAle bound to LPS obtained from 26°C grown cells was significantly smaller than.I: of LPS from 37°C grown cells. No differences were observed which were correlated with the strain or the presence of calcium in the media. The narrow linewidth of Na-LPS from 26°C cells (ave. 4.0 G) was very typical of spectra characterized by 143 classic lifetime broadening (7). The broader signal produced by Na-LPS from 37°C grown cells (ave. 4.6 G) may be indicative of a inhomogenous spin probe environment. The CAle hyperfine splitting parameter of Na-LPS from 26 and 37°C was 4 Gauss. Although the observed line broadening is quite large, it is clearly not as large as would be expected by simple spectral addition of 26°C NaLPS and the rigid component of the 37°C NaLPS. Moreover for magnetically dilute spins, two spin probe populations which differ in their hyperfine splitting parameter by 4 Gauss would be resolved as separate low field peaks rather than as a broadened single peak. Our conclusion from these results is that the head group structure of LPS from 37°C grown 1.ng§31§_is more heterogeneous than LPS from 25°C grown.1. pestis, but the source of that heterogeneity is not known. REFERENCES 1. McNeill, W.H. (1976) in Plagues and Peoples, Anchor Press/Doubleday, Gordon City, New York, 147-149. 2. Ferber, D.M. and Brubaker, R.R. (1981) Infect. and Immun. 31, 839-841. 3. Darveau, R.P., Charnetzky, W.T., and Hurlbert, R.E. (1980) J. Bacter. 143 (2), 942-949. 4. Chapter IV. 5. Hampton, M.J., Floyd, R.A., Clark, J.B., and Lancaster, J.H. (1980) Chem. Phys. Lipids 21, 177-183. 6. Overath, P. and Trauble, H. (1973) Biochemistry 12 (4), 2625-2634. 7. 8. 144 Mason, R.P., Giavedoni, E.B., and Dalmasso, A.P. (1977) Biochemistry 16 (6), 1196-1200. Hartley, J.L., Adams, G.A., and Tornabene, T.G. (1974) J. Bacter. 118 (3), 848-854. 145 Table 1. Elemental Analysis of Yersinia pestis Native LPS cal+, calcium dependent cal’, calcium independent 25°C 37°C 25°C 37°C no Ca2+ +Ca2+ no Ca2+ +Ca2+ no Ca2+ +Ca2+ no Ca2+ +Ca2+ Ca/P 0.28 0.31 0.29 0.32 0.20 0.04 0.10 0.17 Mg/P 0.22 0.25 0.20 0.33 0.25 0.31 0.38 0.40 Ca+Mg/P 0.50 0.55 0.49 0.55 0.45 0.35 0.48 0.57 Na/P 0.45 0.30 0.00 1.56 1.29 0.81 0.55 0.54 FB/P 0.01 0.04 0.02 0.00 0.01 0.01 0.02 0.02 AI/P 0.01 0.02 0.01 0.02 0.01 0.01 0.01 0.02 Zn/P 0.01 0.00 0.02 0.07 0.09 0.08 0.03 0.03 mMP/g LPS 0.58 0.52 0.52 0.52 0.79 0.97 0.58 0.63 charge/P 1.55 1.59 1.11 3.07 2.42 1.72 1.55 1.95 Table 2. 146 Elemental Analysis of Yersinia pestis Sodium LPS calT, calcium dependent cal', calcium independent 37°C no Ca Ca2+ no Ca2+ Ca2+ no Ca2+ Ca2+ no Ca2+ Ca2+ Co]? 0.01 0.07 0.04 0.11 0.04 0.02 0.09 0.03 Mg/P 0.01 0.06 0.00 0.05 0.03 0.02 0.05 0.03 CaTMg/P- 0.02 0.13 0.04 0.16 0.07 0.04 0.14 0.06 Na/P 2.33 2.40 2.90 7.08 1.89 1.66 4.67 1.92 Fe/P 0.01 0.04 0.04 0.02 0.16 0.01 0.02 0.02 Al/P 0.03 0.06 0.03 0.81 0.03 0.02 0.60 0.02 Zn/P 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 mMP/g LPS 0.54 0.61 0.58 0.65 0.55 0.72 0.56 0.56 charge/P 2.46 2.96 3.20 9.89 2.60 1.80 6.81 2.16 147 [:16 cm {are (an) [510] 111:: [1153 {:12}: :13: 5'9: moi Mim:we mam: ETH+~H§ Yersinia pestis Figure 1. Chemical structure of Yersinia pestis LPS based on data obtained from reference 8. 148 £338 822.533.... A.V £55.: ecu A09 5.: 225 Lee eee 23:38 “5355333 AIV ”Sect: ecu ADV 5:: Lee 33% 2595» .593 o 8 use 25.28 822333 A{v uaefiw: use :3: 3:5 +28 ecu 25:38 emu—38% am :1 52:; .28 I: f; L8 as... 252.; =5; 9.3 .8: m5... 5 a S we .awm .eeumfiegea mcwuuwpam ecwwcea»; use we eeceeeeaee eezueceaEew .N we: ww 8.» 228888. on 9. on ON 0. o a d a _ _ 4 ¢m 0 mm .589 cm ..._.N No ¢o mm mm CHAPTER IX CHARACTERIZATION OF THE PHYSICAL PROPERTIES OF CHROMATIACEAE LIPOPOLYSACCHARIDES ABSTRACT Lipopolysaccharides (LPS) isolated from Chromatrum vinosum and Thiocapsa roseopersicina were analyzed by inductively coupled plasma emission spectroscopy and electron spin resonance (ESR) spectroscopy. The LPS ionic compositions and fluidities were compared with those of rough LPS from enteric bacteria. LPS from both §;_vinosum and 1;_ roseopersicina contained a rather large number of cations on LPS weight basis despite low levels of ph05phorus and acidic sugars. Calcium and sodium dominated all other metal cations in the elemental profile of g. vinosum native LPS (nLPS). LPS from I} roseopersicina had a higher phosphorus to weight ratio than LPS from Q. vinosum and contained nearly equal amounts of magnesium and calcium. The total cation binding capacity of I; roseopersicina nLPS greatly exceeded that of enteric rough nLPS. The ESR head group Spin probe CATIZ indicated that the polysaccharide core region of Chromatiaceae nLPS was less restricted or rigid than 021 nLPS from g, 2211- Treatment of Chromatiaceae nLPS with NaEDTA did not dramatically alter LPS head group mobility. 149 150 Surprisingly, the acyl chain mobility of g. vinosum and I. roseopersicina nLPS were identical over the temperature range -5°C to 50°C and an acyl chain structure transition was observed at 24 1 9°C in both preparations. Several mechanisms are presented which may account for the high cation binding capacity of Chromatiaceae LPS. The significance of the physical properties of the lipopolysaccharide from these photosynthetic bacterial are discussed with regard to their low endotoxic activity. INTRODUCTION The recent biochemical analysis of LPS from the family Chromatiaceae has greatly broadened the definition of LPS as a lipid class. These photosynthetic purple sulfur bacteria are composed of ten genera; the LPS of only two, Chromatia and Thiocapsa, have been investigated extensively (1). Their LPS are characterized by a lipid A which has a D-glucosamine-D-mannose rather than the "classic“ diglucosamine sugar backbone and are totally devoid of phosphorus. The fatty acid composition, however, is very much like that of enteric LPS. The polysaccharide chain is low in KDO content and certainly lacks the KDO trisaccharide cage common to enteric bacteria. In addition, the core sugar L-glycero-D-mannoheptose found in enteric LPS is substituted with D-glycero-D-mannoheptose. This unusual sugar has also been detected in LPS from some Yersinia and Salmonella strains. In those strains, however, when the D isomer is present, the O-polysaccharide is synthesized but cannot be attached to the LPS core. Chromatiaceae have apparently evolved the appropriate O-polysaccharide 151 translocase and are able to assemble a complete LPS on to a core containing the D-heptose isomer. Recently analysis of LPS from smooth strains of Salmonella has revealed that approximately 60 mole percent of the total LPS is actually rough LPS (2,3). SDS-PAGE gels of both 9. vinosum and I; roseopersicina indicates that both are composed of purely smooth LPS (4) and there is no indication of significant levels of a rough component. The polysaccharide chains of LPS from both Chromatiaceae species also contain several amine containing sugars. ‘Q. vinosum LPS contains mannosamine and quinovosamine while LPS from I. roseopersicina has glucosamine, mannosamine, and 3-amino-3,6,-dideoxy-D-galactose (5). LPS frail g. vinosum also contains only the D isomer of rhamnose while I. roseopersicina LPS has only the L isomer. Finally, LPS from Q. vinosum is unique in that its O-polysaccharide contains the pentose sugars D-ribose (pyranosidic when internal, furanosidic when terminal), L-arabinose and 3-0-methyl-D-ribose. None of these sugars are present in .1. roseopersicina or in enteric LPS. The anticomplement activity of Q. vinosum LPS has been reported to be only 25% that of S. abortusequi LPS (4). Like enteric LPS, alkaline hydrolysis of g. vinosum LPS significantly decreased anticomplement activity. In contrast, treatment with NaEDTA resulted in no change in endotoxic activity. Conversion of enteric nLPS to a sodium salt is usually accompanied by a sharp increase in endotoxic activity (6). Although Chromatiaceae are not recognized pathogens, the unique biochemical composition of their LPS alone makes their further biophysical characterization worth while. It is easy to become transfixed by the volumous reports on enteric LPS and ignore the 152 potential insight other LPS may supply. The results presented here show that this family of photosynthetic bacteria can greatly broaden our definition of bacterial endotoxins and perhaps provide new perspectives for understanding the pathogenicity of more conventional gram-negative species. MATERIALS.AND METHODS g, vinosum and I. roseopersicina were the generous gift of Dr. Hurlbert, Department of Biochemistry, Washington State University. Inductively coupled plasma emission Spectroscopy and electron spin resonance techniques were performed as before (Chapter III). RESULTS AND DISCUSSION Despite the low phosphorus and acidic sugar content of Q. vinosum LPS, a considerable number of cations were detected bound to the native LPS (Table 1). we can not rule out the possibility that many of these cations simply carried over during the LPS isolation through some undetenuined mechanism (e.g., intrapment within LPS aggregates). There is, however, at least one nontrivial explanation for this behavior. Neutral sugars can form stable complexes with polyvalent cations when two or three hydroxyl groups of a pyran ring are stereochemically arranged to fit into the coordination sphere of the cation. Sugars possessing the axggggax_sequence of oxygen atoms on the six-membered ring (Figure I) have this ability (7). 153 C)" Figure 1 (7H Pyran ring containing the 25:29:25 sequence. The coordination geometry for neutral sugars is much less flexible than that of acidic sugars such as KDO. Although KDO has a slightly higher affinity for calcium than magnesium (8), the specificity of neutral sugars is much greater. A marked preference for cations of 1A ionic radius (La3T, Ca2+, Na+) has been demonstrated (9). For example, a-D-allopyranose has a stability constant for calcium of 6.2/mole and only 0.19/mole for magnesium. It is interesting to note that g. vinosum nLPS retained substantially more calcium and sodium than magnesium (Table 1). Calcium and sodium were both present in ten fold excess over magnesium in nLPS from Q. vinosum. As Table 1 indicates, this is quite uncommon in other gram negative nLPS. It should be pointed out though, that the enterics listed have exclusively rough LPS while the Chromatiaceae LPS is exclusively smooth. Together, the bound cationic charge to LPS weight ratio of Q. vinosum nLPS was comparable to that of enteric nLPS. An analysis of the reported O-polysaccharide sugars of C. vinosum shows that B-D-rhamnose and B-ribopyranose have the required ligand geometry which would preferentially bind Ca2+ and Na+. ' The elemental analysis of g, vinosum LPS also confirmed the early observation by Hurlbert gt a1. (5) that the phosphorus content is quite 154 low (Table 1). In fact, we have calculated the minimum molecular weight of monomeric LPS required for 1:1, phosphorus to LPS molar ratio, to be 41,000. Although this is not an entirely unreasonable value in light of the smooth nature of Q. vinosum LPS, it almost certainly rules out the possibility of more than one phosphorus per LPS. Although LPS from I. roseopersicina was shown to have a much higher phosphorus to LPS weight ratio than.Q. vinosum LPS, it was lower in phosphorus than enteric LPS (Table I). This may actually be a reflection of smooth character and high molecular weight of I. roseopersicina LPS, compared to the LPS of the enteric Species. Furthermore, I. roseopersicina retained an exceedingly large number of cations bound to its nLPS. The total cationic charge to LPS weight ratio was almost twice that of the rough E. coli strain 021. This becomes even more significant when it is remembered that the contribution of the polysaccharide chain in the LPS of I. roseopersicina to its total weight may be as much as 90%. Unlike LPS from C. vinosum, I. roseoperisicina LPS contained about as much magnesium as calcium. Thus, neutral sugars can not totally account for the high cation binding capacity. It is suggested that the bound Mg may be interacted with the phosphate groups in nLPS from I. roseoperisicina whereas nLPS from Q. vinosum lacking such phosphate, also lacks Mg binding sites. Our electron Spin resonance experiments produced equally surprising results. In Chapter III we showed that the lipid head group spin probe CAle has a rather high affinity for specific sites in g. 2211 LPS suspensions. He suggested that likely sites for CAle binding were the phosphorus of lipid A. Melhorn gt al. (10) have shown that CAle has high water solubility and very low affinity for neutral phospholipids 155 regardless of the fluidity of their acyl chains. Nevertheless, the binding affinity of CATIZ for LPS of the family Chromatiaceae was quite high. At the same CAle to LPS weight ratio used with enteric LPS from g. 9911, little free probe signal was evident in the LPS from Chromatiaceae in the temperature range -5 to 37°C. Since the lipid A moiety of LPS from both g, vinosum and I. roseopersicina has been reported to be devoid of phosphorus, the high CATIZ affinity was unexpected. The hyperfine Splitting parameters, 2T", of bound CAle indicated that the nLPS of both C, vinosum and I. roseopersicina were less restricted in head group mobility than nLPS from Er.£211.021- In addition, the head group of LPS from I. roseopersicina was less restricted than that of Q. vinosum (see Figures 2 and 3). Interestingly, electrodialysis and NaEDTA treatment of g. vinosum and I. roseopersicina nLPS did not alter head group mobility. This observation is inconsistent with the usual rigidifying effect of polyvalent cations on anionic lipids (11). It is possible that divalent cations are intramolecularly bound and thus have a minimal effect on LPS-LPS interactions of these Species or at least have a minimal effect on the CAle binding site at the polar-nonpolar interface. 0ur explanation for the location of calcium in C, vinosum is consistant with either of these hypothesis. Hurlbert has also observed that the lethality of Q. vinosum nLPS is unaffected by treatment with NaEDTA (4). This is in marked contrast the with finding that the sodium salt of enteric LPS has increased endotoxic activity when compared to nLPS (6). It is apparent from Figure 4 that the hydrocarbon interior of nLPS from Chromatiaceae are more disordered than that of nLPS from g. coli 156 021. The nLPS from both species of Chromatiaceae appear to undergo an acyl chain melt at 24 i 9°C. The similarity of both the SDS hyperfine Splitting parameter and order parameter data for Q. vinosum and I. roseopersicina nLPS is, however, somewhat surprising. The two Species do have very similar Lipid A moieties, differing only in the presence of oleic acid and B-hydroxycapric acid in I. roseopersicina. Nevertheless, the large difference in polysaccharide composition, phosphorus content, metal ion profile and head group mobility make the similarity in acyl chain mobility difficult to explain. We have observed identical acyl chain mobilities in the sodium salts of g. 9211 021 and 021f2 LPS (12). Here, too, substantial differences exist between these two lipids in both head group mobility and polysaccharide composition. Furthermore, the relative insensitivity of head group mobility in either 9. vinosum or I. roseopersicina_to the loss of divalent cations suggests that the differences in cation composition are inconsequential in the packing of either the polysaccharide or acyl chains. We are left with the conclusion that head group and acyl chain mobilities may often be regarded as separate phenomena. REFERENCES 1. Neckesser, J., Drews, G. and Mayer, H. (1979) Ann. Rev. Microbiol. §§, 215-239. 2. Palva, E.T. and Makela, P.H. (1980) Eur. J. Biochem. 101, 137-143. 3. Goldman, R.C. and Leive, L. (1980) Eur. J. Biochem. 191, 145-153. 4. Hurlbert, R.E. and Hurlbert, I.M. (1977) Infect. and Immun. 16 (3), 983-994. 5. 6. 7. 8. 9. 10. 11. 12. 157 Hurlbert, R.E., Neckesser, J., Mayer, H., and Fromme, I. (1976) Eur. J. Biochem. QB, 365-371. Galanos, 0. and Luderitz, 0. (1976) Eur. J. Biochem. §§, 403-408. Rees, D.A. (1975) Biochemistry of Carbohydrates,_Vol. 5, ed. R.J. Nhelan, MJP Intl. Rev. Sci., Univ. Park Press, Baltimore. Chapter III. Angyal, 5.0. (1973) Pure Appl. Chem. §§, 131. Melhorn, R.J. and Packer, L. (1979) Methods in Enzymology‘ég, 515-526. Vierstra, R. and Haug, A. (1978) Biochem. Biophys. Res. Commun. 84 (1), 138-143. Chapter V. .xuwmcm>w:= macaw campguw: .ucmsgcmamo acumvsmgoovm .Lopc:_;um .2 .go yo uw_m mgu mm: um .um~»pu_uocuoopw camp to; ma; one a__m=oep~m .2.z coco new ma4\¢ s mesmm<« 158 do.~ m~.~ om.~ am.~ ec.H was a\mamtagu coruau ~m.o NH.H HN.H emm.o mewo.o was m\a ~o.o --- oc.o co.o oo.o was m\=N Ho.o --- ~o.o Ho.c mo.o was m\_< ~o.o Ho.o mo.o oo.o Hc.o was m\aL o¢.o ~m.o HH.o m~.o um.c ma; a\az mm.o om.o mm.o mo.H -.o was a\az+au m~.o em.c o¢.o m¢.c ~o.o was m\mz -.o co.o mH.o om.o mN.o was m\au pnu 3..»me omo “ND :00 wcbuwmgwgmmog fiance; 2.9.3.3» #m—pwcos—mm opzotmcumm omdmuott. 32.35950 mm.- m\:o_.umu S: awcmuumm m>puammz Sago paco>om Eocm muvcmgoommxpoqoqw4 o>_aaz we upmx_mc< pavemem_m H o_noh 159 68 67 - 66 65 2T" 6" (o) 63 62 6| 6O 59 1 1 1 1 1 1 '50 5.|0 l5 202530 Temperature ('0) Figure 2. Temperature dependence of the hyperfine splitting parameter 2T.. of CAT12 in Chromatia vinosum native LPS (0) and sodium LPS (0). The probe to LPS weight ratio was 0.074. 160 65 - e41- - (G) 63 - . 62 - 6|- \ \ so \ \ 59 l l l l l l q '5 O 5 IO I5 20 25 30 35 Temperoiure (°C) Figure 3. Temperature dependence of the hyperfine Splitting parameter 2T.. of CATl in Thiocapsa roseopersicina native LPS (0) and sodium LPS To) The probe to LPS weight ratio was 0.074. 161 I. .3533; .8» 6253.: 3 Alv mad: Hmo __oo .u com mama .e~o.o ma: o.umg ugm_wz mm; ow mnocq ms» .mgg: AOV uczpmcwmoomoc 33035 new AOV 58...; «32.8.8 5 mom mo :5 LmumEocuq a:_uuppam o:_$gmaxz us» we mocaucwqmu mczuocmqem» .e mg:m_u .0... 3339.33 3 s a a 2 a ‘ ‘ J ‘ ‘ ‘ ‘. J ‘ d ‘ . q L s .0. . ha 1 at. APPENDIX A USE OF ELECTRON SPIN RESONANCE TO STUDY BACILLUS MEGATERIUM SPORE MEMBRANES* SUMMARY Membranes from dormant and heat-activated spores were labeled with the fatty acid spin probe 5-doxyl stearate and analyzed using electron Spin resonance spectroscopy. Membranes from dormant spores were slightly less fluid above 23° than membranes from heat-activated spores. Also L-proline caused a much larger increase in the upper transition temperature than did D-proline when added to membranes from heat-activated spores. Thus a compound known to trigger germination in this strain may interact stereospecifically to alter the biophysical properties of the spore membranes. INTRODUCTION The mechanism of breaking the dormant state of bacterial spores is not known. In §;_megaterium GM 31551, rapid germination occurs if the spores are first heat-activated followed by the addition of a *Published in Biochem. Biophys. Res. Commun. (1981) 199_(3), 1137-1142. ' 162 163 stereospecific compound like L-proline (1). One model to explain these processes suggests the Spore membrane(s) may be involved (2), and to study this possibility we have used electron spin resonance (ESR) Spectroscopy. We report here what appears to be the first demonstration of biophysical changes in spore membranes that result from heat-activation and L-proline. MATERIALS AND METHODS .E; megaterium spores were grown in supplemented nutrient broth, harvested and stored as previously described (3). All references to spore weights are on a dry weight basis. Spores (50 mg/ml) were extracted with sodium dodecyl sulfate-dithiothreitol and then washed as previously described (4). The details of the isolation and characterization of spore membranes will be described elsewhere (Racine, F.M. and Vary, J.C., in preparation) but briefly the methods were as follows. Sodium dodecyl sulfate-dithiothreitol extracted spores (4) were lysed in 0.1 M HEPES (pH 7.5) containing lysozyme (0.5 mg/ml), RNase (2.5 uglml) and DNase I (2.5 ug/ml) at 30° for 12 min followed by sonication (8 times for 30 5 each) at 0° as previously described (4). Spore membranes were isolated by methods similar to those used for the isolation of §;_ggli_cytoplasmic membranes (5). Membranes from heat-activated Spores were obtained by first heating spores (50 mg/ml) at 60° for 10 min, followed by centrifugation at 5,000 x g for 10 min and then lysis as described above. The final membrane preparation in 10 mM HEPES (pH 7.5) contained 10-20 mg of protein/ml as determined by the method of Lowry gt 31. (6) and was stored on ice for further analysis. 164 All ESR studies were done with freshly prepared membranes (0.4 ml) to which 4-5 pl of 30 mM SDS was added. The labeling techniques that were used have been previously described (7). ESR spectra of such spin labeled preparations allow the determination of the hyperfine Splitting parameter, 2T.., which reports the local fluidity of the membrane lipids (8). High values of 2T" reflect low fluidity. All ESR studies were carried out with a Varian Century Line ESR spectrometer, model E-112, equipped with a variable temperature controller. An external calibrated thermistor probe (Omega Engineering, Inc., Stamford, CT) was used to monitor the temperature of the sample. The data (2T, vs. temperature) were analyzed by an iterative least squares program to be described elsewhere (Coughlin, R.T., Brunder, D.G., and McGroarty, E.J., in preparation). Briefly, a B-spline (9) was used to provide a smooth fit for the ESR data and points of inflection were used to group data. Regressions lines were calculated for each group and then plotted. This analysis allowed the determination of break points. Such breaks in the temperature dependence of 2T" have been correlated with lipid phase separations or lipid phase transitions from gel to liquid crystalline lipid states (7). All temperature dependent ESR parameters were shown to change reversibly up to 46°. For each set of data, at least 3 independently isolated membrane preparations were used. DISCUSSION Membranes isolated from either dormant or heat-activated Spores had the same phospholipids and in the same ratios as whole Spores, similar to previously published data for total phospholipids in this strain (10). 165 The membranes contained no peptidoglycan, a distribution of about 20 proteins ranging from 13,000 to 130,000 daltons, several respiratory associated enzyme activities and a unique carotenoid (unpublished). The tenperature dependence of 2T.. in membranes from dormant spores is illustrated in Figure 1. Of particular importance are the transition temperatures where the slepes of the lines change at 6° and 26° suggesting a change in the relative ratio of gel and liquid crystalline lipid. When the same experiment was done with membranes from heat-activated Spores (Figure 2a) the transition temperatures were slightly different, 7° and 23°. The more noticeable difference, however, is that above the upper transition temperature membranes from heat-activated Spores exhibit a greater slepe (-0.29 gauss/°C) than do dormant spore membranes (-0.25 gauss/°C). This implies that above the upper transition, membranes from heat-activated spores are more fluid than those from dormant Spores. We feel that this change in fluidity should be interpreted with caution with respect to any possible functional role. But it is apparent that there is a physical difference between membranes from dormant and heat-activated Spores which to date has never been reported. Finally, we tested the effect of adding in vitrg a known trigger reagent, L-proline, to membranes from heat-activated Spores. The results (Figure 2) Show a dramatic change in the upper transition temperature from 23° before to 31° after the addition of L-proline. When the same experiment was done with D-proline (30 mM) which cannot trigger germination (1), a much smaller shift occurred in the upper transition temperature (from 23° to 27°). Preliminary experiments suggest that the addition of L-proline to dormant spore membranes caused no dramatic 166 changes in the temperature dependence of 2T". In all of these experiments the low transition temperature did not change significantly. Fran these results, it is apparent that L-proline interacts with the membrane and causes a change in the supramolecular structure. We have no evidence that the transition temperature shift to 31° is fortuitous or significant with respect to 30° being optimal for triggering germination. The important point is that L-proline may interact with membranes from heat-activated Spores in a stereospecific manner to cause a biophysical change in the spore membrane. While these data do not explain the mechanism of triggering germination, techniques described here provide us with a useful tool to further analyze other trigger compounds, membranes isolated at different times during sporogenesis and to determine the role of Spore membranes in development. Of particular interest are the recent data with a proline affinity analog (Rossignol, D.P. and Vary, J.C., submitted) which indicates a possible method to isolate the proline trigger site. Using these present and other biophysical techniques, we hope to study the interactions of L-proline with the trigger site both in vitro and la vivo. REFERENCES 1. Rossignol, D.P., and Vary, J.C. (1979) J. Bacteriol. 13g, 431-441. 2. Vary, J.C. (1978) in Spores VII (Chambliss, G. and Vary, J.C., eds.) 104-108, American Society for Microbiology, Washington, D.C. 3. Shay, L.K. and Vary, J.C. (1977) Biochim. Biophys. Acta §§§, 284-292. 167 4. Vary, J.C. (1973) J. Bacteriol. 116, 797-802. 5. Schnaitman, C.A. (1970) J. Bacteriol. 195, 890-901. 6. Lowry, 0.H., Rosebrough, N.J., Farr, A.L., and Randall, R.J. (1951) J. Biol. Chem. 13, 265-275. 7. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta, in press. 8. McFarland, 3.6. (1972) Chem. Phys. Lipids g, 303-313. 9. Dierckx, P. (1975) J. Comp. Appl. Math. 1, 165-184. 10. Bertsch, L.L., Bonsen, P.P.M., and Kornberg, A. (1969) J. Bacteriol. gg, 75-81. Figure 1. 168 gm- 9... g... ”gentlemen j... Vr'v"'IrV'rfI'r'vrv'VVVVV'VVV'VY'r'rrfr'V'VVf'V' r7 T—f" f - .4» the I.on 16.n- ai.oo than 13.00 E.“ 15.00 35.00 TEMPERflTURE I DEGREES C) Temperature dependence on 2T" in dormant spore membranes. Membranes were isolated and ESR spectra recorded at the indicated temperatures in the presence of 5 US as described in Materials and Methods. Arrows indicate transition temperatures. 169 Figure 2. Effect of L-proline. Membranes from heat-activated spores were isolated and analyzed as described in the legend of Figure 1 in the absence (a) and presence (b) of 30 mM L-proline. 170 g... "q'fflmfifid ““31. 3.. 9.“ I ‘ rVVY'V‘YV’VVV'TY'V"'Tfrff‘v'f"VVV’V'fff'ffv'V'WVTTY'" : ‘ 3.1 I f .m- h o.- o'a 15.. ,i.‘ 6.1-! in a. Si..- ba at. fvuvvvvvf‘vvvlvvv'vvvvvvvrrvvIvvfivrvrrv g.“ g..- 9... "g'ffi'mfifl mm. 5.60 r'Uvjjv'I'YY'T' b - .ao 0'... 0'..- 1b.ol 11.1w Ti... 10.. {on ENTIRE memes C) Si.“ aid- 45.. APPENDIX B THE MODIFICATION OF HUMAN ERYTHROCYTE MEMBRANE STRUCTURE BY MEMBRANE STABILIZERS: AN ELECTRON-SPIN RESONANCE STUDY* ABSTRACT Membrane structure in intact human erythrocytes was analyzed by electron-spin resonance (ESR) Spectroscopy. The spin probes 5-doxyl stearate and 5-doxyl stearate methyl ester revealed thermally-induced structural transitions in the membrane at 37°C and 15°C. The addition of propranolol, diazepam, chlorpromazine, or Pluronic F68 all caused a decrease in temperature of the upper transition, but did not markedly alter the temperature of the lower transition. In addition, diazepam caused a significant decrease in the ordering or packing of the membrane-lipid acyl chains. It is proposed here that the protection from hypotonic hemolysis that has been reported in the presence of these drugs is mediated by a structural rearrangement in the erythrocyte membrane involving a change in protein-lipid interactions. *Published in American Journal of Hematology (1981) 10, 171-179. 171 172 INTRODUCTION Present approaches to the treatment of hemolytic disorders are limited to both in therapeutic principle and in effectiveness. They depend mainly on the use of agents such as corticosteroids, for which the mechanism of action is uncertain, or on the prolongation of survival of damaged cells by splenectomy. It is axiomatic that hemolysis results ultimately fran structural failure in the erythrocyte membrane: attempts by Brewer (1) to diminish irreversible sickling by divalent ion substitution-suggest that erythrocyte resistance to hemolysis may be modified therapeutically. The present study examines the hypothesis that there are topographical sites in the membrane that are especially critical with respect to hemolysis and that might be modifiable with therapeutic effect. The rapid removal from the circulation of cells showing prehemolytic damage mandates an indirect approach to the problem; consequently, we have done an initial investigation in vitro on the structural perturbations of the erythrocyte membrane induced by several agents affecting the mechanical and osmotic resistance of red cells. It is anticipated that this may ultimately lead to the identification of susceptible membrane structures that are affected in common in diverse hemolytic processes. Osmotic fragility and mechanical deformability of human erythrocytes are reportedly altered by low concentrations of a variety of membrane-stabilizing drugs (2). Such drugs have the potential to protect erythrocytes from hemolysis and to increase deformability in the microcirculation, provided that ancillary effects in other systems can be 173 held within acceptable limits. There is at present insufficient understanding of the action of these compounds to define the optimal molecular form for erythrocyte-related effects. Although protein-lipid interactions may be affected, alterations in membrane architecture have not been characterized in detail. One technique useful in studying membrane architecture involves incorporating electron-spin resonance (ESR) probes such as 5-doxyl stearate (505) into the lipid bilayer. The unpaired electron of the doxyl group absorbs microwave energy when the sample is inserted into a magnetic field. The Spectrum of the absorption reflects the structure and fluidity of the membrane in the region of the probe. Since 503 reportedly localizes in a lipid region that is closely associated with proteins and devoid of cholesterol in the human erythrocyte membrane (3,4), electron-spin resonance spectroscopy of erythrocytes labeled with 50$ should reveal structural information Specifically related to protein-lipid interactions. In this paper, we report that SDS- and 5-doxyl stearate methyl ester (SOS-ME)-labeled human erythrocytes exhibit discontinuities in the temperature dependence of ESR spectral parameters. These temperature-induced changes in membrane structure correlate with other reported temperature-dependent phenomena of erythrocytes that are presumed to reflect membrane structural changes (5-9). We report that a variety of membrane stabilizing agents affect the temperature dependence of these measured Spectral parameters. Thus, membrane stabilization may be the result of a structural rearrangement in the erythrocyte membrane predominantly involving a change in protein-lipid interactions. 174 MATERIALS AND METHODS Preparation of Cells Blood was obtained from healthy individuals by venipuncture; informed written consent was obtained from each donor. Anticoagulation was effected by defibrination. Using standard procedures, the buffy coat was removed and the cells were washed three times in 300 mOSm NaCl. To remove serum albumin and other potential binding substances of the Spin probe, the reserve serum was filtered using an Amicon ultrafiltration apparatus and Diaflo ultrafiltration membrane Um10 (molecular exclusion 10,000 daltons). Cells were washed once in serum ultrafiltrate and packed in fresh ultrafiltrate to a hematocrit of 70 e 2 ml/dl. The cells were analyzed within 18 hours of collection. Electron-Spin Resonance Spectroscopy The spin probes 505 and 50$-ME (Syva Corp., Palo Alto, California) were dissolved as a 30 mM solution in absolute ethanol. The labeling procedure used was described previously (10), except that the ethanol was evaporated prior to the addition of erythrocytes. Drugs were added following labeling. To standardize drug concentrations with respect to the stabilizing effect, concentrations were selected that caused approximately equivalent degrees of submaximal protection against hypotonic hemolysis (11,12). The final probe concentration was approximately 0.4 pmoleS/mg of membrane protein. At this concentration, 505 has been reported to minimally perturb erythrocytes (13) and to localize in lipid domains in close association with proteins (3,4). 175 All ESR studies were performed by standard methods as described previously (10). The spin labels used incorporate into the membrane in such a manner that the unpaired electron of the doxyl radical is situated close to, but shielded from, the aqueous phase. A typical spectrum of erythrocytes labeled with SDS is shown in Figure 1. The distance between the low-field and high-field microwave absorption peaks, 2T" (the hyperfine splitting parameter) reports the rotational mobility of the probe and therefore the viscosity of the surrounding environment. With increasing temperatures, intermolecular influences upon the unpaired electron of the probe result in lower values of 2T" which reflect a more fluid environment. The hyperfine Splitting parameter 2T" is therefore directly related to the viscosity of the environment from which the probe is reporting. High values of 2T" indicate rigid environments, while low values of 2T" indicate more flexible environments (14). In studies reported here, the directly measured parameter, ZTL, could be determined in 50$-labeled preparations above about 12°C. This parameter is used along with 2Tn to calculate the order parameter S (14). The order parameter measures the deviation of the observed ESR signal from the case of a completely unifonm orientation of the probe. For a uniformly oriented sample, S = I; for a random sample, S = O (14). The hyperfine coupling constant also calculated from the directly measured 2T" and 2TL_parameters is considered to reflect local polarity (15) and thus reflects the position of the probe within the membrane. 176 Data Analysis The data (2T, vs. temperature, 5 vs. temperature) were analyzed in terms of linear components by an iterative least-squares program to be described elsewhere (Brunder, D.G., Coughlin, R.T., McGroarty, E., in press). Briefly, a B-Spline (a piece-wise set of polynomials that are smooth at the points of connection) was used to provide a fit for the ESR data and points of inflection were used to group data. Regression lines were calculated for each group and break points were determined. This analysis has been Shown in other membrane systems to permit characterization of subtle changes (16). All ESR spectral parameters changed with temperature reversibly up to 48°C. For each set of data, blood samples were examined from at least two healthy, nonsmoking individuals. RESULTS Control Erythrocytes ESR spectra of 50$-labeled, intact erythrocytes showed little or no free probe in the supernatant, indicating that its site is predominantly in the cellular phase (Figure 1). The shape of the spectra indicated that the majority of the fatty acid label was in a single environment over the temperature interval examined (0-48°C). The hyperfine splitting parameter, 2T", decreased in a discontinuous fashion as a function of temperature, and a break point was determined to occur at 37°C (Figure 2a). Spectra recorded at cuvette temperatures above approximately 12°C allowed the determination of the order parameter, S. When 5 was plotted as a function of cuvette temperature, a similar discontinuity was 177 observed (Figure 3). The transition temperature determined with the order parameter agreed quite well with that detenmined using 2T" and presumptively indicates a structural change in the erythrocyte membrane. A second Spin probe 50$-ME was also used to analyze erythrocyte membranes. This uncharged probe causes less perturbation of the cells. Spectra of 50$-ME-labeled cells allowed detennination of the hyperfine coupling constant, which indicated that the probe was in an environment similar to that of SOS-labeled cells at temperatures up to approximately 28°C. Above that temperature, the Spectra changed in such a manner that the spectral parameters were difficult to measure. In Spectra of SDS-ME-labeled cells recorded between 0°C and 28°C, 2T“ was shown to decrease discontinuously with a break at approximately 15°C (data not shown). This discontinuity is presumed to reflect a second structural change in the erythrocyte membrane that occurs at lower temperatures than the transition detected with 505. Therefore, it appears that erythrocyte membranes exhibit two thermotropic transitions, one at 15°C and a second at about 37°C (Table 1). Effects of Membrane StabilizlgggAgents on Membrane Structure To further characterize erythrocyte membrane-mediated phenomena, we analyzed the changes in membrane structure induced by compounds known to protect human erythrocytes from osmotic hemolysis, pr0pranolol, diazepam, chlorpromazine (11), and pluronic polyols (polyoxypropylene-poly-oxy- ethylene condensates) (12). The addition of any of these drugs did not appreciably alter the position of the Spin probes as detennined by the hyperfine coupling constant. Propranolol (Sigma Chemical Corp.) when added to 50$-labeled erythrocytes caused a slight decrease in the upper 178 transition temperature as measured using 2T" and 5 (Table 1). In addition, a low temperature transition could be detected with 505 when propranolol was added (Figure 2b). The temperature of the lower transition detected with 50$-ME was not significantly affected by the addition of propranolol (Table 1). Similar studies were carried out with labeled cells in the presence of diazepam, chlorpromazine, or Pluronic F68. All of these agents reduced the upper transition temperatures but did not appreciably alter the lower transition temperature (Table 1). Furthermore, the addition of diazepam or chlorpromazine (but not Pluronic F68) permitted the detection of the low-temperature transition in 50$-labeled cells. In addition, the presence of diazepam was shown to significantly alter the order parameter as indicated in Figure 3. The other perturbants induced only slight changes in the order parameter. The addition of diazepam caused a decrease in S by as much as 2% at low temperatures; changes greater than 1% in S are regarded as significant (17,18). DISCUSSION Numerous investigations employing a variety of techniques have reported temperature-dependent changes in erythrocyte membranes. Considering the diversity of approaches utilized, the temperatures at which changes were found to occur are remarkably similar to those reported here. Thus quenching of intrinsic tryptophan fluorescence by Spin labels showed discontinuities at 15°C and 35°C (4). Laser-raman Spectroscopy (5) and viscosimetry (6) revealed a discontinuity at approximately 18°C. A discontinuity in the susceptibility of human 179 at 37°C (8). Finally, the preservation of membrane lipid asymmetry by Mg+2 upon lysis has been Shown to diminish above 18°C and disappear at about 40°C (9). These data support the contention that temperature- induced structural transitions occur in erythrocyte membranes and can be detected using spin-labeling techniques. Apparently, these transitions occur in local areas not influenced by cholesterol. Data presented here indicate that there are two thermally induced structural transitions in human erythrocyte membranes that can be detected using the ESR probes described above; one of these is found at 15°C, and the other at 37°C. These transitions are presumed to be associated with changes in the conformation of either membrane proteins, phospholipids, or both. Two independent Studies, one involving quenching of intrinsic tryptophan fluorescence (2,3), the other involving the detennination of binding affinities (19), support the contention that the spin probe 505 is closely associated with erythrocyte membrane protein. The highatenperature transition demonstrated in this study, therefore, appears to be associated with protein-lipid interaction. In all cases, the addition of the membrane stabilizing drugs to the spin-labeled erythrocytes predominantly affected the high but not the low transition temperature. Since 50$ reportedly localizes in close proximity with membrane proteins and since lipids interact more strongly with proteins at elevated temperatures (20), it appears highly probable that these agents perturb protein-lipid interactions. In fact, stabilization of the erythrocytes by other drugs has been reported to require an intact membrane protein structure (21). Addition of propranolol, chlorpromazine, and diazepam (but not Pluronic F68) to erythrocytes allowed for the determination of a 180 erythrocytes to benzyl-lysolecithin has been reported to occur at about 15°C (7). Transfer of phospholipid from hemagglutinating virus of Japan to erythrocyte membrane was reported to begin at about 19°C and saturate low-temperature transition using 505 and caused hemolysis at high concentrations in isotonic saline (data not Shown). Since Pluronic F68 did not cause hemolysis at high concentrations, it is probable that its structural interaction with the membrane has unique features that may be especially important to its effectiveness as an antihemolytic agent. Diazepam was shown to significantly alter membrane order. We are aware of at least one instance in which a drug-induced decrease in erythrocyte membrane order (17) can be correlated with a protein conformational change (22). Propranolol, diazepam, chlorpromazine, and Pluronic F68 have been reported to protect erythrocytes against hypotonic hemolysis. Preliminary evidence (manuscript in preparation) indicates that membrane structural changes induced by some of these drugs are maximal at concentrations that cause maximal protection to hypotonic hemolysis. Thus, it is probable that these compounds affect Specific membrane domains such as Sites of lipid-protein interaction and that these domains are especially significant to cell fragility. The marked reduction of intact cell osmotic fragility and the increase in mechanical deformability that is brought about by an elevation in temperature, or by the presence of perturbants such as the drugs used in this study (17,23,24), might therefore be mediated by specific membrane structural changes. If the relevant changes, as suggested here, are related to lipid-protein interactions in the erythrocyte membrane, there is the possibility of developing more highly specific agents to modify these 181 structures in a controlled fashion. Such agents might find wide application in a variety of disease states in which red cell destruction or erythrocyte perfusion in the microcirculation are critical factors. CONCLUSIONS The human erythrocyte membrane was shown by biophysical probing techniques to undergo structural changes at 15°C and 37°C. Addition of membrane stabilizing drugs caused a decrease in the high-temperature transition but did not appreciably alter the structural change detected at lower temperatures. Furthermore, one of these drugs, diazepam, caused a significant disordering of the membrane lipids. It is proposed that these drugs are altering the membrane structure by changing the lipid-protein interactions within the membrane. REFERENCES 1. Brewer, G.J. (1979) Perspect Biol. Med. 22, 250. 2. Seeman, P. (1972) Pharmacol Rev. 24, 583. 3. Bieri, V.G. and Wallach, D.F.H. (1975) Biochim. Biophys. Acta 496, 415. 4. Bieri, V.G. and Wallach, D.F.H. (1976) Biochim. Biophys. Acta 54;, 198. 5. Verma, S.P. and Wallach, D.F.H. (1976) Biochim. Biophys. Acta 436, 307. 6. Zimmer, G.H., Schinmer, H., and Bastian, P. (1975) Biochim. Biophys. Acta 491, 244, 1975. 7. 8. IO. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 182 Weltzien, H.U., Arnold, B., and Kalkoff, H.G. (1976) Biochim. Biophys. Acta 455, 56. Maeda, T., Asano, A., Ohki, K., Okada, Y., and 0hnishi, S.-I. (1975) Biochemistry 14, 3736. Tanaka, K.-I. and 0hnishi, S.-I. Biochim. Biophys. Acta 426, 218. Janoff, A.S., Haug, A., and McGroarty, E.J. (1979) Biochim. Biophys. Acta 555, 56. Olaisen, B. and Oye, I. (1973) Eur. J. Pharmacol. 22, 112. Gaehtgens, P. and Benner, K.U. (1973) Eur. J. Physiol. 353, R1. Bieri, V.G., Wallach, D.F.H., and Lin, P.S. (1974) Proc. Natl. Acad. Sci. USA 1;, 4797. Griffith, 0.H. and Jost, P.C. (1976) In: Spin Labeling: Theory and Applications, Berliner, L.J. (ed.), p. 454, Academic Press, New York. Hamilton, C.L. and McConnell, H.M. (1968) In: Structural Chemistry and Molecular Biology, Rich, A. and Davidson, N. (eds.), p. 115, W.H. Freeman, San Francisco. Janoff, A.S., Coughlin, R.T., Racine, F.M., McGroarty, E.J., and Vary, J.C. (1979) Biochem. Biophys. Res. Commun. 82, 565. Kury, P.G., Ramwell, P.W., and McConnell, H.M, (1974) Biochem. Biophys. Res. Commun. 56, 478. Huestis, W.H. and McConnell, H.H. (1974) Biochem. Biophys. Res. Commun. 51, 726. Shiga, T., Suda, T., and Maeda, N. (1977) Biochim. Biophys. Acta 466, 231. Mateu, L., Caron, F., Luzzati, V., and Billecoeq, A. (1970) Biochim. Biophys. Acta 598, 109. 21. 22. 23. 24. 183 Mizushima, Y., Sakai, 5., Yamaura, M. (1970) Biochem. Pharmacol. 12, 227 - Meyer, M.B. and Swislocki, N.I. (1974) Arch. Biochen. Biophys. 1_6_4, 544. Mortensen, E. (1963) Acta Med. Scand. 1_7_§, 693. Murphy, J. (1969) J. Lab. Clin. Med. 11, 319. 184 ‘0 fl‘ ‘1 L.— Figure 1. Electron-spin resonance spectra of human erythrocytes labeled with 5-doxyl stearate. The spectra were taken at the temperatures indicated. Scan range was 100 gauss. Absence of free probe is indicated by the lack of a major absorbtion signal at points indicated by arrows. Symmetry of high and low field peaks reveals that the probe is present in a single environment in the membrane. 185 o 16 2'0 3'0 49 5b TEMPERATURE °C 0 1b 2b 30 .40 $0 TEAAFIELAJINUE‘N: Figure 2. Hyperfine Splitting parameter, 2T" (Gauss), as a function of temperature in erythrocytes labeled with 5-doxyl stearaze in the absence of perturbants (A), and in the presence of 5 x 10' M propranolol (B). Arrows indicate the transition temperatures. 186 Figure 3. The temperature dependence of the order parameter, S, in erythrocytes labeled with S-doxyl stearate in the absence of perturbants (closed circles) and in the presence of 3 x 10'4 M diazepam (open circles). 187 00 m¢3p<¢mm< 215 J I A iii 16 15 {a ‘11 20 25 30 35 TeurenATune I‘m " 26 2‘5 3‘0 3'5 Taupennune (°C)