- if: r
Ravi.
‘1. I ..

III].
.I. (Q.- 4. .,

.nri liar-It b
. .. l. I,
..e!ych-.a))l..‘ L.- ‘ uni.-
. 7 .l I O.)

a :rallivn‘whl.
.o I :1.
its... .2

4*

.-
. , 83,563,551... f):
hum: y .
094»... 5"»... #3...) J.
uveti ... 2.3:.)

r)
«3

y 4“,
1.2.1; ..
. 5....
. . Synlfau
. ‘ :0 u,....5
‘ . .. 1.55,...
2.... 3: .

‘1 .
i§§i§:¢:£!

fi!‘
53;!
142%

:ng ."
Jr
’51: §

. Ill. 4
.x. 3!. .Va.
2:5 A-‘Iil

.)

ail o:
57., .v.
:5! .l
.ir .-

 

I.
NIVERSITY LIBRARIES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘Illllllllllllll

This is to certify that the
dissertation entitled

Dissection of Endotoxin Biological Activities
Using Variant Lipid A and Synthetic Antagonists

presented by

Deborah Ann Lill-Elghanian

has been accepted towards fulfillment
of the requirements for

Ph.D. Biochemistry

degree in

 

 

 

Date M] “(In

 

MSU is an Affirmatiw Action ’Equul Opportunity Institution 0 12771

. A

 

 

 

 

 

’ \
LIBRARY
Mlchixan State

L University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

———-

JUN 0 7 19%

q
. v‘.L “ 9' ‘ll
' a

 

[E

 

 

 

'LUS '2. 7 i3}

 

 

WP: n 21998

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU It An Affirmative Action/Equal Opportunity Institution
W

OMS-D. 1

DISSECTION OF ENDOTOXIN BIOLOGICAL ACTIVITIES USING VARIANT
LIPID A AND SYNTHETIC ANTAGONISTS

, By
Deborah Ann Lill-Elghanian

A THESIS
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCF OR OF PHILOSOPHY

Department of Biochemistry
1992

ABSTRACT

DISSECTION OF ENDOTOXIN BIOLOGICAL ACTIVITIES USING VARIANT
LIPID A AND SYNTHETIC ANTAGONISTS

By
Deborah Ann LiIl-Elghanian

The lipid A moiety of lipopolysaccharide is responsible for both the beneficial
and detrimental effects caused by endotoxins in infected mammalian hosts. The
mechanism by which lipid A exerts its effects has not been fully elucidated.

We have investigated the mechanism of endotoxin action by studying a lipid A
molecule that has a variant structure from that of the classical endotoxins. The lipid A we
studied is from a soil bacterium, Rhizobium Info/ii ANU843. The lipid A of R. trifolii
had previously been known to possess some of the classical endotoxin effects, yet it was
soon discovered that this lipid A had a radically different structure. By determining the
structure of the lipid A of this variant form and comparing it to a classical endotoxin, in
terms of structure and conservation of biological effects, we have been able to propose a
mechanism for endotoxin activity.

In studying the lipid A structure of R. trifolii AN U843, we used several chemical
derivatization methods to determine its carbohydrate and fatty acid composition.
Physical methods employed in the structural elucidation of these molecules included 1H-
and 13C—NMR, mass spectrometry, and gas chromatography. Many separation methods
such as TLC, HPLC, and gel permeation chromatography were used.

Several in M9 and i_n m assays were used to answer the question concerning

conservation of endotoxic activity. Endotoxin inhibitors were also tested for their ability

to prevent the effects of lipid A, based upon electrostatic and hydrophobic interactions.

The data generated by the in vitro and i_n vivo assays, and the structural elucidation

 

experiments, provided insight into the mechanism of activity of bacterial endotoxins.

We propose a two-step model for the mechanism of endotoxin action. The first
step involves adsorption of the LPS aggregate to the cell surface governed by charge
interaction. LPS, being negatively charged, associates with a positively charged species
on the cell surface. This association increases the probability of intercalation of LPS
monomers into the cells' plasma membrane. This coalescence is the second step of LPS/
target cell interaction. Intercalation of lipid A decreases the fluidity of the plasma

membrane and this decrease in membrane fluidity leads to cellular activation.

This work is dedicated to my grandfather, the late Joseph W. Ulch, a man who knew
the value of an education.

ACKNOWLEDGMENTS

I would like to thank my mentor, Dr. Rawle I. Hollingsworth, for allowing me to
work with him and learn from him, during the past four years. RIH, thank you for all of
your support. I would also like to thank the members of my guidance committee, Dr.
Estelle McGroarty, Dr. Paul Kindel, Dr. Charles Sweeley, Dr. Walter Esselman, and Dr.
Pamela Fraker, for their help and suggestions over the years. A special thank you to Dr.
McGroarty for just listening and helping me sort things out.

I would like to thank all of my friends from the Hollingsworth lab. You all are a
nice group of people. A special thanks to Mr. Kim for being such a good friend to Robert
and myself.

I would like to thank those persons inside the Biochemistry department and
outside the department who have helped me in this endeavor. Thank you to Patty V055
and Beth Garvy for teaching me about cell culture and photography; to Dr. Sweeley for
allowing me use of your cell culture facilities; to Laurie Iciek and Dr. Kathy Brooks for
helping me out with the immunoglobulin assays; to Dr. Long Le (Long for short!) for the
NMR training; to Dr. JoAnne Whallon for the 3-d imaging and use of the confocal micro-
scope; and to Barb Hamel and Dr. McGroarty for allowing me the use of the autoclave,
pH meter,

I would like to thank my parents, Frank and Janet Lill, my sisters, Rebecca and
Jennifer, and my grandparents, Joseph and Hazel Ulch, for their support and encourage-

ment during this endeavor.

Lastly, I want to thank my husband, Robert Elghanian, for all of his love, support,
and encouragement during the course of our stay here. Robert, it was not easy and I

could not have done it without you.

Thank You All,
DALE

vi

TABLE OF CONTENTS

List of Tables
List of Figures
List of Abbreviations

Chapter One
Literature Review and Historical Background

Endotoxin

Lipopolysaccharides-Structure and Biological Activity
Structure of Lipid A

Lipid A Analogs

Unusual Lipid A

wewwr

Chapter Two
Literature Review of Mechanisms of Endotoxin Action

1. Hydrophobic Interactions

2. LPS Binding Proteins

3. LPS Receptors which are Membrane Localized
4. Mechanism of LPS Action

Chapter Three
Literature Review of Molecular Events Resulting From the

Interactions Between Endotoxin and Target Cells

1. The Macrophage and Cytokine Release
2. Signal Transduction

Chapter Four
Rhizobium Lipid A Structure

1. Introduction

2. Materials and Methods
2.1 Bacterial Cultures and LPS Isolation
2.2 Lipid A Isolation
2.3 Purification of Lipid A
2.4 Fatty Acid Analysis of Crude and Purified Lipid A Components
2.5 Carbohydrate Analysis of Crude and Purified Lipid A Components
2.6 NMR and Mass Spectra

vii

xi

xix

IO
IO
13

l 8
22
24
26

28
29

32
33
33
34
34
35
35
36

2.7
2.8

Phosphate Analysis
Preparation of Phenylcarbamate Derivatives of Crude Lipid A

3. Results and Discussion

3.1
3.2
3.3
3.4
3.5

Phosphate Analysis

Sulfate Analysis

Crude Lipid A IH-NMR

Fatty Acid Analysis

Carbohydrate Analysis of R. trifolii Lipid A

4. Conclusions Regarding Rhizobium trifolii ANU843 Lipid A Structure

Chapter Five
Biological Responses Induced by Rhizobium tnfolii ANU843 LPS

1. Introduction
2. Materials and Methods

2.]

LPS Isolation

USP XXII Pyrogen Assay

Lethal Toxicity Testing

Limulus Amebocyte Lysate Assay

B-Lymphocyte Proliferation

Immunoglobulin Secretion

Production of TN F

Release of Interleukin, IL- 1, IL-6, IL—8, by Rhizobium LPS
Cell Lines Used in LPS Binding Assays

2.10 Conjugation of Fluorescent Dyes to LPS

2.11 Labeling of Cell Lines with LPS- Dye Conjugates
2.12 Fixation of Cells labeled with LPS-Dye

2.13 Fluorescent Microscopy

2.14 Confocal Microscopy

3. Results
3.1

LAL Activity

Proliferation Assays

Immunoglobulin Secretion

TNF Production

Interleukin Production

Pyrogenicity

Toxicity

LPS Binding

Endotoxin Inhibitors (Els) in the LAL Assay

3.10 Effects of HS on RAW264. 7 Cell Growth
3. l 1 The Effect of Els on Labeling with Sal- FITC and R.t.-FITC When

Cells are Grown in the Presence of Els

3.12 Effect of Els on Labeling: Els Used in Binding Assays
4. Discussion and Conclusions

viii

36
37
37
37
38

38
38

7O

List of References 141

ix

Chapter Five
Table 1:

Table 2:

Table 3:

LIST OF TABLES

The mitogenic response of R. trifnlii LPS, harvested at different
growth periods, as measured by 3H-thymidine uptake. Experi-
ments were performed by D. M. Jacobs. K90=R. trifolii 0403
grown to early stationary phase. K50=R. trifolii 0403 grown

to exponential phase. K90 and K50 are both mitogenic, however

K90 is more so than K50. 92

USP XXII Pyrogen Assay comparing fever induction between
R. trifolii ANU843 LPS and S. typhimurium LPS. A result is
considered positive if one rabbit in three shows a temperature
06°C or greater than its control temperature, or if the sum of
the three rabbit temperature rises exceeds 14°C. In either case,
a repeat test is done on five additional rabbits. If not more than
three of the total of eight rabbits have individual increases of
06°C or more, and if the sum of the eight increases does not
exceed 37°C, the material under examination meets the require-

ments for absence of pyrogens. 102

Lethal toxicity in mice comparing R. mfolii ANU843 LPS with
Salmonella typhimurium LPS. l04

LIST OF FIGURES

Chapter One
Figure 1: The structure of a gram—negative bacterial cell envelope. 2
Figure 2: Biological phenomena initiated by lipopolysaccharides as

modified from Westphal, O., Luderitz, O., Galanos, C., Mayer,
H., and Rietschel, E. Th. (1986) Adv. Immunopharmacol. 3: 13. 4

Figure 3: Schematic structure of LPS indicating the hydrophobic and hydro-
philic regions. From "Current Topics in Membranes and

Transport”, Vol. 17, (1982), p.81. 7

Figure 4: The structure of free lipid A from Salmonella minnesota. 9

Chapter Four
Figure 5: Gel filtration profile of R. trifnlii ANU843 LPS on a column
of Sephadex LH-60 resin. Fractions were screened for the

presence of phosphate. 39

Figure 6: Proton NMR spectrum of the crude lipid A of R. trifalii
ANU843. 40

xi

Figure 7:

Figure 8:

Figure 9:

Figure 10:

Figure 11:

GC profile of fatty acid methyl esters derived from R. trifolii.
(A) 3-hydroxytetradecanoic acid, (B) 3—hydroxy-12-methyl-
tetradecanoic acid, (C) 3-hydroxyhexadecanoic acid, (D) 3-

hydroxyoctadecanoic acid, (E) 27-hydroxyoctacosanoic acid. 42

8A: GC profile of fatty acid methyl esters derived from

R. mfolii ANU843. Upper case letters represent the same

fatty acids as in Figure 7.

8B: GC profile of R. trifolii ANU843 fatty acid methyl esters
previously derivatized with phenylcarbamate. Lower case letters

are indicative of the corresponding phenylcarbamate ester. 44

Electron impact mass spectra of both the 9A: phenylcarbamate
derivatized 3-hydroxytetradecanoic acid, and 9B: 3-hydroxytetra-

decanoic acid, indicating it is linked at the 3-OH position. 45

GC analysis of the alditol acetate derivatives of carbohydrates
present in the crude lipid A of R. trifalii ANU843. Peaks A and
B correspond to the alditol acetate derivatives of galactose and

glucosamine, respectively. 47

GC tracings of fatty acid methyl esters of fatty acid components
liberated during KOH saponification. (Letters represent the same

fatty acids as in Figure 7.) 48

xii

Figure 12:

Figure 13:

Figure 14:

CC profile of R. trifolii ANU843 fatty acid methyl esters derived
by treating crude lipid A with methyl sulfonyl anion and methyl
iodide. Note the absence of 27-hydroxyoctacosanoic acid, indic-
ating its amide linkage. Letters correspond to the fatty acids

marked in Figure 7. 50

13A) GC/MS fragmentation pattern for prereduced galacturonic
acid. The ion fragments are two units apart for all major ions,
indicating one end of the molecule was labeled with deuterium,
and therefore the reducing end is tied-up as a glycoside.

l3B) GC/MS fragmentation pattern of glucosamine. The ion
fragmentation pattern shows some of the carbohydrate to be free,
as evidenced by the presence of isotope peaks of equal intensity
for the smaller mass ions. Some carbohydrate was found in gly-
cosidic linkage, as evidenced by identity of peaks with those of

the glucosamine standard not containing deuterium. 53

Iii—NMR spectrum for R. trifalii ANU843 lipid A treated with
phenylisocyanate, indicating the presence of ethanolamine at the

anomeric site. The inset is an expanded spectrum of the R. trifnlii

'Ji
U!

lipid A treated with phenylisocyanate.

xiii

Figure 15:

Figure 16:

Figure 17:

Figure 18:

Figure 19:

Proton NMR spectrum of the methanol fraction obtained from
C18 reverse-phase chromatography of R. trifalii ANU843 crude
lipid A. Fatty acid proton resonances occur between 0.7 and 2.5

ppm. Carbohydrate resonances are between 3.2 and 5.3 ppm. 58

13C-NMR spectrum of the methanol fraction from C 18 reverse-
phase chromatography. Signals at 99.99 and 97.96 ppm are due to
B- andL-forms of the anomeric carbon of the glycosyl component.
The signals in the 60-80 ppm region confirm the presence of one
glycosyl component. There is one carbon signal at 54.1 ppm

attributable to a carbon bearing nitrogen. 61

Proton NMR spectrum of the faster moving band on preparative
TLC analysis of lipid A. The triplet at 5.21 ppm present in the
parent compound is missing, indicating the lack of acyl substitution.
Intense signals between 3.1 and 3.4 ppm are due to traces of

methanol and other hydroxylated species such as moisture. 63

Fast atom bombardment mass spectrum of the faster moving TLC
band. The pseudomolecular ions appear at m/z=616 ([M+H]+)
and m/Z=638 ([M+Na]+). 65

Proton NMR spectrum of the slower moving band of preparative

xiv

Figure 20:

Figure 21:

Figure 22:

Chapter Five

Figure 23:

TLC. Note the presence of a triplet at 5.22 ppm, indicative of

acylation.

Fast atom bombardment mass spectrum of the slower migrating
TLC band. Pseudomolecular ions appear at m/z=842 and
m/z=864.

Structure of 2-amino-2-deoxy—2-N-(27-hydroxyoctacosanoyl)-
3-O-(3-hydroxytetradecanoyl)-gluco-hexuronic acid and its
de—O—acylation product when R=CH3-(CH2)10-CHOH-CH2-CO;

R=H, respectively.

Structure of a possible lipid A component or substructure from
R. trifolii ANU843 LPS. Notice the two shorter acyl chains

meet in the middle of the long chain, transmembrane fatty acid.

67

69

72

76

Standard curve of absorbance versus concentration in endotoxin units

(E.U.s)/mL for E. coli 0111:B4 LPS. The point connected by the

broken lines corresponds to the average of a triplicate ANU843 LPS

sample at 3.8x10'2 ng/mL. The third point on the standard curve is
at a similar weight/mL concentration and has a similar value for

endotoxin units/mL. Absorbance is proportional to the amount of

XV

Figure 24:

Figure 25:

Figure 26

Figure 27:

Figure 28:

Ft gure 29:

endotoxin present. These are the results of the LAL assay. 94

BCLlcloneSBlb lymphocytes were treated with LPS. Salmonella

LPS was capable of inducing I gM secretion whereas Rhizobium

LPS was not. 96

Comparison of TNF production induced by R. trifolii ANU843
lipid A and E. coli J5 lipid A. 98

Comparison of IL-8 production induced by R. trz'folii ANU843
lipid A and E. coli lipid A. 99

27A) Phase contrast and confocal microscopy of RAW264.7 cells
labeled with Salmonella LPS—Rhodamine conjugate.

27B) Phase contrast and confocal microscopy of CHO-K1 cells

labeled with Salmonella LPS—Rhodamine conjugate. 108

Phase contrast and confocal microscopy of RAW264.7 cells labeled
with R. trifolii LPS-FITC conjugate. 110

3-dimensional imaging, shown in various orientations, of
fluorescence localization in a RAW264.7 cell. The Sal-Rho is
localized in vesicles throughout the cell, but appears at higher

frequency in the perinuclear region. 1 l3

xvi

Figure 30:

Figure 31:

Figure 32:

Figure 33:

Figure 34:

The structures of the endotoxin inhibitors and polymyxin B
sulfate are shown. The number 7 used in El-3 and E14 indicates
that 7 sugar residues are linked together in these cyclodextrin

structures. 1 16

The ability of the E15 to inhibit endotoxin activity was measured

by their ability to prevent a positive reaction in the LAL assay. 1 18

32A) RAW264.7 cells grown in the presence of El-l, followed
by labeling with Sal-FITC.

32B) RAW264.7 cells grown in the presence of EM, followed
by labeling with R.t.-FITC. Notice the location of the R.t.-FITC

conjugate. 123

33A) RAW264.7 cells labeled with a mixture of Sal-FITC and

10x EM.

333) RAW264.7 cells labeled with a mixture of R.t.-FITC and

10x EI-l. 125

RAW264.7 cells labeled with Sal—Rho. Notice the vesicles of

LPS-dye conjugate found inside the cell. 128

xvii

Figure 35:

Figure 36:

Attachment of an LPS aggregate to a lipid bilayer, adapted from

Shands (adapted from "Bacterial Toxins and Cell Membranes",

1978, Academic Press).
35A) Disaggregation of an LPS bolus, followed by insertion of LPS

monomers into the target cell bilayer.
35B) Disaggregation of the R. trifolii LPS bolus, followed by

monomer insertion. Notice the position of the 27-hydroxyoctacos-

anoic acid. 134

The lipid A of R. sphaeroia’es ATCC 17023, adapted from

Qureshi, _e_t_._ al_. (50).

xviii

ATCC

B III

BPI
C3H/HeJ
CSF
DPL

El

FA B/MS
FBS
FITC
ELISA

GC/MS
HBSS
IL-1
IL-6

NMR
PMB
RaRBC
RBC

LIST OF ABBREVIATIONS

American Type Culture Collection
Bergensen's medium

bactericidal permeablizing increasing protein
LPS hyporesponsive strain (mouse)
colony stimulating factor

diphosphoryl lipid A

endotoxin inhibitor

fast atom bombardment/mass spectrometry
fetal bovine serum

fluorescein isothiocyanate isomer I
enzyme-linked immunosorbent assay
gas chromatography

gas chromatography/mass spectrometry
Hanks' balanced salt solution
interleukin-1

interleukin-6
3-deoxy-D-mngg—2-octulosonic acid
Lima/us amebocyte lysate
lipopolysaccharide binding protein
lipopolysaccharide

monophosphoryl lipid A

nuclear magnetic resonance

polymyxin B sulfate

rabbit red blood cells

erythrocytes - red blood cells

xix

R.t.-FITC
Rho
Sal-FITC
Sal-Rho
SDS—PAGE

TFA
TLC
TNF

R. trifolii LPS — FITC conjugate

rhodamine B

S. ryphimurium LPS - FITC conjugate

S. typhimurium LPS - rhodamine conjugate

sodium dodecyl sulfate-polyacrylamide gel
electrophoresis

trifluoroacetic acid
thin layer chromatography

tumor necrosis factor

XX

Chapter One: Literature Review and Historical Background

Endotoxin

Life-threatening bacterial infections, that result in the death of an estimated 100,000
individuals per year in the United States alone, make septic shock the nation's 13th.
leading cause of death, as estimated by the Centers for Disease Control (1). The majority
of septic shock cases are caused by gram-negative bacteria. It is the endotoxin found in
the outer membrane of these organisms that brings about the rapid destruction
characteristic of sepsis. Sepsis claims a high number of lives because as yet, there is no
specific treatment for it. Antibiotics are currently prescribed to kill and clear the bacteria
from the bloodstream, yet this results in the release of more endotoxin from the dying
bacterium. It is the body's natural defense against these molecules that leads to the
condition known as sepsis. The original source of the problem is eliminated, but the
resulting inflammatory response continues out-of-control.

Pfeiffer first discovered endotoxins in 1892 and recognized them as being different
from exotoxins. Exotoxins are usually proteins which are heat labile and are inactivated
by heating at 60 - 80°C. They are secreted by both gram-positive and gram-negative
bacteria. Exotoxins can be converted to toxoids and be neutralized by antitoxins (3).
They exert a specific and restricted biological function affecting a specific cell type or
tissue.

Endotoxins are produced only in gram-negative bacteria. They are embedded in the
outer leaflet of the outer membrane where they exist as high molecular weight complexes
composed of polysaccharide and lipid (Figure 1). They remain attached to the cell
surface and are released into their surroundings during cell lysis or during normal surface
blebbing (2). Endotoxins are heat stabile and do not readily form toxoids. It is difficult
to neutralize them by anti-endotoxin antibodies ( l ). Endotoxins affect many different cell

types and tissues.

Matrix protein

Lipo otein
gig EB; \ MN no" ”“900

ddeddd(.dddt£ regattas?

tiiliililiifiit

Figure 1

 

 

 

 

ture of a gram-negative bacterial cell envelo e.

Endotoxins (lipopolysaccharides, LPS) initiate a variety of biological phenomena,
(Figure 2), and many review articles have been written concerning their biological activ-
ities (3,4,5,6). While some of these effects are considered beneficial (7,8,9,10,l 1), such
as induction of immunoglobulin synthesis, B—lymphocyte proliferation and anti-tumor
activity, these molecules are at the same time harmful. Endotoxins are known pyrogens,
responsible for tissue necrosis, vascular damage, and can result in the death of the
infected organism (12,13).

When endotoxins are released from the outer membrane, they consist mainly of lipo-
polysaccharide (LPS), complexed with various amounts of protein and some phospho-
lipid (14). Purified, protein-free LPS (15) was found to exhibit many of the biological
properties exhibited by unpurified endotoxin. Thus, LPS is the active moiety of bacterial
endotoxin.

Lipopolysacch_arides- Structure and Biological Activity

The outer membrane is an effective permeability barrier which is found in all gram-
negative bacteria. It is the exterior surface of these organisms and is attached to the pep-
tidoglycan layer through lipoprotein. The outer membrane is a lipid bilayer comprised of
an inner and outer leaflet, which form a "railroad track" appearance on electron micro-
graphs. LPS is found exclusively in the outer leaflet of the outer membrane of all gram-
negative bacteria.

Gram-negative bacteria are well protected by the outer membrane when they find
themselves in hostile environments. It makes them resistant to host defense mechanisms ,
such as lysozyme and cationic leukocyte proteins, which are successful in destroying
gram—positive organisms (16,17). Enteric bacteria are normally found in the intestinal
tract. Outer membranes of enterics are capable of providing protection to cells from the
detergent actions of bile salts and destruction via digestive enzymes (18). The outer

membrane is also able to provide resistance to many antibiotics that are effective when

BIOACTIVITIES OF LPS OR LIPID A'

Lethal toxicity
Pyrogenicity

Preparative and provocative activity for local Shwanz-

man reaction
Induction of hypothermia in mice
Induction of leukocytosis
Induction of bone marrow necrosis
Depression of blood pressure
Toxicity enhanced by BCG
Toxicity enhanced by adrenalectomy

Toxicity enhanced by galactosamine
Enhanced dermal reactivity to epinephrine

Platelet aggregation

Complement acrivation

Hageman factor activation

Induction of plasminogen activator

Limulus activity (activation of clotting enzyme cas-
cade of amoebocyte lysate of horseshoe crab)

Embryonic bone resorption

Type C RNA virus release from mouse spleen cells

Adjuvant (immunomodulating) activity
Increase of nonspecific resistance to infection
Induction of tumor necrosis

Induction of tumor necrosis factor (TNF)

Induction of interferon (lFN)

Induction of colony stimulating factor (CSF)

Induction of prostaglandin (PG) synthesis

Induction of tolerance to endotoxin

Induction of early refractory state to temperature
change

Somnogenic effect’-"°
Analgesic effect’

Mitogenic activity for B lymphocytes
Macrophage activation '
Polymorphonuclear leukocyte activation
Endotherial cell activation

Induction of mouse liver pyruvate kinase
Inhibition of phosphoenolpyruvate carboxykinase

Modified from Westphal, 0.. Li'tden'tz. 0.. Galanos, C., Mayer, H., and Rietschel, E. ‘11, Adv. Immuno-

pharmacol.. 3. 13, 1986.

Figure 2

Biological phenomena initiated by lipopolysaccharides as modified from Westphal, 0.,
Luderitz, 0., Galanos, C., Mayer, H., and Rietschel, E. Th. (I986) Adv.
Immunopharmacol. 3: 13.

5

used against bacteria lacking an outer membrane, such as macrolides, rifamycins, and
novobiocin (19). Thus the outer membrane serves as a protective barrier against
hydrophobic compounds. The surfaces of gram-negative bacteria are very hydrophilic.
Hydrophilicity is important in evading phagocytosis and allowing resistance to
complement (18). These hydrophilic and barrier protection mechanisms intimately
involve LPS. The LPS enables gram-negative bacteria to survive under hostile conditions
(20).

Lipopolysaccharides (LPS) are unique molecules located in/on the outer membrane of
gram-negative bacteria. LPS are amphiphilic molecules consisting of a hydrophilic
region composed of varying pol ysaccharide chains and a hydrophobic region composed
of lipid, the lipid A, (Figure 3).

The hydrophilic region is composed of the O-specific polysaccharide and the core
oligosaccharide. The O-polysaccharide is very diverse in that it is composed of repeating
oligosaccharide units which vary in composition between LPS isolates from different
species and even strains of the same species. Each repeating unit is usually composed of
four to seven carbohydrates that can be homogeneous or heterogeneous for the sugars
present. The number of repeating units can vary in length from zero to forty. An LPS
preparation of a given bacterial strain is always heterogeneous with respect to the O-
polysaccharide chain length (unless it is a rough mutant) and this is easily demonstrated
by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS—PAGE). SDS-
PAGE of LPS results in development of ladder-like structures. The ladders correspond to
the presence of LPS molecules with O-polysaccharide chains differing in size by
multiples of oligosaccharide repeating units. fit is the O-polysaccharide which is
responsible for determining the O-serological specificity for a particular bacterial strain,
such as Salmonella, Escherichia coli or Shigella (2122)) The O-polysaccharide is also a
receptor for bacteriophage and thus is useful in bacteriophage typing (23).

The core oligosaccharide is the component which links the O-specific polysaccharide

Fig ure 3

Schematic structure of LPS indicating the hydrophobic and hydrophipic regions.
From ”Current Topics in Membranes and Transport”. Vol. 17 (1982), p. 81.

m Bani

use 3.: 382555 9::N 52,.
2:520:25” 8 .3232; 0.312830%? 0

 

 

 

37.23333

 

 

 

 

 

 

 

< 22.. _ ,

(zooooexitla o

 

 

 

J. :35 inflame

 

 

 

 

 

 

 

   

c
00000.
O 0

e. s s
s

   

8

(O-antigen) to the lipid A. The structure of the core is more conserved within any given
genus. The core oligosaccharide and the lipid A are linked by 3-deoxy-D-man_nQ-2-
octulosonic acid (KDO). The KDO linkages are very acid labile and are preferentially
hydrolyzed under mild acid conditions to allow separation of lipid A from the carbohy-
drates of the core and the O-antigen.

The hydrophobic portion of LPS is composed of the lipid A. The lipid A is the most
conserved structure of the LPS molecule. Its structure and composition is very similar
among the Enterobacteriaceae (22,24). Originally, the term lipid A was used to describe
the hydrophobic precipitate obtained when solubilized LPS was treated with mild acid.
Later this term was also used to designate the bound lipid component as it is present in
LPS. Currently the term free lipid A is used for the polysaccharide-free component that
is liberated during acid hydrolysis.

Much work has been devoted to the characterization of LPS. All three components,
the O-polysaccharide, the core oligosaccharide, and the lipid A moiety have been ex-
ensively studied. Let us now focus on the lipid A moiety of LPS.

As was stated earlier, the endotoxic properties of gram-negative organisms are
associated with the LPS. More specifically, it is the lipid A portion that is responsible for
endotoxic activity, whereas the hydrophilic polysaccharide regions are immuno-
stimulatory (24,25). This was first postulated by Westphal and Luderitz (26) in the early
1950's, when it was observed that lipid A was found to be the only component common
to all LPS of the Enterobacteriaceae . It has been shown that free lipid A elicits the same
biological effects as those exhibited by intact LPS (27,28).

Free lipid A isolated from LPS preparations of a wide variety of gram-negative
organisms shows a high degree of conservation of structure. Structure of the lipid A
molecule typically found in the Enterobacteriaceae has only recently been elucidated
(29,30). This molecule is a glycolipid with a general structure which is relatively

invarient from bacterium to bacterium (Figure 4).

Fi ure4

The structure of free lipid A from Salmonella minnesota.

10
Structure of Lipid A

Enterobacterial lipid A typically consists of a B- l ,6—linked glucosamine disaccharide
backbone (31,32) bearing phosphate ester substituents at positions 1 and 4'. Linkage at
the 1 position is alpha. In intact LPS the backbone is attached to KDO at the hydroxyl
group of position 6'. In free lipid A the 6' position is unsubstituted. Lipid A is substituted
by a variety of fatty acids at positions 2,3,2', and 3' on the disaccharide backbone. The
fatty acids are usually 3-hydroxy substituted (R-confi guration), with chain lengths
varying from 12 to 16 carbon atoms and are attached by ester and amide linkages. Many
times these 3-hydroxyl fatty acids are acylated at the 3-hydroxy1 group by other fatty
acids. Sometimes this is referred to as "piggy-backed” fatty acids. There is some
variation among species, with relatively small variations in number, identity, and location
of acyl substituents on the 3-hydroxylated fatty acid substituents. Also varying from
species to species are polar substituents such as phosphate, ethanolamine, or other carbo-
hydrate groups on the 1 and 4' phosphate substituents. Gram-negative organisms contain-
ing this lipid A structure include Salmonella, Proteus, Escherichia coli and Shigella.
There are many others that are included in this grouping of "classical" lipid As (32).
Salmonella lipid A, having been thoroughly studied, was taken as the prototype of lipid A
structure, thus the term "classical" or ”usual" lipid A.

The main structures present in the enteric lipid A can be found in other non-enteric,
gram-negative organisms. These lipid As also exhibit endotoxic activity and contain the
usual diphosphorylated B—l,6—linked glucosamine disaccharide. They also have ester- and
amide-linked fatty acids, but differ in chain length from the enterics. Bacteria such as
Pseudomonas aeruginosa, Vibrio cholera, and Bordetella pertussis fall into this

category.

Lipid A Analogs
The biological effects of LPS have been shown to be associated with lipid A. Many

11

chemical and biological studies have been undertaken, aimed at determining which
structural features of lipid A are responsible for the various biological effects (33,34).
Great emphasis has been placed on separating beneficial from toxic effects based solely
on structure. Information has been derived from the biological, chemical, and physical
analysis of defined natural lipid A preparations, lipid A precursors, and chemically
synthesized analogs based upon the natural isolates. Analogs were also prepared which
differ from their natural counterparts based upon changes in degree of phosphorylation,
acylation, and content of carbohydrate backbone. Most of this work has been performed
by research groups in Germany and Japan (28,32,33,34,40,41). The following is a
summary of their conclusions concerning structure and biological activity.

The conformation an endotoxin molecule adapts is based upon its structure and it is
responsible for its endotoxicity. The conformation is characterized by a balance between
hydrophilic and hydrophobic interactions based upon distribution of acyl and phosphoryl
substituents over the glucosamine disaccharide backbone (28). This amphiphilic
character allows the lipid A to be water soluble and at the same time capable of inter-
acting with cells of the infected host. The major lipid A constituents, acyl groups,
phosphoryl residues, and the glucosamine disaccharide backbone, participate in forming
this stable conformation recognized by the host.

Partially deacylated LPS/lipid A is less biologically active than its native counter-
parts. This has been shown in studies by Munford (25,35) where removal of fatty acyl
substituents resulted in decreased biological activity. Galanos e_t gL. (33) have shown that
a synthetic B- ] ,6-linked glucosamine disaccharide, which carries (R)-3-hydroxytetradec-
anoic acid residues at positions 2,3,2' and 3' and phosphate ester at l and 4' , exhibited
LAL activity, B-cell mitogenicity, and prostaglandin synthesis in macrophage of
comparable activity to lipid A precursor IVA and free lipid A. However, it lacked the
ability to induce a local Schwartzmann reaction and was of moderate pyrogenicity.

Rietschel e1 g1, (28) showed that bacterial and synthetic lipid A precursors containing four

12

3-hydroxy 14:0 residues did not elicit the toxic effects of endotoxin, while those analogs
having at least 5 acyl residues, including one 3-acyloxyacyl group were essential for
eliciting toxic effects such as a local Schwartzmann reaction. In this same study it was
shown that biological activity does not increase with addition of 3-acyloxyacyl groups
beyond two. Rietschel concluded that additional fatty acids disarrange the endotoxic
conformation, resulting in reduced bidlogical activity.

The fatty acids of the classical lipid A range in length from 12 to 16 carbon atoms.
Changing the length of these fatty acids, for instance by increasing carbon chain length,
results in decreased biological activity (36,39).

In regards to phosphorylation, it was observed that the diphosphate (DPL) substituted
analogs were water soluble. Monophosphoryl lipid A (MPL) was less soluble, and the
analogs totally devoid of phosphate were insoluble in water. Biological activity of a
homologous series of analogs increases as degree of phosphorylation increases.
Rietschel concluded that phosphate rendered the lipid A soluble and thus made it
available to interact with host cells.

Scientists at Ribi lmmuno Chem, Hamilton, MT, have done many studies on MPL.
They have found that MPL was the result of acid hydrolysis during lipid A isolation (37).
MPL was missing phosphate at the Cl position of the reducing end sugar of the
disaccharide backbone. It is considered a non-toxic lipid A, yet still acts as a B-cell
mitogen and can stimulate interleukin 1 (IL-1) release. Thus, it still possesses immuno-
logical properties. Scientists in Freiburg and Japan (33) found the monophosphate
analogs to be less toxic by a factor of 100. Whether phosphate was at the C1 or C4'
positions, the toxic properties of the analogs were not greatly different, suggesting neither
phosphate plays a role in toxicity. They were inclined to believe that lower toxicity in
these compounds was again due to decreased solubility.

The glucosamine disaccharide backbone was found necessary for endotoxic

activity. Monosaccharide lipid A precursors such as lipid X and lipid Y and chemically

l3

synthesized monosaccharide analogs elicited immunological activities jg y_i1r_o just as

lipid A did. These activities included acting as mitogens and the capability to gel a
Limulus lysate. These are activities that are also produced by other natural products such
as peptidoglycan and lipoprotein (38). The monosaccharide compounds were not capable
of induction of the toxic i_n m endotoxin effects such as fever production, local
Schwartzmann reaction, and lethal toxicity. It appears that expression of toxic properties
are caused by analogs containing the disaccharide backbone (28).

At this time there are several structural features that are believed to render lipid A
endotoxic. These features are: presence of a glucosamine disaccharide backbone, phos-
phorylation, and the presence of acyloxyacyl substituents. Additional substitutions such
as phosphorylethanolamine, non-hydroxylated fatty acids or carbohydrate substituents are
not required for the biological activities of the lipid A molecule. As of yet, no one has
been able to precisely relate structure with a specific biological activity. In analogs of
naturally occurring lipid A, biological activities were attenuated, but not totally
obliterated. The direct relationship between structure and activity is yet to emerge.
Unusgal Lipid A

Extensive research over the years has lead to the structure elucidation of the lipid
A molecule. Most of this work has been performed on the Enterobacteriaceae and thus
this type of lipid A is termed the usual or classical lipid A. The usual lipid A consists of
a glucosamine disaccharide backbone linked B—1,6. It is phosphorylated at positions 1
and 4' and is substituted with acyl chains in positions 2,3,2', and 3'. The usual lipid A is
heterogeneous in that it may vary slightly between genus and species, yet is still con-
sidered usual. These slight variances include further substitution of the phosphate groups
with polar substituents, variation in fatty acid chain length, and degree of acylation of the
3-hydroxy substituents.

Lipid As with deviations in structure from the usual lipid A, in regards to what

was discussed above, are termed "unusual” lipid As. Some of the structural variations

14
from the classical model found in unusual lipid A are listed as follows: (i) Substitution of

the disaccharide backbone with additional carbohydrate substituents, many times these
sugars are glycosidically attached to C4 of the reducing end glucosamine, (ii) The glucos—
amine disaccharide backbone is replaced by a different amino sugar, or is a mono—
saccharide instead of the usual disaccharide, (iii) Substitution by phosphate no longer
occurs, i.e. no phosphate is present, and (iv) The (R)-3-hydroxy fatty acid substituents are
replaced by the rare 3-(or 4—)oxo fatty acids, unsaturated fatty acids, or fatty acids con-
taining a cyclopropane ring.

Most of the unusual lipid As are considered to be nontoxic. They often have
reduced or are completely devoid of biological activity, as compared to the enteric lipid
A. Unusual lipid A was first discovered in purple phototropic bacteria, more specifically
among the Rhodospirillaceae family (43 ,44,45,46). The following is a description of just
a few of these unusual lipid As.

The lipid A of Rhodopseudomonas viridus and Rhodopseudomonas palustris
consists of a 2,3-diamino—2,3-dideoxy-D-glucose backbone. This lipid A backbone is a
monosaccharide, not disaccharide, and was the first isolation of this carbohydrate (43).
This unusual lipid A contains an amide-linked 3-hydroxytetradecanoic acid, but lacks
ester-linked fatty acids and phosphate. The lipid As of these species show no cross-
reactivity with Salmonella lipid A (47) and toxic effects such as pyrogenicity and lethal
toxicity are very low compared to those of Salmonella (44).

Rhodospirillum tenue contains the Salmonella lipid A backbone, but the phos-
phate groups are further substituted by 4—amino-L-arabinose and D-arabinofuranose
(42,48). An additional glucosamine is linked to C4 of the reducing end sugar. Its amino
group is free. This lipid A cross-reacts with that of Salmonella, and R. tenue LPS
exhibits moderate pyrogenicity and "cryptic” lethal toxicity. Cryptic toxicity refers to the
observation that R. tenue LPS has low toxicity, whereas the isolated lipid A has lOO-fold

higher toxicity (43). The basis of this phenomenon has yet to be identified.

15
Rhodopseudomonas sphaeroides ATCC 17023 has been shown to possess a

glucosamine disaccharide lipid A backbone and phosphate groups at C 1 and C4'.

R. sphaeroides has an unusual fatty acid composition. It possesses an amide—linked
3-oxotetradecanoic acid and 3-hydroxytetradecanoic acid, an ester-linked 3-hydroxy-
decanoic acid and 7-tetradecenoic acid. Serological cross-reactivity between R.
sphaeroides ATCC 17023 and S. minnesota R595 indicate structural similarity between
this unusual lipid A and enterobacterial lipid A. Neither lipid A nor LPS of R.
sphaeroides were found to be lethally toxic or pyrogenic (45).

Yet there are interesting findings concerning the unusual lipid A of R.
sphaeroides. R. sphaeroides lipid A could not induce tumor necrosis factor (TNF)
production in the macrophage cell line RAW264.7. However, the lipid A of R.
sphaeroides was able to block TNF induction by an E. coli deep rough mutant (49).
Qureshi e_t aL (50) have shown that this unusual lipid A does not induce IL-l, yet it does
antagonize the induction of lL—l synthesis by the same E. coli rough mutant, D3lm4.
They have shown that R. sphaeroia’es prevents i_n m TNF release caused by E. coli Re
LPS.

Kirkland e_t a]; (51) have shown that R. sphaeroides ATCC 17023 lipid A could
not stimulate a murine pre-B cell line to synthesize surface immunoglobulin. Yet, it
effectively blocked E. coli LPS-induced activation of the same cell line in a
concentration-dependent manner. Their conclusions suggest R. sphaeroides lipid A,
even though it is considered an unusual lipid A, is competing with E. coli LPS for some
site on the pre-B cell line. This is in agreement with the previously presented data
concerning cytokine release.

Until recently, most unusual lipid As displayed none or few of the classical endo-
toxin activities. Due to this dilemma, they shed little light on the problem of the
structural basis for endotoxin action. Few bacteria have been studied which possess an

atypical or unusual lipid A. When investigations encompassing more organisms have

16

been undertaken, it may be shown that the usual lipid A is not so prevalent.

R. I. Hollingsworth has stated that the molecular basis of endotoxin action rests
upon the realization that our response to these molecules is dictated by their complex
structure (52). The primary role of LPS/lipid A, is to provide a protective barrier for the
bacterium which protects it under harsh environments, not to wreck havoc with our
immune system. Bacteria residing in different environments could develop the same
structural organization using different chemical f unctionalities that are functionally
equivalent. These structurally differing endotoxins may very well have the same surface
topography regarding charge and geometry. They just may be capable of eliciting the
same host response. More unusual lipid A molecules that possess the classical endotoxin
properties need to be isolated and structurally studied, focusing on the important aspects

of structure mentioned above.

Chapter Two: Literature Review of Mechanisms of Endotoxin Action

LPS/lipid A affects nearly all mammalian cells, either directly or indirectly. The
precise molecular mechanisms by which this occurs are not yet known. The net effect of
lipid A on a cell is to stimulate that cell to perform those functions the cell was originally
intended to perform. It seems reasonable to assume that the final intracellular signal
received by a cell following an encounter with LPS/lipid A is not dissimilar to the signal
which the cell normally receives from its environment. However, the molecular
interactions which regulate the LPS-cell encounters are ill defined, as stated above.

To initiate a cellular response, the LPS must first interact with the target cell
plasma membrane. Such an interaction could be highly specific, for instance, it might
require an LPS receptor. On the other hand, this recognition could be non-specific as in
the interaction between the lipid A hydrophobic acyl chains and the lipid bilayer of the
target cell. Either of these scenarios leads to the association of lipid A with the
eukaryotic plasma membrane.

Binding of LPS to the membrane alone is not sufficient to initiate activation of all
cell types, but may be only the initial step of the activation process. This was
demonstrated with experiments in which LPS was bound to B-lymphocytes of normal
mice and the murine LPS-hyporesponsive strain, C3H/HeJ. These experiments showed
that cells from the hyporesponsive strain also bound LPS (53,54). However, a B-cell
response did not occur in the C3H/Hel strain. There are other examples of cells which
definitely bind LPS, but do not appear to be activated by it. This was shown by Morrison
(55) using plasma—free preparations of human platelets and granulocytes. Both seem to
bind LPS from E. coli and Salmonella, but neither appear to be activated.

These data suggest that, in these cell types, the binding of LPS to the cell, by
itself, is not sufficient for activation. Subsequent to the initial binding interaction, some

type of perturbational event leading to the transfer of information must occur.

17

18

Information transfer could come directly, as in the translocation of LPS to the target cell's
interior, or indirectly via one or more second messengers. These events would then lead
to the preprogramrned cellular response (55).

Theories of the mechanism of endotoxin action fall into two major categories.
The first theory is based solely upon intercalation of LPS into the lipid bilayer of the
target cell. This results in a perturbation of the membrane such as a decrease in
membrane fluidity. In the second theory, the signalling event is triggered by the binding
of LPS/lipid A to specific membrane components which have a high affinity for lipid A.
These components are thought of as ”receptors" for the lipid A molecule. Let us first
look at the non-specific interaction regarding intercalation of LPS into the lipid bilayer.
Hydrophobic Interactiong

It has been demonstrated using chemically synthesized lipid A, that the lipid
moiety of LPS is necessary for the induction of the typically observed biological
activities. Hydrophobic interactions between LPS/lipid A and the phospholipids and
other hydrophobic constituents on the cell membrane are known to occur. It is known
that the lipid A in an LPS aggregate is not readily accessible and is buried deep inside
the dense cluster. Thus, in order for lipid A to activate a cell, there must be a molecular
rearrangement of LPS in order for lipid A to become accessible to the cell's plasma
membrane.

There is much evidence suggesting that LPS binds and inserts into lipid bilayers.
One of the first investigations into the role of lipid attachment to cells involved lecithin
and cholesterol (56). Neter showed that both lecithin and cholesterol inhibited the
binding of LPS to human erythrocytes (RBCs) and suggested that hydrophobic
interactions were important in LPS-cell membrane interactions. In 1967, Hammerling
and Westphal (57) made stearoyl derivatives of lipid—free polysaccharides that were
capable of attaching to RBCs, while the native lipid—free polysaccharide could not.

Springer e_t: a_l_. (58) showed isolated lipid A was capable of inhibiting the attachment of

19
native LPS to RBCs and platelets. More recently, studies by Dijkstra et_. §l_. (59) demon-

strate the inhibition of LPS activation of macrophage following incorporation of LPS into
liposomes. They observed the diminished ability of LPS to inducetumoricidal activity
and TNF secretion by murine macrophage. This is consistent with a role for hydrophobic
interaction between LPS and the plasma membrane in LPS-initiated cell activation.

Studies by Munford, which utilize acyloxyacyl hydrolase, a deacylating enzyme
obtained from leukocytes, show a decrease in LPS-dependent cell activation once the
LPS has been deacylated (35,60). Selective deacylation of the nonhydroxylated fatty
acids of LPS by the above mentioned enzyme reduces toxic activities of LPS i_n $19.
The toxic effects of LPS decreased as the degree of LPS deacylation increased. This data
is suggestive of an obligatory role for lipid A in the activation of biological responses.

Morrison _e_t aL (61 ,62,63) has jg yit_rg evidence suggesting a time and temperature
dependent interaction between R-form LPS and rabbit erythrocytes (RaRBC). A brief
explanation of their conclusions follows. Using polymyxin B sulfate-mediated hemolysis
as a probe, they have shown that S. minnesota R595 LPS interacts with RaRBC in two
distinguishable steps. The first step is the adsorption of LPS to the RaRBC membrane,
rendering the cells sensitive to polymyxin B-initiated lysis. This is followed by a time-
dependent decrease in response of the LPS-treated cells to the antibiotic. This is the
result of a time—dependent rearrangement of the LPS within the lipid bilayer of the
RaRBC membrane. At lower temperatures of LPS and RaRBC incubation, the decrease
in antibiotic induced hemolysis is less. The increased membrane viscosity at lower
temperatures allows less rapid rearrangement of LPS within the bilayer. They conclude
by stating that the hydrophobic intercalation of LPS into a cell membrane is important in
the LPS stimulation of responsive cells.

Membrane perturbation caused by LPS binding could result in new interactions
among membrane components which lead to cell activation. Singer e_t a_l; (55,64) state

this could occur by either trans effects or cis effects. Trans effects are described as

20
changes that occur at a localized region on the membrane's surface. They result in the

transmission of a signal across the membrane. Cis effects result in the perturbation at one
or a few points on the membrane surface that produce changes over the entire surface of
the membrane or large areas of it. These effects can result in integral protein associations
or endocytosis of membrane components, both of which are capable of generating a
transmembrane signal.

Larsen et aL (65) provide more evidence in support of a lipid A dependent
membrane perturbation. They examined the effect of LPS and lipid A on the fluidity of
the plasma membranes of monocytes by monitoring the intensity of the fluorescence of
diphenylhexatriene embedded in the monocyte plasma membranes. Preincubation of
monocytes with either LPS or lipid A appeared to increase both the microviscosity of the
cell membrane and the order of the lipid bilayer. These findings suggest that endotoxin
becomes adsorbed onto the surface of the monocyte, presumably via attachment by the
carbohydrate moiety of these molecules and the lipid A becomes directly inserted into the
plasma membrane. This affects its microviscosity and crystalline order. They suggest
that the initial step is reversible and displays features similar to a ligand binding an
external surface receptor. However the subsequent uptake of endotoxin may not be
restricted to a receptor bound molecule uptake, but involves direct intercalation of the
lipid A into the plasma membrane.

Diane Jacobs has proposed a model describing the interaction of LPS with the
plasma membrane as a two-step process. Her model attempts to resolve the conflicting
views of whether or not a receptor is involved in LPS binding (66,67,68). These studies
measured fluorescence depolarization and were conducted by using whole cells
(lymphocytes) and model membranes composed of phosphatidylcholine vesicles. The
first step, adherence, was shown to be ionic in nature, temperature independent, and
inhibited by polycationic species such as polymyxin B. The second step, coalescence, is

the insertion of lipid A into the cell membrane lipid bilayer. It is temperature dependent

21
and not affected by species such as polymyxin B. These studies have shown that

insertion of a large segment of lipid, such as lipid A, caused a perturbation of the lipid
bilayer that was evidenced by the decrease in the fluidity of the bilayer. It is possible that
these LPS-induced changes in fluidity are influencing the cellular response. Therefore,
Dr. Jacobs suggests that the association of LPS aggregates with the cellular surface is
mediated by charge on the LPS and cell surface structures on the target cell, which are,
more than likely, proteins. This association increases the probability of intercalation of
LPS monomers into the target cell lipid bilayer. Intercalation of lipid A decreases the
fluidity of that region of the cell membrane and this change may be the signal for cell
activation.

The observations mentioned above describe mechanisms by which LPS brings
about a membrane perturbation that leads to cell activation without the need for specific
LPS receptors. LPS directly inserts into the hydrophobic inner region of the phospholipid
bilayer of target cells. This causes the lipid A to become buried inside the bilayer while
the polar carbohydrate chains protrude from the outer surface of the plasma membrane.
Shands studied bacterial outer membranes and observed LPS to have such an orientation
(69,70).

The details of LPS cellular activation are as yet unknown, but one would expect
activation to be by a mechanism normally functioning in the cell. It seems highly
unlikely that many types of mammalian cells have evolved, all possessing receptors
specific for LPS. However, cells could evolve with receptors coupled to specific
physiological responses. LPS in its binding is somehow able to interact with these
receptors, directly or indirectly, and initiate cellular responses.

Once LPS has reacted with a membrane in a non-specific manner, the
carbohydrate extending from the cell surface is in a position to react with structures
present on the cell surface. These interactions may create new signals, or block signals

normally generated. This type of scenario may result in a transmembrane signal that

22

could be translated by cells into mitogenesis, initiation of protein synthesis, or
_ transformation (55). Evidence for the model involving lipid A facilitating LPS insertion
into the membrane, followed by other LPS structures involved in specific receptor
recognition is given by Cavaillon e_t a1; (71 ,72). These studies suggest that KDO in the
inner core is needed for the activation of monocytes to release lymphokines such as lL-l.
This is not necessarily so, since other researchers, including ourselves, have found lipid A
to be sufficient in stimulating macrophage.
L‘PS Binding Proteins

Many persons have presented evidence that cell activation is mediated by lipid A
intercalating into the lipid bilayer, however some believe that LPS binds to a specific LPS
receptor. The receptor is most likely a protein, and the binding event is the start of cell
activation. Necessary requirements for specific LPS receptors should include: 1. Binding
of LPS is specific for lipid A. 2. Binding is saturable. 3. The interaction occurs at the
plasma membrane surface. 4. The interaction leads to the induction of transmembrane
signalling. Several laboratories have reported the existence of LPS binding proteins
which may well be candidate LPS receptors. However, let us first look at several proteins
that bind LPS, but are not membrane localized.

There are a variety of host substances which have relatively high affinities for
LPS and have been shown to bind LPS quite well. Mannose—binding protein is known to
bind mannose and glucosamine containing structures and has been shown to bind LPS
quite readily. Binding of lipid A to a variety of proteins with other known functions has
also been established. These serum/plasma proteins include lysozyme, complement Clq,
blood coagulation Factor XII, serum albumin, and high density lipoproteins. Binding of
LPS induces the activation of serine proteases, as in the cases of Cl and Factor XII.

Lipopolysacharide binding protein (LBP) (73,74) is a protein found in trace
amounts in human plasma and that of other endotoxin sensitive species. LBP is a 60 kDa

glycoprotein synthesized by hepatocytes. It will rise from 0.5 ug/mL to approximately

23
50ug/mL within 24 hrs. after induction of an acute phase response. LBP has a binding

site for lipid A and is capable of binding both smooth and rough-form LPS with a high
affinity. LBP is responsible for opsonization of LPS aggregates and intact gram-negative
bacteria. LBP then acts as a carrier protein that brings LPS to the surface of monocytes
and macrophage. The LPS-LBP complex then binds to CD14 (75), a monocyte
differention antigen, which causes the cell to become activated. It is believed that LBP
may belong to a family of proteins that bind lipids and transport them through aqueous
environments. Tobias e_t a_l. believe a principle function of LBP may be to increase the
ability of the infected host to detect LPS early in infection and increase natural resistance
in combatting the infection (76).

Bactericidal/permeability increasing protein (BPI) is a 50-60 kDa (depending on
species of isolation) cationic protein found in the azurophilic granules of neutropils (77).
BPI displays bactericidal activity towards several species of gram-negative bacteria. The
outer membrane is the site of BPI action. BPI causes reversible alterations of the outer
membrane and irreversible alterations that result in death of the bacterium. BPI acts by
increasing membrane permeability to small hydrophobic molecules. Once surface bound
BPI is removed from the bacterium, biosynthesis of new LPS and its transport to the
outer leaflet are sufficient for repair (78). If BPI is not removed, it rapidly kills the
organism.

The effects of BPI in infection may not be limited to bacterial cell death, but may
also control the neutrophil response to LPS. Studies by Maara e_t 9L demonstrate that BPI
inhibits complement receptor (CR1 and CR3) up—regulation stimulated by smooth and
rough-form LPS, as well as lipid A (79). Due to the range of BPI activity against these
different LPS forms, and since lipid A is common to all forms, it was concluded that lipid
A is the component to which BPI reacts and binds. BPI was also shown to inhibit LPS
activity in the LAL assay. Neutralization of LPS by BPI was shown to occur rapidly and

at low doses.

24
BPI and LBP have significant N-terminal sequence homology and have been

shown to be cross-reactive immunologically. Both bind LPS, however they differ in their
effects on LPS and gram-negative organisms, in their sites of biosynthesis, and i_n M
localization (80). This has lead Tobias to conclude that LBP, BPI, and other proteins
such as high density lipoprotein belong to a family of LPS binding proteins that moderate
the host response to LPS, albeit in differing ways.

LPS Receptors which are Membrane Localized

Differing approaches used to characterize specific membrane components present
on target cells have revealed the presence of several components, present on a number of
cells, that bind LPS. These experiments have involved fractionation of cell membranes,
followed by the isolation, purification, and characterization of the LPS binding
substances.

One of the earliest studies identifying specific LPS receptors was carried out by
Springer using human erythrocytes (81,82). A lipoglycoprotein was isolated rich in N-
acetylneuraminic acid, galactose, and hexosamine. This substance was found to bind all
smooth and rough—form LPS of all gram-negative bacteria tested. Although this receptor
is a lipoglycoprotein, neither lipid nor carbohydrate appears to be involved in its binding
activity. The peptide portion appears to be important, for the the receptor is heat labile,
inactivated by aldehydes, and susceptible to proteases. Attempts to locate this lipoglyco-
protein on other blood cells such as platelets and leukocytes were not successful.
Springer concludes that LPS receptors on other cells may differ in nature and there may
be more than one type of LPS receptor. However, they have isolated a cell surface
macromolecule which specifically prevents LPS binding to the RBC, and they were the
first to do so.

David Morrison and coworkers have been very active in the search for LPS
binding proteins. They have synthesized and utilized a radioiodinated, photoactivatable

LPS derivative to detect the specific binding of LPS to membrane components on murine

25

splenocytes. Splenocyte extracts were fractionated by two-dimensional PAGE and
subjected to autoradiography. They found an 80 kDa LPS binding protein present on
murine splenocytes. This LPS binding protein was detected on B—cells, T-cells, and
splenic macrophage. It was also detected on murine B and T cell lines 70Z/3 and YAL— 1.
Interestingly, fractionation of membrane proteins of the C3H/HeJ endotoxin
hyporesponsive mouse strain showed no difference in LPS binding proteins from those
on LPS responsive mice (83,84). This glycoprotein has been found on peripheral blood
mononuclear cells of a variety of species sensitive to endotoxin, humans being one such
species. The 80 kDa LPS binding protein is absent from the mononuclear cells of frog
and chicken, both of which are very resistant to LPS stimulation. This suggests a
correlation between 80 kDa protein expression and endotoxin sensitivity. LPS binding to
the 80 kDa protein is both saturable and inhibited by homologous and heterologous LPS
and lipidA, but not by peptidoglycan or muramyl dipeptide (38,85,86,87).

Kirikae e_t 1L have identified an LPS receptor on murine RBC membranes (88).
Treatment of RBCs with pronase resulted in decreased amounts of 125l-labeled R—form
LPS bound to RBCs. This suggests that this LPS receptor is a membrane protein or
glycoprotein. In this study proteins were electroeluted onto nitrocellulose and
subsequently treated with LPS, followed by rabbit-anti-LPS, and goat-anti-rabbit lgG-
alkaline phosphatase conjugate. This method demonstrated that LPS bound to a single 96
kDa protein.

Hampton e_t al_. used a radioactive, chemically defined, lipid A precursor, 32P-
lipid IVA to demonstrate its specific binding to whole cells, RAW264.7 macrophage-like
cultured cells and to immobilized proteins extracted from these cells and transferred to
nitrocellulose. They found the binding to be saturable, inhibited by excess unlabeled
lipid IVA, and to be proteinase K sensitive. The two major binding proteins are 31 kDa
and 95 kDa in size. Fractionation studies indicate the 31 kDa protein to be enriched in

the nuclear fraction and the 95 kDa protein to be enriched in the membrane fraction.

26
They studied other cell lines and found no binding could be detected over background in

either of two types of fibroblasts, CHO-Kl and L929. The 95 kDa protein was
unobservable in CHO-K1 cells. This observed lack of binding in CHO-K1 cells was
explained as the cells' inability to respond to lipid IVA at all concentrations tested
(88,89).

Kirkland e_t a_l. (90) also used a radioiodinated, photoactivatable derivative of LPS
to label LPS receptors on 702/3 cells. Proteins were extracted and resolved by SDS-
PAGE and viewed by autoradiography. These studies identified 18 kDa and 25 kDa
proteins which bind to the lipid A region of LPS. Labeling was saturable and inhibited
by excess unlabeled LPS and lipid A. These molecules are considered candidates for
LPS receptors.

Mechanism of LPS Action

There is evidence for more than one specific LPS binding protein expressed on
the surface of mammalian cells. However, binding may be mediated through the non-
specific hydrophobic interactions between the lipid A and the plasma membrane bilayer.
The relationship between these ”receptors" or hydrophobic interactions and the
mechanisms of cellular activation have yet to be determined.

In the field of cell biology, much information exists regarding biochemical
mechanisms of signal transduction and cell activation mechanisms. Regarding
endotoxins, there is no conclusive evidence concerning the mechanism of LPS mediated
cell activation. Binding of LPS to a cell, by itself, is not sufficient to cause activation.
To date there is evidence for lipid A activation of calcium fluxes (91), turnover of
phosphatidylinositol intermediates (91 ), and the activation of G-bindin g proteins (92), yet
precise mechanisms of signal transduction mediated by LPS remain to be defined. The
following is a list of events suggested by Morrison (55) concerning LPS triggered cell
activation.

1. The triggering event may well be intercalation of LPS into the lipid bilayer.

27

Intercalation results in membrane destabilization which may be the signal for cell
activation. 2. The LPS may become internalized and placed adjacent to intracellular
structures. This would be the result of direct translocation, pinocytosis, or endocytosis.
As yet, I have not found any information as to the mechanism by which direct
translocation occurs. 3. LPS may interact with specific membrane components, such as
protein/glycoprotein which leads to the generation of a transmembrane signal. 4. LPS
could interact with some additional external species at the cell surface. It is this species

which provides the actual transmembrane signal for the cell.

Chapter Three: Literature Review of Molecular Events Resultinfirom the Interactions

Between Endotoxin and Target Cells

The Macrophage and Cytokine Release

The cell most widely studied, which interacts with endotoxin and shows the
greatest changes resulting from this interaction, is the macrophage. When treated with
endotoxin, the macrophage undergoes morphological changes and begins synthesis of
various proteins, collectively termed cytokines. The cytokines most studied are lL-l, IL—
6, TNF, and colony stimulating factor (CSF). The cytokines are mediators of endotoxin
activity by acting as second messengers carrying regulatory instructions to cells, organs,
and entire systems.

Studies by Michalek _e_t gl_. (93) demonstrated the importance of these cells in
mediating cellular activities of endotoxins. This group used adoptive transfer to evaluate
the importance of lymphoreticular cells to the effects of endotoxin on the host. Effects
studied included lethality, immunogenicity, induction of interferon, and CSF induction.
C3H/HeJ mice became sensitive to each of these effects after adoptive transfer of bone
marrow cells from the C3H/HeN, LPS responsive mouse strain. Efficiency of transfer
was found to be directly proportional to dose of X-irradiation and inversely proportional
to the number of surviving host stem cells. C3H/HeN-C3H/Helx chimeras became
sensitive to LPS for each parameter tested. C3H/HeN mice were also rendered
unresponsive to LPS by adoptive transfer of C3H/HeJ bone marrow. This group
suggested that lymphoreticular cells, either directly or via their soluble mediators, are
essential for the expression of endotoxicity in the hostg They suggest that LPS acts
initially and directly to stimulate some bone marrow derived cell types. As a result of
this stimulation, a variety of mediators, such as interferori, TNF, and CSF are produced.
These mediators go on to induce secondary effects on the host such as dehydration,

hypotension, vascular damage, e_t cetera When of sufficient magnitude, these effects

28

29

become irreversible and result in death of the host. In other studies, mice which were
LPS responsive, but T-cell or B—cell deficient, responded normally to the toxic effects of
LPS (94,95).

There is strong evidence that LPS-stimulated macrophage products are the direct
mediators of endotoxin induced responses. As noted earlier, macrophage produce a
variety of factors including IL—l, TNF, CSF, interferon, and E—series prostaglandins jg
£19. in response to stimulation by LPS. Isolated macrophages from LPS-responsive mice
produce a similar spectrum of soluble factors (96). When purified and injected into
animals, each of these soluble factors induces one or more of the same toxic effects
induced by LPS, such as fever, hypotension, shock, and death.

Compounds which affect macrophage by increasing or decreasing their state of
activation result in enhanced or attenuated LPS responsiveness, respectively. Admini-
stration of anti-cytokine antibodies to normal mice who also received LPS has been
shown to attenuate endotoxin effects, and even prevent death in the case of anti—TNF
administration (94,97). Such studies are currently underway in human trials (98).

These findings support a role for the macrophage as a major cellular mediator of
endotoxin induced effects. However, other cells are also involved in LPS phenomena.
LPS stimulation of platelets has been recognized to be one of the most rapid host
responses to endotoxin. The result is platelet aggregation and release of vasoactive
amines (99). Endothelial cells have also been shown to be involved in these phenomena,
such as in the condition known as disseminated intravascular coagulation.

Signal Tragsduction

It is agreed that the cytokines and other humoral factors play a most important
role in the mechanism of endotoxin action. However, what are the intracellular events
that result in the release of these cell products following these changes?

Nakano gt gl_. used the C3H/HeJ mouse and its normal counterpart to look at

these events. In their earlier studies, they looked at the relation between endotoxin

30

initiated intracellular signal transduction and the calcium—calmodulin system (100,101 ).
They indicated that the calmodulin system relates to LPS induced IL-1 production by
murine macrophage. They have since proven that cAMP and cGMP do not enhance the
production of lL—l in murine macrophage, regardless of having or not having LPS
stimulation. This suggests that these cyclic nucleotides do not work as second
messengers in the intracellular signal transduction for IL-1 production resulting from LPS
stimulation. This same group has shown that LPS induces the phosphorylation of a
distinct set of proteins located in the membrane and in the cytosol of macrophage prior to
the synthesis of lL-l and TNF by these cells. They successfully purified to homogeneity
an LPS induced cytosolic phosphorylated protein in the macrophage which is
interestingly lacking from the cytosol of the C3H/HeJ LPS hyporesponsive mouse strain.
An LPS induced phosphorylating protein has yet to be identified (102).

Judy Spitzer studied the effects of LPS on hepatocyte signal transduction. Her
studies involved inducing LPS low-responsiveness by continuous infusion of non-lethal
doses of LPS into rats (i.e. endotoxin-induced tolerance). The conclusion of these experi-
ments can be summarized as follows. 1. Endotoxemia perturbs the calcium-dependent
and diacyl glycerol (DAG)-activated arms of signal transduction pathways. 2. Specific-
ally, they attributed the low responsiveness to reduction in intracellular Ca2+ mobil-
ization, decrease in DAG levels, and decrease in protein phosphorylation. 3. Protein
kinase C (PKG-mediated events were implicated in the molecular mechanism of LPS
action.

Ding gt gl_. suggested that microtubules may be targets of endotoxin action on
macrophage. Taxol, a plant derived anti-tumor drug, binds irreversibly to polymerized
tubulin, thus preventing its depolymerization. Ding showed taxol was able to act on
macrophage and mimic two effects of LPS: decrease in TNF receptors and induction of
TNF release. Both actions of taxol were absent in C3H/HeJ mice. They concluded a

protein associated with microtubules may be a cellular target of LPS (103).

31
Tanke (99) looked at GTPase involvement in sepsis. They used lipid X to induce

low-responsiveness to endotoxin. Lipid X has been identified as the monosaccharide
precursor in lipid A biosynthesis. Lipid X treatment was assumed to block access of LPS
to target cells and thus prevent mortality. Inhibition of GTPase activity occurred, and it
was concluded that a G-protein might be activated by interaction between LPS and a
receptor, thus triggering signal transduction.

Cavaillon e_t a_l. ( 104) compared lipid A and Re-LPS in their abilities to induce IL-
1 synthesis in macrophage and act as a mitogen for lymphocytes. Lipid A was mitogenic,
but unable to induce IL-l release by macrophage. They concluded that the interaction of
LPS/lipid A with the cell membrane differs between different cell types.

There are several targets for LPS. It does not interact with a single organ or cell
type. LPS stimulation results in the initiation of multiple metabolic pathways,
simultaneously, which normally occur in the cell. Unfortunately, the metabolic processes
rapidly reach a point where they run on ”out-of~control". The result is the synthesis of
many cytokines which are ultimately responsible for the mechanisms of endotoxin action.
A conclusion concerning the nature of what LPS recognizes, or is recognized by, and how
this recognition is transformed into a signal for cell activation has yet to be reached.
Further experimentation is needed to determine how the above scenarios fit together.
Endotoxins have been the subject of controversy for many years past and will surely be

cause for discussion long into the future.

Chapter Four: Rhizohium Lipid A Structure

Introduction

Endotoxin literature cites three major structural features of lipopolysaccharides
which are necessary for the induction of endotoxin effects. These features are the
presence of a g1 ucosamine disaccharide as the carbohydrate lipid A headgroup,
phosphorylation of the disaccharide, and the presence of fatty acid substituents. No one
has been able to correlate specific structural elements with a specific biological activity.

Unusual lipid A molecules have been isolated lacking the phosphorylated,
glucosamine disaccharide (43 ,44,45,46,47,48,49, 105). These molecules display
attentuated or none of the typical endotoxin sequelae. They offer little to be learned
concerning the structural basis for endotoxin action. It would be beneficial to have an
unusual lipid A that still possesses some of the classical endotoxin consequences, but has
a different lipid A structure. In this way, one could compare differences in structure and
bioactivity to explain the structural basis for endotoxin action.

The LPS of the soil bacterium, Rhizohium trifolii, was thought to have the same
lipid A structure as that of the Enterobacteriaceae ( 106). In experiments conducted by F.
B. Dazzo, R. trifolii 0403 was isolated at different stages of growth and examined for
immunochemical properties. This LPS was found to give a positive response in the LAL
assay, and was capable of acting as a mitogen (107). Further chemical analysis of the
fatty acids on this species by R. l. Hollingsworth gt a_l.(108) showed the presence of 3—
hydroxylated fatty acids exclusively, and an unusual 27-hydroxyoctacosanoic acid in the
LPS (109). They were the first to isolate this unusual long chain fatty acid, which
indicated that the structure of this lipid A was different from that of the enterics and thus
could be considered unusual.

Throughout the course of this project, we have worked with LPS from strain

ANU843 of the same species of Rhizobium . This organism infects clover and is non-

32

33

pathogenic to humans. However, LPS from Rhizobium trifolii was shown by Dazzo to
produce a positive response in the LAL assay, as mentioned above. There are conflicting
reports in the literature on the structure of the lipid A of the genus Rhizobium . This
conflict concerns the degree of phosphorylation, fatty acid composition, and composition
of the lipid A backbone (110,11 1,] 12,113). Several of the features of enteric lipid A
structure which were thought to be critical to endotoxin activity are not found in rhizobial
LPS (109,110,111,114).

This research project was divided into two areas. Firstly, determining the lipid A
structure of R. tnfolii ANU843. Secondly, determining which biological activities
triggered by classical endotoxin are conserved in the "non-classical" endotoxin obtained
from Rhizohium.. The data obtained from these areas will help to determine what
structural features elicit a particular biological response, and develop a more
encompassing view of endotoxins. It also provides insight into the mechanism of

endotoxin action.

Materigls and Methods:
Bacterial Cultures and LPS Isolation

Rhizobium trr'folii ANU843 was grown aerobically to stationary phase in 6L of
Bergensen's (BIII) broth at 29°C, with shaking at 142 rpm (115). The 6L of culture was
used to innoculate 120L of medium in a B. Braun fermenter, model Biostat UEIOOD.
The cells were grown to early stationary phase at 29°C, aerobically, with stirring at 200
rpm. The cells were harvested using a Sharples A512 centrifuge at 5100 rpm. After
harvesting, the cells were again centrifuged for 30 min. at 8000 rpm.

The cell pellets were washed 3 times with 0.5M NaCl for 30 min. at room temp-
erature and centrifuged for 30 min. at 8000 rpm. Cells were extracted 3 times using the
hot phenol/water method of Westphal (116). Aqueous layers were pooled into 12,000—

14,000 molecular weight cut-off dialysis tubing and dialyzed extensively against many

34

changes of distilled water. The volume of the dialysate was reduced to 200mL using a
rotory evaporator and treated with 2.5mL of an aqueous solution of DNase I, RNase A,
and MgClz - 6H20 to remove contaminating carbohydrate. The enzyme solution was
prepared by dissolving 250mg MgClz, 25mg RNase A, and 19mg DNase I (40,000 units)
in 5mL distilled water. The enzyme treated dialysate was ultracentrifuged for 4hrs. at
40,000 rpm at 4°C, using a Type 50.2 Ti fixed angle rotor. The LPS pellet was
resuspended in water and centrifuged at 6500 rpm for 10 min. The supernatant liquid
containing LPS was frozen, then lyophilized.

Alternatively, R. trifolii ANU843 was grown in Blll broth (6L) at 30°C until
early stationary phase was reached (115). The cells were then centrifuged for 40 min. at
7000 rpm. The cell pellets were treated as previously described to obtain the LPS. The
combined extracts were extensively dialyzed against water, the total volume reduced to
15mL, and was treated as before with enzymes. This solution was then subjected to gel
permeation chromatography on a Sepharose 48 column, using 0.05M EDTA/triethyl-
amine buffer at pH7.0. Fractions were assayed for carbohydrate content using the phenol
—sulfuric acid assay (117). Two peaks were obtained, pooled, and dialyzed. The first
eluting peak corresponded to the one obtained when the LPS was isolated by ultracentri-
fugation.

Lipid A Isolation

The lipid A was cleaved from the O—polysaccharide and core oligosaccharide by
hydrolysis with 1% acetic acid for 3 hrs. at 100°C. After cooling the hydrolyzed sample
to room temperature, it was extracted three times with twice its volume of 5:1 chloroform
-methanol. The organic layers, which were removed after centrifugation at 8000 rpm for
10 min., were pooled. The combined extracts were dried by passage through absorbant
cotton , and concentrated to dryness under a stream of nitrogen.

Purification of Lipid A

The crude lipid A preparation was fractionated on a C 18 reverse-phase column

35
(11 x 1.2 cm) with water (7mL fraction), 2:1 water/methanol (7mL), 2:1 methanol/water

(l4mL), 4:1 methanol/water(14mL), pure methanol (l4mL), and pure chloroform
(28mL) in that order. The pure methanol fractions were combined and subjected to
preparative chromatography on C18 reverse-phase layers (10 x 10 cm) using 4: 1 MeOI-I/
water as the mobile phase. Two bands which reacted positively with orcinol/sulf uric acid
reagent (118) were obtained. These were scraped and eluted with 1:1 chloroform/MeOH.
There were other slowly-mi grating bands at and close to the origin.
Fatty Acid Analysis of Crude and Purified Lipid A Components

Lipid A components were converted to their fatty acid methyl ester derivatives by
methanolysis with 5% HCl in methanol for 16 hrs. at 70°C. Methyl ester derivatives
were subjected to gas chromatography (GC) analysis on a 25m Supelco SPBI capillary
column or a JW DB-l capillary column of 30m, using helium as the carrier gas and a
temperature program of 150°C initial temperature, 0.00 hold time, 3.0 degree/min. rate,
final temperature of 300°C, final hold time of 15 min. and a run length of 65 min.
Electron impact mass spectrometry coupled to GC (GC/MS) for fatty acid analysis
utilized a DB—l, 30m, capillary column, with a temperature program of 190°C initial
temperature, 0.00 initial hold time, rate of 3.0 deg/min., final temperature of 300°C, final
hold time of 30 min. Mass spectra were recorded on a Jeol JMA DA5000 spectrometer.
Carbohydrate Analysis of Crude and Purified Lipid A Commnents

Lipid A was treated with 5% HCl in methanol at 70°C for 16 hrs. After
concentration to dryness, the methanolysis products were dissolved in methanol and
subjected to reduction by using a sodium borohydride solution (lOmg/mL). The
reduction proceeded for 5 hrs. at room temperature, after which time 3M HCl was added
dropwise until effervescence ceased. The oxidant was concentrated to dryness and the
remaining residue was treated with 0.3mL of 2M trifluoroacetic acid for 3hrs. at 120°C.
When cool, the hydrolysate was extracted with 2 times lmL chloroform. The chloroform

extracts were discarded and the aqueous layer was concentrated to dryness under

36

nitrogen. The aqueous residue was dissolved in methanol and reduced a second time with
sodium borohydride for 1hr. 3M HCl was then added to neutralize any remaining base.
The solution was concentrated to dryness and washed several times with methanol. To
the residue was added 0.1mL pyridine and 0.1mL acetic anhydride. The solution was
briefly sonicated and left at room temperature for l6hrs. The acetylated mixture was
concentrated to dryness, then lmL chloroform and lmL 3M HCI were added to the
residue. The chloroform layer was washed once with lmL 0.5M NaCl, dried over anhy-
drous sodium sulfate, and subjected to GC analysis on a D8225 capillary column with an
initial temperature of 200°C, hold time of 5.00 min., rate of 2.0 deg/min., final
temperature of 230°C, final hold time of 55.00 min., and a run length of 75 min.
Retention times were compared with alditol acetate derivatives of a number of sugar
standards. GC/MS analysis of lipid A derivatives were analyzed on a Jeol JMA DA5000

mass spectrometer using the DB225 capillary column.

NMR and Mass Spectra

Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker WM250
spectrometer operating at 250MHz for protons and 62.8 MHz for 13C. Varian VX-NMR
spectrometers operating at 300MHz and 500MHz were also used for recording lH
spectra. Spectra were obtained in either deutero—chloroform containing 10% D-4
methanol, deutero—chloroform, or deuterium oxide. Chemical shifts were quoted relative
to the chloroform resonances at 7.24ppm for 1H and 77me for 13C measurements
and at 4.65ppm relative to the water resonance for 1H when using deuterium oxide. Mass
spectra (GC/MS) were recorded on a Jeol J MA DA5000 spectrometer. Fast atom
bombardment mass spectrometry (FAB/MS) was recorded on a Jeol HX-lOO-HF
spectrometer. In all of the FAB/MS experiments, glycerol was used as the matrix.
Phosphate Analysis

Total phosphate content was analyzed by converting organic phosphate to

inorganic phosphate by the method of Lowry (1 l9). LPS of R. trifolii ANU843 was

37
separated on a Sephadex LH-60 gel filtration column using 1: 1:2 n-propanol/acetonitrile/

water as the eluting solvent. This resulted in the separation of LPS into three peaks as
judged by testing of the individual fractions using the phenol-sulfuric acid assay for
carbohydrate. Each fraction was screened for the presence of phosphate.
Preparation of Phenylcarbamate Derivatives of Crude Lipid A

Phenylcarbamate derivatives of the lipid A were made by dissolving 54mg of R.
Irifolii ANU843 lipid A in 7mL pyridine and 70uL phenylisocyanate. The sample was
briefly sonicated and then heated at 70°C for 14hrs. It was then concentrated, dissolved
in 40mL chloroform and 40mL methanol. To this was added lOOuL of TFA, then the
mixture was heated at 100°C for 10min. The lipid stood at room temperature for 24hrs.,
to ensure complete conversion of all carboxylate groups to methyl esters. The product
was concentrated to a small volume and subjected to flash column chromatography using
silica gel. Eluting solvents were chloroform (IOmL), chloroform (30mL), chloroform
(30mL), 5:1 chloroform/methanol (25mL), 5:1 chloroform/methanol (lOmL), 5:2 chloro-
form/methanol (lOmL). All fractions collected were screened for carbohydrate using the
orcinol method and bands were visualized under UV light. The sample testing positive
for carbohydrate was subjected to fatty acid methyl ester analysis as described previously.
GC/MS was performed on the phenylcarbamate derivatized methyl esters using an initial
temperature of 65°C, initial time 0.00, rate 3deg/min., final temperature of 300°C, final

time of 30 min., using a 30m DB-l capillary column.

Results and Discussion:
Phosphate Agggsis

LPS submitted to gel permeation chromatography was resolved into three peaks
as evidenced by carbohydrate screening using orcinol. Two of these peaks were major
and the last peak was minor. The major components did not test positive for the presence

of phosphate (119), however the minor peak was positive. The minor peak did not

38
contain the typical carbohydrate and fatty acid components of rhizobial LPS. It was

concluded that this minor peak was definitely not LPS, (Figure 5). This experiment
clearly showed that phosphate was absent from the LPS of Rhizobium .
Sulfate Anglvsis

Sulfate has also been reported as a polar substituent in LPS (120), and it has also
been found in the LPS of R. melilon' (121). Several species of Rhizobium were grown in
the lab of R. I. Hollingsworth, in the presence of 35S-MgSO4. Samples were screened by
scintillation counting and it was found that R. trifolii ANU843 was completely devoid of
sulfate.
Crude Lipid A 1H-NMR

The 1H—NMR spectrum of crude lipid A isolated from R. trifolii ANU843 can be
described as follows, (Figure 6). There are signals at 0.80 and 12me that are
characteristic of methyl and methylene signals, respectively, indicative of long chain fatty
acids. Signals at 2.38ppm correlate with methylene groups adjacent to a carbonyl
functionality. Farther downfield are two groups of sextets, one at 4.17ppm and the other
at 4.92ppm. From multiplicity and chemical shifts, these signals correspond with
methine protons attached to a carbon bearing a hydroxyl and an acyloxy function,
respectively. There are unresolved resonances between 2.00 and 4.50ppm which are
assigned to protons on the carbohydrate backbone of the lipid A molecule. These
resonances are broad and not well resolved, for the lipid A molecules form strong
aggregates and the polar groups do not have much motional freedom.
Fatty Acid Analysis

Fatty acid substituents were identified by preparing fatty acid methyl ester
derivatives of the rhizobial fatty acids and comparing their retention times on a gas
chromatogram with those of a known methyl ester standard and a rhizobial standard.
Fatty acid methyl ester analysis by GC and GC/MS showed the presence of the following

fatty acids: 3-hydroxytetradecanoic acid, 3—hydroxy-12-methyltetradecanoic acid, 3-

39

TOTAL “£1058

o—----o mosrmare

Donn _
let’..

..—-—°—’"

.o—DOH

 

A I. u
M_D“OI_IOOOI*OCDIO_ 0... -00000— OI.

 

    

 

Figure 5

Gel filtration profile of R. trifolii ANU843 LPS on a column of Sephadex LH-60 resin.
Fractions were screened for the presence of phosphate.

 

x8 1

 

 

 

pm

Fr ure 6
Proton NMR spectrum of the crude lipid A of R. trifolii ANU843.

41

hydroxyhexadecanoic acid, 3-hydroxyoctadecanoic acid, and 27—hydroxyoctacosanoic
acid, (Figure 7).

We have looked at the extent of fatty acid acylation (i.e. acyloxyacyl linkages).
Phenylisocyanate was used to generate carbamate esters of free hydroxyl groups present
on fatty acids and the carbohydrate backbone (122). This procedure was used to
determine whether or not the hydroxyl group was esterified by another acyl chain in the
case of 3-hydroxy fatty acids. Fatty acid methyl esters were made of the phenyl
carbamate derivatized material. Comparisons in GC retention times were made with an
R. trifolii fatty acid methyl ester standard. Phenylcarbamate derivatives have retention
times that are slightly shorter than the corresponding 3-hydroxy compound. GC/MS was
also used in determination of the extent of acyl substitution of fatty acids. MS of
phenylcarbamate derivatives contain fragments assignable to the phenylcarbamate group:
mlz=93 aniline, m/z=l 19 phenylisocyanate, mlz=l37 phenylcarbamic acid. Without the
phenylcarbamate group, 103 is the major ion. Comparisons of the rhizobial methyl ester
standard for fatty acids and that of the phenylcarbamate derivatized material are shown in
Figure 8.

Based upon comparison of GC retention times with a known standard and in ion
fragmentation patterns from GC/MS, we concluded the following concerning R. trifolz'i
ANU843 lipid A fatty acids: (i) The 27-hydroxyoctacosanoic acid is always linked at the
27-hydroxyl position, (ii) 3—hydroxytetradecanoic acid is found in the form of a free 3-
hydroxyl group, and in a form substituted at the 3-hydroxyl position. GC analysis
intensities indicated that the free hydroxyl is the predominant form (Figure 9), and (iii) 3-
hydroxy-l2-methyltetradecanoic acid is free at the 3-hydroxy position as evidenced by a
change in CC retention time and the presence of characteristic ion fragments in GC/MS.
We believe the 3-hydroxyhexadecanoic acid and the 3-hydroxyoctadecanoic acid are free
at the 3-hydroxyl positions. Both were modified by phenylisocyanate, however the mass

spectra lacked the characteristic ions of phenylcarbamate derivatives and-indicated that

42

 

 

 

 

. n
q C
c. K
-3 a.
l
.. D
n v
9 a
' C
" o
" fl
0‘
0
K
B 8
C
O
I" V
0
fl
0
O
..:
on.
(1“ .l
‘0
an:
“I 1" L; 1
~
I
W Lu TILL. ~L_.A__A

 

flaw

GC profile of fatty acid methyl esters derived from R. trifolii. (A) 3-
hydroxytetradecanoic acid, (B) 3-hydroxy-12-methyltetradecanoic acid, (C) 3-
hydroxyhexadecanoic acid, (D) 3-hydroxyoctadecanoic acid, (E) 27-hydroxyoctacosanoic
acid.

43

Figure 8
8A: GC profile of fatty acid methyl esters derived from R. m'folii ANU843.

SB: GC profile of R. mfolii ANU843 fatty acid methyl esters previously derivatized with
phenylcarbamate. Lower case letters are indicative of the corresponding
phenylcarbamate ester.

oo

2

:3

MI Sll'.’ -———< h:
SDI’IC —-tl
9.0'33 --
IlO'OC —-

1

83

 

 

0. 960'!

21“,

60"!

 

u Cll'.’

A A

8A

L‘AL

 

 

a pro-n on")!

OCI‘CI

o,“"‘ ”‘v“
x ml" —_’2

new
11 L—

( $0"!

 

45

"955 SPECTRLH Dot. rile: MZIQIOI . - . 214833-91 [3:45

: Rim-C (RE _
3:”‘5'59‘ Cl (Pos.) CC 450.61:l IP: (of: 119.0000 Int. 24.9m0 Lv 0.00

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sean! ($96) - ($09! (eeet. 1.00 (9A)
1% r 9— '-
R
c‘ r
. m‘
t
r 59
v 9?
f £0
A 1 2'17
0:
u 1»
n 40‘
d
O
" L
C ' ‘
e 2e- 9 if
1
a L Mf¥ALi¥~A ‘— 4 A 1* v V V
I“ 200 are 400 see see no see
“/2
“S SPEC?” Data File: mzrerer Zim-Ol 13:45
If |;i$" ET $90..) X 450.66 HP; 0"! 1.3.“ lat. £31.49?! Lv C.“
$0.04 ($41) - (695! teeel. 1.“) (98)
3 tbs
R .
i
Q u‘
t
I
V
¢ ‘0‘
a .
b
0
n 40‘
d
C
n
C
c a 14 .
1
”Lil. Jiffy . , , - . -
18 an ace 4“ 5“ 6“ m :2

Figure 9

Electron impact mass spectra of both the (9A) phenylcarbamate derivatized 3-
hydroxytetradecanoic acid, and (9B) 3-hydroxytetradecanoic acid, indicating it is linked
at its 3-hydroxyl position.

46

they were converted to unsaturated acids by an elimination process.
Carbohydrate Analysis of R. trifolii Lipid A

Carbohydrates were identified as their alditol acetate derivatives by comparison of
their retention times on the gas chromatograph with those of known alditol acetates and
by analysis of their mass spectra. GC/MS of alditol acetates of lipid A without a
carboxylic acid prereduction step, following conversion to the methyl ester, gave no
desirable peaks. When lipid A was prereduced with sodium borohydride before
hydrolysis and conversion to the alditols, two peaks were visible on the GC tracings. One
of the peaks corresponded to glucosamine and the other one to galactose, (Figure 10).
Since the glycosyl components could not be visualized as alditol acetate derivatives
without prereduction, it was evident that they occurred as uronic acids. This was
confirmed by deuterium labelling with sodium borodeuteride during prereduction before
conversion to alditol acetates. The lipid A of R. trifolii contained galacturonic acid and
comparitively large amounts of glucosamine. The relative composition of the crude lipid
A preparation was 1:1 galacturonic acid to total amino sugar.

We wanted to examine the carbohydrate backbone of R. trifolii lipid A. The
carbohydrate resonances were indistinguishable in lH-NMR spectroscopy because of the
degree of aggregate formation. Several attempts were made to remove the fatty acid
components from crude lipid A in order to liberate the free carbohydrate backbone.
Saponification using 1N KOH in n-butanol was not successful and led only to partial
saponification. No release of the 27-hydroxyoctacosanoic acid was observed. The 3—
hydroxytetradecanoic acid, 3-hydroxy-lZ-methyltetradecanoic acid, 3-hydroxyhexa-
decanoic acid, and 3-hydroxyoctadecanoic acid were releaSed, indicating that they are 0-
linked and confirming the belief that 27-hydroxyoctacosanoic acid is N-linked and more
difficult to release. The GC tracings of the free fatty acids are shown in Figure 11. This
suggests that the lipid A molecules formed aggregates in which the polar head group was

inaccessible to nucleophiles. The aggregates were strong enough to prevent a very polar

47

 

 

 

 

. n
.. e
. fl
- ~ A
P.
P
>n
.1
o
o
h
. B
I)
n
r.
r.
a '; 0
Q . .-
tn 2 ‘. g
o ‘.
' u
a
d a a
:
o
. a
. K
" .'

 

 

Figure 10

GC analysis of the alditol acetate derivatives of carbohydrates present in the crude lipid A
of R. mfolii ANU843. Peaks A and B correspond to the alditol acetate derivatives of
galactose and glucosamine, respectively.

 

 

 

J
. 3
‘r o
r. .
J 2
k
n
’3
I
B
n
r.
.13
C
; II
~ '
a l
a . n
I o ..
‘ 1 '" 2 a
w
' '.
{LL—J '
_.I

Figure 11

GC tracings of fatty acid methyl esters of fatty acid components liberated during KOH
saponification. (Letters represent the same fatty acids as in Figure 7.)

49

nucleophile such as hydroxide anion from getting past the strong hydrophobic inter-
actions of the fatty acyl groups attached to the lipid A head group.

We tried removal of fatty acyl groups with hydrazine. Hydrazine is not as polar
as hydroxide anion, but it is a much better nucleophile. Hydrazine has also been effective
in the cleavage of fatty acyl groups (123). We had hoped that hydrazine, being less polar,
would bypass the strong hydrophobic interactions of the acyl groups and get near the
carbohydrate head group to cleave the acyl moieties. We found hydrazine to be just as
ineffective as hydroxide anion in deacylating rhizobial lipid A.

We tried a more non-polar reagent, tertiary butyl ammonium hydroxide, believing
that the counterion would be non-polar enough to bypass the negative interactions of the
hydrophobic acyl chains. Again, we report that tertiary butyl ammonium hydroxide was
not effective in promoting saponification.

Crude lipid A was treated with methyl sulfonyl anion and methyl iodide, in an
attempt to cleave ester/amide bonds and methylate the carbohydrate backbone to give
linkage information. This procedure gave essentially the same results as KOH treatment.
On this occasion, the free fatty acids were isolated directly as fatty acid methyl esters.
1H-NMR spectra looked essentially the same as crude lipid A as far as carbohydrate
resonances were concerned. The GC profile of fatty acids liberated as methyl esters is
shown in Figure 12. Again, the 27-hydroxyoctacosanoic acid was missing, indicating the
difficulty in which to remove it.

We have tried obtaining NMR spectra at higher temperatures (50°C) and in
different solvents such as d-DMSO and d—pyridine. This enhanced the carbohydrate
signals somewhat, but they were still broad since the long chain fatty acids were still
present. The methods used on the "typical" lipid A were not effective on our lipid A
molecules. R. trifolii ANU843 lipid A molecules seem to have very strong hydrophobic
attractions for each other, thus causing the formation of very stable aggregates. It is the

unique 27—hydroxyoctacosanoic acid that is the cause of these effects.

|.SOO

IOI77I

II.‘.I
‘\

five 9- I90! IOII‘IS‘
3.791

.070

 

 

 

new;

GC profile of R. trifolii ANU843 fatty acid methyl esters derived by treating crude lipid
A with methyl sulfonyl anion and methyl iodide. Note the absence of 27-hydroxyocta-
cosanoic acid, indicating its amide linkage.

This aggregation problem has also made it difficult to obtain mass measurements
of the rhizobial lipid A using fast atom bombardment mass spectrometry. Rhizobial
samples have been submitted many times for mass analysis, and yet nobody was able to
obtain a spectrum. This was again attributed to aggregate formation of lipid As, making
it difficult for the sample to leave the matrix on the probe. The only person to
successfully obtain a lipid A FAB/MS spectrum was Dr. R. I. Hollingsworth, and this
data will be presented later.

Information regarding the anomeric position of the rhizobial lipid A was obtained
from GC and GCfMS analysis of alditol acetates prereduced with sodium borodeuteride.
The GC/MS fragmentation pattern for prereduced galacturonic acid indicated that the
reducing end is tied-up as a glycoside. The ion fragments are two units apart for all major
ions, indicating that one end of the molecule was labeled with deuterium. The
fragmentation patterns for glucosamine showed that some of this carbohydrate was free,
while some of the carbohydrate was found in glycosidic linkage at the anomeric position.
This was evidenced by the presence of isotope peaks of equal intensity for the smaller
mass ions. During the prereduction step using sodium borodeuteride, any free anomeric
position would have been reduced to the alditol and thus contain one deuterium. This
made the ion fragments from this end one unit greater than those molecules that were
glycosidically linked and did not pick up a deuterium at this point. Furthermore, the ion
fragments were identical to those of the glucosamine standard. Since the glucosamine
standard was not subjected to a prereduction, and contains two protons at position Cl ,
this corresponded to a glucosamine that was linked at its glycosidic position (Figure 13).

Our structural information suggests that R. trifolii lipid A carbohydrates are free
and glycosidically linked at the anomeric carbon. Close examination of the 1H-NMR
spectrum for the lipid A that was treated with phenylisocyanate indicated that

ethanolamine was present as a substituent at the anomeric site (Figure 14). lH-NMR

Figure 13

13(A) GC/MS fragmentation pattern for prereduced galacturonic acid. The ion
fragments are two units apart for all major ions, indicating one end of the molecule was
labeled with deuterium, and therefore the reducing end is tied-up as a glycoside.

13(B) GC/MS fragmentation pattern of glucosamine. The ion fragmentation pattern
shows some of the carbohydrate to be free, as evidenced by the presence of isotope peaks
of equal intensity for the smaller mass ions. Some carbohydrate was found in glycosidic

linkage, as evidenced by identity of peaks with those of the glucosamine standard not

containing deuterium.

In 23m.

mm—

<2

N\:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

now own son 0mm OGN and no“
.IrLlL >4 Will. Jack J . _.
A mmm ~mm g. bk“ 6 Mi. “—
. an m
. .. 0.: .
4 . o a ram
m L o
A 0 d . .. 0
L . u
. m N .
L m m vfi-H . a
A v a
4 5m c
e v D
. v >
. w
«m .00 o
. a
F . ._
.tttllt 11.11 litll:%ooa
N\Z .
00m 80m 08v 00m OQN Gad o
. f - m1 . it Fiii.- iiil rid Q
. . mwm a H
m m . h .
H mm WEN
H _ an m a .
. am .
fir. .av
L P .
a a ..
.4 P Ru 0w
H e a H
. s
.. .00
. m: e. .

 

 

 

 

mo—uu-«>o (Isaac-oecuo

Figure 14

lH-NMR spectrum for R. trrfolii ANU843 lipid A treated with phenylisocyanate,
indicating the presence of ethanolamine at the anomeric site.

55

3133: 1.39! no: «-

 

8?? o
0&3 _ .8“; 2 aflogfi Suntan n.
V— .m— en? 2 a5 a
than .
.09. a 38.3.3393! «.0 n
.. tennis . tonne «
:8. a banana a
sac m w n u n m m
P. by L IF bl bl? r P F p h p h b — P FL — b n P b F b P Ll Dd H L
1 Iii <III .It 4‘ «airfiti
u o s a 2. S
w «r
Eco 0." n4 o.m n.m o.n n.n o.v n.v 0.6. 06 o o
hLL t— r p P b — rb - n —b 5L b F b r? P n h - bL 1P 1P P b p — n n b h — D Pr FLI - b D b F ~1P

 

 

 

 

 

 

 

 

 

 

 

.32:- . In» sent; .38

56
signals correspond to: signals at 1.9ppm (CH3 on the end of fatty acyl chains), 1.0 -

1.4ppm (CH2 of acyl chains), 1.0ppm (CH3—CHO-CH2 for C28), 1.5ppm (CH2—CH2-
CHO), 2.3ppm (CH2-CO for C28), 2.4—2.7ppm (CHO-CHz-CO), 3.5—3.6ppm (OCH3),
3.6-3.7ppm (NH-CHz-CHz-O), 4.2-4.4ppm (O—CHz-CHz-NH), 4.8—4.9ppm (OCH-CH3)
5.2-5.4ppm (OCH-CHz-COO), 7.0-7.8ppm (ring 1Hs of phenylcarbamate). Selective
proton decoupling of the above peaks gave coupling patterns that agree with the above
peak assignments.

In order to prove the presence of ethanolamine at the anomeric position, we
attempted to synthesize a structural analog from glucose containing ethanolamine
functionalized as a phenylcarbamate derivative at the anomeric position. In this way we
could compare the chemical shifts of NMR signals of the synthesized product with that of
rhizobial lipid A. Unfortunately, we were not successful in this endeavor.

Studies on purified lipid A components of crude lipid A were much more
rewarding. Crude lipid A was fractionated on a C18 reverse-phase column by HPLC.
Fractions were eluted from the column with water, 2:1 water/methanol, 2:1 methanol/
water, pure methanol, and pure chloroform. The pure methanol fractions were combined,
subjected to preparative TLC on C18 reverse—phase plates, and two bands reacting
positively with orcinol spray were scraped and eluted with 1:1 chloroform/methanol.

The presence of carbohydrate and fatty acyl components were seen in the 1H-
NMR spectrum of the methanol fraction from C18 reverse-phase separation of crude lipid
A, (Figure 15). Resonances between 0.7 and 0.9ppm, 1.1 and l3ppm, 2.1 and 2.5ppm,
and 3.2 and 5.3ppm correspond to methyl resonances of fatty acyl components,
methylene groups of fatty acyl groups, methylene groups next to carbonyl functions, and
glycosyl resonances and resonances due to protons attached to other hydroxylated
carbons, respectively. The presence of resonances for only one glycosyl residue support
a monosaccharide component in the methanol fraction of the lipid A components

obtained from HPLC separation. GC/MS analysis of the fatty acid methyl esters liberated

57

Figure 15

Proton NMR spectrum of the methanol fraction obtained from C18 reverse-phase
chromatography of R. trifolii ANU843 crude lipid A. Fatty acid proton resonances occur
between 0.7 and 2.5 ppm. Carbohydrate resonances are between 3.2 and 5.3 ppm.

gm

Emu m w
r N n v
o

— — — DDIIII-Pl ’7' rP'
Err
Eh?»
I DD. Err
DI 'F'PPFFP
II I?
PFFFF
rrP’
p Err?
Fr
PPR
pr?
R?
FF

1 Wales? ,

 

 

 

 

 

 

59

by acid catalyzed methanolysis indicated that the major fatty acyl components of the
methanol fraction were 27-hydroxyoctacosanoic acid and the 3—hydroxytetradecanoic
acid.

13C-NMR analysis supported the presence of a monosaccharide component in the
methanol fraction, (Figure 16). Resonances were seen at 99.99 and 97.96ppm,
corresponding to B and‘ carbon resonances of a reducing glycosyl residue. No other
glycosyl carbon resonances were observed. A signal at 54.10ppm was assigned to a
carbon atom attached to nitrogen. This indicated the presence of a single amino sugar
and not a glucosamine disaccharide.

As was previously mentioned, TLC of the methanol fraction yielded two major
components. The 1H-NMR spectrum of the faster moving band appeared very similar to
the parent compound, (Figure 17). A downfield triplet present in the spectrum of the
parent compound is missing in the spectrum of this fraction. This proton could be
assigned to an axial ring proton. Its chemical shift may be due to acylation at the carbon
this proton was attached to. Absence of this resonance in the faster migrating TLC band
was indicative that this molecule lacked an acyl residue. Fatty acid analysis indicated
that 27-hydroxyoctacosanoic acid was the only fatty acid present. Pseudo-molecular ions
at m/z=616 and m/z=638 ([M+H]+ and [M+Na]+) were observed in the FAB/MS of the
faster moving TLC component, (Figure 18). This was indicative of a molecular weight of
615.

The second TLC component was the major band obtained. In the 1H-NMR
spectrum, the downfield triplet was present as in the parent compound, indicating the
presence of an O-linked acyl residue, (Figure 19). GC analysis of the alditol acetates in
this fraction showed glucosamine after prereduction. This indicated the presence of the
uronic acid of glucosamine in the lipid A. This sugar residue correlated with the
FAB/MS data of this second TLC component. The spectrum contained ions at m/z=842

and m/z:864. This corresponded to [M+H]+ and [M+Na]+ to give a molecule with a

Figure 16

13C-NMR spectrum of the methanol fraction from C 18 reverse-phase chromatography.
Signals at 99.99 and 97.96 ppm are due to B- andoL-forms of the anomeric carbon of the
glycosyl component. The signals in the 60-80 ppm region confirm the presence of one
glycosyl component. There is one carbon signal at 54.1 ppm attributable to a carbon
bearing nitrogen.

61

gleam

 

 

 

 

 

 

 

 

62

Figure 17

Proton NMR spectrum of the faster moving band on preparative TLC analysis of lipid A.

The triplet at 5.21 ppm present in the parent compound is missing. indicating the lack of

acyl substitution. Intense signals between 3.1 and 3.4 ppm are due to traces of methanol
and other hydroxylated species such as moisture.

63

C Eswfi

'I’IIIIIrP'FIFIIIIPIR’l'IF’IPIII, I
P’IP' IVPPDI’D by F
>>>>>
-

 

 

.33.

 

 

 

— ——_~-O-

 

 

 

 

EM

Fast atom bombardment mass spectrum of the faster moving TLC band. The
pseudomolecular ions appear at m/z:616 ([M+H]+) and mlz=638 ([M+Na]+).

81mm: - cum" -

Ilulfl 2: a

 

 

 

 

mm

 

.wum

 

 

 

 

WN.§

 

GDN

 

 

 

 

 

 

we.

 

 

 

mo-I«--)o «83:14:00

figural?

Proton NMR spectrum of the slower moving band of preparative TLC. Note the presence
of a triplet at 5.22 ppm, indicative of acylation.

67

o w

'PF'p'k’FIIPFFbI

1L1.

 

 

9 Baum

2am
N n

——h —..-

I IOOOI

 

 

v

m

rrPI’IIIP’FFrP>P>>IP

rIP’rFPPF

1—

o

FPFFP

Figure 20

Fast atom bombardment mass spectrum of the slower migrating TLC band.
Pseudomolecular ions appear at m/z=842 and mlz=864.

69

ON 23E

 

 

 

 

 

n~.§

 

“a.

 

 

 

 

W

O
n

 

W
n

 

are

 

C0¢Cfl~>0 «83“‘600

70

molecular weight of 841 , (Figure 20). This corresponded to the molecular weight of the
first TLC component after substitution by a 3-hydroxytetradecanoyl residue. Mild
methanolysis of the second TLC fraction liberated a 3-hydroxytetradecanoyl residue, thus
confirming that an amide linkage binds the 27-hydroxyoctacosanoyl residue to the
carbohydrate backbone of R. trifolii ANU843 lipid A.

FAB/MS showed that neither of these molecules are phosphorylated. This agreed
with our results of the absence of phosphate in R. trifolii ANU843 LPS and that of other
literature citings (1 10,1 11).

Conclusions Refiggrdinfig R. trifolii ANU843 Lipid A Structure

The lipid A component of endotoxin is responsible for the diverse biological
effects brought on by this molecule. Whether lipid A acts directly or indirectly remains
to be determined. Lipid As extensively studied from enteric organisms such as
Salmonella and E. coli and non-enteric genuses such as Pseudomonas and Vibrio all
possess the same basic structure. Lipid As with varient structures are considered non-
toxic and do not induce biological activity or induce attenuated effects.

Free lipid A obtained from chemical treatment of LPS contains several
components which vary in structure due to natural heterogeneity and also because of
degradation. Hence, the classical lipid A will typically contain monophosphorylated and
diphosphorylated backbone variations of head group substitution by polar substituents,
heterogeneity of chain length and variation in degree of and site of substitution by fatty
acids. These factors make purification of pure homogeneous species very difficult and a
synthetic route to pure species is very often preferred (34,124,125).

Rhizobial LPS was previously shown to elicit some of the classical endotoxin
activities such as elicitation of positive LAL and Schwartzmann reaction responses and
the ability to act as a mitogen. At the time, it was assumed that Rhizobium lipid A had
the same structure as enteric lipid A, based upon its ability to bring about these classical

endotoxin effects.

71

One of the major questions addressed by this research was : Does the classical
lipid A structure apply to the genus Rhizobium ? We have previously mentioned
conflicting reports in the literature addressing this issue. Data obtained from our study
showed that the usual lipid A model does not apply to the lipid A of Rhizobium mfolii
ANU843.

Work in this lab on crude lipid A has confirmed earlier observations that R. trifolii
ANU843 LPS contains an unusual, branched fatty acid as well as another unusual fatty
acid component, 27-hydroxyoctacosanoic acid (109). Gas chromatography of fatty acid
methyl esters of the LPS of R. trifolii clearly shows that this fatty acid is a major
component of the lipid A of this organism. Neither of these fatty acids had been reported
before in gram-negative bacteria, thus making the lipid A of R. trifolii unusual in its
structure. This finding has been confirmed by Urbanik-Sypniewska e_t_. a_l., who found
this same fatty acid in Rhizobium meliloti LPS ( 126). The long chain fatty acid has also
been found in other nitrogen-fixing organisms (127).

The lipid A we obtained from R. trzfolii ANU843 contains several components
when analyzed by TLC. We have isolated and completely deduced the structure of a 2-
amino-2—deoxy-2-N-(27-hydroxyoctacosanoyl)-3-O-(3-hydroxytetradecanoyl)-glgpp—
hexuronic acid and its de-O-acylation product from the free lipid A of R. trifolii
ANU843 (114). These components were found in the methanol fraction from C18
reverse-phase chromatography of the crude lipid A and are known to be minor
components, especially since they lack the other 3-hydroxy fatty acid components,
(Figure 21).

Other evidence of these structures being minor components comes from the
phenylcarbamate derivatization of fatty acids. In this experiment the 27-hydroxyocta-
cosanoic acid was shown to be linked to another fatty acid at its 27-hydroxyl position.
There was no evidence of the free 27-hydroxyoctacosanoic acid as was isolated in the

minor lipid A components fully characterized. In experiments utilizing a sodium boro-

72

 

Figure 21

Structure of 2-amino—2-deoxy-2—N-(27-hydroxyoctacosanoyl)-3-O-(3-
hydroxytetradecanoyl)-gluco-hexuronic acid and its de-O-acylation product when
R=CH3-(CH2) 1 o—CHOH-CHz-CO; R=H, respectively.

73

deuteride prereduction step to determine the status of the anomeric carbon of the rhizobial
lipid A carbohydrates, no evidence of the uronic acid of glucosamine was found.

After prereduction of the crude lipid A of R. trzfolii ANU843, followed by
hydrolysis, reduction, acetylation, and GC/MS analysis, we found another major
carbohydrate component of the lipid A to be galacturonic acid as a major constituent of
the lipid A head group (52). Since then, another group has also found galacturonic acid
to be a major constituent of rhizobial lipid A (127). Prereduction using sodium
borodeuteride showed that glucosamine was also a major component of the amino sugar
found in R. trifolii . The composition of our crude lipid A preparation is approximately
1:1 galacturonic acid to amino sugar.

I propose that one of the major lipid A structures found in R. trifolii consists of a
disaccharide composed of glucosamine as the reducing end sugar, linked glycosidically to
the anomeric position of galacturonic acid. This is based upon the following results.
When crude rhizobial lipid A is not prereduced before derivatization to the alditol acetate,
nothing is seen by GC analysis. Uronic acids are known to be difficult to cleave under
acidic conditions. Thus, glucosamine was not liberated during acid cleavage and
remained bound to galacturonic acid. This type of molecule would not be derivatized
properly and if it were, would not be easily eluted from the GC column. When the uronic
acid was converted to the methyl ester and then prereduced, both galactose and
glucosamine alditol acetates were liberated. Remember, experimental evidence showed
galacturonic acid to be linked at its C 1 position, and therefore it cannot be at the reducing
end. Glucosamine was shown to be linked at the reducing end in the same experiment.
We also know that ethanolamine was shown to be a substituent in NMR experiments.
The ethanolamine most likely is linked to the reducing sugar, glucosamine, at position
C1. This substituent would have been hydrolyzed during the derivatization procedure for
alditol acetates and would not have shown up in GC/MS analysis. This scenario also

applies to galacturonic acid-galacturonic acid-ethanolamine.

74

Another possibility for a rhizobial head group is a disaccharide of galacturonic
acid and glucosamine, but this time they are linked through both anomeric positions, as
are glucose and fructose in sucrose. Again, absence of prereduction would result in no
loss of glucosamine. When prereduction does occur, one would liberate the alditols of
galacturonic acid and glucosamine. Again, this fits the model of galacturonic acid and
glucosamine being linked at their anomeric positions. One might also have galacturonic
acid linked to galacturonic acid through their anomeric positions.

There is also evidence of glucosamine being free at the anomeric position. This
could be explained by galacturonic acid-glucosamine, free at the reducing end. We do
not have a glucosamine disaccharide nor a g1 ucosamine-galacturonic acid. Without
prereduction one would still liberate the alditol acetate of glucosamine. This did not
happen, so they cannot be present.

R. trifolii LPS was found to be completely devoid of phosphate. This result is in
agreement with the findings of Zevenheuzen e_t pl; (111), who studied the hexosamine
content of Rhizobium LPS. They found a low hexosamine content of rhizobial LPS as
compared with corresponding values for enteric LPS. They suggested that the lipid A
portion of Rhizobium is essentially different from that of the Enterobacteriaceae and
that it must have a different backbone. They also concur with the results of Humphrey
and Vincent (1 10), who found a low phosphorous content in rhizobial LPS and
furthermore suggested that the small amount of phosphorous present was due to
contamination by phospholipids and nucleic acids. All of these phosphorylation results
are at variance with the model of the typical lipid A.

All of our results differ with those obtained by the group of Russa e_t fl, (1 12).
This group of investigators studied two wild type and two mutant strains of R. trifolr'i and
concluded that the lipid A of Rhizobium contains g1 ucosamine disaccharide, phosphate,
and no unusual fatty acids. Their study relied heavily upon chromatographic

comparisons of their lipid A sample with enteric lipid A samples. Their approach does

75
not allow for possibly finding structurally different lipid As if the differences do not

significantly contribute to chromatographic or electrophoretic mobility of the molecule.

As stated previously, R. trifolii lipid A was found to elicit some of the classical
endotoxin phenomena such as LAL response, Schwartzmann reaction, and the ability to
cause B-cell proliferation. However, we now know that R. trifolii has a radically
different structure from that of the classical endotoxin with respect to carbohydrate
content, fatty acid content, and degree of phosphorylation. How does this lipid A varient
form relate to the classical endotoxins?

Structural analogs of endotoxin have shown that the lengths of fatty acyl chains
are important in rendering the molecules toxic. It has long been accepted that the acyl
chains may vary in length between 12 and 18 carbons in length. The presence of a 28-
carbon long fatty acid is inconsistent with the traditional endotoxin model. Rhizobial
lipid A molecules, being membrane derived, must contribute to the architecture of the
outer membrane of these organisms. We propose the 27-hydroxyoctacosanoic acid
normally spans the outer membrane of Rhizobium in a trans-membrane orientation. This
is consistent with what is known about the ultrastructure of these organisms (128). It has
been shown to be impossible to effectively freeze fracture the outer membrane of these
organisms (52). R. I. Hollingsworth had previously demonstrated that the 27—hydroxyl
group in 27-hydroxyoctacosanoic acid was esterified by another 3-hydroxy fatty acid
(109). Our phenylcarbamate derivatization also supports this, as it showed the 27—
hydroxyl position to be esterified. An arrangement where a 3-hydroxytetradecanoyl
residue esterifies the 27-position of this long chain fatty acid would give a structure
where two fatty acid chains meet in the middle of the membrane and one chain
completely traverses it, (Figure 22).

This unusual arrangement of fatty acids in the membrane of Rhizobium has been
proposed as an obligatory requirement for their viability in their natural environment.

This structure may pack very well into the membrane and serve as a permeability barrier

76

OH

o Liv/MW
o o /\/ i
W ’\/\/\/ .
on on 0 0H Ion

Figure 22

Structure of a possible lipid A component or substructure from R. trifolii ANU843 LPS.
Notice the two shorter acyl chains meet in the middle of the long chain, transmembrane
fatty acid.

77

to oxygen while nitrogen fixation occurs. From our chemical analyses, we know that
these molecules form very stable aggregates. The usual procedures to chemically modify
endotoxins do not work on these molecules. The length of the hydrophobic region of the
molecule displayed in Figure 22 is 37 Angstroms, exactly the dimensions of the
hydrophobic region of a typical membrane bilayer. A structural requirement necessary
for a transmembrane component would be presence of polarity at both ends. This
requirement is met by the structure shown in Figure 22.

Rhizobial lipid A completely violates the "law" that endotoxins must possess
phosphate substituents. When comparing the rhizobial and enteric lipid As however,
there is a unifying feature of negative charge. Both the classical lipid A molecule and the
rhizobial lipid A have charged backbones. LPS serves to protect gram-negative
organisms from their environment. If the matrix required is a negatively charged mass,
cross-linked by divalent cations, then the bacterium will use those building materials
more readily available to it. A soil bacterium might choose to use a carbon source as its
charged element, such as carboxylate, since phosphate would be more importantly used
for nucleic acid biosynthesis. A bacterium which exists in the digestive tract has an
unlimited source of phosphate from plant and animal materials. There is a greater chance
that phosphate would be used as the charged element, in this case. The final product
synthesized by both organisms could function in the same manner. An external receptor
system on target cells may have difficulty distinguishing between the two.

This has led us to propose a model for the interaction of LPS with target cells. It
is the same model as that proposed by Diane Jacobs (67) and Larsen e_t gI_._ (65). This is
not the model that is most widely accepted by the endotoxin community.

On the cell there are mediators/receptors (probably proteins) which recognize
certain structural features of the LPS complex. Not every LPS aggregate from varying
bacteria will be identical, but the mediators will recognize certain structural features of

the LPS complex. The target cell will react to a three-dimensional matrix of charge,

78

which can be factored out of both the rhizobial and enteric aggregates.

After recognition, the lipid A aggregates fuse with the target cell plasma
membrane. Analogs having greater than 18 carbon atoms or less than 12 atoms, have
been shown to induce little or no activity. This does not appear to be the case for R.
mfolii LPS. For the analogs, stable integration into the bilayer is not possible since the
chains are too long for one leaflet and too short to span the membrane or get stuck in the
middle of a leaflet since they are too short for it. They create disorder which leads to
improper packing and a marked increase in fluidity. This is not true for the rhizobial lipid
A, for it has been shown that its fatty acids successfully cross the membrane, based upon
computer analyses.

We are looking at charge, packing, and overall conformation of the LPS/lipid A
molecule, instead of just which sugars are present, or how long the fatty acid chains are.
We have studied how these molecules interact with mammalian cells and what these
radically different structural features do to the ability of rhizobial LPS to behave as a
typical endotoxin molecule. The biological aspects of this project are discussed in the

following chapter.

Chapter Five: Biological Responses Induced by Rhizohium trifolii ANU843 LPS

Introduction

 

It is known that LPS is capable of causing a wide variety of biological effects in
mammalian systems. Some of these effects are beneficial such as proliferation of B-cells,
immunoglobulin synthesis, and tumor necrosis (7,8,9,10,1 1). On the other hand, these
molecules are potent toxins capable of fever induction, vascular damage, and causing
death of the infected host (12,13). Molecules with the classical endotoxin structure, as
previously described in this text, are able to bring about all of these effects, both good
and detrimental. Those molecules having varient lipid A structures are considered non-
toxic or less toxic compared to their enteric counterparts.

LPS from R. trifolii , as shown by Dazzo e_t a_l., exhibited some of the classical
endotoxin activities (52,106,107). We now know that this lipid A has a radically
different structure from that of the enterics, and this was described in the previous section
of the text. We set out to determine which of the classical endotoxin responses are
conserved by the LPS of R. trrfolii . The data obtained provide insight into the
mechanism of LPS action.

In order for LPS to exert its effect upon mammalian systems, it must successfully
interact with the plasma membrane of its target cell. One set of experiments comparing
enteric and rhizobial LPS involved the binding of fluorescently labeled LPS to various
eukaryotic cell lines and examining the extent of this binding.

We followed the theory which suggests that LPS associates with cells in a manner
that is indicative of a two-step process (66,67). The first step involves association of the
LPS aggregate with the cell surface as mediated by electrostatic interaction. LPS, being
negatively charged, associates with positively charged species on the cell surface. This
association increases the thermodynamic probability of intercalation of LPS monomers

into the cell's plasma membrane. This coalescence is considered to be the second step.

79

80

Intercalation of the lipid A will decrease the fluidity of that region of the plasma
membrane, and this decrease in membrane fluidity leads to cellular activation. Increase
in microviscosity of the lipid bilayer, as caused by LPS, has been demonstrated by Larsen
e_t e_tL and Price e_t gl_. (65,68). Both groups monitored the intensity of polarized light
emitted by diphenylhexatriene, a probe embedded in the membrane of mononuclear cells
previously treated with LPS/lipid A. From their experiments they determined that
pretreatment with LPS/lipid A increased the apparent microviscosity of the lipid bilayer
which resulted in increased order within the membrane.

All of our binding studies were performed at low temperature to ensure
observation of the initial binding step followed by coalescence and not just fusion of
lipids. We chose to label several cell lines to show that LPS is capable of binding cells
that are not just of bone marrow origin. We chose cell line RAW264.7 because it has
previously been used in LPS activation studies. CHO-Kl is a Chinese hamster ovary cell
line chosen because it was previously shown by Raetz (89) not to bind LPS and thus
provided evidence suggestive of the presence of an LPS receptor. BCLlclone5Blb was
chosen because it is a transformed cell line that was supposedly responsive to endotoxins
and could be used in studies that look at immunoglobulin secretion and cell proliferation.

In this section we will present data involving biological responses brought about

by R. trifolii LPS/lipid A, or the lack thereof. Several i_n vitro and i_n vivo assays were

 

used to answer the question concerning conservation of endotoxic activity. l_n gigp
assays included the induction of IgM, IgG, lgA, TNF, IL-1, and IL-6 synthesis and
secretion, besides the extent to which LPS binds varying cell lines. The Limulus
Amebocyte Lysate assay was also used as a screen for endotoxicity. The ability of R.
trifolii to act as a mitogen was also tested. g1 v_iv_9_ testing included the ability to generate
fever and lethal toxicity testing.

There have been several instances where polycationic molecules have been used

to block the activity of endotoxin (142). Polymyxin B sulfate and polylysine are two

81

examples of molecules with net positive charge that are capable of interacting with LPS
electrostatically and preventing it from binding to cells such as B-lymphocytes (66).
Molecules having a net negative charge, such as dextran sulfate or heparin, are also
capable of blocking LPS activity by interacting with positively charged species on cell
surfaces, thus preventing LPS from doing so. Besides electrostatic forces, hydrophobic
interactions are also important. For example, polymyxin B nonapeptide, a derivative of
polymyxin B lacking the hydrophobic tail, inhibits LPS binding to a lesser extent than
polymyxin B. Endotoxin inhibitors were synthesized to interact with LPS both electro-
statically and hydrophobically. These molecules were tested for their ability to prevent
binding of LPS to macrophage-like cells and for their ability to prevent a positive

response in the LAL assay. The data generated by all of the above i_g vitro and i_g vivo

 

experiments provides insight into the mechanism of bacterial endotoxins.

Medals and Methods:
LPS Isolation

The same procedures used to grow R. trifolii ANU843 and isolate its LPS were
used as those described in the Materials and Methods section of Chapter 4. Salmonella
typhimurium LPS was purchased from Sigma Chemical Co. (St. Louis, MO). Both LPS
types were converted to their triethylammonium salt forms by passage over a Sepharose
4B size exclusion column (129,130) using 0.05M fonnate buffer, pH adjusted by addition
of triethylamine.
USP XXII Pyrogen Assay

This assay involved measuring the rise in temperature of rabbits following the
intravenous injection of an endotoxin test solution. S. typhimurium LPS was injected at a
dose which equaled 0.05 pg/kg body weight. R. trifolii ANU843 was injected at a dose
of 0.05 rig/kg and 1.0 yg/kg body weight. Both LPS types were dissolved in pyrogen-
free diluent and were injected into an ear vein of each of three rabbits in volumes of 1.0
mL. Rectal temperature-measuring probes, remaining inserted throughout the testing
period , were used to record body temperature at 1, 2, and 3 hrs. subsequent to injection.
Lethal Toxicity Testing

Healthy, previously unused albino mice, weighing 17-23 grams, were used.
Water and food were provided fl mgt_ur_n_. The endotoxin solution and a corresponding
negative control were injected in groups of five mice each. All LPS samples were
dissolved in pyrogen-free phosphate buffered saline and administered intraperitoneally by
injecting 0.5 mL of material into each mouse. R. trifolii was administered at
concentrations of 1000, 500, 100, and 50 yg/animal. S. typhimurium LPS was
administered at the same concentrations as rhizobial LPS. After injection, the animals
were observed immediately and at 0.25, 4, 24, 48, and 72 hrs. After 72 hrs., the animals

were weighed.

Limulus Amebocyte Lysate Assay

R. trifolii LPS was screened for its ability to be recognized as an endotoxin by
utilizing the Limulus amebocyte lysate assay. We purchased the Whittaker Bioproducts
QCL-lOOO Quantitative Chromogenic LAL , Whittaker Bioproducts, Inc. (Walkersville,
MD). This was a chromogenic assay which relies upon the ability of endotoxin to
activate an enzyme, a serine protease, found in the lysate, which will then cleave p-nitro-
aniline conjugated to a peptide substrate. This resulted in chromophore release. The
amount of chromophore released is proportional to the amount of endotoxin present. The
R. trzfolii ANU843 was tested in parallel with E. coli 01 1 1:B4 endotoxin, the reference
standard. The assay was performed following the test kit directions.
B-Lymphocyte Proliferation

R. trifolii 0403 was grown to early stationary phase (K90) and to exponential
phase (K50). LPS was extracted and tested for its ability to act as a mitogen with
splenocytes isolated from LPS-responsive mice, C3H/St and from LPS-hyporesponsive
mice, C3H/HeJ. Measurement of tritium—labeled thymidine uptake was used as a means
of visualizing B-cell proliferation. Assays were performed in the presence and absence of
fetal bovine seum (FBS). Tests were run in parallel with LPS from two different strains
of Salmonella LPS. The data from these experiments was kindly provided by F. B.
Dazzo, with the actual experiment performed in the lab of Dr. D. M. Jacobs.

Alternatively, proliferation assays were performed using a transformed, B-cell
leukemia cell line, BCLl cloneSBlb, purchased from American Type Culture Collection
(ATCC), using the procedure described in "Selected Methods in Cellular Immunology"
(131). Cells were prepared to a final concentration of 2.5x105 cells/mL in RPMI-1640
medium, 15% FBS. 100 14L of the cell suspension was added to wells in a 96-well
micro'titer plate. 10 FL of LPS solutions of R. trifolii ANU843 and S. typhimurium were

added to the appropriate wells to give final concentrations ranging from 4.1— 16.4 pg/mL.

84
10 pL of medium was added to negative control wells. The cells plus LPS were

incubated for 48 hrs. at 37°C and at a 5% C02 atmosphere. 3H—thymidine (ICN
Radiochemicals, Division of ICN Biomedicals, Inc., 6.7Ci/mmol, 248GBq/mmol,
1.0mCi, 1.0 mL sterile aqueous solution) was diluted 1:10, and 10 pL of diluted 3H-
thymidine was added per well. Cells were pulsed 16 hrs., after which time they were
harvested onto filters, washed with distilled water, and then lysed with 10% TCA,
followed by further water washes. The filters were dried, cut, and placed into
scintillation vials. 10 mL of scintillation cocktail was added to the filters, and each vial
was kept in the dark, in the cold, for 45 min. Vials were counted on the 3H-channel of an
LKB Wallac 1209 Rackbeta liquid scintillation counter for 2 min. (LKB Wallac,
Finland).

Proliferation assays using the BCLlcloneSBlb cell line were also performed with
the following modifications: (i) We used a more dilute cell suspension, 2.5x104 cells/
mL, (ii) 3H-thymidine was added after 24 hrs. incubation of LPS with cells, then pulsed,
followed by harvesting, (iii) 3H-thymidine was added at the same time as LPS addition,
followed by the usual workup, (iv) 3H-thymidine was diluted 1:5 with cold thymidine
and then further diluted 1:9 with RPMI-l640 medium. This was then added to the cells
incubating with LPS, and (v) The cell medium, RPMI-l640, was prepared without 2-
mercaptoethanol and FBS, and without only FBS. The assay was performed as originally
described.

Immunoglobulin Secretion

Cell line BCLlcloneSBlb, a B-cell leukemia which is LPS responsive (132), was
stimulated with LPS from S. typhimurium and R. trifolii ANU843. The media in which
the cells grew were assayed for the secretion of I gM using the ELISA (enzyme linked
immunosorbant assay) technique. lgG and IgA secretion were also investigated. BCLI
cloneSBlb was resuspended to a final concentration of approximately 3.0x104 cells/mL.

One milliliter of this suspension was placed into each well. Into each well was also

85
placed 100 pL of each LPS solution, or RPMI-l640 for the negative control. Final LPS

concentrations ranged from 8-80 yg/mL of LPS. The cells were allowed to continue
growing for 5 days at 37°C in 5% C02 atmosphere, at which time 250 ”L of the medium
was removed for IgM screening. The removed medium was centrifuged and 200 pL of
the supernatant was removed and frozen until assayed. The cells were allowed to
continue growing for a total of 7 days, at which time 250 pL medium was removed for
lgG and IgA screening. Again, the medium was centrifuged and the supematants were
removed and frozen until tested. Alternatively, the BCLlclone5B 1b cell line was
allowed to grow in the presence of LPS for 24, 48, and 72 hrs., at which times media was
removed, centrifuged, and frozen until ready for use.

All supematants were assayed for immunoglobulin release in the following
manner. The appropriate standard (100 pL) and 100 ”L of the supematants diluted 1:10
with 1% FBS/PBS/0.02% NaN3 were added to 96-well plates, previously coated with
GAMI g (goat-anti-mouse I gG, IgM, IgA) at a final concentration of 0.015 mg/mL. Plates
were covered with parafilm and incubated 1.5 hrs. at 37°C. The plates were next washed
3 times with 0.05% Tween/PBS, followed by 4 washes of distilled water. Plates were
patted dry. To each well was added 100 ”L of the appropriate alkaline phosphatase
conjugated antibody, diluted in 1% FBS/PBS/NaN3. We incubated the plates at 37°C for
1.5 hrs., covered with foil. All steps hereon were light sensitive. Again, the rinsing step
was repeated. The phosphatase substrate, Sigma 104 Phosphatase Substrate (1 mg/mL),
was added to all wells and the plate was again incubated at 37°C for 1 hr. 100 yL of 1M
NaOH was added per well to stop the enzyme reaction. The plate was then read at 405nm
and absorbances were recorded. Amount of antibody secreted was calculated from the
standard curve.

Production of TNF
R. trifolii ANU843 lipid A was sent to XOMA Corporation, Berkeley, CA, to test

its ability to cause macrophage to release TNF. The assay in which rhizobial lipid A was

86
tested was run in parallel with E. coli J5 (Re) lipid A, obtained from Ribi ImmunoChem

Research, Inc., Hamilton, MT. Recombinant human TNF-ii (20 U/ng) and rabbit anti-
human TNF-a polyclonal antibody were purchased from Genzyme (Cambridge, MA).
Freshly drawn blood from healthy human donors was collected into Vacutainer tubes
containing anticoagulant (ACD, Benton Dickinson, Rutherford, NJ). Aliquots of blood
(225;:L) were mixed with 20 pL of RPMI medium containing varying concentrations of
lipid A. The mixture was incubated at 37°C for 46 hrs. and the reaction was stopped by
the addition of 750 pL of RPMI, followed by centrifugation at 500g for 7 min. Super-
natants were assayed for TNF levels using the Biokine ELISA test ,T Cell Sciences,
(Cambridge, MA).
Release of Interleukins, lL-1,IL-6, and IL—8 by Rhizohium LPS

R. trifolii ANU843 lipid A was sent to XOMA Corporation, Berkeley, CA, to
test for its ability to bring about production and secretion of IL-1 and IL—6 by
macrophage. Results are forthcoming from XOMA Corp. A measure of the ability of R.
mfolii ANU843 lipid A to induce IL-8 secretion was run in parallel with that of E. coli
lipid A at XOMA Corp. Freshly drawn blood from healthy human donors was collected
as described for the TNF assay reported in this section of text. Aliquots of blood were
mixed with 20 pL of RPMI medium containing varying concentrations of lipid A. The
mixture was incubated at 37°C for 4—6 hrs. and the reaction was stopped by the addition
of 750 yL of RPMI, followed by centrifugation at 500g for 7 min. Supernatant liquids
were assayed for lL-8 levels using the Quantikine Human IL—8 Immunoassay (ELISA), R
& D Systems, Inc., (Minneapolis, MN).
Cell Lines Used in LPS Binding Assays

We have made a comparative study of the binding of rhizobial LPS and enteric
LPS to several eukaryotic cell lines. RAW264.7 is a murine, Abelson leukemia virus
transformed macrophage. RAW264.7 was grown in Dulbecco's modified Eagle's

medium, 90%, FBS 10%. Subcultures were prepared by scraping and resuspending in

87
fresh DMEM, then aspirated and dispensed into new flasks.

CHO-Kl is a Chinese hamster ovary cell line, Cricetulus griscus . CHO—Kl was
grown in Ham's F-12 medium, 90%, FBS 10%. These cells were subcultured by rinsing
the cell sheet twice with fresh trypsin (0.25%) solution. The trypsin was removed and the
cells were incubated at 37°C for 5—10 min., until the cells detached. Fresh culture
medium was added, and the cells were aspirated, then dispersed into new flasks.

BCLlcloneSBlb is a murine B-cell leukemia lymphoma cell line. Cells were
cultured in RPMI-1640 medium with 2.0 mM L-glutamine and 0.05mM 2-mercapto-
ethanol, 85%, FBS 15%. Subcultures were prepared by scraping and resuspending the
dislodged cells in fresh RPMI-1640. Aliquots were aspirated and discharged into new
flasks.

All three cell lines were purchased from the American Type Culture Collection
(ATCC). They were purchased under the ATCC No.: TIB 71, CCL 61, and TIB 197,
which correspond to the RAW264.7, CHO-Kl , and BCLlcloneSBIb cell lines, respec-
tively.

Coniugation of Fluorescent Dyes to LPS

R. trifolii ANU843 and S. typhimurium were chemically linked through their
hydroxyl groups to the fluorescent dyes, fluorescein isothiocyanate isomerl (FITC) and
rhodamine B (Rho), so that LPS binding to cells could be visualized using fluorescent
microscopy. LPS (IS-25 mg, depending on which LPS type was used) was sonicated
briefly in 1.5 mL pyridine until the LPS went into solution. 4—dimethylaminopyridine (5
times the weight of LPS) was added. This was followed by the addition of FIT C or Rho
(1/5th. the weight of LPS). The solutions were covered in foil, sonicated, and left to set
overnight at room temperature. The LPS-dye conjugates were placed on a rotary
evaporator to remove pyridine. They were then placed on an LH-20, lipophilic gel

permeation column, and eluted with 1:1 water/methanol, collecting 80 drops/fraction.

88

Elution of FITC-LPS conjugates resulted in the resolution of 2 peaks containing
FITC. 50 pL of each fraction was screened for the presence of carbohydrate using 5%
aqueous phenol/H2804. Only the first peak contained dye plus carbohydrate and these
fractions were pooled, frozen, and lyophilized. The second peak did not contain carbo-
hydrate and was discarded as only containing free FITC. Elution of the Rho-LPS
conjugates gave two peaks containing dye. The fractions containing rhodamine were
screened for carbohydrate by using the phenol/H2804 reagent. Only those fractions
containing both dye and sugar were pooled, frozen, and lyophilized. This time, much
free dye remained on the column and the free rhodamine was eluted off the column using
pure methanol.
Labeling of Cell Lines with LPS-Dye Coniugates

Pre-cleaned coverslips were placed into 6-well plates and cells suspended in
medium were added at concentrations ranging from 3x105 to 1x106 cells/well. Cells were
allowed to grow and adhere overnight, after which time they were washed in a balanced
salt solution and then treated with LPS-dye. Cells were labeled with the LPS-dye by
placing the coverslips face-down onto 10 yL of the LPS—dye solution, ranging from 5.7 to
5.9 yg/mL, (that had been previously cooled at 6°-8°C) for 5 hrs. Labeling was
performed in the cold at 6°-8°C. In some instances the cells were pretreated with a
IWM solution of sodium azide before labeling with LPS-dye.
imon of Cells labeled with LPS-Dye

The following steps were done at 6°—8°C, being careful to keep the cells out of
the light as much as possible. The coverslips with the adherent cells now fluorescently
labeled were transferred to clean, 6-well plates, and washed once with 4.0 mL of
balanced salt solution and twice with 4.0 mL NaN3 (lOmM) solution. The cells were
fixed for 10 min. in 4.0 mL 3.7% formaldehyde in HBSS. They were washed once in 4.0
mL of distilled water and then mounted onto slides in 1 drop of glycerol (Difco FA

mounting fluid, pH 9.0). The coverslips were sealed using clear nail polish.

89

Fluorescent Microscopy

Slides were viewed using a Leitz fluorescent microscope. Cells were viewed as
phase contrast images and fluorescent images. Images were magnified 128x.
Confocal Microscopy

Slides were viewed using a Zeiss 10 Laser Scanning Confocal Microscope.
Magnification was set at 2800x. Three-dimensional imaging utilized the scanning
confocal microscope plus a Silicon Graphics 4D—30 personal Iris computer. Software

was VoxelView E by Vital Images (Fairfield, IA).

82%

The LPS of a closely related strain, R. trifolii 0403, which we know to have the
same lipid A composition as that of R. trifolii ANU843, was studied under the
assumption that it had the same lipid A structure as the Enterobacteriaceae (106). In this
study, the LPS of this strain was shown to give a positive response in the LAL assay (data
not shown). It was also capable of acting as a splenic B-cell mitogen with the same
activity per p g as standard S. ryphimurium LPS preparations (Table l) in the absense of
serum and in LPS responder and hypo-responsive mouse strains. R. Irifolii 0403 also
tested positive for the local Schwartzmann reaction. This was an i_g yiyp assay involving
the formation of lesions after intravenous injection of LPS at the site of a previous intra-
derrnal injection of the same LPS. Hemorrhage and/or lesions appearing at the site of
original injection at approximately 4 hrs. after intravenous injection are indicative of a
positive reaction (133). The amount of R. mfolii 0403 LPS (on a per weight basis)
needed to elicit a response of equal magnitude to the enterics, was much greater than the
amount of Salmonella LPS necessary to induce the same response (134).

R. trifolii ANU843 possesses structural features radically different from that of
classical endotoxin molecules. The question asked was, "What do the radically different
structural features of this LPS molecule do to its ability to behave like the typical
endotoxins?”

LAL Activipy

The LAL assay is an endotoxin-induced coagulation reaction of the blood of the
horseshoe crab, Limulus polyphemus (135,136). It is a rapid method that is widely used
in the detection of endotoxin. We utilized a chromogenic assay in which R. trifolii
ANU843 was tested in parallel with E. coli 0111:B4, the reference endotoxin standard.
The same magnitude of response, per mg, was obtained for the LPS of both organisms,
even though R. trifolii LPS is lacking in phosphate, and has a different carbohydrate and

fatty acid composition. In fact, the LPS of strain ANU843 demonstrated a slightly higher

91

Table l

The mitogenic response of R. trifolii LPS, harvested at different growth periods, as

measured by 3H-thymidine uptake. Experiments were performed by D. M. Jacobs.
K90:R. Info/ii 0403 grown to early stationary phase.
K50=R. trifolir' 0403 grown to exponential phase.
K90 and K50 are both mitogenic, however K90 is more so than K50.

 

 

 

Table l
LPS
Cells dose, ug K90 K50 St 41 8142
C90

Serum-free medium

C3N/8t 0.05 1173 700 1033 1340
0.5 3756 2147 2260 2694
5.0 3168 1610 3500 3179

bkg - 833 epn

C3H/HeJ 0.05 127 115 129 148
0.5 297 151 567 287
5.0 511 140 438 609

bkg II300

5% F85

C$Hl8t 0.05 195 155 903 920
0.5 527 221 1803 2138
5.0 2243 542 4177 4306

bkg - 304

C3N/HeJ 0.05 68 51 35 45
0.5 104 52 184 70

' 5.0 239 53 162 235_
bkg a 350

St #1, St#2 = Salmonella LPS

C3H/St = LPS-responsive mouse strain
C3H/HeJ = LPS-hyporesponsive mouse strain

93

activity per mg compared to the standard E. coli 01 1 l:B4 LPS. Results are shown in
Figure 23.
Proliferation Assays

As stated previously, the LPS of R. trifolii 0403 was capable of inducing a
mitogenic response of equal intensity, as measured by 3H-thymidine uptake, to that of
Salmonella LPS (Table 1). Since it is known that R. trifolii 0403 and R. trifolii
ANU843 possess identical lipid A structures, and knowing the lipid A is responsible for
induction of biological response, we concluded that R. m’folii ANU843 LPS would also
be capable of causing B-cells to proliferate.

We attempted to show this with R. trifolii ANU843 LPS by using the transformed
cell line, BCL1clone5Blb. ATCC recommended BCLlclone5Blb as being useful for
studying the mechanisms and biochemistry of cell surface receptors and the mechanisms
underlying signalling of B-cells to differentiate and replicate, as the cell line is hyper—
responsive to LPS (137,138). Again, we were measuring uptake of 3H-thymidine as an
indicator of proliferation in response to LPS. Both R. trifolii ANU843 and S.
typhimurium LPS were tested in parallel.

Cell line BCLlcloneSBlb was not well suited for studying proliferation. Since it
is a transformed cell line, these cells divided too rapidly on their own, at a rate which was
independent of the presence or concentration of LPS. We tried modifying the
experimental conditions to accomodate for rapid proliferation as described in the
Materials and Methods section. None of these modifications helped to give an accurate
account of the effects of LPS upon B-cell proliferation with this cell line.

Looking back on the literature cited concerning BCL1 cells, the original leukemia
cells from which BCLlcloneSB l b was derived, these tumor cells were originally isolated
from the animals and used as a primary culture. Obviously, sometime during the
conversion from BCL1 to the final BCLlcloneSBIb, the cells have lost their ability to

respond to LPS supranorrnally jg vitro.

94

ABS‘OG .32.

024'

    
 

sundardtgggjorutaa 0.74 x 10'2 ng/mL
1.35 x 10‘? nq/Inl.
3.70 at 10': rig/tar.
7.40 x ro-2 rig/tar.

or g.tr1tolljm1843 3.80 x 10'? rig/ml.

 

   

 

0.074 rig/EU
o l :2 U .U‘ I .1. I .1. I k I ‘I
5.0-81ml.
Fi ure 23

Standard curve of absorbance versus concentration in endotoxin units (E.U.s)/mL for E.
coli 0111:B4 LPS. The point connected by the broken line corresponds to the average of
a triplicate ANUS43 LPS sample at 3.8xlo-2 ng/mL. The third point on the standard
curve is at a similar weight/mL concentration and has a similar value for endotoxin
units/mL. Absorbance is proportional to the amount of endotoxin present. These are the
results of the LAL assay.

95

Immunoglobulin Secretion

One of the beneficial effects of LPS is its ability to stimulate lymphocytes to
synthesize immunoglobulin. BCLlclone5B lb had previously been shown to synthesize
and secrete IgM in response to LPS (132). We looked at the ability of R. trifolii
ANU843 LPS and S. typhimurium LPS to stimulate these cells to secrete IgM. We tested
over a final concentration range of 8-80 pg/mL of LPS.

Originally the cells were left to incubate with LPS for 5 days before the medium
was harvested for I gM. It appears that the longer the cells are allowed to grow, the more
apt they are to secrete I gM. Negative control cells which did not receive LPS secreted
IgM at relatively high levels. The cells stimulated with Salmonella LPS gave higher
readings for IgM than did the negative control cells, however there was no difference in
IgM secretion levels between the negative control cells and those treated with rhizobial
LPS. The BCLlclone5B 1b cell line has IgM bound to its surface. The supematants were
not contaminated by fragments from the cell membranes, because the medium was
centrifuged and the supematants were removed for testing. Therefore, it appeared that
the cells themselves were secreting IgM without stimulus. We decided to harvest the
medium after 72, 48, and 24 hrs. Again the cells secreted IgM, but to a lesser extent as
the incubation time was reduced. At 72, 48, and 24 hrs. there was stimulation of the B-
cells to secrete IgM by Salmonella LPS. The longer the cells were allowed to incubate,
the lower the difference between background and stimulation. The cells treated with
LPS for 24 hrs. gave the greatest difference between background IgM and Salmonella
induced IgM secretion. R. trifolli LPS did not induce IgM secretion under any condition.
The results of LPS stimulation of B-cells, for 24hrs., are shown in Figure 24.

We also looked at the ability of both LPS types to stimulate release of IgG and
IgA by BCLlcloneSBlb. These cultures were allowed to incubate for 7 days at which

time the medium was harvested and assayed for the presence of lgG and IgA. Neither S.

Induction of IgM Secretion by LPS

 

 

 

 

 

2.5-
- Salmonella 7
E . W11"!
1.5-
E“
2
.99
o.s~
Oj'r'lvrrfi‘rwv'I‘l‘r
o rozoao,~4ososovoso9o
Lipopolysaccharidemg/mL]
Figure24

BCLlcloneSB 1 b lymphocytes were treated with LPS. Salmonella LPS was capable of
inducing I gM secretion whereas Rhizobium LPS was not.

97

typhimurium nor R. trifolii LPS were capable of inducing lgG or IgA secretion by this
cell line at the concentrations tested (data not shown).

It is interesting to note that proliferation of B-cells and stimulation to secrete
antibody are not necessarily coupled. Remember, both Salmonella and Rhizohium LPSs
were able to cause B-cell proliferation, however only the Salmonella LPS was able to
bring about IgM secretion.

TNF Production

R. trifoll'i ANU843 lipid A was sent to XOMA Corporation, Berkeley, CA, to test
its ability to cause macrophage to release TNF. R. trifolii was tested in parallel with the
lipid A of E. coli .15. The lipid A of R. trifoll'i ANU843 was able to induce TNF
production, However it did so at a concentration much greater than that of E. coli lipid A
(Figure 25). For example, 200 ng/mL of rhizobial lipid A was needed to give the same
response as that produced by less than 0.5 ng/mL of E. coli lipid A. We conclude that R.
trifolii can induce TNF production, however it is not very efficient in doing so, as
compared to the classical lipid A.
lnterleufin Production

A sample of R. trifolir' ANU843 lipid A was also tested for its ability to induce
macrophage to release IL-1 and IL-6. These assays were performed at XOMA Corp.
The results of these assays are forthcoming. Assays measuring the release of IL-8 from
macrophage in response to rhizobial and enteric lipid As were performed at XOMA Corp.
These results showed that macrophage responded to R. trifolii lipid A in the same
manner as they did in the TNF assay. Again, rhizobial lipid A was less effective in the
synthesis of IL-8 as compared to its enteric counterparts (Figure 26).

Pyrogenicity

Endotoxins are known pyrogens (139,140). Samples of R. trifolii ANU843 LPS

and S. typhl'urium LPS were sent to the Biological Test Center, Irvine, CA, to facilitate a

comparison of their abilities to act as pyrogens. S. typhimurium LPS, injected into

98

TNF Production

 

 

 

 

 

TNF Production (pg/mL)

0 I I w I ‘ l ‘ l

0.0 0.5 1.0 1.5 2.0 2.5
E. coli .15 Lipid A (ng/mL)

 

1500

 

 

 

1

‘ 1

450 500
R trifolii ANU843 Lipid A (ng/mL)

V

TNF Production (pg/mL)

' I l I
0 100 200 300

Figure 25

Comparison of TNF production induced by R. trifolii ANU843 lipid A and E. coli .15
lipid A.

IL-8 Production

 

3500 ‘
3000 " _ ..
3 j —9_ R. trtfolu
E —’— E. coli
B 2500 "
3 .
:
~2- 2000 -
0
§ 1
a; 1500 -
a; I
a 1000 -
500'
0 nut
.01 .1 l 10 100 1000 10000

Lipid A Concentration (ng/mL)

Figure 26
Comparison of lL—8 production induced by R. trifolii ANU843 lipid A and E. coli lipid
A.

100

rabbits at a concentration of 0.05 yg/kg body weight, induced fever in 3 out of 3 rabbits,
and was considered pyrogenic at this concentration. R. trifolii ANU843 was non-
pyrogenic when injected at the same dose that gave a positive response with Salmonella
LPS (Table 2). When the concentration of R. trifoll'i was raised to 1.0 rig/kg, it was
capable of fever production in 3 out of 3 rabbits. This was a dose proportional to the
amount of MPL or lipid X (on a weight basis) necessary to cause fever. Remember, both
MPL and lipid X are non-toxic. It was concluded that the LPS of R. trifolii was non-
pyrogenic.

It should be noted that the LAL assay is more frequently being used instead of the
jg yjyp test for pyrogens. Companies that market the kit claim it is as effective in
screening for endotoxins as is the pyrogen assay. Our i_n vi_vq pyrogen results showed
there was a difference between the classical LPS and the varient LPS in their abilities to
be endotoxic, whereas the results of the LAL assay demonstrated there was no difference
between the classical and varient LPSs. Obviously the two assays involve different
mechanisms of action.
101m

The Biological Test Center, Irvine, CA, tested for the toxic effects of both R.
trifolii ANU843 LPS and S. ryphimurium LPS in mice. The results are tabulated in
Table 3. Salmonella LPS was lethally toxic in 100% of the animals into which it was
injected, at the higher concentration (1000 p g), and 40% lethal at half this concentration.
The other 60% of mice receiving the lower concentration were adversely affected as
evidenced by lethargy and weight loss. R. trifolr'i LPS induced lethargy in all animals at
both concentrations tested, however this state was transitory and these mice showed no
significant change in body weight. We concluded R. trifolii LPS was non—toxic
compared to the enteric LPS.

LPS Bindino

 

In order for LPS to induce its biological effects it must first interact with the target

101

Table 2

USP XXII pyrogen assay comparing fever induction between R. trifolir‘ ANU843 LPS
and S. typhimurium LPS. A result is considered positive if one rabbit in three shows a

temperature 06°C or greater than its control temperature, or if the sum of the three rabbit
temperature rises exceeds l.4°C. In either case, a repeat test is done on five additional
rabbits. If not more than three of the total of eight rabbits have individual increases of
06°C or more, and if the sum of the eight increases does not exceed 3.7°C, the material
under examination meets the requirements for absence of pyrogens.

102

Table 2

—————p

USP XXII Pyrogen Assay

' ' LPS 0.05 /k
S. ryphrmurrum ( Fg g) Temperature = degrees C

Rabbit Wt. (kg) Control 1st. Hour 2nd. Hour 3rd. Hour Max. Rise

NZ-B 2.63 39.62 40.80 40.53 39.87 I . 18
428-V 3.26 39. 13 40.26 40.04 39.64 I. 13

; LPS 324-D 2.74 3930 39.3) 39.88 39.56 0.58

0“ a, Total Rise : 2.89 0c

arabblt

itional

585 01 R. trifolr'r' LPS (0.05 its/kg)

laterial
Rabbit Wt. (kg) Control 1“. Hour 2nd. Hour 3rd. Hour Max. R'me
I93-X 1.91 39.07 39.37 39.3 39.09 0.30
MY 2.59 39.04 40.17 39.63 39.11 . 1.13
222-2 2.50 38.73 38.86 38.78 38.65 0.13

TotalRIse: 156°C

Rabbit Wt.(|tg) Control [afloat Milan: 3rd.I-Ionr MaLRise

236-0 2.64 38.96 39.35 39.22 39.14 0.39
241-F 2.49 39.05 39.20 39.1 I 38.95 0.15
242-C 2.43 39.34 39.62 39.46 39.27 0.28
253-W 2.63 38.96 39.40 39.16 39.00 0.44
281-W 2.25 3932 39.82 39.62 39.43 0.50
Total Rise : 1.76 0C
+ 1.56 °C
_
3.32 °C

R. mfilii LPS (1.0 lag/kg)

Rabbit Wt.(kg) Conu'ol Inflow Mflonr “Hour MnRise
237

13D 39.42 40.15 40.13 40.05 0.73

4742 2.15 39.47 40.66 40.48 4036 1.19

190-A 2.24 39.07 39.85 39.59 39.52 0.78
Total Rise : 2.70 °C

 

103

Table 3

 

Lethal toxicity in mice comparing R. trifolii AN U843 LPS with S. typhimurium LPS.

Test Sample

RHDL—A

RHDL—B

RHDL—C

RHDL-D

d=lethargy

RHDL-A
RHDL-B
RHDL-C
RHDL-D

104

 

Animal # flejght b

19g
21 g
20g
22g
19g

UI-D‘UJN—

O

0.

O

O

O

0

O

21 g 0
20g 0
20g 0
19g 0
O

O

0

O

O

O

O

O

O

0

O

MADON—

17g

22g
20g
20g
18g
17g

MADON—

Ut-hWNF—
2
0a

ooooo ooooo ooooo ooooo a

T=death Observations are in hours.

S. typhimurium LPS lOOOyg/animal
S. typhimurium LPS 500pg/animal
R. trifolii LPS lOOOyg/animal
R. mfolii LPS SOOyg/animal

manna manna gamma manna 5
ooooo manna aaaqa qqq4% fi

m
(D
2
9?.
O
:3

(fl

OOOOO OOOOO 0.9.0.:

OOOOO OOOOO nan.

Weight

105

cell. The following experiments were done to make a comparative study of the binding
of rhizobial LPS and enteric LPS to eukaryotic cells. We wanted to determine if both
LPS types bound to a macrophage cell line, that had previously been shown to bind LPS
(59,89), to the same extent. We determined whether binding was specific, if one LPS
could inhibit binding of the other. We looked at the extent to which both LPS types
could interact with cell lines that did not originate from bone marrow. Lastly, we studied
the effects of molecules which we synthesized to block LPS activities, such as binding to
target cells. These molecules were termed endotoxin inhibitors (E13) and their effects
will be discussed last.

We have made a comparative study of the binding of rhizobial LPS and enteric
LPS to a macrophage-like cell line, RAW264.7. As stated in the Materials and Methods,
the LPSs of S. ryphimurium and R. trifolii were chemically linked to fluorescein isothio-
cyanate (FITC) and rhodamine B (Rho), so that binding could be visualized using
fluorescence microscopy. The lowest concentrations at which the LPS—dye could be
visually detected were: R. trifolii LPS-FITC, 5.87 mg/mL (Rt-FITC), Salmonella LPS-
Rho, 5.82 mg/mL (Sal-Rho), and Salmonella LPS-FITC, 5.47 m g/mL (Sal-FITC).
Concentrations at one tenth these working concentrations could no longer be detected
visually on the Leitz microscope.

Since RAW264.7 is an adherent cell line, the best results were obtained when the
cells were grown directly on coverslips. The RAW264.7 cells on coverslips were labeled
with the LPS—dye by placing the coverslips face-down onto 10 pL of the LPS-dye
solution (that had been cooled to 6°C) for 5 hrs., a time which had been determined to be
optimum for binding. The Sal-FITC and Sal—Rho could be visualized after 15 min., but
the R.t.-FlTC was faintly visible at 2 hrs., and best seen at 5 hrs., so we opted to do all
labeling for 5 hrs. to be consistent. Labeling was performed at 6° - 8°C. We envision
that LPS binds to eukaryotic cells in a two-stage process. The first stage is an

electrostatic interaction between the negatively charged LPS and the positive charge on

106

the cells’ surface (possibly a protein). This is followed in stage 2 by intercalation of the
lipid A long chain fatty acids into the plasma membrane. We thought that at low
temperatures we would detect only the electrostatic component. Low temperature should
also prevent uptake by endocytosis/phagocytosis, as this process is less effective at
temperatures below 18°C. After allowing contact between LPS-dye and the cell, the cells
were treated with a lOmM solution of sodium azide to prevent further active metabolic
processing resulting in dye uptake. The results of labeling were the same whether the
cells were treated with azide before adding LPS-dye conjugate or after labeling. Both
Sal-FITC and Sal-Rho were capable of crossing the plasma membrane and getting into
the interior of the cell. R.t.-FITC was detected only at the cell surface. Further proof of
this was provided by confocal microscopy. It was shown that cells labeled with Sal-dye
conjugate fluoresced at the cell surface, as well as in the cell's interior, whereas R.t.-F1TC
was found only at the cell surface, or just inside the plasma membrane. Results are
shown in Figures 27 and 28.

We know that the Salmonella LPS did not disrupt the plasma membrane, making
it leaky so that more Sal-dye conjugate gets inside the cells. RAW264.7 cells labeled
with free LPS for O, 3, and 7 hrs. were treated with trypan-blue dye and cell viability was
determined in different fields based upon the ability of cells to exclude dye. It was found
that these macrophage-like cells had 90% viability or greater in some instances.

From the initial binding studies we concluded that Salmonella and Rhizobium
LPSs interact with RAW264.7 cells differently. Salmonella LPS was capable of crossing
the plasma membrane and was found in the cells' interior. R. trifolii ANU843 LPS was
found to bind only at the cell surface, as evidenced by normal fluorescence and confocal
fluorescence microscopy.

Previous literature utilizing radiolabeled LPS (89,141) has shown that unlabeled
LPS can reduce the extent to which labeled LPS binds. This would be suggestive of the

presence of an LPS-receptor on the target cell. We have performed experiments utilizing

107

Figure 27( A)

Phase contrast and confocal microscopy of RAW264.7 cells labeled with Salmonella
LPS-Rhodamine conjugate.

Figure 27(B)

 

Phase contrast and confocal microscopy of CHO-K1 cells labeled with Salmomella LPS-
Rhodamine conjugate.

hm Esz
mum
<NN

109

Fioure 28

 

Phase contrast and confocal microscopy of RAW264.7 cells labeled with R. Info/ii LPS-
FITC conjugate.

110

F9 L189

 

Figure 28

lll

unlabeled Salmonella and Rhizobium LPSs in attempts to determine if one LPS can
prevent the binding of the other type. RAW264.7 cells preincubated with 10 yL of
unlabeled rhizobial LPS (LPS stock concentration was 5 and 20 mg/mL) for 5 hrs. After
this time the cells were placed in 10 yL of Sal-Rho for an additional 5 hrs. Fluorescent
microscopy showed that rhizobial LPS had no effect on binding of Sal-Rho. The cells
still bound Sal-Rho as they had before. In another instance, cells were preincubated with
unlabeled Salmonella LPS (5 and 20 mg/mL) for 5 hrs., followed by 5 hrs. incubation
with either R.t.—FlTC or Sal-Rho. In both cases and at either concentration the results
were the same. The unlabeled Salmonella LPS had no effect on binding of Sal-Rho or
R.t.—F1TC. The R.t.-FIT C was still found to bind the cells' surfaces. This suggests the
Salmonella LPS does not disrupt the plasma membrane since the R.t.-FITC stays at the
surface and does not move to the cells' interior. The Sal-Rho bound to cells after
pretreatment with unlabeled Salmonella LPS and the dye was again found throughout the
cells.

Binding studies using Sal—Rho and R.t.-FITC were carried out with two other cell
lines under the same conditions. CHO-Kl and BCL1clone5Blb were Chinese hamster
ovary and mouse B-cell leukemia lymphoma cell lines, respectively. As with the
RAW264.7 cell line, Sal-Rho moved into the interior of these cells, whereas R.t.—FITC
bound each cell type only at the surface (Figure 27). Experiments using unlabeled
Salmonella LPS followed by labeled Salmonella and Rhizohium LPSs had the same
results as those of the RAW264.7 cell line.

The confocal microscope was used to generate a 3-dimensional image of Sal-Rho,
specifically where it was localized inside the cell (Figure 29). Results indicated that the
dye was localized in vesicles which were distributed throughout the cell, but tended to
appear in higher frequency near the nucleus (in the perinuclear region). The dye was
definitely excluded from the nucleus. Location of dye was the same in both the

macrophage and Chinese hamster ovary cell lines. We did not look at the B-cell

112

Figure 29

3-dimensional imaging, shown in various orientations, of fluorescence localization in a
RAW264.7 cell. The Sal-Rho is localized in vesicles throughout the cell, but appears at
higher frequency in the perinuclear region.

113

 

 

Figure 29

114

leukemia cell line.

Considering that the Sal—Rho was capable of moving into the cells' interior, and
R.t.-FITC was not, we set out to determine if the Salmonella LPS was modified in any
way. Such a modification may have involved cleavage of the LPS into smaller units,
which would allow its entry into the cells' interior. We labeled the macrophage with Sal—
Rho as usual, then extracted the LPS-dye conjugate out of the cells. We had hoped to
resolve this LPS on a size exclusion column and then compare its elution profile with that
of intact Sal—Rho conjugate. Unfortunately the small amount of LPS—dye conjugate used
in labeling could not be detected in collected fractions.

Endotoxin Inhibitors (Els) in the LAL Assay

Researchers in our lab have synthesized several molecules which have a net
positive charge, hydrophobic moieties, or a combination of these two. The molecules are
referred to as endotoxin inhibitors (E15), and are denoted as El— 1 , E1-2, E1-3, and EI-4.
Their structures are given in Figure 30. The Els were designed to interact with LPS both
electrostatically and hydrophobically, and in doing so should block endotoxin activity.

The extent to which these molecules function to inhibit endotoxin activity was
measured by their ability to prevent a positive reaction in the LAL assay. EI-l, EI-2, El-
3, and E1-4 were tested at various concentrations in the presence of a fixed amount of E.
coli endotoxin. The results of Figure 31 shOw El-l and EI-2 to inhibit at concentrations
similar to that of polymyxin B. El-4 will inhibit at higher concentrations. El-3 does not
inhibit at all. In fact, its values suggest enhancement of LAL activity. This was
explained by the presence of yellow coloring in El-3 that could not be removed by
decolorization. The normal presence of color in E13 absorbed in the wavelength range
of 402-410 nm, where the LAL assay is measured. What we actually saw were false
positive values in the case of EI-3.

Effects of Els on RAW264.7 Cell Growth
A sample of El-l was sent to XOMA Corp. to test its ability to prevent TNF

Fioure 3O

 

The structures of the endotoxin inhibitors and polymyxin B sulfate are shown. The
number 7 used in El-3 and E1-4 indicates that 7 sugar residues are linked together in these
cyclodextrin structures.

116

   

£1 _ l R = alkyl E1 - 2
+-
+ NHz-(CHQTCH3
NH, CH, C112 OCH, 0
0
H0
H O OH
OH
7
7
El - 4
NH,
a - 3 . l
DAB-Phe-leu

R-DAB-thr—DAB-DAB

NH2 NH2 thr-DAB-DAB

NH; NH,
R = (rmethyloctanoic acid

POLYMYXIN B

Figure 30

 

 

117

Figure 31

The ability of the E15 to inhibit endotoxin activity was measured by their ability to
prevent a positive reaction in the LAL assay.

118

flimsy. n 99.9%.“ Bows.”

 

 

33.4335 coafiucoucoo a

 

J
r .. u u .‘p -

  

 

 

 

 

 

m-_m

28329; coacucoucoo .w

 

 

 

5.60m . u Bows." 8+m8d

. «.0

 

 

 

 

 

 

 

 

 

 

~-_w

96.0

.. 8.6

.. 83
.. 83

.. 8nd

 

«moo :2 memosqv

"'“809 u mwv

is.“

98835 8:350:00 a

gd

5180.“

0d

«5.3 m 63%..

and

 

856

Bad

8.6

..o

 

8 - § § §
6 d 0' °'
manor 1: mmv

g

g

wusov :2 WV

Absorbance at 408nm

g

0:100

E

119

El-4

 

Figure 31 (cont'd)

120
release by macr0phage. They reported that E1-1 killed the cells used in the assay. We

tested all Els and polymyxin B sulfate for their ability to cause destruction of cell line
RAW264.7. EI—l and El-2 caused cell lysis at concentrations of 9.5x10’2 mg/mL. This
was the same concentration previously used in an attempt to determine if these molecules
had bacteriocidal ability. (They did not whereas polymyxin B did.) At concentations of
one tenth and one hundredth of this value, El-l and El-2 did not cause lysis of the
RAW264.7 cells. EI-3 at 9.5x10'2 mg/mL did not affect macrophage cell growth.
Similarly, polymyxin B (PMB) and E14 at concentrations of 8.2x10‘2 mg/mL had no
effect on cell growth. It appeared that EL] and E1-2 were capable of making the cell
membranes of RAW264.7 very fluid, thus resulting in cell destruction.
The Effect of Els on Labeling with Sal-FITC and R.t.-F1TC When Cells are Grown in the
Presence of Els

RAW264.7 cells were treated with El—l, El-2, EI-3, and PMB at concentrations of
9.5x10-2 mg/mL for the E13 and 8.2x10’2 mg/mL for PMB. The cells remaining and the
cellular debris were labeled with Sal-FITC and R.t.-F1TC. Fluorescent microscopy
indicated the following. With EI—l and EI-2, most of the cells were destroyed, cellular
debris fluoresces, as do the remaining cells (Figure 32). There was no difference between
cellular labeling with Sal-FITC and R.t.-FITC. Both fluorescent labels were found in the
interior of the cells. For E1—3 and PMB, no cell lysis occurred. In these cases, Sal-FITC
moved across the plasma membrane, while R.t.—F1TC stayed on the cell surface.
Effect of Els on Labeling: Els Used in Binding Assays

We wanted to ascertain the ability of the E15 to prevent the binding of
fluorescently labeled LPS to cells from the cell line RAW264.7. The Els were mixed
with the labeled LPSs and left in the cold for 30 min. before having contact with the cells.
In the LAL assay system, the inhibitors were used at a concentration of 104 ng per assay,
compared to E. coli LPS which was used at a concentration of 10'3 ng per assay. Due to

this large difference in concentration between endotoxin and inhibitor, and the

121

concentration of LPS-dye conjugate that must be used in order to view it under the
microscope, we chose to use Els at concentrations of 10x, 0.1x, and the same
concentration, x, as endotoxin. What we observed can be described as follows.

EI-l: 10x El-l caused increased fluorescence in both the Salmonella and R.
rrifolii LPS treated cells (Figure 33). R. trifolii moved into the cell interior. At x El-I,
there was no change in degree of fluorescence for either LPS type. At 0.1x El—l there
was a small decrease in fluorescence for Rhizobium and no change for Salmonella .

EI-Z: 10x E1-2 caused increased fluorescence for both LPS types. EI-2 used at
the same concentration as LPS had no effect. 0.1x EI-2 appeared to decrease
fluorescence in both Salmonella and Rhizobium LPS-treated cells.

EI-3: At 10x, x, and 0.1x EI-3 appeared to decrease the degree of fluorescence for
both Salmonella and Rhizohium as compared to normally labeled cells. There appeared
to be no difference between the degree of fluorescence in all LPS solutions tested.

EI-4: E1-4 did not change the degree of fluorescence for Salmonella at any
concentration. 10x El-4 gave a slight increase in rhizobial fluorescence. E1—4 at the other
two concentrations for Rhizobium had no effect.

PNIB: At 10x and x PMB there was an increase in the degree of fluorescence for
Salmonella and at 0.1x it decreased. R. rrtfolii fluorescence decreased at x and 0.1x

concentrations. 10x showed no difference.

122

Figure 32(A)

 

RAW264.7 cells grown in the presence of El-l, followed by labeling with Sal-FITC.
Figure 32 (B)

RAW264.7 cells grown in the presence of EM, followed by labeling with R.t.-FlTC.
Notice the location of the R.t.-F'1TC conjugate.

123

32A

 

323

Figure 32

124

Figure 33(A)
RAW264.7 cells labeled with a mixture of Sal-FITC and 10x El-l.

Figure 331 B)
RAW264.7 cells labeled with a mixture of R.t.-F1TC and 10x El-l.

125

 

33B

 

Figure 33

126

Discussion and Conclusions

From the binding of the fluorescent LPS conjugates to various mammalian cells 1
summarize the following observations. The two LPSs, rhizobial and enteric, bind in
different fashions to all cells tested. The Salmonella LPS was capable of crossing the
plasma membranes of all cell lines studied and labeled the exterior surface as well as
being localized in discrete vesicles found throughout the cytosol (Figure 34). These
vesicles were localized heavily around the nucleus, and the dye was excluded from the
nucleus. Our results agree with those of another group which labeled RAW264.7 cells
with an FITC-LPS conjugate (S. minnesora ) (59). They also report FITC-LPS to be
taken up by each individual cell and be localized in the perinuclear regions of the cells.
When they focused through their cells they found fluorescence to be associated with
intracellular vacuoles and it also did not cover the nuclear regions. LPS from R. trifoll'i
behaved differently to the enteric LPS in all cell lines studied. The rhizobial LPS bound
to the surface of cells only. Perhaps possession of the 28 carbon, long chain fatty acid by
R. Irifolii LPS is responsible for preventing the LPS from crossing the plasma membrane
and getting into the interior of the cell.

The use of unlabeled LPS to prevent binding of the fluorescent conjugates
produced two interesting results. Firstly, the Salmonella LPS was not disrupting the
membrane by making it leaky so that more LPS gets in. This was based on the
observation that cells prelabeled with Salmonella LPS still label R. trifolii LPS only on
the cell surface. If Salmonella LPS disrupted the membrane, one would expect the
follow-up label with R.t.—FITC to find its way into the interior of the cell. Trypan blue
exclusion also showed the cells to be intact.

Secondly, unlabeled Salmonella and Rhizobium LPSs were unable to prevent
binding of either labeled conjugate. These results support those of Zimmerman (143),
who could not demonstrate saturation of binding on murine leukocytes. Morrison reports

of the "lack of a saturable limit" to the total amount of LPS which can be associated with

127

Fig ure 34

RAW264.7 cells labeled with Sal-Rho. Notice the vesicles of LPS—dye conjugate found
inside the cell.

 

Figure34

129

a cell. This statement was based upon results of different investigators (55).

These results, and the fact that we could observe LPS binding on different cell
types, point to the lack of a specific LPS receptor. It is highly unlikely that all three cell
types we tested (plus all cells LPS has been shown to bind in the literature) would possess
a receptor specific for LPS and that it would be identical in all cells. Our results suggest
that endotoxin will intercalate into the mammalian cell plasma membrane by hydrophobic
interaction in a variety of cell lines. Endotoxin is known to interact with phospholipid,
protein, glycoprotein, and lipoprotein, all of which are found on the surface of
mammalian cells, and all of which can be polar molecules containing net positive
charges. With LPS being a charged species, electrostatic interaction between LPS and
the surface of the cell is also important in order for LPS to get near the surface of the cell
so that lipid A can intercalate. Our results may very well be influenced by the
concentrations of LPS which we must use in order to visualize binding. We are working
at concentrations which are much higher than would be encountered under physiological
conditions. If an LPS receptor were present, it would indeed be saturated and the excess
labeling would occur by fusion. Again, we stress that it is unlikely that a receptor for
LPS would be found on many cell types and be identical. As was mentioned in the
literature review (73 ,74,77,78,81 ,82,83,84,88,89,90), many proteins/receptors have been
isolated that bind LPS, yet have different protein sequences and molecular weights. For
such a molecule to have so many different receptors on different cell types is rather far-
fetched. More likely it is the ability of these proteins to adsorb the LPS aggregate onto
the cells' surfaces, followed by the disaggregation of LPS monomers and fusion of the
lipid A moiety with the plasma membrane.

Binding to the Chinese hamster ovary cell line demonstrated that LPS has the
ability to bind many cell types which may not appear to be activated by LPS. This
suggests that binding alone is not sufficient to initiate activation, but is only the initial

step of the activation process. However, we cannot fairly state that binding to CHO-Kl

130

cells, or any other type of non-immunological cell results in zero activation. It is possible
that these cells are activated to proceed with functions that follow once these cells
become stimulated. They just are not a part of the endotoxin cascade. More importantly,
the results of R. mfolii ANU843 LPS binding and the biological activities we tested for
demonstrated that binding alone is not sufficient to initiate activation. We will discuss
this in more detail later.

What can we conclude about the endotoxin inhibitors and how does this correlate
with the mechanism of endotoxin action? The endotoxin literature cites LPS binding as
the initial step towards endotoxin activation of cells. Binding is followed by a signal
which causes the cell to become activated. There have been suggestions that endotoxin
affects the fluidity of target cells and it is this membrane perturbation that activates the
cell. Studies have shown that lipid A can increase the order of the lipid bilayer
(59,65,67,68).

Our results involving endotoxin inhibitors suggest that some of them act by
increasing membrane fluidity. EM and E1-2 act by making the cell membranes more
fluid. In other words, they are disruptive to cell membranes. This is supported by the
following observations concerning EL] and EI-2. EI-l was sent to XOMA Corp. to
determine how well it prevented TNF release by macrophage. During this study it was
determined that El-l lysed the cells. These experiments were performed at a con-
centration of EM similar to that used in the LAL assay. Both El-l and E1-2 killed cell
line RAW264.7 cells when the cells were grown in their presence at concentrations equal
to those used in the LAL assay. EI-3, E14, and polymyxin B had no effect at similar
concentrations. Cells labeled in the presence of El-l and E1-2 had increased dye uptake
at 10x their concentrations. Normal concentrations of EI-l and El-2 had little effect on
prevention of binding. At one tenth their concentrations there was decreased binding.
This suggests there is some interaction between LPS and the E15 that makes less LPS

available for binding, yet at higher concentrations the increase in membrane fluidity

131

overrides any prevention of binding based on interaction between the inhibitor and LPS.
E1-3 lacks a hydrophobic moiety and appears to be the best suited for binding prevention.
There was interaction between LPS and El-3 which prevented LPS interaction with the
cell membrane. Since El-3 lacks hydrophobicity, one does not see the increase in
fluorescence seen with the other inhibitors. E1-3 does not inhibit LAL, yet its inherent
color is responsible for making it test falsely positive.

The results of the ML assay involving the E15 shed some light onto the mode of
endotoxin action. It is known that membranes must be present within the lysate in order
for the LAL assay to function correctly. If membranes are lacking from the lysate, the
positive results cannot be had. 1f El-l and EI-2 function by increasing membrane
fluidity, then they will be capable of disturbing the integrity of the membrane component
of the lysate. If endotoxins function by making the membrane more rigid, then activity
will not be detected if the membrane was fluid due to El effects.

The critical test involved EI-4. E14 is similar to EI-3 in structure, except for the
octyl chains on E14. At low concentrations El-4 does not inhibit, but if El-4
concentration was raised to a higher value (105 to 10°), then EI-4 was capable of
inhibition. We believe the inhibitors act by making the lysate membrane so fluid that the
endotoxin can no longer increase the packing order of the bilayer and hence no activation
of the cascade occurs.

The central dogma of the endotoxin literature is the structural requirement
necessary for endotoxic activity. Great importance has been placed on the diphos-
phorylated, glucosamine disaccharide with its fatty acid chains of specific length. Many
studies have been devoted to correlating a specific biological activity with a specific
structure, yet no one has been able to successfully accomplish this.

R. trifoliz' AN11843 has now been shown to possess a lipid A structure which
deviates from the classical endotoxins. 1t possesses a lipid A composed of sugars other

than glucosamine, lacks phosphate, and contains unusual fatty acids, yet it is still capable

132

of inducing some of the classical endotoxin effects. Obviously the classical endotoxin
structure is not necessary. A structure which has a similar 3-dimensional geometry and
charge distribution which can bind a membrane and induce the same effect on that
membrane, such as decreasing membrane fluidity, will show the same net result.

Our results show that R. Irifolii LPS is not toxic compared to that of the enterics.
It was able to induce fever and a local Schwartzmann reaction at concentrations much
higher than the classical LPS. It will cause mice to become transiently lethargic, but it
will not kill them as Salmonella LPS does. We believe these differences in endotoxic
response to be attributed to the fact that R. trifolii LPS does not induce cytokine
production. It has been suggested that TNF and IL-1 are the species responsible for the
toxic effects of endotoxins (144,145). Since R. trlfolii LPS does not readily induce TNF
and IL-1, it therefore does not cause the detrimental effects of endotoxin.

Why does R. trifolii LPS induce some responses and not others? Could it be that
different cell types respond to LPS in different manners and therefore one cannot make
comparisons between cell types? This is an example where binding alone is not
sufficient to induce a cellular response. We believe that the lipid A must intercalate into
the plasma membrane and this results in decreased membrane fluidity. For a mitogenic
response to occur, maybe the LPS does not have to insert into the membrane as it does for
protein synthesis to occur, and this is why we see proliferation, but not synthesis of IgM,
TNF, or interleukins, which are ultimately responsible for the toxic effects of endotoxin
such as fever response and lethal toxicity.

Our proposed mechanism for endotoxic activity is described as follows. In the
1960's, Shands proposed a model for LPS insertion into the lipid bilayer which involved
end attachment of the LPS aggregate. This resulted in disaggregation which allowed
individual LPS monomers to insert into the bilayer (Figure 35). We propose that a
mechanism similar to the Shands model is indeed what occurs. The negative charge of

the LPS aggregate is attracted to the cell surface by net positive charges present on

Figure 35

 

Attachment of an LPS aggregate to a lipid bilayer, adapted from Shands. (Adapted from
"Bacterial Toxins and Cell Membranes" 1978, Academic Press.)
35( A) Disaggregation of an LPS bolus, followed by insertion of LPS monomers into the
target cell bilayer.
353) Disaggregation of the R. mfolii LPS bolus, followed by monomer insertion.
Notice the position of the 27-hydroxyoctacosanoic acid.

35A

gg'gg fiéfifig 1.... e... h... Mm...

135

species such as protein. This first step is independent of temperature and can be
reversible. The second step involves disruption of the LPS aggregate so that LPS
monomers can insert into the lipid bilayer. This second step is irreversible and temper—
ature dependent. Now that the lipid A is in the membrane, a membrane perturbation
occurs which results in an increase of the microviscosity of the bilayer and thus causes an
increase in the order of the lipid bilayer (59,65,67,68). In other words, that area of the
membrane becomes less fluid, more rigid. This is the signal which is triggered causing
the cell to become activated.

Other information supporting this model comes from the lipid A structure of the
non-toxic, devoid of biological activity, Rhodopseudomonas sphaeroides ATCC 17023
LPS (45). This molecule possesses a glucosamine disaccharide backbone, contains phos-
phate at positions Cl and C4', and possesses amide and ester linked fatty acids with
lengths of 10-14 carbon atoms. This molecule possesses all of the classical endotoxin
”hardware”, but it is totally devoid of biological activity. If you examine its structure in
Figure 36, you will notice it has 2 unusual fatty acids : 3-oxotetradecanoic acid and a
7—tetradecenoic acid (50).

R. sphaeroicles ATCC 17023 is known to bind to eukaryotic cells (51), but it
does not stimulate them. (Here is another example where binding alone is not sufficient
to induce biological activity.) R. sphaeroides possesses fatty acids containing double
bonds. Degrees of unsaturation are known to make membranes more fluid. We propose
that when the R. sphaeroides lipid A fuses with the plasma membrane, its unsaturated
fatty acids make that area of the membrane more fluid and thus the signal for activation
does not occur.

Our labeling studies show that a longer time must pass before R. Irl'folii LPS can
be detected visually on the surface of target cells. Salmonella LPS binding can be
visualized at 15 min., whereas R. trifolii appears at 2—3 hrs. We know that R. trifolz'l' has

a C28 fatty acid which has been shown to traverse a cell membrane by results from

136

OH

HO\ 0

043-0 0 +10%

/ 0 [OH

HO O NH NH -P\'°O
:0 =0 '0 '0 OH
OH O OH =0

=0
1R.) (R31

(R4)

(R2)

R. sphaeroides Lipid A

 

Figure 36
The lipid A of R. sphaeroides ATCC 17023, adapted from Qureshi, e_t. _a_l_. (50).

137

electron microscopy of freeze fractured membranes. It would be more difficult for an R.
trifolii LPS molecule to pull out of its LPS aggregate and insert into the target cell
membrane, for the C28 fatty acid must be ripped out of the bolus and then inserted all the
way through the plasma membrane (Figure 358). A Salmonella lipid A, lacking any
transmembrane fatty acids, could more easily move from LPS aggregate to insertion into
the plasma membrane bilayer. This would explain why the R. trifolii LPS labels more
slowly. From our chemical analyses we know that the R. trzfolii LPS forms very stable
aggregates which are difficult to disperse. This idea also points to the difficulty of release
of the LPS from the aggregate followed by membrane insertion.

Results for TNF and interleukin production indicate that a higher concentration of
rhizobial LPS is needed to induce secretion. Since rhizobial LPS does not labelas readily
as enteric LPS , it makes sense that by increasing the concentration of rhizobial LPS you
increase the chance that more of it will bind and thus cause the biological effects.

Once the LPS has bound, it is possible that the Salmonella LPS causes a
membrane fluidity change that is different from that of the change caused by R. trifolii
LPS binding. Possibly the rhizobial LPS does not change membrane fluidity to the extent
that enteric LPS does. This would explain the lack of subsequent biological response.

From our labeling studies we see that Salmonella LPS is endocytosed and finds
its way into the cell interior. Is this important for activation? The concentration of
rhizobial LPS necessary to stimulate TNF and interleukin synthesis is much less than the
concentration used in labeling. At the higher concentration rhizobial LPS is still present
on the surface of the membrane and thus it appears that the ability to transverse the
membrane is not important. The exception to this would be if a small amount of R.
triflflii LPS were able to cross the membrane, but at such a small concentration that its
fluorescence could not be detected. This scenario would not fit the membrane fluidity
model.

Our data does not allow us to define the intracellular processes which are directly

138

triggered by the attachment of endotoxin to the target cell. This ultimately stimulates the
various metabolic events, such as Q m synthesis and secretion of cytokines, leading to
the condition known as an endotoxic crisis. We hope that our studies shed some light
onto the mechanism of endotoxin action.

There are aspects of the project that need to be further investigated. We were able
to isolate and characterize a minor component of the lipid A of R. trifolii ANU843 LPS.
As stated earlier, TLC separation of the rhizobial crude lipid A yields several
components. The other components need to be isolated and characterized. Most likely a
disaccharide will be found. 1 suggest that the methods used to fractionate the minor
component be used again, with the exception of using larger amounts of LPS. It is quite
easy to grow cells on a large scale and then isolate the LPS. HPLC can be used to
separate the components as it was used before. This would again be followed by further
fractionation using preparatory TLC if necessary. We had great difficulty in chemically
manipulating R. trifolii LPS. Researchers in our lab have since developed a method to
deacylate both ester— and amide-linked fatty acids. This method utilizes NaOH and
thiophenol and has been successful in removing all acyl groups from the 2011 strain of
Rhizohium meliloti . R. meliloti also contains the 27-hydroxyoctacosanoic acid, which
you will remember, was difficult to cleave from R. trifolii LPS. This method of
deacylation should work on R. mfolii LPS and should definitely be tried.

Once the lipid A backbone has been completely freed of acyl groups, then the
sugar can be subjected to linkage analysis by using methylation procedures followed by
acetylation. Further involved chemistry should be performed, however these ideas are a
starting point.

As far as further work on the biological aspects are concerned, several things can
be done. I believe that the biological assays should also be performed with other species
of Rhizohium such as R. melilon' , which is also known to possess the 27-OH, C28 fatty

acid and uses sulfate as its charged species instead of phosphate, or carboxylate in the

139

case of R. Irifolii LPS. It would be interesting to determine if R. meliloti LPS behaves
the same way in the binding assays as R. trifolii LPS. 1 would suspect that it does
knowing about its long chain fatty acid and the fact that preliminary work shows it to
have reduced activity like R. trifolii LPS in the TNF assay.

The question concerning decrease in membrane fluidity should be addressed. We
should determine whether R. trifolii LPS has the same effect as Salmonella LPS in
changing the membrane viscosity of artificial membranes composed of phospholipid. A
probe such as the one used in the Jacobs' experiment (68) could be used to measure
changes in membrane fluidity. It would also be interesting to get R. sphaeroides LPS
and test this one in parallel with the other LPSs.

It would be interesting to determine if all "usual" lipid As get incorporated into
vesicles and taken into the cell as measured by fluorescence microscopy. What about an
LPS like R. sphaeroides which has a lipid A similar to the classical lipid A, but lacks its
activity? Further work should be done to determine if LPS uptake is important. Must it
get into the cell andjust how does it get in? Maybe LPS induces the cell to phagocytize
it. Studies have shown that Shigella is capable of doing this by mobilizing actin
filaments (146), and Shigella, being a gram-negative organism, has LPS at its surface.
Could treating cells with substances that prevent actin from polymerizing, such as
cytochalasins, prevent LPS uptake, while still keeping LPS bound at the cell surface?

One could also use radiolabeled LPS more effectively than the fluorescent
conjugates. Using a 3H-labeled Salmonella LPS, one could label the cells and then try to
extract the LPS out again. As we tried with Sal-Rho, one could separate the LPS based
on size and compare this with 3H-Salmonella LPS that was not used to label. Hopefully
this would give information regarding whether or not the LPS needed to be cleaved in
order to get inside of the cell.

Since R. meliloti LPS contains sulfate, it can be 35S-labeled and used in binding

assays at concentrations which are physiological instead of at the higher concentrations

140

needed to be visualized by the fluorescence microscope. One could perform competition
binding assays at more reasonable concentrations to determine if binding is really
saturable.

As we said before, much work has gone into studying the structures of
endotoxins and their mechanisms of action since Pfeiffer first discovered them. Much

effort will go into studying these unique molecules in the future.

List of References

@991”?

ll.

12.

13.

14.

16.

17.

18.
19.
20.

Ezzell, C. (1992) Science News 142: 97-112.

Marvin, H., ter Beest, M., Witholt, B. (1989) J. Bacteriol. 171: 5262-5267.

Kabir, S., Rosenstreich, D., Mergenhagen, S. (1978) in "Bacterial Toxins and Cell
Membranes", (Jeljaszewicz, J. and Wadstrom, T., eds.) p. 59-87, Academic Press,
London, New York, San Francisco.

Burrell, R. (1990) Crit. Rev. Micro. 17: 189-208.

Bradley, S. (1979) Ann. Rev. Microbiol. 33: 67-94.

Morrison, D., Ryan, J. (1987) Ann. Rev. Med. 38: 417-432.

Gery, 1., Kruger, J., Spiesel, S. (1972) J. Immunol. 108: 1088-1091.

Andersson, J ., Melchers, F., Galanos, C., Luderitz, O. (1973) J. Exp. Med. 137:
943-953.

McKearn, J., Paslay, J., Slack, J., Baum, C., Davie, J. (1982) Immunological Rev.
64: 5-23.

Beutler, B., Krochin, N., Milsark, D., Leudke, C., Cerami, A. (1986) Science 232:
977-979.

Michie, H., Manogue, K., Spriggs, D., Revhaug, A., O'Dwyer, S., Dinare, C.,
Cerami, A., Wolff, S., Wilmore, D. (1988) N. Eng. Jour. Med. 318: 1481-1486.

Walker, R., Casey, L. (1985) in "Handbook of Endotoxin”, Vol. 3, (Berry, L. J.
ed.) p. 225-238, Elsevier Science Publishers B. V. , Amsterdam.

Galanos, C., Freudenberg, M., Reutter, W. (1979) Proc. Natl. Acad. Sci. USA 76:
5939-5943.

Leive, L., Shovlin, V., Mergenhagen, S. (1968) J. Biol. Chem. 243: 6384-6391.
Westphal, O., Luderitz, O., Bister, F. (1952) Z. Naturforsch 7b: 148.

Donaldson, D., Roberts, R., Larsen, H., Tew, J. (1974) Infect. Immun. 10: 657-
666.

Patterson-Delafield, J ., Martinez, R., Lehrer, R. (1980) Infect. Immun. 30: 180-
192. ,

Nikaido, H., Vaara, M. (1985) Micro. Rev. 49: 1-32.
Nikaido, H., Kakae, T. (1979) Adv. Microb. Physiol. 20: 163-250.

Vaara, M., Nikaido, H. (1984) in "Handbook of Endotoxin", Vol. 1, (Rietschel, E.
Th., ed.) p. 1-45, Elsevier Science Publishers B. V., Amsterdam.

141

21.

22.

23 .
24.

25.

26.
27.

29.

30.

31.
32.

33.

34.

35.
36.

142

Nghiem, H-O., Staub, A., Galanos, C., Luderitz, O. (1982) Eur. J. Biochem. 125:
431-436.

Rietschel, E. Th., Mayer, H., Wollenweber, H., Zahringer, U., Luderitz, O.,
Westphal, O., Brade, H. (1984) in " Bacterial Endotoxin - Chemical, Biological,
and Clinical Aspects", (Homma, J., Kanegaski, S., Luderitz, O., Shiba, T. and
Westphal, O., eds.) p. 11-22, Verlog Chemie.

Lindberg, A. (1973) Ann. Rev. Microbiol. 27: 205-237.

Rietschel, E. Th., Brade, H., Brade, L., Kaca, W., Kawahara, K., Lindner, B.,
Luderitz, O., Tomita, T., Schade, U., Zahringer, U. in "Bacterial Endotoxins:
Structure, Biomedical Significance, and Detection with the Limulus Amebocyte
Lysate Test", p. 31-50, Alan R. Liss, Inc.

Munford, R. (1988) in "Bacterial Host - Cell Interaction", p. 123-140, Alan R.
Liss, Inc.

Westphal, O., Luderitz, O. (1954) Angew. Chem. 66: 4074-117.

Johnson, A. (1985) in "Handbook of Endotoxin", Vol. 3, (Berry, L. J., ed.) p. 216-
221, Elsevier Science Publishers B. V., Amsterdam.

Rietschel, E. Th., Brade, H., Brade, L., Bradenberg, K., Schade, U., Seydel, U.,
Zahringer, H., Galanos, C., Luderitz, O., Westphal, O., Labischinski, H.,
Kusumoto, S., Shiba, T. (1987) Prog. Clin. Biol. Res. 231: 25-53.

Rietschel, E. Th., Wollenweber, H., Russa, R., Brade, H., Zahringer, U. (1984)
Rev. Infect. Dis. 6: 432-438.

lmoto, M., Kusumoto, S., Shiba, T., Naoiki, H., Iwashita, T., Rietschel, E. Th.,
Wollenweber, H., Galanos, C., Luderitz, O. (1983) Tetrahedron Lett. 24: 4017-
4020.

Gmeiner, J., Luderitz, O., Westphal, O. (1969) Eur. J. Biochem. 7: 370-379.

Rietschel, E. Th., Wollenweber, H., Brade, H., Zahringer, H., Lindner, B., Seydel,
U., Bradeczek, H., Bamickel, G., Labischinski, H., Giebsbrecht, P. (1984) in
"Handbook of Endotoxin", Vol. 1, (Rietschel, E. Th., ed.) p. 187-200, Elsevier
Science Publishers B. V., Amsterdam.

Galanos, C., Lehmann, V., Luderitz, O., Rietschel, E. Th., Westphal, O., Brade,
H., Freudenberg, M., Hansen-Hagge, T., Luderitz, T., McKenzie, G., Ulrich, S.,
Strittmatter, W., Tanamoto, K., Zahringer, U., lmoto, M., Yoshimura, H.,
Tanamoto, K., Yasuda, T., Kusumoto, S., Shiba, T. (1984) Eur. J. Biochem. 140:
221-227.

Homma, J., Matsuura, M., Kanegaski, S., Kawakubo, Y., Kojima, Y., Shibakawa,
N., Kumazawa, Y., Yamamoto, A., Tanamoto, K., Yusuda, T., lmoto, M.,
Yoshimura, H., Kusumoto, S., Shiba, T. (1985) J. Biochem. 98: 395-406.
Munford, R., Hall, C. (1986) Science 234: 203-205.

Myers, K., Truchot, A., Ward, J., Hudson, Y., Ulrich, J. (1990) in "Cellular and

37.

38.

39.

41.

42.

43.

45.

53.

143

Molecular Aspects of Endotoxin Reactions", Vol. 1, (Nowotny, A., Spitzer, J .,
and Ziegler, E., eds.) p. 145—156, Elsevier Science Publishers B. V., Amsterdam.

Ribi, E., Cantrell, J., Feldner, T., Myers, K., Peterson, J. (1986) in "Microbiology"
(Levie, L., Bonventre, P., Morello, J., Silver, J., and Wu, H., eds.) p. 9-13, Am.
Soc. Micro., Washington D. C.

Dziarski, R. (1991) J. Biol. Chem. 266: 4719-4725.

Kumazawa, Y., Nakatsuka, M., Takimoto, H., Furuya, T., Nagumo, T.,
Yamamoto, A., Homma, J., Inada, K., Yoshida, M., Kiso, M., Hasegawa, A.
(1988) Infect. Immun. 56: 149-155.

Takada, H., Kotani, S., Tsujimoto, M., Ogawa, T., Takahashi, 1., Harada, K.,
Katsukawa, C., Tanaka, S., Shiba, T., Kusumoto, S., lmoto, M., Yoshimura, H.,
Shimamoto, T. (1985) Infect. Immun. 48: 219-227.

Kotani, S., Takada, H., Tsujimoto, M., Ogawa, T., Takahashi, I ., Ikeda, T.,
Otsuka, K., Shimauchi, H., Kasai, N., Mashimo, J ., Nagao, S., Tanaka, A.,
Harada, K., Nagaki, K., Kitamura, H., Shiba, T., Kusumoto, S., lmoto, M.,
Yoshimura, H. (1985) Infect. Immun. 49: 225—237.

Mayer, H., Weckesser, J. (1984) in "Handbook of Endotoxin", Vol. 1,
(Rietschel, E. Th., ed.) p. 221—244, Elsevier Science Publishers B. V., Amsterdam.

Mayer, H., Salimath, P., Holst, O., Weckesser, J. (1984) Rev. Infect. Dis. 6: 542—
545.

Weckesser, J ., Drews, G., Mayer, H. (1979) Annu. Rev. Microbiol. 33: 215-239.

Strittmatter, W., Weckesser, J ., Salimath, P., Galanos, C. (1983) J. Bacteriol. 155:
153-158.

Krauss, J., Seydel, U., Weckesser, J., Mayer, H. (1989) Eur. J. Biochem. 180:
519-526.

Galanos, C., Roppel, J ., Weckesser, J ., Rietschel, E. Th., Mayer, H. (1977) Infect.
Immun. 16: 407-412.

Tharanathan, R., Weckesser, J ., Mayer, H. (1978) Eur. J. Biochem. 84: 385-394.

Takayama, K., Qureshi, N... Beutler, B., Kirkland, T. (1989) Infect. Immun. 57:
1336-1338.

Qureshi, N., Takayama, K., Kurtz, R. (1991) Infect. Immun. 59: 441-444.
Kirkland, T., Qureshi, N., Takayama, K. (1991) Infect. Immun. 59: 131-136.
Hollingsworth, R., Lill-Elghanian, D. (1990) in "Cellular and Molecular Aspects
of Endotoxin Reactions", Vol. 1, (Nowotny, A., Spitzer, J ., and Ziegler, E., eds.),
p. 73-84, Elsevier Science Publishers B. V., Amsterdam.

Chaby, R., Metezeau, P., Girard, R. (1984) Cell. Immunol. 85: 531-541.

55.

61.
62.

69.
70.
71 .

72.
73.
74.
75.

76.

144

Sultzer, B. (1976) Infect. Immun. 13: 1579-1584.

Morrison, D., Rudbach, J. (1981) in "Contemp. Topics in Molecular Immun-
ology", Vol. 8, (Mandy, W., Inman, F., eds.), p. 187—218, Plenum Publishing Co.,
New York.

Neter, E., Westphal, O., Luderitz, O., Gorzynski, E., Eichenberger, E. (1956)
J. Immunol. 76: 377-385.

Hammerling, V., Westphal, O. (1967) Eur. J. Biochem. 1: 46-50.

Springer, G., Adye, J. (1975) Infect. Immun. 12: 978—986.

Dijkstra, J., Larrick, J., Ryan, J., Szoka, F. (1988) J. Leuk. Biol. 43: 436-444.
Pohlman, T., Munford, R., Harlan, J. (1987) J. Exp. Med. 165: 1393-1402.
Carr, C., Morrison, D. (1984) Infect. Immun. 43: 600-606.

Carr, C., Morrison, D. (1984) Rev. Infect. Dis. 6: 497-500.

Carr, C., Morrison, D. (1985) Infect. Immun. 49: 841-89.

Singer, S., Nicolson, G. (1972) Science 175: 720-731.

Larsen, N., Enelow, R., Simons, E., Sullivan, R. (1985) Biochimica et Biophysica
Acta 815: 1-8.

Jacobs, D. (1984) Rev. Infect. Dis. 6: 501-505.

Jacobs, D., Price, R. (1984) in "Recent Advances in Mucosal Immunology",
Vol. 216A, (Mestecky, J., McGhee, J ., Bienenstock, J., Ogra, P., eds.) p. 691-699.

Price, R., Jacobs, D. (1986) Biochimica et Biophysica Acta 859: 26-32.
Shands, J., Graham, J., Nath, K. (1967) J. Mol. Biol. 25: 15—21.
Shands, J. (1965) J. Bacteriol. 90: 266-270.

Haeffner-Cavaillon, N., Caroff, M., Cavaillon, J-M. (1985) Mol. Immunol. 26:
485-494.

Lebbar, S., Cavaillon, J-M., Caroff, M. (1986) Eur. J. Immun. 16: 87-91.
Tobias, P., Soldau, K., Ulevitch, R. (1986) J. Exp. Med. 164: 777-793.
Tobias, P., Soldau, K., Ulevitch, R. (1989) J. Bio. Chem. 264: 10867-10871.

Wright, S., Ramos, R. Tobias, P., Ulevitch, R., Mathison, J. (1990) Science 249:
1431-1433.

Schumann, R., Leong, S., Flaggs, G., Gray, P., Wright, S., Mathison, J., Tobias,
P., Ulevitch, R. (1990) Science 249: 1429-1431.

77.

78.

79.

81.

82.

85.

89.

91.

93.

95.

8

145

Weiss, J., Elsbach, P., Olsson, 1., Odegerg, H. (1978) J. Biol. Chem. 253: 2664—
2672.

Weiss, 1., Muello, K., Victor, M., Elsbach, P. (1984) J. Immunol. 132: 3109-3115.

Marra, M., Wilde, C., Griffith, J., Snable, J., Scott, R. (1990) J. Immunol. 144:
662-666.

Tobias, P., Mathison, J., Ulevitch, R. (1988) J. Biol. Chem. 263: 13479-13481.

Springer, G., Adye, J., Bezkorovainy, A., Jirgensons, B. (1974) Biochemistry 13:
1379-1389.

Springer, G., Adye, J. (1975) Infect. Immun. 12: 978-986.

Lei, M-G., Morrison, D. (1988) J. Immunol. 141: 996—1005.

Lei, M-G., Morrison, D. (1989) J. Immunol. 141: 1006-1011.

Morrison, D. (1989) Microbial Pathogenesis 7: 389-398.

Morrison, D. (1990) in "Cellular and Molecular Aspects of Endotoxin Reactions",
Vol. 1, (Nowotny, A., Spitzer, J., and Ziegler, E., eds.) p. 183-189, Elsevier
Science Publishers B. V., Amsterdam.

Roeder, D., Lei, M-G., Morrison, D. (1989) Infect. Immun. 57: 1054—1058.

Zoeller, R., Wightman, R, Anderson, A., Raetz, C. (1987) J. Biol. Chem. 262:
17212-17220.

Hampton, R., Golenbock, D., Raetz, C. (1988) J. Biol. Chem. 263: 14802-14807.

Kirkland, T., Virca, G., Kuus-Reichel, T., Multer, F., Kim, S., Ulevitch, R.,
Tobias, P. (1990) J. Biol. Chem. 265: 9520-9525.

Prpic, V., Weiel, J., Somers, S. (1987) J. Immunol. 139: 526—533.
Dziarski, R. (1989) Eur. J. Immunol. 19: 125-130.

Michalek, S., Moore, R., McGhee, J., Rosenstreich, D., Mergenhagen, S. (1980)
J. Infect. Dis. 141: 55—63.

Vogel, S., Henricson, B. (1990) in "Cellular and Molecular Aspects of Endotoxin
Reactions", Vol. 1, (Nowotny, A., Spitzer, J., and Ziegler, E., eds.), p. 465-473,
Elsevier Science Publishers B. V., Amsterdam.

Vogel, 8., Hanson, C., Rosenstreich, D. (1979) J. Immunol. 122: 619-622.
Nathan, C. (1987) J. Clin. Invest. 79: 319—326.

Beutler, B., Milsark, 1., Cerami, A. (1985) Science 229: 869-871.

Bone, R. (1991) in "Infectious Disease Clinics of North America", Vol. 5,
(Young, L. and Glauser, M., eds.), p. 793-805, W. B. Saunders Co., Philadelphia.

100.
101.

102.

103.
104.

105.
106.
107.
108.
109.
110.
111.

112.
113.

114.
115.

116.

117.

118.

146

Nowotny, A., Nakano, M. (1990) in "Cellular and Molecular Aspects of
Endotoxin Reactions", Vol. 1, (Nowotny, A., Spitzer, J ., and Ziegler, E., eds.)
p. 171-182, Elsevier Science Publishers B. V., Amsterdam.

Shinomiya, H., Nakano, M. (1987) J. Immunol. 139: 27302736.

Terada, Y., Shinomiya, H., Nakano, M. (1989) Biochem. Biophys. Res. Comm.
158: 723-729.

Nakano, M., Shinomiya, H. (1990) in "Cellular and Molecular Aspects of Endo-
toxin Reactions", Vol.1, (Nowotny, A., Spitzer, J., and Ziegler, E., eds.) p. 205-
214, Elsevier Science Publishers B. V., Amsterdam.

Ding, A., Porteu, F., Sanchez, 13., Nathan, C. (1990) Science 248: 370-372.
Cavaillon, J-M., Munoz, C., Fitting, C., Couturier, C., Haeffner-Cavaillon, N.
(1990) in "Cellular and Molecular Aspects of Endotoxin Reactions", Vol. 1,
(Nowotny, A., Spitzer, J., and Ziegler, E., eds.), p. 257-267, Elsevier Science
Publishers B. V., Amsterdam.

ROppel, J., Mayer, H., Weckesser, J. (1975) Carbohydr. Res. 40: 31-40.
Hrabak, E., Urbano, M., Dazzo, F. (1981) J. Bacteriol. 148: 697-71 1.

Personal communication with Diane M. Jacobs.

Hollingsworth, R., Dazzo, F. (1988) Anal. Microbiol. 7: 295—302.
Hollingsworth, R., Carlson, R. (1989) J. Biol. Chem. 264: 9300-9304.
Humphrey, 8., Vincent, J. (1969) J. Gen. Microbiol. 59: 411-425.

Zevenhuizen, L., Scholten-Koerselman, I., Posthumus, M. (1980) Arch.
Microbiol. 125: 1-8.

Russa, R., Luderitz, O., Rietschel, E. Th. (1985) Arch. Microbiol. 141: 284-289.

Mayer, H., Krauss, J ., Urbanik-Sypniewska, T., Puvanesarajah, V., Stacey, G.,
Auling, G. (1989) Arch. Microbiol. 151: 111-116.

Hollingsworth, R., Lill-Elghanian, D. (1989) J. Biol. Chem. 264: 14039—14042.

Dazzo, F. (1982) in "Experimental Microbiol Ecology", (Burns, R., Slater, J.,
eds.), p. 431—446, Blackwell Scientific Publications.

sttphal, 0., Jann, K. (1965) in "Methods in Carbohydrate Chemistry",
(Whistler, R., ed.), p. 83-91, Academic Press.

Dubois, M., Gilles, K., Hamilton, J., Rebers, P., Smith, F. (1956) Anal. Chem.
28: 350.

Kesler, R. (1967) Anal. Chem. 39: 1416—1422.

119.

120.
121.
122.
123.

1 24.
l 25.
1 26.

127.

128.

129.

130.
I31.

132.

133.

134.
135.

136.

137.
138.

147

Lowry, O., Roberts, N., Leiner, K., Wu, M., Farr, A. (1953) J. Biol. Chem. 260:
1-17.

Rivera, M., McGroarty, E. (1989) J. Bacteriol. 171: 2244—2248.

Personal communication with Rawle I. Hollingsworth.

Hollingsworth, R., Dazzo, F. (1981) J. Microbiol. Methods 7: 295-302.
Rietschel, E. Th., Sidorczyk, Z., Zahringer, H., Wollenweber, H., Luderitz, O.
(1983) in " Bacterial Lipopolysaccharides-Structure, Synthesis, and Biological
Activities", (Anderson, L., Unger, F., eds.), p. 195—218, American Chemical
Society.

Inage, M., Chaki, H., Kusumoto, S., Shiba, T. (1981) Tetra. Lett. 22: 2281-2284.
Inage, M., Chaki, H., Kusumoto, S., Shiba, T. (1980) Tetra. Lett. 21: 3889-3892.

Urbanik-Sypniewska, T., Seydel, U., Greck, M., Weckesser, J., Mayer, H. (1989)
Arch. Microbiol. 152: 527-532.

Bhat, U., Mayer, H., Yokota, A., Hollingsworth, R., Carlson, R. (1991) J.
Bacteriol. 173: 2155-2159.

Tsien, H. (1982) in "Nitrogen Fixation", Vol. 2, (Broughton, W. ed.) p. 182-198,
Clarendon Press, Oxford.

Galanos, C ., Rietschel, E. Th., Luderitz, O., Westphal, O. (1971) Eur. J. Biochem
19: 143-152.

Galanos, C., Luderitz, O. (1975) Eur. J. Biochem. 54: 603-610.

Bradley, L. (1980) in "Selected Methods in Cellular Immunology", (Mishell, B.,
Shiigi, S., eds.) p. 158-161, W. H. Freeman & Co., New York.

Severinson-Gronowicz, E., Doss, C., Howard, F., Morrison, D., Strober, S.
( I980) 125: 976-980.

Galanos, C., Hansen-Hagge, T., Lehmann, V., Luderitz, O. (1985) Infect. Immun.
48: 355358.

Personal communication with Dr. F. B. Dazzo.
Levin, J., Bang, F. (1964) Bull. Johns Hopkins Hosp. 115: 265-274.

Takayama, K., Qureshi, N., Raetz, C., Ribi, E., Peterson, J., Cantrell, J., Pearson,
F., Wiggins, J ., Johnson, A. (1984) Infect. Immun. 45: 350—355.

Slavin, S., Strober, S. (1978) Nature 272: 624-626.

Knapp, M., Severinson-Gronowicz, E., Schroder, J., Strober, S. (1979) J.
Immunol. 123: 1000-1006.

139.

140.

141.
142.
143.
144.

145.

146.

148

Galanos, C., Luderitz, O., Freudenberg, M., Brade, L., Schade, U., Rietschel, E.
Th., Kusumoto, S., Shiba, T. (1986) Eur. J. Biochem. 160: 55-59.

Takayama, K., Qureshi, N., Raetz, C., Ribi, E., Peterson, J., Cantrell, J., Pearson,
F., Wiggins, J., Johnson, A. (1984) Infect. Immun. 44: 421-426.

Tahri—Jouti, M., Chaby, R. (1990) Mol. Immunol. 27: 751-761.

Vaara, M., Viljanen, P. (1985) Antimicrob. Agents Chemother. 27: 548-554.
Zimmerman, P., Gregory, S., Kern, M. (1977) J. Immunol. 119: 1018-1023.
Blatteis, C. (1990) in "Cellular and Molecular Aspects of Endotoxin Reactions",
Vol. 1, (Nowotny, A., Spitzer, J., and Ziegler, E., eds.) p. 447-454, Elsevier
Science Publishers B. V., Amsterdam.

Loppnow, H., Libby, P., Freudenberg, M., Krauss, J ., Weckesser, J ., Mayer, H.
(1990) Infect. Immun. 58: 3743—3750.

Maurelli, A. (1992) ASM News 58: 603—608.