=21». .. ii T
A 2.. Elsfififiwflgutxflv
:9 an! , .Hvt ..
.51

AV

a.»
flag.
a.
5.24;. I!
1 “129%
rfinnma...‘
c ID.
“mm... H. nth} “WM...
. an. .1 ‘

afar» I
0%.
. n 4.4.0". 1.0..
541...... “Wimw.
.0. . xv lfle. fl
- .- ll i In
1.1.1.: 3mm“ 1. wmhfimlvn‘fiqhm

‘ ‘4‘
2r, , .. may. 5 3%
z.
. , . 5‘

ix .(1
r. ... :1. 2
.l‘

{pi-1 ur
AWN?
, .. it”.
..

v

v.1 ,
grub”)
,3 a);

f:

 

{HESlS

IIIIIIIIIIIIIIIIIIIIIIIII IIIIIIIIIIIIIII IIIIIIIIIIIIIIIIIII

01554 4533

LIBRARY

Michigan State
Unlverslty

 

 

 

 

This is to certify that the

dissertation entitled

Investigations of the Molecular Mechanisms
for Host Specificity and Infection in
the Rhizobium/Legume Symbiosis

presented by

Robert A. Cedergren

has been accepted towards fulfillment
of the requirements for

Ph .D . degree in Biochemistry

 

\ L lYI/gI'KWIl

Major professor
Date 0 flw / 9‘

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE ll RETURN BOX to movothh chockoutflom your record.
TO AVOID FINES mum on or bdon‘dato duo.

DATE DUE DATE DUE DATE DUE

       

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Atflmatlvo mum/Equal Oppommlty Inflation
Wmt

INVESTIGATIONS OF THE MOLECULAR MECHANISMS
FOR HOST SPECIFICITY AND INFECTION IN
THE RHIZOBIUM/LEGUME SYMBIOSIS

By
Robert A. Cedergren

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Biochemistry

1996

ABSTRACT
INVESTIGATIONS OF THE MOLECULAR MECHANISMS

FOR HOST SPECIFICITY AND INFECTION IN
THE RHIZOBI UM / LEGUME SYMBIOSIS

By
Robert A. Cedergren

The Rhizobium/legume symbiosis represents an important
agricultural and ecological interaction between prokaryotes and
eukaryotes. Despite years of study, the molecular mechanisms critical to
various aspects of this relationship, especially host specificity, have not yet
been fully elucidated. The studies presented here explore the criticality of
membrane chemistry in ensuring a viable symbiotic interaction.

Several common links between the structural chemistry of the
chitolipooligosaccharides (Nod factors) of Rhizobium and the general
rhizobial membrane lipid and lipopolysaccharide chemistry of these
bacteria are uncovered. Aspects of common chemistry include sulfation,
methylation and the position and extent of fatty acyl chain unsaturation.
In the case of sulfation, common genetic control is shown by using strains
with mutations in the associated nod genes. A large degree of spatial
overlap between these molecules is also demonstrated since it is revealed

that the chitolipooligosaccharides are normal rhizobial membrane

components. These results contradict the current dogma and allow us to
begin to develop a comprehension of the highly integrated membrane lipid
and glycolipid chemistry of rhizobia.

Novel aspects of the cell surface chemistry of the unusual Rhizobium
strain R. leguminosarum biovar trifolii 45 have also been uncovered.
Differences in this strain are manifested by a unique LPS core region and
O-antigen as well as a distinct extracellular polysaccharide; compared to
other R. trifolii strains, these cell-surface components are uniquely devoid
of galactose. The discovery here of a galactose—containing lipid-linked
oligosaccharide with a composition that resembles the typical R. trifolii 0-
antigen may allow us to understand better these abnormalities. These
findings suggest that this strain is an effective natural isolate that has
surface chemistry similar to some ineffective mutants.

The plant sulfolipid, sulfoquinovosyl diacylglycerol, was identified
in the membranes of various strains of the family Rhizobiaceae. Until this
study, this lipid was thought to be restricted to plants, photosynthetic
bacteria or diatoms. The presence of this lipid in rhizobia, therefore, has
significant implications to the mechanisms of symbiosis. Investigations
into the heretofore unresolved biosynthetic pathway of this lipid were also
performed. Two molecules, thiolactaldehyde and thioquinovose, are

proposed to be involved.

To Mom and Dad.

Thank you for everything.

iv

ACKNOWLEDGMENTS

First and foremost, I owe my sincerest thanks to my advisor,
colleague, and friend, Dr. Rawle I. Hollingsworth. You have taught me
not only about the technical aspects of science but also about how
problems should be approached. You have shown me the importance of
going against the ”establishment” and that science needs skeptics. Your
encouragement, support, and infectious optimism made this a great place
to earn a PhD. Thanks again R.I.H. and remember that it was always
”the best of times”.

I also owe my gratitude to the members of my committee, Dr. Paul
Kindel, Dr. Estelle McGroarty, Dr. William Reusch, and Dr. Jack Watson,
for the time they dedicated to my project. Your advice was well taken and
is deeply appreciated.

I would like to thank Bao Jen, Ben, Carol, Chuck, Debbie, Du,
Gabriela, Gang, Guangfei, Guijun, Hussen, Jeongrim, Jie, Jim, I], Jung,
Kim, Luc, Maria, Steve, Yin, Ying, and Yuanda, i.e., the Hollingsworth
group, past and present. The technical assistance I received from all of
you was a key component of my success and the ”occasional” diversionary
discussions made this a very pleasant place to work.

I would also like to thank members of the Mass Spectrometry
Facility, especially Bev and Dr. Huang, as well as Dr. Long Le at the

NMR Facility for all of your assistance. I also owe thanks to the Dazzo

V

lab, Frank, Saleela, and Guy, for helpful discussions and the occasional
use of their GC-MS.

Last, but definitely not least, I would like to thank my wife, Kerry,
and daughter, Alexandria, for their unconditional love, encouragement

and motivation. Without you, this thesis would not exist.

Thanks again everybody,
Rob.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................. ix
LIST OF FIGURES ............................................ x
LIST OF ABBREVIATIONS ..................................... xiv
CHAPTER 1
Overview of rhizobial cell surface chemistry and its possible roles in
symbiotic nitrogen fixation ..................................... 1
Introduction .............................................. 2
Literature Review ......................................... 5
The Process of N odulation ................................ 5
Membrane Associated Carbohydrates of Rhizobia ........... 9

Involvement of Cell Surface Components in Host Specificity . . 17
Involvement of Cell Surface Components Throughout Infection 20

Rhizobial Genetics ....................................... 22

The Nod Factor Model for Host Specificity ................. 24

Conclusion ............................................... 32
CHAPTER 2

Common links in the structure and cellular localization of Rhizobium
chitolipooligosaccharides and general Rhizobium membrane

phospholipid and glycolipid components ......................... 34
Abstract ................................................. 35
Introduction .............................................. 37
Materials and Methods .................................... 46
Results and Discussion ..................................... 52
Conclusions .............................................. 81
Acknowledgments ......................................... 83

CHAPTER 3

The "missing" typical Rhizobium leguminosarum O-antigen is
attached to a fatty acylated glycerol in R. trifolii 48, a strain that also

lacks the usual tetrasaccharide "core" component ................. 84
Abstract ................................................. 85
Introduction .............................................. 87

Materials and Methods .................................... 92

Results and Discussion ..................................... 95
Acknowledgments ......................................... 116

CHAPTER 4

Occurrence of sulfoquinovosyl diacylglycerol in some members of the

family Rhizobiaceae. ........................................... 117
Abstract .................................................. 118
Introduction .............................................. 120
Materials and Methods .................................... 123
Results ................................................... 129
Discussion ................................................ 154
Acknowledgments ......................................... 157

CHAPTER 5

Investigation of the biosynthesis of sulfoquinovosyl diacylglycerol in

Rhizobium meliloti. ............................................ 158
Abstract .................................................. 159
Introduction .............................................. 160
Materials and Methods .................................... 170
Results and Discussion ..................................... 173
Conclusion ............................................... 185
Acknowledgments ......................................... 187

CHAPTER 6

Retrospect .................................................... 188

BIBLIOGRAPHY ............................................... 199

viii

LIST OF TABLES

Table 1.1: Some rhizobial species and their associated host

plants ............................................. 3
Table 2.1: List of fatty acids normally present in the lipids of

R. meliloti membranes .............................. 70
Table 3.1: Reported carbohydrate composition of LPS of various

Rhizobium trifolii strains ............................ 105

ix

Figure 1.1:

Figure 1.2:

Figure 1.3:

Figure 1.4:

Figure 1.5:

Figure 2.1:

Figure 2.2:

Figure 2.3:

Figure 2.4:

LIST OF FIGURES

Representation of the events leading to nitrogen

. fixation in the rhizobia/ legume interaction ..........

Basic structure of typical enteric bacterial
lipopolysaccharide ................................

Glycosyl sequences of the two major BPS structures
found in R. meliloti ...............................

Four nod gene inducing compounds from various
systems .........................................

Putative pathway for the biosynthesis of chitolipo-
oligosaccharides in Rhizobium meliloti and the
functions of some of the associated nod factors .......

Diagrammatical representation of the putative signal
exchange in the early events of the rhizobia legume
symbiosis .......................................

The major reported structures of chitolipooligo-
saccharides produced by various rhizobial strains . . . .

Sepharose 4B gel filtration profile of the crude phenol
extract of 35S-labelled R. meliloti 2011 cells grown in
the absence of luteolin ............................

Sepharose 4B gel filtration profile of the crude phenol
extract of 35S-labelled R. meliloti 2011 cells grown in
the presence of luteolin ...........................

8

11

15

25

31

38

43

54

56

Figure 2.5:

Figure 2.6:

Figure 2.7:

Figure 2.8:

Figure 2.9:

Figure 2.10:

Figure 2.1 1:

Figure 2.12:

Figure 2.13:

Figure 2.14:

Figure 3.1:

Autoradiograph of a sodium dodecylsulfate gel

showing the separation of 35S-1abelled LPS isolated

from R. trifolii ANU843 and R. meliloti 2011 in the

presence and absence of flavone inducers ........... 59
f

Sepharose 48 gel filtration profiles of LPS isolated

from 35S-labelled R. meliloti 1021 cells and three Tn5

mutants assaying for total carbohydrate and

radioactivity .................................... 62

Sepharose 48 gel filtration profile of material isolated
from 35S-labelled R. meliloti 1021 cells carrying a Tn5
mutatation in the nodH gene assayed for total
carbohydrate and radioactivity .................... 63

Sepharose 43 gel filtration profile of extract from 358-
labelled R. sp N GR234 cells showing the

correspondence of radioactivity to the LPS marker
component KDO ................................. 65

Total ion chromatogram showing the typical fatty acid
species in membrane lipids from R. meliloti 2011
derivatized as methyl esters ....................... 69

Proton (A) and 13C (B) NMR spectra of the total
membrane extract from R. meliloti 2011 ............. 72

lH-“C multiple quantum coherence spectrum (HMQC)
of the total membrane extract from R. meliloti 2011 . . . 73

Proton NMR spectrum of sulfoquinovosyl diacyl
glycerol from mutant JSSl6 ....................... 75

Proton NMR spectrum of chitolipooligosaccaride
fraction from R. meliloti 2011 membranes of cells
grown in the absence of flavone ................... 78

Electrospray (negative ion) mass spectrum of a chito-
lipooligosaccharide preparation from R. meliloti 2011 80

Biogel P2 chromatogram of LPS hydrolysate of R.
trifolii 4S ....................................... 96

xi

Figure 3.2:

Figure 3.3.:

Figure 3.4:

Figure 3.5:

Figure 3.6:

Figure 3.7:

Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Figure 4.6:

1H NMR spectrum of fractions C (A) and D (B) of
peak 2 from Figure 3.1 ........................... 98

1H NMR spectrum of mild acid hydrolysis products
of combined R. trifolii 4S fractions eluted from C18
column with 50 and 100% acetonitrile in water ...... 101

Total ion chromatogram of alditol acetates of the
R. trifolii 4S extract eluted from a C18 column with
100% acetonitrile ................................ 104

Total ion chromatogram of fatty acid methyl esters of
the same fraction as in Figure 3.4 .................. 107

Orcinol/ sulfuric acid stained TLC plate of (A) a total

lipid extract of R. trifolii AN U843, (B) a total lipid

extract of R. trifoliz' 4S, and (C) the combined R. trifolii

48 fractions eluted from the C18 column with 50 and

100% acetonitrile in water ........................ 110

Possible mechanism of transfer of LPS O-antigen
repeat unit from the carrier lipid to diacylglycerol . . . 112

1H-NMR spectrum of the purified 35S-labelled major
component (Rf=0.22) from the membrane extract of R.

meliloti 2011 .................................... 131

13C-DEPT-NMR spectrum of the purified 35S-labelled
component (Rf=0.22) ............................... 134

1H-HOHAHA NMR spectrum of the purified 35$-
labelled component (Rf=0.22) ...................... 135

1H-DQF—COSY spectrum of the purified 358-1abelled
component (Rf=0.22) ............................. 137

lH/13C-HMQC spectrum of purified 35S-labelled lipid
component (Rf=0.22) ............................. 139

1H-NMR spectrum of purified 3SS-labelled lipid
component (Rf=0.22) ............................. 141

xii

Figure 4.7:
Figure 4.8:
Figure 4.9:

Figure 4.10:

Figure 5.1:
Figure 5.2:
Figure 5.3:
Figure 5.4:

Figure 5.5:
Figure 5.6:

Figure 5.7:

Figure 6.1:

Negative ion fast atom bombardment mass spectrum
of purified lipid component ....................... 143

Orcinol stained TLC plate of membrane extracts from
members of the Rhizobiaceae ..................... 148

Autoradiogram of TLC plate of membrane extracts
from members of the Rhizobiaceae ................. 150

Fourier transform infrared spectra of (A) SQDG
standard and (B) extract from putative SQDG TLC
band of the R. meliloti 2011 sample ................. 153

Compilation of some of the proposed pathways for
the biosynthesis of sulfoquinovosyl diacylglycerol . . . . 163

Proposed pathway for the biosynthesis of
sulfoquinovosyl diacylglycerol ...................... 167

Phosphorimager scan of paper chromatograph of
extract of R. meliloti ............................. 174

Phosphorimager scan of paper chromatograph of
extract of R. trifolii .............................. 175

Phosphorimage of one-dimensional paper
chromatogram of ethanol extracts of various
organisms ...................................... 177

Phosphorimage of one-dimensional paper
chromatogram of ethanol extract of R. meliloti
after various treatments .......................... 182

Proposed scheme of thioquinovose formation from
thiolactaldehyde condensation with DHAP ......... 185

Thin layer chromatogram of membrane extracts of

three 11on mutant strains along with the parent wild
type strain ...................................... 194

xiii

ATP

BIII
CDP
CLOS

C M
CPS
DAG
DEPT
DFQ-COSY
DHAP
DPG
EPS
FAB-MS
FAME

FID

LIST OF ABBREVIATIONS

adenosine 5’-diphosphate

adenosine 5’-phosphosulfate

adenosine 5’-triphosphate

modified Bergensen’s media

cytidine 5’-diphosphate

chitolipooligosaccharide

carboxymethyl cellulose

capsular polysaccharide

diacylglycerol

distortionless enhancement by polarization transfer
double quantum filtered-correlated spectroscopy
dihydroxyacetone phosphate
diphosphatidylglycerol

extracellular polysaccharide

fast-atom bombardment mass spectrometry
fatty acid methyl ester

free induction decay

xiv

GC / MS
GDP
GlcNAc
HMQC
HOHAHA
HPLC
KDO

LPC

LPS
MDO
NMR
PAPS

PC

PE

PG
PMME
SDS-PAGE
SQDG
TEAE

TFA

TLC

gas chromatography / mass spectrometry
guanosine 5’-diphosphate

Nsacetyl glucosamine

heteronuclear multiple-quantum coherence spectroscopy
homonuclear Hartmann-Hahn spectroscopy
high performance liquid chromatography
2-keto-3-deoxyoctulosonic acid
lysophosphatidyl choline

lipopolysaccharide

membrane derived oligosaccharides

nuclear magnetic resonance
3’-phosphoadenosine—5’~phosphosulfate
phosphatidyl choline
phosphatidylethanolamine

phosphatidyl glycerol

phosphatidyl monomethyl ethanolamine
sodium dodecyl sulfate-polyacrylamide gel electrophoresis
sulfoquinovosyl diacylglycerol
triethylaminoethyl cellulose

trifluoroacetic acid

thin layer chromatography

uridine 5’-diphosphate

XV

CHAPTERI

OVERVIEW OF RHIZOBIAL CELL-SURFACE CHEMISTRY

AND ITS POSSIBLE ROLES IN

SYMBIOTIC NITROGEN FIXATION

2
INTRODUCTION

Symbiotic nitrogen fixation is a classical illustration of an
interaction between eukaryotes and prokaryotes that encompasses many
fields of science; biochemistry, chemistry, genetics, microbiology, botany,
physiology and agricultural science. The relationship between bacteria
belonging to the genera Rhizobium, Bradyrhizobium and Azorhizobium
(collectively termed rhizobia) and their host leguminous plants (family
Leguminosae) has interested scientists for over 100 years. The importance
of symbiotic nitrogen fixation to the global nitrogen cycle has been well
documented. It has been estimated that the rhizobial legume symbiosis
accounts for over 50% of natural nitrogen fixation and when taking into
account the nitrogen fixation of agricultural value, this number jumps to
over 70% (1). The interest in this unusual relationship, however, has
stemmed not only from the various agricultural ramifications but also
because the information gleaned from this system might provide insight
into the processes involved in other infections as well.

One of the most studied aspects of the symbiotic relationship, and
also one of the most intriguing, is the specificity in host range exhibited by
most species of rhizobia. With a few notable exceptions, each species or
biovar (i.e., subspecies) of bacteria is only able to infect a very limited

number of plants. For instance, Rhizobium meliloti strains are capable of

3
infecting alfalfa plants but lack the ability to form a symbiosis with clover

(Table 1.1). Conversely, strains belonging to Rhizobium leguminosarum
biovar trifolii (hereafter referred to as R. trifolii) infect clover but not

alfalfa. Fundamental differences must therefore exist between the

Table 1.1: Some rhizobial species and their associated host plants.

 

Rhizobial Species Host Plant
Rhizobium meliloti Alfalfa (Medicago)
R. leguminosarum biovar trifolii Clover (Trifolium)

biovar phaseoli Bean (Phaseolus)
biovar viciae Vetch (Vicia), Pea (Pisum)
Rhizobium species strain NGR234 Tropical legumes, Paraspom'a (non-legume)

Bradyrhizobium japonicum Soybean (Glycine)

various species of rhizobia with regard to host recognition. Many
theories have been advanced over the years to explain the mechanisms
responsible for initial as well as other necessary interactions throughout
the rhizobia / legume symbiosis. The studies performed in this dissertation
were aimed at testing some of these theories and hopefully unraveling the
molecular mechanisms involved in host specificity as well as other
important aspects of the infection process.

The literature review that is presented in Chapter 1 is a general

overview on the history of rhizobial research. An emphasis is placed on

4
the role of membrane components as well as diffusable metabolites in the

process of nitrogen fixation. Evidence of common structural and
biosynthetic characteristics between these two cell surface related
elements is presented in Chapter 2. It is shown that rhizobial
chitolipooligosaccharides and their general membrane lipid components
have several common structural features and that the same genes are
involved. Chapter 3 presents the discovery of a new class of lipid from a
strain of R. trifolii with unusual membrane chemistry. This lipid may
answer questions as to why this strain is able to infect with a membrane
that is so unique to its species. The isolation and characterization of
sulfoquinovosyl diacylglycerol from R. meliloti is shown in Chapter 4.
This is a lipid that heretofore was thought to be virtually nonexistent
outside of photosynthetic organisms. The occurrence of this lipid amongst
various rhizobial species as well as implications for a role in symbiosis are
addressed. Experiments aimed at deciphering the biosynthesis of this lipid
are given in Chapter 5. Based on the data presented in this chapter, two
compounds are suggested that may be involved in one of the several
possible synthetic pathways that have been delineated over the years. The
final chapter encompasses all of these studies by offering a broad
perspective on the role of the membrane in assuring a viable infection in

the Rhizobium/ legume symbiosis.

5
LITERATURE REVIEW

The Process of N adulation

The process of symbiotic nitrogen fixation is characterized by the
ability of the Gram-negative soil inhabiting rhizobia to invade and
participate in a mutualistic relationship with a legume host. A wide
variety of microorganisms, both free-living and symbiotic, possess the
ability to fix nitrogen. The rhizobia/ legume system is by far the most
studied, however, due to the agricultural importance of legumes. The
infection by rhizobia culminates in the formation of a novel structure on
the root, termed nodule, within which the bacterium transforms into a
nitrogen fixing bacteroid state fueled by the plant. This reduction of
molecular nitrogen into ammonia that is performed by the bacteroid is of
utmost agronomical and ecological importance. Nitrogen is one of the
most limiting nutrients to plants since it is not readily available in most
soils. Atmospheric nitrogen is not assirnilable by plants and the only other
usable sources besides biological fixation are the small amounts in
rainwater, the waste and decay products of other organisms or artificial
fertilizers (1). The latter of these, although accounting for increased
agricultural productivity, has had an adverse ecological impact. Run-off
from fertilized lands has contaminated drinking water supplies with

nitrates and contributed to the eutrophication of lakes and rivers (1). The

6
application of liquid ammonium fertilizers and the associated release of

excess ammonia also increases the load of greenhouse gases in the
atmosphere. An additional problem related to fertilizers is that the
reduction of nitrogen by abiological means is an extremely energy-
expensive procedure. With the growing world population and subsequent
demand for food, these concerns will become increasingly more and more
significant. Both the economic as well as environmental expense of
nitrogen fertilizers will eventually make them prohibitive and biological
nitrogen fixation that much more important.

A basic diagrammatical representation of the events leading to
nitrogen fixation is shown in Figure 1.1. The first obvious indication of an
impending infection occurs with the appearance of distinct morphological
changes in the root. These changes begin with deformations and curling
of the root hairs (Shepard crooks) (2, 3) along with simultaneous
mitogenic activity in the cortical cells of the root (4). The bacteria, which
were bound to the root after being attracted by plant chemotaxants (5, 6),
become trapped by the contorted root hair. These trapped bacteria then
begin to degrade (or induce the degradation of) the plant cell wall and
subsequently influence the development of a new structure called the

infection thread (7). The infection thread, which can be compared to a

Figure 1.1: Representation of the events leading to nitrogen
fixation in the rhizobial legume interaction. After chemotaxis
to the host root hair, the bacterium binds and the plant cell
wall degrades. At this point begins the formation of the
infection thread through which the bacteria are able to
traverse to the rapidly dividing cortical cells. The bacterium is
released from the infection thread into the plant cell
simultaneously becoming surrounded by a plant derived
peribacteroid membrane. When the plant cell is nearly full of
bacteria, the outer membrane and peptidoglycan of the
bacteria are lost and the enlarged bacteroid form is able to

reduce nitrogen. Adapted from references 1 and 10.

Rhizosphere
I nodule organogenesis

binding degradation of
release of followed plant cell wall

((
flaVOHOIdS a -- by root hair infection 7) - \
attracts deformations thread I

l bacteria developmen -tIl\\\\ \) 2

signal to increase

cried cell { T \
dwision I /‘ 1‘\

Host Legum . . '—' QQA‘.“
Root e g "2:353? fig ‘91 \‘ II
--

Infection
thread

 

 

Plasmalemma

     
   

Bacteria

 

 

 

Mature bacteroid-
cell division stops,
loss of outer
membrarie and
can,
875% 13331.:
increase in volume.

 

 

 

Plant
Nucleus

Peribacteroid
membrane

 

Dividing
bacterium

Figure 1.1

9
root hair growing inward, is composed of plant derived material and acts

as a tunnel which allows the bacterium to navigate its way to the inner
cortex of the root. These densely populated cortical cells are then
penetrated by the ramified infection thread and infected by the bacterium
(8). As they leave the infection thread, the rhizobia are enveloped by a
plant-like plasma membrane that is at this point designated as the
peribacteroid membrane (9). Eventually the plant cells in this area have
divided and become infected to such a great extent that they create a bulb-
like structure on the root. This structure is called the nodule and this
entire process termed nodulation. At this stage the invading organisms
are in their bacteroid state and capable of relieving the host plant of the

need for external sources of nitrogen.

Membrane Associated Carbohydrates of Rhizobia

The general cell envelope organization of rhizobia is fairly typical of
Gram-negative bacteria. It consists of an inner cytoplasmic membrane, a
periplasmic space in which resides the peptidoglycan and an outer
membrane of which the outer leaflet is mainly composed of
lipopolysaccharide (LPS). The LPS is generally composed of three
structural domains: the lipid A, which acts as the membrane anchor, the
core region, an oligosaccharide that is fairly conserved between species

and the O—antigen, a repeating polysaccharide of varying length that is

10

Figure 1.2: Basic structure of typical enteric bacterial
lipopolysaccharide. Three structural domains exist in normal
LPS: the lipid A membrane anchor, the inner and outer core
regions that are fairly conserved amongst species and
typically contain KDO and heptose as marker carbohydrates,
and an O—antigen repeat that is highly strain specific.

Adapted from reference 137.

11

l I O-Antigen
Repeating
Units

 

 

 

 

 

 

 

 

 

 

 

Core Marker Sugars
HO
OH
OH
Inrciler ,0
an
Outer HO
Core OH
Heptose
|
O
”0—1;
0‘
_/ O
o o. -
0 l
0— 0'
0' LipidA
Y

 

Figure 1.2

12
highly variable in composition between species (137). In the ”typical”

enteric Gram-negative bacteria, the LPS has been very well characterized
(Figure 1.2). The lipid A in these bacteria consists of a disaccharide of 1-6
B—linked glucosamine substituted with 3-OH fatty acids in either ester or
amide linkages at the 2, 3, 2’ and 3’ positions (137). There are also fatty
acids in ester linkages at the 3-OH functionalities. Additionally, there is a
net negative charge on lipid A resulting from phosphate esters at the 1 and
4’ positions. The core region is generally characterized by the presence of
two marker carbohydrates, 2-keto-3-deoxyoctulosonic acid (KDO) and
heptose (137).

The structural elucidations performed thus far on rhizobial LPS
have shown it to contain certain features distinct from the ”typical” LPS.
For instance, the lipid A from R. trifolii ANU 843 was found to contain an
unusual long chain fatty acid accounting for about 50% of the total fatty
acid content (11). This fatty acid, 27-hydroxyoctacosanoic acid, was
subsequently found to be a constitutive component in members of the
Rhizobiaceae (12). Another unusual fatty acid, 29-hydroxytriacontanoic
acid, was also isolated from a rhizobial species (13). Besides the fatty acid
composition, the headgroup of lipid A in rhizobial strains has been shown
to differ from that found in the enteric bacteria. It was found that the lipid
A from R. trifolii ANU 843 lacks the phosphate groups and instead a

carboxyl group accounts for the requisite negative charge (14, 15). This

13 _
may not be a critical structural feature, however, since another study

found the typical phosphorylated diglucosamine backbone in this species
(16). In subsequent studies, glucosamine and galacturonic acid were
identified as the lipid A sugars in this strain (12, 15). Yet another unusual
rhizobial lipid A backbone was identified as a trisaccharide containing two
uronic acids and a glucosamine (17). A later study indicated that this
structure was incorrect though, and the purported glucosamine aldonic
acid component was not present (23).

The LPS core region is usually conserved between strains of the
classical Gram-negative bacteria and this is thought to be generally true
of rhizobia as well. Two distinct core oligosaccharides are known to exist
simultaneously in the biovars of R. leguminosarum (trifolii, phaseoli and
viciae) (18-22). One is a trisaccharide consisting of two galacturonic acid
residues and a KDO while the other is a tetrasaccharide containing
galactose, galacturonic acid, mannose and KDO. Neither contains
heptose, a "typical" core marker (along with KDO) found in the
Enterobacteriaceae.

The basic O-antigen structure of rhizobial LPS is very different from
the classical Gram-negative bacteria as evidenced by its behavior in
sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The enteric
LPS appears as a "ladder"; discrete bands which reflect various degrees of

polymerization of linear oligosaccharide subunits (24, 25). With rhizobial

14
LPS there are typically only two major bands. These have been attributed

to the so-called LPS 1, with an intact O-antigen (smooth), and LPS II, with
an attenuated O-antigen (rough) (26-31). The reason for this major
difference is not known but it may be that the rhizobial O-antigen is a
branched polysaccharide as opposed to a linear repeat. The only rhizobial
O-antigen characterized to date has been from the unusual R. trifolii 48
strain (32). The question of branching was not addressed in this study.
Most rhizobial strains also synthesize a capsule or capsular
polysaccharide (CPS) as well as an excreted acidic extracellular
polysaccharide (BPS). The structure of these two glycoconjugates are
generally similar but the difference between the two lies in the fact that
the CPS is associated with the cell surface while the EPS is sloughed off
into the extracellular milieu. Much of the earlier work done on rhizobial
EPS was performed on R. meliloti. The major form of R. meliloti EPS,
also known as succinoglycan or EPS I (Figure 1.3), is an acidic repeating
octasaccharide carrying certain non-carbohydrate substituents (a
succinate, pyruvate and O-acyl group) that account for its acidity. A
family of genes (exo) important for the synthesis and export of the
succinoglycan has been well characterized (for review see reference 33).
Certain mutant strains of R. meliloti have been found to produce an
alternate EPS, termed EPS II, that has a structure considerably distinct

from the succinoglycan (Figure 1.3) (34, 35). The EPS of the biovars of R.

15

+Glc L” (Inc L1" Glc 5*” can” n
6

 

 

I} 1’6 OAc
Glc
I...
Glc
I...
Glc — OCOCHZCHZCOO '
IB-la
4/O\ ,CH3
Glc\ /C\
6 O COO"
Succinoglycan
Gal “'13 Glc 9—2:”
+7 \6 I6
O\ /O OAc
C
/ \
COO" CH3
EPS 11

Figure 1.3: Glycosyl sequences of the two major EPS
structures found in R. meliloti, succinoglycan, or EPS I, and

EPS II. OAc = acyl group, Glc = glucose and Gal = galactose.

Adapted from reference 33.

16
leguminosarum have been found to be fairly similar to that of R. meliloti.

These bacteria typically produce an octasaccharide repeating EPS
consisting of glucose, glucuronic acid, and galactose (5:2:1) with a
pyruvate acetal attached to the terminal galactose (36). Another non-
carbohydrate substituent, an ether linked 3-hydroxybutanoic acid group,
has also been found in certain strains of R. trifolii (37, 38).

Other classes of carbohydrate-containing molecules have been
found to be membrane associated in the rhizobia. One class, the cyclic B-
(1,2)-glucans, is related to the membrane derived oligosaccharides (MDO)
of E. coli. These cyclic oligosaccharides are composed entirely of glucose
and are localized in the periplasmic space (39, 40). It has been shown that
these components are related to osmotic balance (41). Another E. coli
related rhizobial polysaccharide recently characterized is one containing a
high degree of KDO that is not excreted (i.e. not an EPS) and is
structurally distinct from the LPS (42, 43). Isolated from R. meliloti and R.
fredii, this polysaccharide consists of repeating units of galactose and
KDO and is structurally analogous to the group 11 K antigens (CPS) of E.
coli. Another class of glycolipid, the digycosyl diacylglycerols, has been
recently found in strains of R. trifolii (44, 45). Although their appearance
in Gram-negative bacteria is rare, they are widespread components of

plant cell membranes.

17
Involvement of Cell Surface Components in Host Specificity

The earliest models presented to implicate cell surface chemistry in
the host recognition process involved lectin/ carbohydrate binding. The
first paper on this subject, by Hamblin and Kent, presented evidence that
the plant lectins might bind the bacteria to the roots in a general fashion
(46). Using a single Rhizobium strain these researchers found that
bacteria treated with lectins were capable of erythrocyte agglutination.
However, the question of host specificity was not addressed. In a
subsequent study, Bohlool and Schmidt used a fluorescently labelled
soybean lectin to bind specifically to all but 3 of 25 soybean nodulating
Rhizobium strains tested (47). In addition, the lectin did not bind to 23
strains from 5 various species in which soybean is not a host. Therefore
they postulated that legume lectins presented a site on the root surface
which recognized a specific polysaccharide on the appropriate bacteria.
The studies that followed attempted to identify the bacterial
polysaccharide receptor that was binding to the lectin. The first such
series of experiments, by Dazzo et al., implicated the capsular
polysaccharide (CPS) as the determinant of symbiotic specificity. It was
shown that the CPS of the clover symbiont R. trifolii carried determinants
which were antigenically cross-reactive with a surface component on
clover seedling roots (48, 49). These researchers proposed that the host

lectin was a diffusable component excreted from the plant that joined R.

18
trifolii to clover root hairs by recognizing common antigenic determinants

on each. In later studies, they were able to identify the clover lectin
trifoliin A, show it to be present at discrete sites near the tip of the root
hair cells, and show that the specific encapsulated rhizobial symbiont
bound at these sites (50-52). Various workers were subsequently able to
identify other specific bacterial capsular polysaccharides and their
associated lectins in the R. meliloti/ alfalfa (50), R. leguminosarum/ pea
(53) and B. japonicum/ soybean (54, 55) systems.

The role of the acidic extracellular polysaccharide (BPS) in host
specificity has proven to be somewhat controversial. One series of
experiments has shown that the O-acylation pattern of the EPS of R.
trifolii was influenced by host specificity genes and therefore the
interactions of the EPS with lectins may contribute to the specificity (36,
58). The authors make the claim that "these alterations would be
pronounced in the host root environment and contribute to host
specificity" (58). This data was refuted by others, however, who found no
differences in the structures of EPS repeating subunits from the various
biovars of R. leguminosarum (59-61). These authors state unequivocally
that "the discernible structural features of the acidic polysaccharides
secreted by Rhizobium biovars cannot be the determinant of host

specificity" (59). This debate has not been settled.

19
Besides the CPS and EPS, the lipopolysaccharide (LPS) has also

been shown to be involved in host specific lectin binding. Experiments by
Wolpert and Albersheim implicated the O-antigen containing LPS by
showing binding to specific lectin columns (56). A later study explored the
growth phase dependence of trifoliin A binding and LPS structure (57). It
was found that the relative amount of various glycosyl components in the
LPS changed from the exponential to stationary phases of growth and
these changes correlated to the ability of the bacteria to bind trifoliin A and
to infect.

Most of the recent studies in the area of membrane polysaccharides
and host specificity have been aimed at determining precise structural
features required by the carbohydrate receptors as well as determining the
particular genes involved. Researchers have characterized the complete
structure of the CPS that binds trifoliin A (62) and demonstrated a
correlation between relative levels of non-carbohydrate side groups in the
CPS and their effects on lectin binding (63, 64). Studies have also been
performed relating the bacterial culture age to surface polysaccharide
structure and bacterial attachment to roots (65-67). In an important
experiment, Diaz et al. were able to isolate a key pea lectin gene. After
introducing the gene into white clover roots, the transgenic plants were
able to be nodulated by a strain usually specific for peas (68). Similar

results were obtained by altering the bacterial genes. Introduction of host

20
specificity genes from R. trifolii into a another Rhizobium species resulted

in a hybrid strain with efficient nodulation of the R. trifolii host clover,
production of the R. trifolii EPS and an increase in binding to the trifoliin A
lectin (58).

Although there is overwhelming evidence for the involvement of
carbohydrate/lectin interaction in host specificity, other results seem to
suggest that this interaction is necessary but not sufficient for an effective
symbiosis. For instance, it was found that a genetically engineered strain
of Azotobacter vinelandii, carrying the trifoliin A receptor from R. trifolii,
acquired the ability to specifically adhere to clover roots but not infect
them (69, 70). In addition, R. trifolii mutants were found that were able to
bind the host lectin and attach specifically to clover roots but were also
non-infective (71, 72). However, even though the carbohydrate/lectin
interaction may not exclusively account for host specificity, the plethora of
positive results from over twenty years of research implicating various

cell surface components in the recognition process cannot be ignored.

Involvement of Cell Surface Components Throughout Infection

Besides the recognition and binding stages of infection, various
defects in normal membrane chemistry have been correlated with a
dysfunction at other stages of the nodulation process as well. It was first

determined that the presence of LPS was critical for a viable symbiosis in a

21
study by Maier and Brill in which they noticed the absence of a surface

antigen associated with the LPS O-antigen in non-nodulating mutants of
Bradyrhizobium japonicum (73). Russa, et al., found that LPS isolated
from closely related nodulating and non-nodulating strains of R. trifolii
were vastly different in carbohydrate composition (74). A later study
showed that in a non-nodulating mutant strain of B. japonicum the LPS
seemed to lack an O-antigen (75). These results were echoed in another
study of a similar mutant of R. phaseoli (30). In this case, it was
demonstrated that the symbiotic defect was a flaw in infection thread
development. Subsequently, the ability of the bacteria to synthesize the
correct LPS structure was implicated in seemingly every stage of the
symbiotic process. These stages included competitiveness for nodulation
(76), infection thread development (77), bacterial release from the
infection thread (27), avoidance of the plant defense mechanisms (44) and
the development of complete nitrogen fixing nodules (22, 26, 29, 78).
Despite the overwhelming evidence, however, one study has proclaimed
that LPS does not play a major role in symbiosis (28). Using various
strains of R. meliloti with mutations in all of the major LPS genes these
researchers found that none of the mutant strains were defective in
symbiosis.

A definitive involvement of BPS in a viable symbiosis has been

shown as well. Most of the work in this area has centered upon resolving

22
phenotypic and structural changes from various mutations in the exo

genes (i.e., the genes responsible for BPS synthesis). The first such study
showed that a mutant strain of R. meliloti which failed to succinylate its
BPS I and did not produce any BPS II was defective in nodule invasion
(79). The ability of these strains to nodulate was partially rescued by the
addition of exogenous EPS II (45, 80). In a similar experiment performed
by Djordjevic et al., it was shown using both Rhizobium species strain
NGR234 and R. trifolii that the EPS or even the EPS oligosaccharide from
the parent strain of the Bx0' mutant partially restored nodulation ability
(81). Many other studies have been performed demonstrating the
significance of the BPS to infection. These have included the importance
of the regulation of EPS production (82), the variation of the infection
ability of Ex0' mutants depending on the host plant genotype (83) and the
effect of pH on restoring nodulation ability to Exo" mutants (84). The
general consensus on the role of the EPS in infection is that it ensures

bacterial release from the infection thread and nodule invasion.

Rhizobial Genetics

It is generally thought that several major classes of genes exist in
rhizobia that are related to the infection process. Some of these, (i.e. exo,
nif and fix) are important in the formation of functional nodules but

supposedly not in the determination of host range. Another family of

23
genes, known as the nod genes, have been proposed to be solely

responsible for specificity. These genes have been classified as either
common or host specific. Common nod genes, such as nodA, 210618 and
nodC, are interchangeable between various species of rhizobia that infect
diverse hosts and are also apparently conserved in function (85-88). It has
been shown that these genes are responsible for causing the increased
cortical cell division in the roots during infection (89) and for the
deformation of root hairs (85, 90). The host specific genes are particular to
a species or biovar and are believed to be the determinants of the bacterial
range of hosts. Mutations in these genes result in loss of root hair
deformation ability on the homologous host and may result in the
acquisition of this ability on heterologous hosts (91, 92). Examples of host
specificity genes are nodPQ and nodH of R. meliloti, which were shown to
specify nodulation of alfalfa (93, 94), and nod E of R. leguminosarum
biovars viciae or trifolii, which is responsible for infection of pea or clover,
respectively (95). The nod genes have been shown to be located on a
megaplasmid called pSym which has been estimated at 15 kb (96-98).

It has been observed that most of the nod genes are not expressed in
free-living cells rather only when in the presence of the plant host or plant
host exudates or extracts (99, 100). Exceptions to this observation are the
family of nodD genes, which are expressed constitutively. It appears that

the product of the nodD gene is responsible for the induction of the other

24
nod genes in the presence of plant exudates (101, 102). The compounds

isolated from the plant exudates found to be most active on nodD were a
family of flavonoids (or isoflavonoids). It also was found that distinct
flavonoids seemed to induce the genes of a particular species or biovar of
rhizobia. For example, luteolin (3',4',5,7-tetrahydroxyflavone), was
shown to induce nodD in R. meliloti (103, 104) (Figure 1.4). Additionally,
one of the inducer compounds from soybean, daidzein (105), has been
shown to be an inhibitor of nodD induction in R. trifolii, a symbiont of

clover (106).

The Nod Factor Model for Host Specificity

The first quantitative evidence that legume root hair deformations
could be induced by incubating the roots with a cell-free culture filtrate of
the appropriate rhizobial symbiont was made about sixty years ago (107).
Since then, many studies were performed that implicated diffusable
metabolites as the basis for nodule formation (4, 108, 109). Several
different classes of molecules including bacterial oligosaccharides (124,
125), polysaccharides (121-123), nucleic acids (123) and proteins (121, 123)
were implicated as the initial root hair deforming factors. None of these
components, however, were isolated in quantities sufficient for a

thorough structural analysis.

   

Luteolin Narigenin
3',4',5,7-tetrahydroxyflavone 4',5,7-trihydroxyflavanone
(R. meliloti ) (103) (R. viciae) (131)

   

Daidzein
4',7-dihydroxyisoflavone
(B. japonicum ) (105)

4', 7-dihydroxyflavone
R. trifolii (132)

Figure 1.4: Four nod gene inducing compounds in various systems.

26
In later studies, the involvement of the common nodABC genes as

well as the host specific nodPQ and nodH genes in the production of the
diffusable metabolite from R. meliloti was shown (110-112). At this point
this unidentified component was termed the ”Nod factor”. Using wild-
type R. meliloti strains grown in media containing the inducer luteolin, the
cell-free culture supernatants exhibited root hair deformation (Had)
activity on the homologous host alfalfa but not on the heterologous host
vetch (110). An assay of Had activity, after application of a particular
compound to the root surface, is thought to be indicative of the ability of
that compound to induce nodulation. The sterile filtrate was extracted
with organic solvents and analyzed by high performance liquid
chromatography (HPLC). However, again the problem of insufficient
material for detailed structural determination was encountered. Lerouge,
et al., attacked this problem by genetic overexpression; they introduced a
plasmid into R. meliloti that contained extra copies of the nod genes (113).
The plasmid was comprised of all of the common, specific, and regulatory
nod genes at 5—10 copies per cell. This resulted in a greater than 100-fold
increase in the Had activity of the culture filtrate of this genetically
enhanced organism over that of the wild type R. meliloti (113). This
activity also was not affected by mutations in the common nod genes.
These workers then attempted to purify large quantities of these Nod

factors by introducing the plasmid into a strain deficient for BPS

27
production, presumably to facilitate the filtrate extraction (113). Using

reverse-phase HPLC they were able to isolate approximately 4mg of
purified Nod factors out of ten liters of luteolin induced culture. The
structure of the primary Nod factor was determined utilizing a
combination of mass spectrometry, 1H and 13C-nuclear magnetic
resonance spectroscopy and chemical modification. It was identified as a
tetrasaccharide composed of B—1,4-N-acetyl glucosamine with a sulfate
group at the 6-OH position of the reducing sugar and a hexadecadienoic
acid attached in an amide linkage at the 2-OH position of the non-
reducing residue.

When this paper appeared it was considered to be a breakthrough in
the field of rhizobial research not only because it showed that root
deformation activity was attributable to a characterized molecular species
but also because it demonstrated host specificity. When the authors
applied the Nod factor in a solution (10'3-10'11M) to roots of the
homologous host alfalfa as well as a heterologous host vetch, only the
alfalfa showed Had activity (113). These authors made the claim that this
particular Nod factor was the single molecular determinant of host
specificity in R. meliloti. They did mention, however, the presence of other
"minor" Nod factor components differing in fatty acid distribution. Later,
one of these other components was isolated and identified as an

acetylated derivative of the original Nod factor (114). In fact, it was

28
eventually determined that R. meliloti produces a family of sulfated

chitolipooligosaccharides (CLOS) that exhibit a host specific effect on
plant roots (116). These compounds range in degree of polymerization
from three to five sugar residues and contain a distribution of fatty acid
chain lengths and unsaturations. Therefore, the general consensus in the
field is that the only constant structural element, the sulfate linkage, is the
essential basis for the activity of these molecules. Indeed, it was shown
that hydrolysis of the sulfate group resulted in a strong decrease in the
morphogenic activity of the Nod factor (115) whereas lack of the acetate
group, changes in the number of glucosamine residues or in the fatty acid
distribution resulted in a less pronounced reduction in these activities
(115-117). It was also demonstrated that synthetic R. meliloti CLOS with
or without this sulfate group was active or inactive for Had activity on
clover roots, respectively.

After the isolation of the Nod factors, many studies were aimed at
determining the exact roles of the common and host specific nod genes
which, as discussed above, had been shown to be required for the
production of the alfalfa specific signals. A depiction of the biosynthesis of
Nod factors as well as the roles of some of the more important nod genes
is presented in Figure 1.5. It was first discovered that the host specific
nodP and 11on genes displayed a great deal of homology to the cysD and

cysN genes of E. coli (118). These E. coli genes encode proteins that are

29
involved in the synthesis of adenosine-5'-phosphosulfate (APS) (119). This

ATP sulfurylase activity is the first step to the formation of activated
sulfate which, of course, would be necessary for addition of the sulfate
group to Nod factors that is required for specificity. The function of the
other host specificity gene, nod H, was then proposed by Roche, et al., to
be involved in the transfer of the activated sulfate moiety to the nascent
Nod factors (120). These authors showed that strains with mutations in
the nodH gene produced chitolipooligosaccharides that were nonsulfated
and that these strains also lost their ability to nodulate alfalfa. They then
suggested, using sequence homology with known proteins, that the gene
product of nodH was a sulfotransferase.

The common nod genes, nodA, nod B and nod C, have been
implicated in the assembly of the glycolipid backbone that receives the host
specific modifications. It was first shown that the product of nodC
functioned as a UDP-GlcNAc transferase based on sequence similarity to
chitin synthases (126, 127). This was later supported by experiments
showing that bacteria expressing the N odC protein incorporated GlcNAc
from UDP-GlcNAc into chitin oligomers (128). The NodB protein was
shown to de-acetylate the non-reducing terminal GlcN Ac on chitin
oligomers and also was shown to share sequence homology to proteins

with similar functions (129). Lastly, the product of nodA is thought to be

30

Figure 1.5: Putative pathway for the biosynthesis of
chitolipooligosaccharides in Rhizobium meliloti and the

functions of some of the associated nod factors.

 

31

OH
[1061M HO O
Frc-6-P—> GlCN-6-P -> -> HO OUDP
N80
CH3

nodC 1

OH
NH H
CH QgH 8H
3 2_3 3
\OCOCHa 0303/ nodH, P. Q
NH NH
"0‘“ B\N‘ c: 0 90 CO
CH3 2_3 CH3

R,
nodE, F/

 

nodA
nodB
nodC
nodD
nodE
nodF
nodH
nodL
nodM
nodP
71on

proposed N-acyl transferase (130)

evidence for deacetylase activity and sequence homology (129)
chitin synthase homology, proposed UDP-GlcN Ac transferase (126-128)
family of nod gene transcriptional activators (101, 102)
proposed to synthesize fatty acyl chain, ketoacyl synthase (133)
acyl carrier protein homology (134)

proposed transfer of activated sulfate (120)

homology to acetyl transferase (95, 135)

proposed glucosamine synthase (136)

proposed ATP sulfurylase, involved with non (118, 119)
proposed APS kinase, involved with nodP (118, 119)

Figure 1.5

 

32
an N-acyltransferase. Cells expressing this protein acylate the

deacetylated GlcN Ac oligomers while nodA’ mutants do not (130).

CONCLUSION

For many years, the molecular mechanisms involved in host
specificity and infection in the rhizobial legume symbiosis has been a well-
studied area of research in the field of nitrogen fixation. All of the major
cell-surface carbohydrates and glycolipids of rhizobia have been
implicated to a certain extent in ensuring a viable infection or determining
host range (or both). These components have included the lipopoly-
saccharides, capsular polysaccharides, extracellular polysaccharides,
cyclic B-(1,2)-glucans and most recently chitolipooligosaccharides.
Regardless of the wealth of results involving all of these classes of
molecules, the latter has been heralded as the single determinant of host
specificity. Additionally, one family of genes (nod) is thought to exist for
the exclusive purpose of synthesizing these glycolipids. Mutations in
these genes, which have been shown to cause defects in
chitolipooligosaccharide structure, eliminate the ability of the bacteria to
discriminate the proper host plant. The effect of these mutations on the

other cell-surface components, however, has never been assessed. Given

33
the proven involvement of these molecules in all stages of infection, this

seems quite odd. The experiments presented in this dissertation will
address this oversight as well as attempt to provide a framework for a
new model in which the entire membrane structure is the critical

determinant for infection and host specificity in symbiotic nitrogen

fixation.

CHAPTERZ

COMMON LINKS IN THE STRUCTURE AND CELLULAR
LOCALIZATION OF RHIZOBIUM CHITOLIPOOLIGOSACCHARIDES
AND GENERAL RHIZOBIIUW MEMBRANE PHOSPHOLIPID AND

GLYCOLIPID COMPONENTS

35

ABSTRACT

Several common links between the structural chemistry of the
chitolipooligosaccharides of Rhizobium and the general rhizobial
membrane lipid and lipopolysaccharide chemistry of these bacteria have
been uncovered. Aspects of common chemistry include sulfation,
methylation and the position and extent of fatty acyl chain unsaturation.
It was found that bacteria which are known to synthesize sulfated
chitolipooligosaccharides (such as R. meliloti strains and the broad-host-
range Rhizobium species strain NGR234) also have sulfated
lipopolysaccharides. Their common origins of sulfation have been
demonstrated by using mutants which are known to be impaired in
sulfating their chitolipooligosaccharides. In such cases, there is a
corresponding diminution or complete lack of sulfation of the
lipopolysaccharides. The structural diversity of the fatty acids observed in
the chitolipooligosaccharides is also observed in the other membrane
lipids. For instance, the doubly unsaturated fatty acids which are known
to be predominant components of R. meliloti chitolipooligosaccharides
were also found in the usual phospholipids and glycolipids. Also, the

known functionalization of the chitolipooligosaccharides of R. sp.

36
NGR234 by O- and N-methylation was also reflected in the

lipopolysaccharide of this organism.

The common structural features of chitolipooligosaccharides and
membrane components is consistent with a substantial degree of
biosynthetic overlap and a large degree of cellular, spatial overlap
between these molecules. The latter aspect is clearly demonstrated since it
is shown here that the chitolipooligosaccharides are, in fact, normal
rhizobial membrane components. This increases the importance of
understanding the role of the bacterial cell surface chemistry in the
rhizobia/ legume symbiosis and developing a comprehension of the highly

integrated membrane lipid and glycolipid chemistry of rhizobia.

37

INTRODUCTION

Nod factors are substituted chitolipooligosaccharides (CLOS) that
have been isolated from the cell-free supernatants of rhizobial strains
genetically engineered to overproduce them. This class of molecules has
been proclaimed as the absolute determinant of host specificity in the
rhizobial legume symbiosis. When applied to roots of the homologous
host of the rhizobial strain from which they were isolated, these lipid
molecules effect morphological changes on the root that mimic those seen
in the early stages of an authentic infection. The production of these
glycolipids also seems to be enhanced by molecules made by the plant
(flavones). It is therefore believed that a molecular signaling dialogue
exists that accounts for host specificity (Figure 2.1). The plant produces a
specific flavone that stimulates production of CLOS in the appropriate
bacteria; these glycolipids are excreted from the bacterium, diffuse across
the rhizosphere and are recognized by the distinct host plant root. The
plant responds by beginning to prepare itself for an upcoming symbiotic
event. This model was widely accepted with great enthusiasm by the
majority of the scientists working in this area. However, perhaps due to

the fervor associated with the discovery of the Nod factors, some very

38

COflICGI CO" H
RHIZOSPHERE Dlvlslon / ‘
cn,on cn,on ergo on.
\ o o o o
(— “We 0 o
I“ N" “H “N “o
no I no no . no .
co co co so
a: on, on, on,

 

I
I
SOIL I
I
I I
I
when <L
l l
I

sell
rhizobia nodulation genes
_ are activated by plant
/) \ clonal: and extracellular

 

 

 

 

I
Phenolic I lactou are generated
9..," l pSym
signal} I /I
l / Model Rhizoblum Cell
I \
l ,\ \
Root ‘ \
cap \
LEGUME ROOT

 

Figure 2.1: Diagrammatical representation of the putative
signal exchange in the early events of the rhizobial legume
symbiosis. Flavonoids released by the plant induce the
nodulation genes located on the Sym plasmid to produce
specific chitolipooligosaccharides which begin the cascade of
morphological events leading to a successful infection.

Figure taken from reference 10.

39

major inconsistencies regarding this model have been overlooked. Some
of these oversights are addressed by the experiments presented in this
chapter.

A major conceptual problem with the Nod factor model lies in the
notion of scale and the concentration of these molecules needed to bring
about a reaction. The chitolipooligosaccharides have not been isolated
from wild type cells in any substantial quantities, rather only through
methods of genetic augmentation. Between five and ten extra copies of
the entire set of nod genes along with additional regulatory genes were
included in these strains (113). This resulted in more than a 100-fold
increase in root hair deformation activity. It seems odd that a molecule so
important to the assurance of a viable symbiosis is not produced by a wild
type bacteria in an amount adequate for isolation. This is especially
strange considering the fact that this molecule must be made in quantities
sufficient for diffusion throughout the rhizosphere in the soil in order to
signal the plant from a distance. The original manuscript on the isolation
of CLOS reported a yield of four milligrams of purified glycolipid from
ten liters of induced culture (113). This gives a concentration in the high
nanomolar range; this is the same range that these molecules have been
shown to exhibit their mitogenic effect on the root cortex (115). It seems
ridiculous, therefore, to envision the same density of cells in the vicinity of

a plant root in the wild as would be found in a liquid culture. In addition,

4O
remembering that the CLOS has to diffuse through the soil and of course

that the lab strains are genetically supercharged makes this proposal all
the more absurd.

A definitive experiment to perform in order to prove that CLOS is
indeed the exclusive determinant for host specificity would be to
exogenously add the appropriate chitolipooligosaccharide to a Nod'l Fix“
strain, i.e., one which fails to produce the correct CLOS and hence fails to
nodulate the proper host. If the theory were correct, one would expect
that the presence of the exogenous CLOS would restore the ability of
these bacterial mutants to infect. This experiment has been tried in
various laboratories using innumerable methodologies and to date has
not been successful. This raises serious questions as to the strict
exclusivity of the chitolipooligosaccharides in host specificity.

Another confounding aspect of the Nod factor model is the
tremendous variability in chitolipooligosaccharide structure, even in the
same strain. The first studies on R. meliloti CLOS promoted the idea that
each rhizobial species synthesized, almost exclusively, only one molecule
with no variation in carbohydrate or fatty acid chain length, degree of
sulfation or acetylation, or position or extent of alkyl chain unsaturation
(113). Subsequent studies, however, have demonstrated that any given
bacterial species, including R. meliloti, can produce CLOS containing a

variety of different fatty acids. These glycolipids may also contain

41

between zero and four double bonds in their acyl chains, differing degrees
of sulfation, varying numbers of glucosaminyl residues and various
substituents ranging from methyl and carbamoyl to entire glycosyl groups
(Figure 2.2). In one case, 18 distinct chitolipooligosaccharides were
identified or characterized in one bacterial strain (138). The structural
diversity of CLOS is such that the relatively sparse number of nod genes
cannot account for it. The only location in the cell, known to contain the
vast repository of enzymatic activities capable of producing such a myriad
of structures is the membrane. In order to use this machinery during their
biosynthesis, chitolipooligosaccharides would have to be membrane
localized. If this were true, the structural modifications one finds in
CLOS should also be found in the other membrane components of that
strain. If these lipids were membrane localized, the more complete picture
that emerges would have the bacterial cell surface and membrane
chemistry as the critical factors in determining host range and the

outcome of a potential symbiotic event between rhizobia and legumes.

The importance of rhizobial cell-surface chemistry to various stages
of the symbiotic nitrogen-fixing process between these bacteria and
legumes was delineated in Chapter 1. The lipopolysaccharides (LPS) of
rhizobia have been implicated in various stages of the symbiotic nitrogen-

fixing process between these bacteria and legumes. These stages have

42

Figure 2.2: The major reported structures of
chitolipooligosaccharides produced by various
rhizobial strains. Upper: The N-acetyl glucosamine
oligosaccharide backbone. Lower: Various

substituents found in different strains.

 

43

 

 

 

 

R3 OH R6
R5 0 o
R 4 HO HO OR7
’NRZ H H
R1 0=< 0:
_ CH3 — H CH3

Strain R1 R2 & R4/R5 R5, R7 n Ref.
R. meliloti

”wild type” 16:2 H H/O-acetyl H 5031-1 H 2,3 113

nodPQ‘ 16:2 H H /O-acetyl H H/SOgH H 2 ,3 ”

nodH' 16:2 H H /O-a cetyl H H H 2,3 ”

AK41 16:3 H H/O-acetyl H 503H H 1-3 116
Rhizobium sp. 16:0/ 18:1 Me H/carb H / carb 2-O-Me-()—Fuc H 3 138
strain NGR234 (3-O-acetyl)

(4-O-sulfa tyl)

R. viciae

”wild type” 18:4/18:1 H O-acetyl H H H 2,3 95

nodL‘ 18:4/18:1 H H H H H 2,3 ”

nodE" 18:1 H O-acetyl H H H 2,3 ”
B. japonicum

USDA 110 18:1 H H H 2-O-Me-Fuc H 139

USDA 135 16:0/ 16:1 H H /O-acety1 H 2-O-Me—Fuc H 140

/18:1
USDA 61 18:1 H/ H /O-acetyl H/carb 2—O-Me-Fuc/ gly 2,3 140

Me = methyl; carb = carbamoy1;Fuc = fucosy1;g1y = glycerol

Me

Figure 2.2

Fuc

44

included competitiveness for nodulation (76), infection thread
development (77), bacterial release from the infection thread (27),
avoidance of the plant defense mechanisms (44) and the development of
complete nitrogen fixing nodules (22, 26, 29, 78). A definitive involvement
of capsular polysaccharide biosynthesis and structure is also well
established. Specific mutations of bacterial genes involved in the various
stages of capsular polysaccharide biosynthesis show a high correlation
with symbiotic deficiencies in nodulation (33). The general consensus on
the role of the EPS in infection is that it ensures bacterial release from the
infection thread and nodule invasion (45, 79, 80, 81). The cell surface
components of rhizobia, then, have always been considered as essential in
the very specific interaction of these bacteria with legumes. This
interaction is known to be controlled from the bacterial side by the host
specific nod genes. These genes, along with the common nod genes which
are essential for nodulation in all rhizobial species, have been shown to be
involved in the synthesis of CLOS. In the case of R. meliloti and
Rhizobium species strain NGR234 these molecules contain a sulfate group
that has been shown to be critical to host specificity (113, 114, 138). In R.
meliloti, two of the more important host specific genes, nodPQ and nodH,
are known to be involved in the sulfation of the CLOS (118, 120). The
function of nodPQ in the biosynthesis of the activated sulfate donor has

been clearly demonstrated (118, 141). The nodH gene has been shown to

45

be critical in determining host specificity and has also been implicated in
transfer of the activated sulfate to the Nod factor in R. meliloti (120).

The experiments presented in this chapter were aimed at
determining what, if any, structural relationship might exist between
rhizobial chitolipooligosaccharides and other more general membrane
components. The main structural feature that was targeted was the
sulfate ester on the chitolipooligosaccharide found in R. meliloti since this
was thought to be the crucial element in recognition and also because
strains carrying mutations in the putative genes involved (nodPQ and
nodH) were readily available. Another important goal in these
experiments was to show that these lipids are in fact membrane localized

as opposed to being specifically excreted from the cell.

46

MATERIALS AND METHODS

Strains and bacterial cultures.

R. meliloti 1021 mutants were obtained from Dr. Sharon Long
(Stanford University, CA). All bacterial strains were grown in liquid
cultures using modified Bergensen‘s media as previously described (142)
except that 35S-labelled sulfate in the form of sulfuric acid (New England
Nuclear) at a level of 100 uCi/ liter of culture was added before adjusting
the pH of the medium to neutrality. This medium contains sulfate as the

only form of sulfur.

Nuclear magnetic resonance spectroscopy and mass spectrometry.
Nuclear magnetic resonance (NMR) spectra were recorded on a
Varian VXR-SOO spectrometer operating at 500 MHz for protons. Spectra
were acquired in deuterium oxide containing 10% deuterated methanol
with chemical shifts being referenced relative to external TMS. Gas
chromatography / mass spectrometry (GC/ MS) was performed on a
JEOL 505 mass spectrometer interfaced to a Hewlett Packard 5890 gas
chromatograph. Spectra were recorded in the positive ion mode using
electron impact with an ionization voltage of 70eV. For the analyses of

carbohydrate derivatives, a 30-meter capillary column (I 8: W, D3225)

47

was used with a temperature program of 180 to 230°C at 3°C per minute
and a 30 minute hold at the upper temperature. Fast atom bombardment
mass spectrometry (FAB-MS) was recorded on a JEOL 110-HX-HF
instrument in both the negative and positive ion modes. The matrix used
for CLOS analysis (performed in the negative ion mode) was 1:1
glycerol/4—nitrobenzyl alcohol. For phospholipid analyses, the matrices
were 4-nitrobenzy1 alcohol / 15-crown-5 / camphorsulfonic acid for positive
ion mode and 4-nitrobenzyl alcohol/15-crown-5/tert-butyl ammonium
hydroxide for negative ion mode. Electrospray mass spectrometry was
performed on a Fisons Platform I instrument operating in the negative ion
mode. The sample was infused into the instrument in 1:1 acetonitrile

water.

Lipopolysaccharide isolation.

Lipopolysaccharide (LPS) was isolated and purified by gel filtration
chromatography on Sepharose 48 columns as previously described (22,
143). Column effluents were monitored for radioactivity by liquid
scintillation counting and for carbohydrate content by the phenol/ sulfuric
acid method (149). The purity of LPS preparations was established by
compositional analysis for fatty acids and carbohydrates by GC and
GC/ MS (see below). Sodium dodecylsulfate polyacrylamide gel

electrophoresis (SDS-PAGE) was performed using 16% acrylamide gels.

48
LPS on gels was detected both by silver staining and by exposure of the

dried, 2,5-diphenyloxazole-impregnated gels to Kodak x-omat x-ray

plates at -75°C for 4 days.

Chitolipooligosaccharide isolation and purification.

Two liters of broth culture were typically grown for CLOS
isolation. Cells were harvested in late stationary phase by centrifugation
and lipids in the cell pellet were extracted by stirring either with a mixture
of chloroform, n-butanol, methanol and water (221:1:4) or chloroform,
methanol, water (4:15). The organic layers were removed and the
aqueous layers were concentrated to dryness and then redissolved in 5 ml
of water and adsorbed onto a short column of C-8 or C-18 reverse phase
packing (5cm X 2.5 cm). The column was eluted sequentially with 30 ml
each of water, 4:1 water / methanol, 1:1 methanol / water, 4:1 methanol /
water and pure methanol. Fractions were analyzed by proton NMR
spectroscopy; those containing the characteristic N-acetyl methyl signals
at about 2 ppm were subjected to further purification by reverse phase
high performance liquid chromatography on a C-8 reverse phase column
using ultraviolet (UV) detection and monitoring at 220 nm. A linear
solvent gradient from 10% acetonitrile in water to 80% acetonitrile in
water was used for the separation. The chitolipooligosaccharides could

also be purified by gel filtration on Biogel P6 using acetonitrile/ water (1:1)

49

as the solvent. Peaks were analyzed for glucosamine content by gas
chromatography while radiolabelled samples were also analyzed by
scintillation counting. The typical yield of product from 2 liters of cells
was 8-12 mg and was not significantly affected by the addition of flavone
(luteolin). Products were analyzed by NMR spectroscopy and

electrospray mass spectrometry.

Compositional and linkage analysis of chitolipooligosaccharides.

The carbohydrate compositions of chitolipooligosaccharides were
determined by hydrolysis of approximately 1 mg sample with 2M
trifluoroacetic acid (1 ml) at 120°C for 2 hours followed by extraction of
the free fatty acids from the monosaccharides with chloroform. The
aqueous layer was then concentrated to dryness under a stream of
nitrogen and the product redissolved in water and reduced to alditols
using sodium borohydride (1 mg). The excess borohydride was
decomposed by treatment with 30 uL of acetic acid and the resulting
solution concentrated to dryness. Methanol (2 ml) containing 2% acetic
acid was then added and the solution concentrated to dryness again. This
process of adding acetic acid in methanol and concentrating was repeated
5 times to remove all traces of boric acid as the volatile methyl ester. The
residue was finally peracetylated by treatment with pyridine (0.2 ml) and

acetic anhydride (0.2 ml) at room temperature for 24 hours. The reagents

50

were removed under a stream of nitrogen and the residue was analyzed
by GC/ MS. Alternatively, both the fatty acid and the carbohydrate
compositions were determined simultaneously by methanolysis of a
sample of lipooligosaccharide (0.5 mg) with 3% HCl in methanol at 70°C
for 24 hours, peracetylation of the mixture with acetic anhydride and
pyridine and performing GC and GC/MS analyses directly on the
mixture using the same column conditions as described above.
Methylation analysis was performed using silver oxide as the base and
methyl iodide as the methylating agent in dimethylformamide (DMF) as
the solvent. A sample of lipooligosaccharide was dissolved in dry DMF (1
ml) and methyl iodide (0.2 ml) and silver oxide (50mg) were added. The
mixture was stirred at room temperature for 36 hours and then diluted
with chloroform (2 ml) and filtered. The filtrate was concentrated to
dryness and then the residue was hydrolyzed, reduced with sodium
borohydride, peracetylated and analyzed by GC and GC/MS as
described above. Analyses for KDO were performed by the thiobarbituric

acid assay (148).

Phospholipid isolation and characterization.
Phospholipids were separated and isolated by thin layer
chromatography (TLC) on silica gel using a solvent system composed of

chloroform, acetone, methanol, acetic acid and water (10:4:2:2:1, vzv).

51

Bands were detected by removing the edges of the plates and spraying
them with 5% phosphomolybdic acid followed by heating at 120 °C for 5
minutes. The bands were removed from the layers by scraping and eluting
with chloroform/ methanol (2:1). The components thus isolated were
analyzed by FAB-MS in both the positive and negative ion modes. The

fatty acids were analyzed by GC and GC/ MS after methanolysis.

52

RESULTS AND DISCUSSION

The Sepharose 43 gel filtration chromatogram of the LPS of R.
meliloti 2011 showed it to be labelled by 358 (Figures 2.3 and 2.4) in a
flavone-independent manner. The ratio of 358 counts to carbohydrate
absorbance units for the LPS peak (fractions 45-60) in induced and non-
induced cultures is approximately the same in each case. The addition of
luteolin does have some (albeit slight) effects on the size distribution of the
lipopolysaccharide as well as the extent of membrane lipid synthesis and
cell encapsulation. This latter point is reflected in the reduction of
carbohydrate content in the fractions preceding those of the LPS. These
components have been described (42) as KDO-rich capsular antigens
related to the E. coli K antigen and are structurally distinct from LPS. The
reduction in amount observed when the cells are grown in the presence of
luteolin is not surprising since luteolin is, in fact, bactericidal at higher
concentrations and should provoke stress responses from the bacteria
even in moderate concentrations. In order to reinforce these results and
verify that the label was attached to the LPS and not free, a sample of the
LPS was subjected to sodium dodecyl sulfate-polyacrylamide gel
electrophoresis (SDS-PAGE) in parallel with a similar preparation from

R. trifolii ANU843. The gels were examined by silver staining and by

53

Figure 2.3: Sepharose 43 gel filtration profile of the crude
phenol extract of 35S-labelled R. meliloti 2011 cells grown in
the absence of luteolin. Fractions were assayed for
radioactivity by scintillation counting and for total
carbohydrate. The second eluting peak (fractions 35-60) is
due to LPS; note the presence of radiolabel. The first eluting
peak (fraction 28-34) is due largely to a high molecular weight
capsular polysaccharide. There is, however, a small amount
of a coincident very high molecular weight LPS form. Note
the presence of radioactivity in these components. The late
eluting peak (fractions 68 to 82) contains free sulfate and a {3-
1,2-glucan as well as low molecular weight sulfur containing

metabolites.

54

 

 

 

 

 

 

 

 

__¢m__m_.counts per min. 4
total carbohydrate "
A
3 . no luteolin ~ 0.06
2 - 0.04
counts
per Abs.
min. ' g 490 nm
(310001.
1 }\~/\J - 0.02
0 “*‘A_:"
10 20 30 100

FRACTION NUMBER

Figure 2.3

55

Figure 2.4: Sepharose 4B gel filtration profile of the crude
phenol extract of 35S-labelled R. meliloti 2011 cells grown in
the presence of luteolin. Note that the LPS is still
radiolabelled and that the level of radiolabelling
(counts/absorbance) is essentially the same as that observed
in the absence of luteolin. Note that the presence of luteolin
also has an effect on the extent of encapsulation as evidenced

by the decrease in intensity of the first eluting peak.

56

 

 

 

 

 

12I __. _____ .___ counts per min. ~ 0‘3
_____._.._ total carbohydrate. I
a II
??
+ luteolin I I -

8l : _ 0.2
323““ I I Abs.
min. I a 490 nm
(x1000) . '

[I 4 .1 \. .. 0'1

I l.
r\. J x
O A AMA/mfg“ M," ' \mw‘i

 

10 26 30 4b 56 60 7o 80 90 100
FRACTION NUMBER

Figure 2.4

57
autoradiography. These analyses also indicated that the LPS of R.

meliloti was radiolabelled and that the level of incorporation of label was
independent of flavone (Figure 2.5). In contrast, the R. trifolii LPS was
found to contain no sulfur. These results also reflected the
aforementioned changes in molecular weight distribution due to the
presence of the flavone. Silver-stained gels reflected exactly what is
portrayed in the autoradiograms. In SDS-PAGE, the LPS of wild type
rhizobial strains generally appears as only one major band with a smaller
amount of a faster moving component which is thought to be rough LPS.
The form of the sulfur containing species was definitively identified as
sulfate by liberating it from a Sepharose 4B purified LPS sample by mild
acid hydrolysis and precipitating the free sulfate as barium sulfate. N o
radioactivity was detected in the solution containing the carbohydrate
components. Based on the level of radioactivity and the amount of LPS
isolated (assuming a molecular weight of about 30,000 daltons for a
smooth LPS) a conservative estimate of 1.3 sulfate groups per LPS
molecule was obtained. It should be noted that because of the extreme
lability of sulfate groups, this estimate is likely to be lower than the actual
value since desulfation is likely to have occurred during the isolation and
purification processes. Both UV and SDS-PAGE analyses indicated that
there was no protein contamination in the LPS preparations. In addition,

free sulfate was not present since it would, of course, be separated from

58

Figure 2.5: Autoradiograph of a sodium dodecylsulfate gel
showing the separation of 35S-labelled LPS isolated from R.
trifolii ANU843 and R. meliloti 2011 in the presence and
absence of flavone inducers. The samples for each condition
were applied in triplicate using decreasing amounts (3:2:1) for
each lane. Note that the R. meliloti LPS is radiolabelled
independently of flavone (luteolin). Note also that there is no
incorporation of label in the R. trifolii LPS. Only a slight shift
of one of the minor bands is observed for the R. meliloti LPS
isolated from cells grown in the presence of flavone
compared to the result obtained for cells grown in the absence

of flavone.

59

 

Figure 2.5

60

the polysaccharide during the chromatography and would also run with
the ion front in the SDS gel. The LPS carbohydrate and fatty acid
compositions of R. meliloti 2011 have been reported before (12, 143).

The effect(s) of impairment of the nod Q1 and 11on2 genes on the
sulfation of the LPS was next examined. The nonl and non2 mutants
used in the study nodulate more slowly than the parent strain R. meliloti
1021 while the 11on] Q2 double mutant is ineffective in nodulation (141).
These mutant strains were all derived from R. meliloti 1021 by Tn5
insertion. Assaying fractions of the Sepharose 4B gel filtration column of
the parent strain clearly showed that the LPS was sulfated (Figure 2.6,
fractions 10-15). A similar analysis on the nonI mutant (ISS12) showed
a decrease in the level of sulfation of the LPS in comparable fractions. The
same result was observed with the LPS of the nonZ mutant (JSSl4).
However, this same LPS region in the profile of the nonl Q2 double
mutant (JSS16) contained no sulfate. The effect of mutation of the nodH
gene was examined in a similar manner. This mutant displays a
phenotype identical to the nonl Q2 double mutant; it produces a CLOS
devoid of sulfate and is ineffective in nodulation of the homologous host
alfalfa. This strain was also derived as an insertion mutant of R. meliloti
1021. Inspection of the Sepharose 4B gel filtration profile revealed the
absence of sulfate in the same LPS region as the nonI Q2 mutant (Figure

2.7).

61

' Figure 2.6: Sepharose 43 gel filtration profiles of LPS
isolated from 35S-labelled R. meliloti 1021 cells and three
11on mutants assaying for total carbohydrate and
radioactivity. The small, earlier eluting peak lacking
radioactivity (fractions 7-10 in the first panel) has been
attributed to a capsular polysaccharide related to the E. coli K
antigen (42). Loss of sulfate groups on the LPS of all three
mutants results in increased levels of synthesis of the non-
sulfated protective capsular polysaccharide. (Note: A late
eluting peak of totally included free radiolabelled sulfate and

B-1,2-glucan has been omitted for clarity.)

62

III 369.055 330

III. 32332323

 

 

 

 

6
cu _
2
_ .
. 0
I o 2
2 O
. O
O. N
.. n A... ...... a
.0. C M
R
«4.. F M. .l
S “I S _ 0
m» ._o ..o J. A
n..
.o. II...IOImm< ....I.O.I. .o..l0nl mm< I
.IrllfI Amocouaocsiao III Auucaoaocs—zao
5. 5
_ . .,
_ .
l0
2
o O
n m
m .. m
R 4 R
F 1
m .s F
s
W _ . J. _ __l
2. . l. o

...I|e_I.....IeIII mm<

 

....I.OI.. ....IOI. mm<

Figure 2.6

 

I
A
~ CPMIthouaanda) ”H‘-

Aes —---—-~

 

 

0" 00.0.0.0;

FRAC. NO-

Figure 2.7: Sepharose 4B gel filtration profile of material
isolated from 35S-labelled R. meliloti 1021 cells carrying a Tn5
mutation in the nodH gene assayed for total carbohydrate

and radioactivity.

64

These results indicate that the current models explaining the
molecular basis of host specificity should be revised in order to factor out
the role(s) of LPS from those of the glucosamine-containing
lipooligosaccharides. This work shows unequivocally that the
lipopolysaccharide is a substrate for enzymes encoded for by nod genes.
The fact that the sulfation of LPS is flavone independent but the non and
nodH genes are (supposedly) flavone regulated requires explanation.

In order to investigate further the relationship between
chitolipooligosaccharide and typical membrane lipid and glycolipid
synthesis, the promiscuous Rhizobium species strain NGR234 was also
examined. This strain nodulates at least 70 different genera of legumes
and has been shown to synthesize chitolipooligosaccharides bearing a
wide assortment of functionalities (See Figure 2.2). The most notable
features of the NGR234 CLOS is the presence of a methylated 6-
deoxyhexosyl residue on the reducing end (that also may or may not
contain a sulfate moiety) and the methylation of nitrogen on glucosamine
(138). R. sp. NGR234 was grown in the presence of 35S-labelled sulfate
and the LPS isolated and purified. Assays of the Sepharose 4B column
eluate showed radioactivity coinciding with the marker LPS component 2-
keto-3-deoxyoctulosonic acid (KDO) (Figure 2.8). This fraction was
pooled and rechromatographed on Biogel P6 from which it eluted as one

radioactive fraction which tested positive for LPS by the KDO assay. The

 

 

 

 

0.015 ‘ 40°
—u— K00
—-O— DPM
~300
0.010 -
g -2... E
:4 ‘ O
0.005 - '
' I- 100
0.000 - fl - 1 ' °
10 15 20 25

Fraction Number

Figure 2.8: Sepharose 4B gel filtration profile of extract from
35S—labelled R. sp NGR234 cells showing the correspondence
of radioactivity to the LPS marker component KDO
(measured in absorbance units). The late eluting peak of
radioactivity is due to free sulfate and other small sulfur-

containing substituents.

66
peak was analyzed by GC and GC/ MS to determine its carbohydrate

content. Like the chitolipooligosaccharide isolated from this strain, the
lipopolysaccharide was also sulfated and both 0- and N-methylated. The
GC / MS analyses indicated that, like the CLOS, a 2-O-methyl-6-
deoxyhexose was a component of the LPS. In addition, methylation of the
nitrogen of a 2-amino-2,6-dideoxyhexose component was also observed.
The predominant glycosyl component was glucose.

The fatty acid components of rhizobial CLOS are often cited as
being the primary determinants of their ability to induce host specific
responses in legume plants. In the case of R. meliloti, for instance, much
importance has been placed on the presence of a 2,9-doubly unsaturated
C-16 fatty acid in which one of the double bonds is conjugated to the
carbonyl group and has a trans-configuration (113, 114). These
unsaturated fatty acids are still thought to be critical even though it was
also demonstrated that there were several other fatty acid components,
including 0c,f3,y,8- chains, and saturated chains of varying lengths, in
chitolipooligosaccharides from R. meliloti (116). The fatty acid
composition of the membrane lipids and the number and stereochemistry
of the unsaturated linkages therefore required special attention. The
modifications of the membrane lipids were generally similar to those
reported for the chitolipooligosaccharides. The presence of C18:2 fatty

acids could be discerned by GC and GC/ MS analyses. A total ion

67

chromatogram from a representative membrane lipid fatty acid methyl
ester analysis of R. meliloti 2011 is shown in Figure 2.9 and a list of the
components is given in Table 2.1. The high incidence of methoxy fatty
acids can be attributed to artifacts due to the addition of methanol to
unsaturations and to methanolysis of cyclopropane groups.

The NMR spectra of total membrane lipids gave definitive
information on the types of fatty acids which were present in the total
lipid extracts of the bacteria especially with regard to the number and
types of double bonds. Hence in the case of R. meliloti 2011, the proton
spectra (Figure 2.10A) and the 13C NMR spectra (Figure 2.10B) both
indicated that unsaturated fatty acids were major components of the
phospholipids. Unsaturation was indicated in the proton spectrum by the
presence of signals downfield of about 5 ppm and in the 13C spectrum by
signals between 130 and 150 ppm. One common site of unsaturation was
0013- to the ester carbonyl group. This was indicated by the presence of a
doublet (I = 12 Hz) at 6.10 ppm, attributable to the a—proton and a doublet
of triplets (J = 12 + 7 Hz) at 6.88 Hz attributable to the B-proton. The
proton-13C multiquantum coherence spectrum (HMQC) confirmed the
assignments for the vinylic protons and carbons (Figure 2.11). Hence the
signal assigned to the B-proton of the 0013- unsaturated system showed a
connectivity to a 13C resonance at the characteristic position of 149.6 ppm.

The signal at 6.10 in the proton spectrum (assigned to the a-proton)

68

Figure 2.9: Total ion chromatogram showing the typical fatty
acid species in membrane lipids from R. meliloti 2011
derivatized as methyl esters (see Table 2.1 for assignments).
Note the high occurrence of bisunsaturated fatty acids. The
large number of isomers could arise from isomerization
during the derivatization. Note also the frequency of
methoxylated methyl esters. These are expected to arise from

addition of methanol to unsaturations and scission of

cyclopropyl rings.

.02 50m

 

 

 

 

 

 

 

 

 

 

 

r . £2 8.8 82 8?. 8n
I<I 1 4 2'43le O
VN HN c D. ‘ fl; an. m vm N H ..
cN 5H w< A: w .
mm nu mu 1 cm
r
NM 1
w.“
am 1 0v
9 .
6 .
u on
. Ow
Q v
2 C
9. om ow om . 8.

HM

HGHOGmoHH>u

Mommam>o

Figure 2.9

70
Table 2.1: List of fatty acids normally present in the lipids of

R. meliloti membranes.

 

Peak # Molecular
gfiig. 6) Structure Weight—

1 Unidentified

2 CH300C(CH2)7COOCH3 216
3 OHC(CH2)9COOCH3 214
4 CH300C(CH2)8COOCH3 230
5 CH3(CH;_)12COOCH3, 14:0 242
6 CH300C(CH2)9COOCH3 244
7 Unidentified

8 CH3(CH2)XCH=CH(CH2),COOCH3 (x+y=12), 16:1 . 268
9 CH3(CH2)14COOCH3, 16:0 270

CH2
10 / \ 282
CH3(CH2),‘CH-CH(CH2),COOCH3 (x+y=12), 17:0A
11 CH3(CH2)15COOCH3, 17:0 284
12 CH3(CH2)XCH=CH(CH2),COOCH3 (x+y=14), 18:1 296
13 CH3(CH2)15COOCH3, 18:0 298
14 Unidentified 310
15 CH3(CH2),CH=CH(CH2),CH=CH(CH2)ZCOOCH3 294
(x+y+z=12), 18:2
16 Same as above 294
17 Same as above 294
18 Same as above 294
CHz
19 / \ 310
CH3(CH2)xCH-CH(CH2),COOCH3 (x+y=14), 19:0A

20 Methoxylated FANIE

21 Methoxylated FAME

22 Methoxylated FAME

23 Unsaturated keto FAME 310
2A Methoxylated FAME

25 Unidentified

71

Figure 2.10: Proton NMR spectrum (A) of the total
membrane extract from R. meliloti 2011. Note the presence
of a sharp singlet at 3.22 ppm due to methyl protons on the
choline headgroup. The methylene signals appear at

approximately 3.61 and 4.25 ppm. 13C NMR spectrum (B) of

the total membrane extract from R. meliloti 2011.

72

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.1141

 

 

I ‘l I r_]' ffi’ I fi lit—'1'"! I "'U"1 W l’ I 1"1'“l"'|"'[-'I"‘|"T"1"|“'I""t"'F”!*|—T"l'—'l

I
7 6 5 4 3 2 1 ’0 09m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figures 2.10

73

 

 

 

 

L
‘ no
-N
.4
. .-
I"
I.
o
. 0 ~53
<~<®> 02 _
ob ‘9 .
l L“
OD I3
r A
~ 5°:
L2:
.4
n
—v
H
F
O
00 #3
I II IT I
i; .1. I. 1. I. .. o .,
g a; o 0 l~ F 0 a

P2

Figure 2.11: 1H-“C multiple quantum coherence spectrum
(HMQC) of the total membrane extract from R. meliloti 2011.
The degree and type of unsaturations is the same as reported
for the CLOS of this strain. Connectivities are explained in

the text.

74
displayed a connectivity to a signal at 131.5 in the 13C spectrum. The

HMQC spectrum also showed a connectivity between a signal at 5.32
ppm in the proton dimension and 130.8 ppm in the 13C dimension. This
was assigned to the presence of an isolated double bond. The NMR
spectra confirmed the TLC and mass spectrometric results indicating that
the predominant lipid component was phosphatidyl choline as evidenced
by the characteristic signals for its headgroup (Figure 2.10A). The
information obtained from NMR spectroscopy on the extent, position and
type of unsaturation was especially important since it cannot be obtained
by mass spectrometric analysis of fatty acid methyl esters. One
complication in the latter type of analyses is that 0t,B-unsaturated esters
easily add nucleophiles such as alcohols and hydroxide ions under acid or
base catalysis and a large proportion of such fatty acids is lost during their
liberation.

Since LPS, CLOS and other glycolipids are likely to share similar
cellular localizations (spatial overlap), the possibility that other rhizobial
glycolipids might have similar types of fatty acids as were reported for the
chitolipooligosaccharides was next examined. One such glycolipid is
sulfoquinovosyl diacylglycerol (144). Hence the proton NMR spectrum of
sulfoquinovosyl diacylglycerol from non mutants contained signals
between 6 and 8 ppm indicating that 0013,78 unsaturated fatty acids were

present (Figure 2.12).

75

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ ‘I V 1 V V ‘ ‘ V V I V V V T V V V V V V V V V Y

 

1
.1
Cl
.1
.1
.4
d
.1

i

Figure 2.12: Proton NMR spectrum of sulfoquinovosyl
diacylglycerol from mutant JSS16. Note the very high level of
fatty acid unsaturation giving rise to the signals downfield of
about 5.0 ppm. This is very unlike the situation with the
parent strain again underlining the high degree of coupling

and interdependence of membrane lipid chemistry.

76

Finally, the last goal was to obtain more direct evidence that
chitolipooligosaccharides were actually membrane components as the
above studies indicate, and not really excreted metabolites. If this were
true, it would then be possible to obtain substantial quantities from cells
after disrupting the membrane structure. This was accomplished by
extracting cells with an organic solvent mixture to remove phospholipids
and disrupt the membrane thus releasing the more polar lipids (such as
chitolipooligosaccharides) which, along with the lipopolysaccharides,
should partition into the more aqueous layer. The components of the
aqueous layer were absorbed onto a C-18 reverse phase cartridge. The
cartridge was then eluted with an increasingly non-polar mixture of
water and methanol. Using radioactivity counts (from 35S) and
glucosamine content as an index, further purification yielded material
which contained glucosamine as the only carbohydrate component.
Methylation analysis under the mild silver oxide conditions indicated a
1,4-linkage between the glucosamine residues. There was a minor
component with a different methylation pattern which was attributed to
the non reducing terminal. Under these conditions, the reducing terminal
is expected be methylated at the 1-position and suffer some degree of
desulfation to give the same methylation pattern as the internal residues.
The proton NMR spectra of the chitolipooligosaccharides obtained from

Rhizobium meliloti strains 2011 grown in the absence of flavone is shown

77

in Figure 2.13. The level of unsaturation of the fatty acids in
chitolipooligosaccharides from bacteria grown in the absence of flavone
was generally lower than when the bacteria were grown in the presence
of flavone. There was also a significant difference between the levels of
unsaturation of the fatty acids of the 2011 and 1021 lipooligosaccharides.
The latter tended to have a lower degree of unsaturation. The
electrospray mass spectrum of the CLOS from strain 2011 grown in the
absence of luteolin is shown in Figure 2.14. Trisaccharide species seem to
predominate. The presence of a sulfate group was confirmed by
radiolabelling with 35S-labelled sulfate. Typically, several milligrams of
chitolipooligosaccharides were obtained per liter of culture. Significant

(though lower) quantities were obtained in the absence of flavone.

78

 

 

 

 

 

.21“ . _  K...

Figure 2.13: Proton NMR spectrum of chitolipooligo-
saccharide fraction from R. meliloti 2011 membranes of cells
grown in the absence of flavone. The spectrum was
measured in D20 at 50°C. The anomeric protons can be seen
at the edge of the residual water line at 4.60 ppm. Note the
prominent acetyl group signals at ~2 ppm. The fatty acid

signals appear between 1.1 and 1.6 ppm.

79

Figure 2.14: Electrospray (negative ion) mass spectrum of a
chitolipooligosaccharide preparation from R. meliloti 2011.
The complexity of peaks indicates a high level of
heterogeneity due to differences in substitution. The peak at
m/z 1197 corresponds to the M-H ion of an unacetylated
tetrasaccharide species bearing a sulfate group, a
octadecanoyl fatty acid chain and a sodium adduct. Lower
masses correspond to species with varieties of sulfation, fatty

acid type, acetylation, and carbohydrate chain length.

80

100- 1047.4

705.6 937.1

   

1083.2 “97-“

     

641.6 674.0 ‘

 

 

1 1

IIIII IHIIIIIJ IIIIII “III “III IIIMIIlIn. 1111M 1411111011 '1 1.1111111 1.11.. 111‘. 1I11.I11111.111i:11111

Dale

 

.n 60 7M 70 80! 80 ‘ll 90 1000 1050 1100 1150 1200|1250 1300 1350 I400 1450

Figure 2.14

81

CONCLUSIONS

The results of this study clearly demonstrate biochemical and
spatial links between the chitolipooligosaccharides of rhizobia and the
other membrane lipid and glycolipid components of these organisms.
There are important parallels between normal lipid structure and
chitolipooligosaccharide structure especially with respect to unsaturation.
The important feature of chitolipooligosaccharide sulfation is that it is
coupled to lipopolysaccharide sulfation. Chitolipooligosaccharides
themselves are membrane components. It has been suggested that the
unsaturated fatty acids, which eventually become part of
chitolipooligosaccharide structure, are first put on phospholipids and then
transferred to the chitolipooligosaccharide (145). Whereas this would
have been quite remarkable if chitolipooligosaccharides were synthesized
and excreted as was first thought, this overlap in fatty acid structure is
hardly surprising in light of the evidence presented here that these
molecules are actually membrane components. The shuffling and transfer
of fatty acids between membrane lipid components is a well documented
phenomenon (146, 147) as are common modifications, such as
hydroxylation, desaturation and methylation, of fatty acids from different

lipid species in the same membrane. The nod genes, therefore, encode

82

functions that clearly affect and determine bacterial membrane and
general surface chemistry. In light of this finding, it is reasonable to
conclude that bacterial membrane and surface chemistry are critical
determinants of the outcome of the symbiotic relationship between

rhizobia and legumes.

83

ACKNOWLEDGMENTS

I would like to thank Dr. Sharon Long for providing the bacterial
mutants used in this study. The analyses of the total lipid extracts of R.
meliloti were performed by Jeongrim Lee. Much of the work on the
non’ mutants was done by Kerry Ross. The NMR studies in this chapter
were performed by Rawle Hollingsworth. This work was supported by
grant #DE-FGOZ-89ER 14029 from the US. Department of Energy to
Rawle Hollingsworth. The GC/ MS and FAB mass spectral data were
obtained at the Michigan State University Mass Spectrometry Facility
which is supported in part by grant #DRR-00480 from the Biotechnology
Research Technology Program, National Center for Research Resources,
NIH. The NMR data were obtained on instrumentation that was
purchased in part with funds from NIH grant #1-SlO-RRO4750, NSF

grant #CHE-88-00770, and NSF grant #CHE-92-13241.

CHAPTER 3

THE ”MISSING” TYPICAL RHIZOBIUM LEGUMINOSARUM O-
ANTIGEN IS ATTACHED TO A FATTY ACYLATED GLYCEROL IN
R. TRIPOLH 48, A STRAIN THAT ALSO LACKS THE

USUAL TETRASACCHARIDE ”CORE" COMPONENT

84

85

ABSTRACT

Novel aspects of the cell surface chemistry of the unusual Rhizobium
strain R. leguminosarum bv. trifolii 48 have been uncovered. Typically,
strains of R. trifolii produce two distinct core regions in their
lipopolysaccharide, one is a trisaccharide consisting of galacturonic acid
and 2—keto-3-deoxyoctulosonic acid (KDO) while the other is a
tetrasaccharide made up of galactose, mannose, galacturonic acid and
KDO. Strain 45 is shown here to be devoid of the latter, galactose-
containing moiety. This result, coupled with previous data showing this
strain to produce an O-antigen and extracellular polysaccharide that are
both lacking galactose and unique to the rest of the R. trifolii strains,
suggested some sort of defect in galactose biosynthesis or incorporation in
this strain. However, the discovery of a galactose-containing lipid-linked
oligosaccharide with a composition that resembles the typical R. trifolii 0-
antigen is also presented in this chapter. This lipid contains glycerol, 2,3-
di-O-methyl~6~deoxyhexose, fucose, mannose, galactose, glucose,
quinovosamine, glucosamine and KDO which are all typical carbohydrate
components of R. trifolii lipopolysaccharide. The main fatty acid
constituents are stearic and oleic acids with smaller quantities of 3-

hydroxy fatty acids and 27-hydroxyoctacosanoic acid. This molecule is

86 .
most likely an O-antigen-like repeating unit linked to a mono or

diacylglycerol lipid anchor. There is evidence for the presence of a
truncated version of the core tetrasaccharide identical to a disaccharide
found in rough mutants. These findings suggest that strain 48 is an
effective natural isolate that has modifications that are similar to some
ineffective mutants. They also demonstrate that, contrary to previous
indications, synthesis of an LPS with an intact core tetrasaccharide is not

mandatory for a strain to be effective in nodulation.

87

INTRODUCTION

Cell surface chemistry has long been considered to be a major
contributing factor in the very specific interaction between rhizobia and
legumes. Acidic exopolysaccharides (EPS) (79, 80, 81, 150) capsular
polysaccharides (CPS) (49, 63, 64, 70), lipopolysaccharides (LPS) (29, 73, 74,
75, 77, 151), chitolipooligosaccharides (CLOS) (113, 115, 116, 152, 153),
cyclic-B-1,2-glucans, (154) and diglycosyl diacylglycerols (155, 156) have all
been implicated to varying extents in infection. The structural
interrelation between these various components as well as their common
genetic origins is not at all well understood although it has now been
documented. For instance, it was shown in Chapter 2 that several
structural links exist between the LPS and CLOS produced by Rhizobium
meliloti and that these structural elements (e.g. sulfation) are affected by
the same genes (11on and nodH). It has also been reported that
introduction of a gene normally encoding for LPS biosynthesis in R.
meliloti (lpsZ) will restore nitrogen fixing ability to a mutant that fails to
produce EPS (143). Recently it has been shown that expression of this
gene causes some modification of normal CPS production (157). The
structural overlap between LPS and CPS has also been very well

documented for many years in various Gram-negative bacteria (158-160).

88
Rhizobium leguminosarum biovars trifolii and viciae have been well

studied with regards to EPS and LPS structure and composition. In the
Rhizobium leguminosarum biovar trifolii cluster, hereafter referred to as
R. trifolii, strain 45 has been particularly well characterized and has been
found to be quite distinct from other members of the species with respect
to CPS and EPS structure. It has been shown that R. trifolii strains
typically produce EPS consisting of an octasaccharide repeating unit
comprised of glucose, glucuronic acid and galactose (5 : 2 : 1 respectively)
with a carboxyethylidene (pyruvate acetal) group attached to a terminal
side chain galactose and another to the adjacent glucose residue (59, 61).
The BPS/CPS of some strains also contain acetate groups and 3-
hydroxybutyryl groups (37, 38, 58, 161). The terminal galactose residue
may also bear one 3-hydroxybutyryl group and sometimes an acetate
group (62). R. trifolii 45, however, produces a heptasaccharide repeat
unit that lacks the typical terminal galactose with the 3-hydroxybutyryl
and carboxyethylidene groups (36, 162). It therefore appears that this
strain somehow lacks the ability to synthesize activated galatose (UDP-
galactose) or is defective in the synthesis of a specific galactosyl
transferase. In the former scenario, some very important consequences
would result for LPS synthesis since both the core and O-antigen
components of these lipopolysaccharides typically contain galactose (19,

20, 57, 163-165).

89
The lack of galactose incorporation in the EPS of R. trifolii 48 is also

evident in the O-antigen of the LPS. Recently, the complete structural
characterization of the O-antigen of R. trifolii 4S LPS was determined
(32). The structure was shown to be a pentasaccharide consisting of
rhamnose, N-acetylglucosamine and N-acetylmannosamine (3:1:1). This
O-antigen is vastly different in composition to that of any other R. trifolii
strains studied to date. The "typical" R. trifolii lipopolysaccharides (like
those of the other biovars) are rich in methyl deoxyhexoses,
deoxyhexoses, mannose, galactose, glucose, 2—amino-2,6odideoxy glucose
(quinovosa-mine), uronic acids, heptose and 2-keto-3-deoxyoctulosonic
acid (KDO) (57, 163-165). This total lack of galactose in the O-antigen is
more than just curious and definitely supports the idea that this strain is
either defective in the biosynthesis of UDP-galactose or in synthesis of a
galactosyl transferase which appears to be quite general. Another fact
that emerges is that there is an important overlap in LPS and EPS/ CPS
biosynthesis. This is completely consistent with what we know of
Rhizobium meliloti mutants that are also impaired in integrating
galactose into their EPS; some of these mutants are now known to be
impaired in LPS structure as well (33).

One area of rhizobial lipopolysaccharide chemistry and
biochemistry in which galactose biosynthesis and galactosyl transfer

activity is critical is in the core region. There are two core

9O
oligosaccharides in wild type R. trifolii LPS; one is a trisaccharide that

contains two molecules of galacturonic acid linked to KDO (18) and the
other is a tetrasaccharide in which galactose, mannose and galacturonic
acid are coupled to KDO (19, 20). One general observation is that bacteria
that are impaired in the synthesis of this galactose-containing
tetrasaccharide are unable to attach an O-antigen and, as a consequence,
have a ”rough” phenotype (22, 78, 166). This is generally taken as an
indication that the tetrasaccharide is the point at which the O-antigen is
attached to the lipopolysaccharide. In the case of R. trifolii 43, the fact
that the O-antigen is entirely non-classical is completely consistent with
the fact that the LPS might not contain the galactose-containing
tetrasaccharide and that an alternative compatible core and O-antigen
had evolved. Questions which then could be raised are ”W hat is the fate
of the classical O-antigen repeat? Is it not made at all or is it made but
never transferred or, perhaps, transferred to some other acceptor?" It
was of interest, therefore, to determine whether R. trifolii 4S synthesized
either of these two core oligosaccharides, especially the critical galactose-
containing one. If the impairment is largely an inability to incorporate
galactose into these polysaccharides, then a truncated version of the
tetrasaccharide, such as the mannose-KDO disaccharide found in some

mutants (22, 166), should be present. It was also of interest to look for the

91
remnants of the classical O-antigen biosynthesis and transfer pathway,

especially the fate of the classical O-antigen repeat unit.

92

MATERIALS AND METHODS

Bacterial cultures.

R. trifolii 48 was grown at 29°C in 4L shaken flasks containing 2L of
modified Bergensen's media as previously described (142). The cultures
were supplemented with 4M 4',7—dihydroxyflavone (DHF). Cells were
grown to late exponential phase and harvested by centrifugation at

10 krpm on a Sorvall RC2-B centrifuge equipped with a GSA rotor.

Lipopolysaccharide (LPS) isolation.

LPS was isolated by phenol extraction, purified by Sepharose 4B gel
filtration chromatography and the lipids released by mild acid hydrolysis
and extraction as described previously (14, 32). The aqueous
carbohydrate—containing fraction was separated on a Biogel P2 column
(120 cm X 1.5 cm) using 0.1% formic acid as the eluant. Ten milliliter
fractions were collected and 100 pl. aliquots removed and assayed for

carbohydrate content using the phenol / sulfuric acid method (149).

Membrane lipid isolation.
The cell pellet was extracted with 200mL of 1:2:2:3 chloroform/1-

propanol/ methanol/ water by vigorous stirring for 24 hours at 37°C. The

93
membrane extract was dried by flash evaporation, redissolved in 20%

acetonitrile and loaded onto a LiChroprep RP-18 column (EM
Separations, Gibbstown, N]; 8 x 2 cm). Successive elutions were made

with 30 ml portions of 20, 50 and 100% acetonitrile in water.

Carbohydrate compositional analysis.

Portions of the isolated fractions were methylated and reduced with
NaBD4. Each sample (~0.1mg) was treated with 21111 of 1% methanolic
HCl, sonicated for 1 min., heated for 15 min. at 68°C and kept overnight at
room temp. After drying the sample under a stream of N 2, 5mg of NaBD4
in 0.4mL of 1:1 methanol/ water was added and the solution kept at room
temperature for 24 hours. Excess borohydride was destroyed by dropwise
addition of 20% acetic acid in methanol until effervescence ceased. The
sample was dried with N2 and treated with lmL of 2M CF3C02H for 2
hours at 120°C. Reduction with NaBH.; and peracetylation with acetic
anhydride/ pyridine was performed as previously described (32). CC
analyses were performed on a Hewlett-Packard 5890 gas chromatograph
using a 30m capillary column (I 8: W, DB225) with He as the carrier gas. A
temperature program of 180°C to 220°C at 3°C/ min and a final hold of 60
min was used. GC-MS analyses were performed on a IEOL 505 mass
spectrometer using electron impact ionization (70eV) in the positive ion

mode. The same temperature program and column as described above for

94
CC analyses were used. Samples were also assayed for 2-keto-3-

deoxyoctulosonic acid (KDO) by using the thiobarbiturate procedure as

previously described (167).

Fatty acid compositional analysis of lipid fraction.

Lipid fractions were analyzed for fatty acid content by converting
them to the methyl ester derivatives. Portions of each fraction were
dissolved in lmL 5% methanolic HCl and sonicated for 15 min. To each
sample, 2mL of chloroform was added and the solution kept at 70-75°C
for 36 hours with intermittent sonication. After cooling to room
temperature and drying with N2, the sample was partitioned between
water and CHC13. The organic layer was removed and the aqueous layer
extracted twice with CHCl3. These organic layers were combined,
reduced to ~IOuL and used for CC and GC-MS analysis. These analyses
were performed as above except the temperature program used was
150°C to 300°C at 3°C/ min with a final hold of 20 min. and the column

used was a DB1 (J 8: W).

Nuclear magnetic resonance spectroscopy.
Proton nuclear magnetic resonance spectroscopy (1H NMR) was
performed on a Varian VXRSOO spectrometer (500 MHz). Spectra were

recorded in D20 at 25, 30 or 70°C as indicated.

95

RESULTS AND DISCUSSION

The Biogel P2 chromatogram (Figure 3.1) of the water-soluble
fraction of the mild acid hydrolysate of the LPS of R. trifolii 43 showed
some major differences to that typically obtained from most wild-type R.
leguminosarum biovars. These chromatograms typically show three very
well defined peaks, the first corresponding to the O-antigen, the second to
the core-tetrasaccharide and the last to the trisaccharide (22). In this case,
the first peak did correspond to the O-antigen but the second peak was
broad and appeared to contain more than one component. It was divided
into four sections based on the general shape and each section was
analyzed separately by proton N MR spectroscopy and by GC-MS of the
alditol acetates. In the 1H NMR spectra (Figure 3.2A), the three
characteristic doublets between 4.8 and 5.4 ppm attributable to the
anomeric protons of the tetrasaccharide hexosyl residues were not
observed. In addition, GC-MS analyses indicated that galactose was
absent from all of the fractions. The NMR spectrum of the material in the
fourth section of the second peak (Figure 3.2B) indicated that it contained
a previously described disaccharide composed of mannose and KDO. This
was readily identifiable by the anomeric proton at 5.0 ppm and the other

carbohydrate signals. This molecule has been characterized before in a

96

ABSORBANCE

 

 

 

   

”if 'c I. 9"n
‘3}. I 3 3 c I 1 gamIgzt
I A I 100' .
FRACTION#

Figure 3.1: Biogel P2 chromatogram of LPS hydrolysate of
Rhizobium trifolii 48. The small peak after 3 was devoid of

carbohydrate material.

97

Figure 3.2: 1H NMR spectrum (A) of fraction C of peak 2
from Figure 3.1. This is representative of fractions A and B.
The 1H NMR spectrum (B) from fraction D. Note the
mannose anomeric proton at ~5.0 ppm. Both spectra were

recorded in D20 at 25°C.

98

 

 

 

 

 

 

 

 

 

 

* 1 I r r 1 Y 1

1 4 3 2 1 W
to 1' r T 1 Tfifi ' I *j I
7 0 4 3 pp

Figures 3.2

99
rhizobia mutant that is incapable of making the core tetrasaccharide (64).

The NMR spectrum shown in that report is consistent with the one
collected here. The spectrum of this fourth section contained signals
indicating that smaller amounts of another component might also be
present. The proton NMR spectrum of the third peak was readily
identifiable as that of the trisaccharide core component which lacks
galactose. These results definitively showed that the galactose-
containing tetrasaccharide was absent.

The 20% acetonitrile fraction from reverse phase chromatography
of the R. trifolii 43 membrane extract when subjected to CC and GC-MS
analyses was found to be composed exclusively of glucose. The proton
NMR spectrum compared favorably to published spectra of cyclic-134,2-
glucan (168). This fraction was not analyzed further. The 1H NMR
spectra of the other two fractions, eluting with 50 and 100% acetonitrile,
although very similar to each other were both broad and ill-defined.
Spectra of the aqueous fractions from the mild acid hydrolysis were,
however, much more informative (Figure 3.3). There were resonances
that indicated the presence of methoxy, deoxy and amino sugars. These
were, specifically, sharp singlets between 3.2 and 3.4 ppm that indicated a
relatively large amounts of O-methylation while the signal at ~2.0 ppm
indicated the presence of an N-acetyl group. Additionally, signals at ~2.6

ppm indicated the presence of KDO. There were also signals for fatty acid

100

Figure 3.3: 1H NMR spectrum of mild acid hydrolysis
products of combined R. trifolii 48 fractions eluted from C18
column with 50 and 100% acetonitrile in water. The spectrum
was measured in D20 at 70 °C. Note the presence of intense
peaks between 3.2 and 3.4 ppm indicative of a high degree of
methylation. Multiple signals due to anomeric protons can be
seen between 4.7 and 5.2 ppm. The 6-deoxy group signals are
at ~1.1 ppm and the N -acetyl group signals at ~2.0 ppm. The
intense peak at ~1.83 ppm is from residual acetic acid. The
intense signals at ~3.5 ppm are from a low molecular weight

contaminant that is released from the reverse phase packing.

101

 

 

 

 

 

 

 

 

 

Figure 3.3

102
groups between 0.9 and 2 ppm. The carbohydrate composition was

determined definitively by GC-MS. The composition of the two fractions
were virtually identical and the total ion chromatogram of the last eluting
fraction is shown in Figure 3.4. The major components were glycerol, 2,3-
di-O-methyl-6-deoxyhexose, fucose, mannose, galactose, glucose,
quinovosamine and glucosamine. The thiobarbituric acid assay also
indicated the presence of KDO in these samples. These are the typical
glycosyl components of rhizobial LPS O-antigens and have been described
in the various R. leguminosarum biovars that have been analyzed over the
years. The data in Table 3.1 summarizes some of these publications and
illustrates the great similarity of the glycosyl composition of the LPS of
these strains to that of this newly discovered lipid. As noted earlier, most
of these glycosyl components, especially the methylated deoxyhexoses and
quinovosamine, are missing in the LPS of strain 48 (32). GC-MS analyses
of the methyl ester derivatives of these fractions revealed that the
predominant fatty acids were typical phospholipid fatty acids, namely
stearic and oleic acids (Figure 3.5). However, other fatty acids which are
characteristic of lipopolysaccharides, such as the 3-hydroxy fatty acids
and 27—hydroxyoctacosanoic acid (11, 12, 14) were also present. The fatty
acids in the anchor, therefore, are a curious mix of LPS and other
membrane lipid fatty acids. The question then arose as to what is the lipid

anchor for these oligosaccharides. The possibility that it was the typical

103

Figure 3.4: Total ion chromatogram of alditol acetates of the
R. trifolii 4S extract eluted from a C18 column with 100%
acetonitrile. The identity of the major peaks were determined

by co-elution with standards and analysis of mass spectra.

GMOm

 

 

104

 

 

 

 

 

 

 

 

 

comm ooom oomH coca Com
5 u o ~
~ .
u o v
gun—8830 .w n v .
03585550 .5 .
8830 6 o N 1 o m
39830 .n .
03582 .v
885m .m .
8330830295 3 .N .
"Boom—O A u o m
. o H a .
. . . a . . . . _ . . I m . . . . . o o H

Figure 3.4

‘cQ'JG'UCUf-IOO

105

Table 3.1: Reported carbohydrate composition of LPS of

various Rhizobium trifolii strains.

R. trifolii strain

 

 

Component 0403a Ar-3b Cory?) K-8b 16257c 23¢
KL

3-O-methylhexose nr nr nr nr 0 1.1
2-O-methyl-6-

deoxyhexose 8 l .4 2. l 6.4 4.5 0
3-O-methyl-6-

deoxyhexose nr nr nr nr 0 3.6
3,4-di-O-methyl-6-

deoxyhexose nr nr nr nr 0.7 0
3-N-methyl-3-amino—

3,6-dideoxyhexose 15 nr nr nr 0
Rhamnose 4 40.3 0 2 1.9 0.6 7 S
Fucose 13 20.2 0 8 7.2 4.0 7 5
Mannose 7 0.9 7.9 3.2 1.9 l 7
Galactose tr 0 4.9 2.8 1.1 2 2
Glucose 4 26.6 28 4 7.1 0.9 1 0
Heptose 21 O 1.0 0 0 0
Hexosamines nr 0.8 2.1 1.2 l 5 1.6
Uronic acids 27 2.2 1 1.7 5.3 2 2 7.0
2-Keto-3-deoxyoctonoic

acid nr 1.3 2.9 4 1 7.2 2.7

 

Taken from references 1643, 165b, and 163C; results are expressed as percent of dry weight of LPS.
nr = none reported, tr = trace.

106

Figure 3.5: Total ion chromatogram of fatty acid methyl
esters of the same fraction as in Figure 3.4. The identity of the
major peaks were determined by co-elution with standards as

well as analyses of mass spectra.

107

cmom

 

 

 

 

 

 

 

 

 

 

 

 

ooom comm oodm coma coon
‘ ION
N .
10v
oeufiosua .
92.... .8
32m .
98a .
9.23.58.” .
L;
m o: .
. _. ....2:
3 mm om mm om 3

«QSGUMGOO

Figure 3.5

108
lipid A was easily ruled out by two major facts. First, there was the

relatively low abundance of hydroxy fatty acids compared to the other
fatty acids. The fatty acids of R. trifolii LPS are almost exclusively
hydroxy fatty acids. The second and most important fact is that all GC
and GC-MS analyses clearly indicated that galacturonic acid was absent
from the samples. Galacturonic acid is a component of the lipid A (15).
The presence of glycerol could indicate that the glycolipid was
contaminated with phospholipids. This was not likely, however, because
phospholipids normally do not elute from a C-18 reverse phase column
with 50% acetonitrile in water. In any event, the absence of phospholipids
was definitively proven by thin layer chromatography analysis using a
total lipid extract from the same strain and another as references (Figure
3.6). From this it could be seen that there were no contaminating
phospholipids. Since glycerol is a component of the glycolipid, and could
be freed by acid hydrolysis, it follows that it has to be linked glycosidically
to the reducing terminus of the oligosaccharide chain where it could serve
as a membrane anchor through acylation by fatty acids. This is the usual
configuration of glycerol-containing glycolipids.

A structure that is consistent with the data is shown in Figure 3.7. It
could arise by the simple transfer of the activated LPS O-antigen repeat
from the carrier lipid to diacylglycerol as indicated. This could happen

since the usual acceptor core oligosaccharide is not present and the

109

Figure 3.6: Orcinol/ sulfuric acid stained TLC plate of (A) a
total lipid extract of R. trifolii ANU843, (B) a total lipid
extract of R. trifolii 4S, and (C) the combined R. trifolii 43
fractions eluted from the C18 column with 50 and 100%
acetonitrile in water. Note that the fastest moving
component in each total lipid extract is diacylglycerol (arrow).
Solvent system used was chloroform, acetone, methanol,
acetic acid, water (10:4:2:2:1, vzv). Total lipid mixtures were
obtained by extraction with chloroform, methanol, water

(4:1:5, vzv).

110

 

Figure 3.6

111

Figure 3.7: Possible mechanism of transfer of LPS O-antigen
repeat unit from the carrier lipid to diacylglycerol. The
product is a chimeric lipid. The O-antigen repeat might be

polymerized by successive transfers to a hydroxyl group on

the end sugar residue.

112

 

 

Figure 3.7

113
membrane of this strain has a high proportion of diacylglycerol (See

Figure 3.6). The transfer of activated carbohydrates to diacylglycerol is
very common in Rhizobium since two classes of glycolipids,
sulfoquinovosyl diacylglycerols and diglycosyl diacylglycerols are both
synthesized by these organisms (144, 155, 156) and both involve the
transfer of activated sugars to diacylglycerol. It is quite reasonable that in
the event of aberrant LPS and EPS biosynthesis, as is present in R. trifolii
48, there could be some surrogate function for other glycolipids and
glycosyl acceptor molecules. It is also quite possible that a fair amount of
this hybridization of molecular structure occurs in normal strains. In the
event that the carbohydrate chain in the glycolipid becomes further
polymerized by the addition of more O-antigen chains, the resulting
polymer would have the usual carbohydrate composition of
lipopolysaccharides (including the presence of KDO) as well as some of
the characteristic fatty acids. Such a hybrid LPS could restore the proper
functionality of LPS/BPS defective strains. The "LPS" molecule whose
production is controlled by the lpsZ gene in R. meliloti (143, 169) fits the
profile of such a hybrid molecule.

The discovery of this class of LPS-like lipid-linked oligosaccharides
in R. trifolii 43 has other significant implications. R. trifolii 48 has been
considered to be an unusual but important member of the clover

nodulation group. The recent discovery that the LPS O-antigen of R.

114
trifolii 45 has a vastly different structure to the rest of the R. trifolii strains

raised an interesting question about the role of LPS in infection of this
strain. It is well documented that LPS plays an important role in various
stages in the nodulation process. As reviewed in Chapter 1, these stages
have been demonstrated to include infection, bacterial release from the
infection threads, and bacteroid transformation and maturation (26, 27,
29, 75, 77, 143). It seemed puzzling then that 48 could have such a different
LPS O-antigen structure and still belong to the same clover nodulation
group. The results presented in this report, however, resolves this
apparent conflict. If LPS is required to play a direct role, it is quite possible
that this new lipid somehow performs the same function in the nodulation
process for 48 that the "typical" LPS would in the other strains of trifolii.
It is also clear that the synthesis of a core tetrasaccharide is not a
prerequisite for a strain to be effective in nodulation, contrary to earlier
observations (22, 166).

Another salient point that emerges from this study is whether or not
R. trifolii 4S possesses the capability of synthesizing galactose. There are
two possibilities, the first one is that R. trifolii 48 is incapable of
synthesizing UDP-galactose and the galactose in this glycolipid is made by
the action of an epimerase on glucose after its incorporation. The second
is that UDP—galactose is synthesized but there is a specific galactosyl

transferase for this glycolipid that is functional and the others (or some

115
other general transferase) are defective. Attempts to detect the presence

of UDP-galactose in this strain were unsuccessful.

116

ACKNOWLEDGMENTS

The experiments on the R. trifolii 4S lipopolysaccharide core region
were performed by Ying Wang. This work was supported by grant #DE-
FG02-89ER 14029 from the US. Department of Energy to RIH. The
GC/ MS data were obtained at the Michigan State University Mass
Spectrometry Facility which is supported in part by grant #DRR-OO480
from the Biotechnology Research Technology Program, National Center
for Research Resources, NIH. The NMR data were obtained on
instrumentation that was purchased in part with funds from NIH grant
#1-SlO-RRO4750, NSF grant #CHE-88-00770 and NSF grant #CHE-92-

13241.

CHAPTER 4

OCCURRENCE OF SULFOQUINOVOSYL DIACYLGLYCEROL IN

SOME MEMBERS OF THE FAMILY RHIZOBIACEAE

117

118

ABSTRACT

A radiolabelled component of a membrane extract of Rhizobium
meliloti 2011 cells grown in the presence of 35S-labelled sulfate was
isolated by silica flash chromatography and purified by HPLC. Based on
1- and 2-dimensional nuclear magnetic resonance (NMR) spectroscopic
and mass spectrometric analyses, the structure of the compound was
determined to be sulfoquinovosyl diacylglycerol (SQDG). NMR analyses
indicated substantial heterogeneity in the fatty acid composition and that
an important group was the cyclopropyl fatty acids. This first report of
the occurrence of SQDG outside of the plant kingdom, photosynthetic
bacteria or diatoms deserves special attention since, in this case, the
bacterium is one which can fix nitrogen in symbiosis with plants. The
origins of the ability of the bacteria to synthesize this class of membrane
lipids is an important question. Membrane extracts of other strains of the
family Rhizobiaceae were screened for the presence of SQDG. The
occurrence of SQDG in the symbiotic organisms was confirmed while no
SQDG was detected in either the Agrobacterium tumefaciens or the
Escherichia coli strains tested. The current function of these lipids in

symbiosis and the commonality of the ability of bacteria which function as

119

plant symbionts to synthesize such molecules are all germane to studies of

the Rhizobium / legume symbiosis.

120

INTRODUCTION

The rhizobial cell surface has always been considered to be an
essential factor in our attempts at understanding the interactions between
these bacterial symbionts and their legume partners. The tremendous
wealth of information pertaining to the role of membrane components in
this relationship is reviewed extensively in Chapter 1. Much debate and,
admittedly much controversy, has revolved around the roles of capsular
polysaccharides (36, 58-61). The roles of lipopolysaccharides in ensuring a
successful infection process is much more generally accepted. It is well
established that bacterial mutants having defective lipopolysaccharide
structures are incapable of infections which progress into viable, occupied
nodules which are capable of supporting nitrogen fixation (22, 29, 73, 166).
The bacterial cell envelope, in which of course the lipopolysaccharide
resides, has still not been rigorously characterized, despite many years of
study. For instance, it was only relatively recently shown that 27-
hydroxyoctacosanoic acid, a 28-carbon fatty acid, was found to be a major
component of the lipopolysaccharide of virtually all rhizobia (11, 12).
There have been several studies attempting to elucidate the total lipid
composition of the membranes of some of the members of the family

Rhizobiaceae. A major consistent finding in these studies is the presence

121
of phosphatidylcholine (PC), a typical plant lipid, in the membrane of all of

the species investigated thus far (170-172). This finding may suggest
either a role for this lipid in the infection process or the possibility of gene
exchange between the bacterium and the plant in the distant past.
However, although PC is not common amongst Gram-negative bacteria,
it is also not exclusive to the Rhizobiaceae (172).

As described in detail in Chapters 1 and 2, the Nod factors have been
advanced as the sole determinants of host specificity in the
Rhizobium/ legume symbiosis. The Nod factors are a class of closely
related mono-N-acylated chitin oligomers substituted with various side
groups. The type of fatty acid and the nature of the side group have been
thought to account for the host specificity (173). In the case of R. meliloti it
has been shown that the critical feature that determines host specificity is
the sulfate group on the reducing terminal glucosamine residue (115, 120,
141). The emergence of sulfated lipid-linked carbohydrates as mediators
of the infection process in the R. meliloti / alfalfa symbiosis raised the
question of a possible link of this phenomenon to the bacterial surface
chemistry. Therefore, a systematic study of the bacterial membrane was
performed with a focus on finding sulfur-containing lipid components.

Bacteria were grown in the presence of 35S-labelled sulfate and
labelled membrane components were purified chromatographically and

characterized. The isolation, purification and characterization of a

122

sulfur-containing lipid, whose occurrence was thought to be restricted to
plants, algae and photosynthetic bacteria, is presented in the first part of
this work (174). Some questions that arise due to this finding are: "Was
this a trait that was captured from plants but has no symbiotic
significance?" or, "Is the ability of the bacteria to make plant-like lipids a
prerequisite for the close interaction (possibly membrane fusion) between
bacteria and plants during symbiosis?" In the second part of this study
these questions were begun to be addressed by screening various members
of the Rhizobiaceae for the presence of this sulfolipid. These experiments
were conducted in an attempt to probe the relatedness of the various
genera and to, perhaps, provide a clue to the possibility of a requisite

membrane compatibility in rhizobial infection.

123

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The various bacterial strains which were used are as follows:
Rhizobium meliloti 2011 (from Ethan Signer, Massachusetts Institute of
Technology), R. sp. strain NGR234 (from Frank Dazzo, Michigan State
University), R. leguminosarum bv. trifolii ANU843, R. leguminosarum bv.
trifolii ANU845 (from Barry Rolfe, Australian National University),
Bradyrhizobium japonicum USDA110 (from Frank Dazzo, Michigan
State University), Agrobacterium tumefaciens C58 (from Frans DeBruijn,
Michigan State University), and Escherichia coli TG-l (from Gregory
Zeikus, Michigan State University). All strains of bacteria (with the
exception of E. coli) were grown at 30°C in liquid cultures containing
modified Bergensen's (BIII) media as previously described (142) except
that in addition, 35S—labelled sulfate at a level of 100uCi/ liter of culture
was added in the form of carrier free sulfuric acid (New England
Nuclear). The E. coli strain was grown at 37°C in MSX4 media with 358-

labelled sulfate added as above.

124

Lipid isolation and purification.

Lipids were extracted by stirring the bacterial cells (after harvesting
by centrifugation) with 100 ml of a mixture of chloroform, methanol, n-
butanol and water in the ratio of 2:1:1:4 at room temperature for 24 hours.
The cell debris was removed by centrifugation and the supernatant
allowed to partition into two layers. The lower organic layer was
recovered and the aqueous layer was recombined with the cell debris and
the mixture re-extracted with 200 ml of a 2:1:1 mixture of chloroform,
methanol and n-butanol, respectively. The organic 1ayer was again
recovered and combined with the one from the first extraction. The
combined organic layer was concentrated to dryness and
chromatographed on a silica column (4 cm x 2 cm) with a mixture of
chloroform, acetone, methanol, acetic acid and water in the ratio of
10:4:2:2:1, respectively. Fractions of 4 ml were collected and a 100 pl
aliquot of each subjected to analysis for radiolabel by scintillation
counting. One major fraction (from a plot of fraction number vs. d.p.m.)
was recovered and subjected to further purification by high performance
liquid chromatography (HPLC) on an ALTEX ultrasphere C-8 reverse
phase column (1 cm x 25 cm). A linear gradient was employed starting
with a combination of 1:1 acetonitrile/ water (40%) and n-propanol (60%).
This ratio was maintained for 10 minutes at a flow rate of lml/ min. The

flow rate was then increased to 1.2ml/ min and the gradient was run to a

125

final composition of the two solvent systems of 10 and 90%, respectively.
The column effluent was monitored at 230 nm and the peaks were
collected and counted. The entire isolation and purification was repeated
using unlabelled sulfate to obtain material for use in structural
characterization. Purity of the product was assessed by thin layer
chromatography (TLC) on silica layers using a mixture composed of
chloroform, acetone, methanol, acetic acid and water in the ratio of

10:4:2:2:1, respectively.

Chemical characterization.

Fatty acids were released as methyl esters from the purified lipid by
methanolysis of a sample (~50 ug) at 75°C with 1 m1 of methanol
containing 2% HCl. The methanolysate was concentrated to dryness and
partitioned between 2 ml of hexane and 1 ml of water. The upper hexane
layer was recovered and subjected to gas chromatography and gas
chromatography/ mass spectrometry.

The glycosyl component of the purified lipid was obtained by
treating a sample (3 mg) of the component with ethanolic potassium
hydroxide (0.5 M) at 100°C for 30 minutes. The mixture was treated with
10 ul of acetic acid and then concentrated to dryness. The fatty acids were
removed by partitioning the residue between 4 ml of a 1:1 mixture of

water and hexane. The aqueous layer was subjected to gel filtration

126

chromatography on a Biogel P2 column (20 cm x 0.5 cm) in pure water.
One milliliter fractions were collected and assayed for the presence of
carbohydrates by spotting on silica plates, spraying with orcinol/ sulfuric
acid and heating at 120°C for 5 minutes. The fractions giving a positive
reaction were pooled and lyophilized. The optical rotation was measured

in water at the sodium D line on a Perkin Elmer Model 141 polarimeter.

Nuclear magnetic resonance (N MR) spectroscopy.

All NMR spectra were measured in CD3OD solution at 500 MHz
for the 1H or 125 MHz for the 13C nucleus. The 13C-DEPT spectrum (175)
was measured at 75 MHz for 13C. Double quantum filtered J-correlated
2-dimensional spectra (phase sensitive mode) (176) were obtained using a
total of 512 data sets (32 transients) over a spectral width of 3000 Hz. The
total number of points used was 2048. The FIDs were multiplied by a
gaussian function in the f2 dimension and by a shifted gaussian in f]. The
final data set was symmetrized by triangular folding. Data for the
HOHAHA experiments (177) were obtained using similar acquisition and
processing conditions. A mixing time of 80 msecs was used. Spectra were
not symmetrized. For the HMQC experiment (178), a spectral width of
19,000 was employed for the 13C dimension. A total of 80 transients were
acquired for each of a total of 512 data sets. A total of 1024 points was

used in these analyses.

127

Mass spectrometry.

Fast atom bombardment mass spectra were recorded on a JEOL
HX-110 HF mass spectrometer using Xenon neutrals (6 keV) as the
primary beam. Triethanolamine was used as matrix and spectra were
obtained in the negative ion mode. Gas chromatography/ mass
spectrometry was performed on a JEOL 505 mass spectrometer using a
D31 capillary column. The gas chromatography program used was: 150°

to 300° at 3°/ minute with a 10 minute hold at 300°C.

Screening for sulfoquinovosyl diacylglycerol (SQDG).

Cells were harvested by centrifugation and lipids were extracted
using the procedure described above. The crude organic extracts were
subjected to thin layer chromatography (TLC) on silica gel using isolated
lanes to prevent cross-contamination and employing the solvent system
described above. Sample extracts were chromatographed along with a
standard of SQDG, the isolation and characterization of which is
presented in this chapter. Bands containing glycosyl components were
visualized by spraying with an orcinol/ sulfuric acid mixture and heating
at 120°C for approximately 10 min. Radioactive bands were detected by
exposure of the 2,5-diphenyloxazole-impregnated TLC plate to a Kodak
x-omat x-ray plate at -80°C for three weeks. Additionally, TLC plates

were developed in which the radioactive glycosyl-containing bands were

128

scraped and the lipids extracted from the silica with chloroform, methanol
(1:1, vzv). Verification of the presence of SQDG was obtained by FTIR
spectroscopy using a Nicolet 710 Fourier Transform Infrared
Spectrometer. The samples were analyzed as thin films on a NaCl

substrate.

129

RESULTS

Analysis of purified lipid.

TLC analysis of the membrane lipids of Rhizobium meliloti 2011 on
silica layers followed by autoradiography revealed the presence of two
major radiolabelled components (Rf=0.22 and 0.18). These components
were eventually isolated using a combination of flash chromatography on
silica and reverse phase chromatography. The 1H-NMR spectra of the
two components were very similar except for slight differences due to
fatty acid distribution. The spectrum of the slower migrating component
on TLC (first eluting on reverse phase HPLC) is shown in Figure 4.1. It
contained signals between 2.8 and 4.7 characteristic of carbohydrate
groups as well as signals between 0.8 and 2.4 ppm characteristic of fatty
acid groups. A triplet at 5.35 ppm was assigned to the vinyl protons of an
unsaturated fatty acid. A multiplet at 2.02 ppm was assigned to the
methylene groups adjacent to the unsaturation. Signals for the methylene
groups adjacent to the carbonyl carbons of fatty acid groups appeared as a
multiplet at 2.32 ppm. The signals for the terminal methyl groups of fatty
acid chains appeared at 0.89 ppm. One interesting feature was the
presence of two sets of very upfield signals. These were assigned to

signals for protons on cyclopropyl groups. The most upfield signals, at

130

Figure 4.1: 1HwNMR spectrum of the purified 35S-labelled
major component (Rf=0.22) from the membrane extract of R.
meliloti 2011. Signals between 2.8 and 4.8 ppm are due to a
carbohydrate and a glyceryl component. The other signals
are due to fatty acids. Note the presence of a high degree of
unsaturation (vinyl proton resonances at 5.35 ppm) and the
presence of cyclopropyl group (most upfield signals).
Assignments are made in the text. The strong resonances at
3.30 and ~4.8 ppm are due to residual CHDZ' and Ol_i_
signals, respectively, of the solvent. The signals between 0.5

and 2.4 ppm are due to resonances in the fatty acids.

131

Eco 0.0 m6 o.« m.« o.m

FL—beP—bbehthrbethIbI_PPhb—Pup

9m -o.m

h b h F FLIP »

m.v

hpIp

oh

m.m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.1

132
-0.33 ppm, were assigned to the methylene group of the cyclopropyl

function. A broad multiplet at 5.32 ppm indicated the presence of a
glyceryl moiety. This signal was assigned to H-2 of the glyceryl moiety.
The other signals between 2.8 and 4.8 ppm were generally too complex to
assign directly. The 13C-DEPT spectrum of the same slower migrating
component (Figure 4.2) contained signals between 10 ppm and 40 ppm,
consistent with the aliphatic resonances of fatty acid chains. The presence
of unsaturations was confirmed by resonances at 130 ppm. A resonance at
99.2 ppm was assigned to an anomeric proton of a glycosyl residue. Based
on the chemical shift of a proton (assignable to an anomeric proton) at
4.76 in the 1H--NMR spectrum, the a-configuration was assigned. This
proton displayed a coupling constant of 2-3 Hz indicating that the
neighboring proton was axial. Signals between 50 and 110 ppm were
sufficient for only 1 glycosyl unit. The 13C-DEPT spectrum indicated that
the molecule contained three primary carbons bearing heteroatoms. One
of these at 53.5 ppm was assigned to carbon linked to sulfur. The other
two were assigned to a glyceryl moiety. Five carbons bearing oxygen
atoms gave rise to signals 62 and 76 ppm, giving a total of 9 carbon
attached to heteroatoms.

The HOHAHA spectrum of the slower migrating major lipid
component (Figure 4.3) conveniently broke down the spectrum into spin

systems for the glyceryl, carbohydrate and fatty acid components. Based

133

Figure 4.2: 13C-DEPT-NMR spectrum of the purified 353-
labelled component (Rf—7.0.22). The anomeric carbon of a
glycosyl residue appears at 99.2 ppm. The number of signals
in the range of 50-110 ppm and the number of anomeric
carbons are consistent with the presence of only one glycosyl
unit. The signal at 130 ppm is due to vinyl carbons. Note the
very upfield primary carbon resonance at 53.5 ppm in the

carbohydrate region of the spectrum, indicative of a carbon-

sulfur linkage. r.

Sea 0 0« ON

00" 0: ON“

.5..o...—.P-._—..._...P-...-_....-_...—.-.._...-—....—.-..—.-.p_r.+.

 

.1..- 3.. :....:1 _._.. 2443111117115

mZOgU 44¢.

11.. 111

 

 

 

 

O
1 I111 LDPDTII'»! {PI >101 vi. 11E [1117b P1111»? >1>bptp>1111n -.LIPD pp
1 1 I 111111: 1#« _ 311111.111 1.1111 11141441 111 N111 111' 1- «Ig II11111 11
‘

 

940520 2:52

11F111I11urtr. 1.1.11. §§§§i§§

 

 

1|- Inn» IDI
-11 1

 

[r I 11 1411111411 41114— 4141.4 qi‘h_1‘l ‘Dfi
‘

mZOmM<U mZmA>IPmE

 

mzommtd A>IEZ

 

Figure 4.2

135

 

 

 

 

 

 

 

 

 

 

.‘
. ‘ .
q o
d “‘
4 ‘ u
‘5 ' ‘ i.‘ "'
‘ -' O
‘ ‘ . o o > . .‘ .f:I.Vp,I I"°“ '
4 ' \ ltpAv
A 1|
' 4 --- 0
4 .-
‘ '
4
v < -
4
‘ I
4 .. a
. ‘ .
q
4
. ‘ .
4
< . a. .
1
’ q
< u
4 L
. I
. ‘ '
‘ I .
<
Q.
$‘fl:
' 4
<
4 .
4
. , ‘
4 .- , .—
jfiav - a: «a T - h t-uvh
5 5« 1|! . cl! 3'! n .
4 VvVYTYVYY‘IYVYYTWYYY‘ITTV‘WYYYYIYAV‘YIVVVVIYVYVIVVYI‘Y

VT’VYVVI'VVYI
55504541335302.52111.51.00.c) ‘O.5

 

Fl (Dom)

Figure 4.3: 1H-HOHAHA NMR spectrum of the purified 35$-

labelled component (Rf=0.22). Note the connectivities for

glycerol (gly) and for the fatty acid components (fa). The

carbohydrate connectivities are not marked.

136
on this spectrum, the signals at 5.32 ppm, 4.50 ppm, 4.17 ppm, and 3.58

ppm were assigned to the glyceryl moiety. The other signals between 2.8
and 4.8 ppm were assigned to the glycosyl residue. The DQF-COSY
spectrum (Figure 4.4) allowed the assignments of the spin connectivities in
the spectrum. Hence, the signal at 3.39 ppm was assigned to H-2 (a
double of doublets with J=7.3 Hz and 2.0 Hz). Based on these coupling
constants, it was determined that the glycosyl H-3 signal at 3.62 ppm
(triplet with J=7.3 Hz) was axial and that the hydroxyl group at this
position was therefore equatorial. The large coupling with both H-4 and
H-2 indicated the gluco configuration. H-4 appeared further upfield (3.08
ppm) as a triplet (127.3 Hz). The signal for one of the H-6 protons (dd,
J=13.5 + 7.9 Hz) appeared at 2.90 ppm, indicating that it was not attached
to a carbon atom bearing oxygen. It was concluded that this carbon was
attached directly to sulfur. Based on the chemical shift a sulfonic acid
moiety was proposed. The proton assignments agreed well with the 13C-
NMR assignments. These were confirmed by an HMQC experiment
(Figure 4.5). The carbon attached to sulfur appeared at 53.5 ppm as
expected. The glyceryl methylene carbons appeared between 62 and 67
ppm while the remaining glyceryl resonance, as well as those for the
carbohydrate carbons, was further downfield. Assignments for the shifts
in the fatty acid region also appear as expected. The complete proton

assignments are given in an expanded 1H-NMR carbohydrate region in

137

 

 

 

 

 

 

 

 

 

  

u . -==-vasmmm

MW...
.2“ -- i? a.

 

 

W117.VI'I‘V'IIY'IIVY—YYIY'VII'T‘Y‘IYYI'IVYYVIYYV.IYTT—Y—IYT‘YIY
5.5 5.0 4.5 4.0 5 3.0 2.5 2.0 .5 .0 0 5 -0 5

F1 (ppm)

Figure 4.4: 1H-DQF-COSY spectrum of the purified 35S-

labelled component (Rf=0.22). Note the assignments for the

carbohydrate and glycerol components. Primed numbers

refer to the carbohydrate component.

138

Figure 4.5: 1H/13C-HMQC spectrum of purified 358-labelled
lipid component (Rf=0.22). All cross peaks due to the
carbohydrate and glycerol components are assigned. Note
the proton and carbon chemical shifts at the 6-position, the

site of sulfonylation. Primed numbers refer to the

carbohydrate components.

139

.566. «u

 

 

o o" om om ow om om om om om co" 0"" cm“ om“ ova om"
karhhbrPLHFFlebbPhhbhhb~>hFP—Phbb—P>>P_>bbb—>PrbyPP>>h—PFPF.—[PPP>L>b—>F>>— .

o m
m.m

.

m 06
o.m

1 u . u L.“ o .n. a h
a: a. m.v
g:
cm m o.v
\

am.

‘
. . . 8 \ . n , . . -
.00 n O 4... out.“ .-00 u ('0 .000 O- 0.1+“? nll’. Ill." MI '9. 0| .O‘O’N‘OIOQ‘. ‘03... ‘9! \n‘ .Oi‘oll'

Figure 4.5

.IY
« ...
m. . m
mm m
I l0\ n
O .vIO.N
«:de m n
O Fm"

 

‘1'

I.
v

d

O
.5

I
:P
O
J...
I

‘Y
O

 

 

140

Figure 4.6. Based on the NMR analyses, the molecule was determined to
be a diacylglyceryl derivative of a 6-deoxy-a-glucopyranose-6-sulfonic
acid. These assignments agree with previously published results (204).
According to integration of NMR signals in analyses of crude lipid extract,
SQDG comprises approximately 5-10% of the total lipid in Rhizobium
meliloti.

The 1H-NMR spectrum of the faster migrating component (Rf=0.22)

had essentially the same features as that of the slower migrating other
major component (Rf=0.18). The fatty acid distribution was, however,
different since signals for multiple unsaturation were more dominant.
This splitting of TLC components for these lipids is quite common and has
been demonstrated and studied by other workers who also attribute
differences in mobility to fatty acid class (179). These workers also
observed differences in the orcinol response for the two TLC components.

The fatty acids were determined by gas chromatography/ mass
spectrometry (CC/MS) to be predominantly hexadecanoic acid,
octadecenoic acid and a methyleneoctadecanoic acid. There were other
minor peaks assignable to a bis-unsaturated C-18 fatty acid, tetradecanoic
acid and tetradecenoic acid.

The proposed structure was confirmed by performing fast atom
bombardment mass spectrometry on the slower migrating component

(Figure 4.7). The major ion in the pseudo-molecular ion region, i.e.,

141

 

 

 

 

 

 

 

 

 

I
3
| “’38, b 2 6a" 4! ,
1a 5 3 U 5”
l l ‘
ll * '
I I I l I I I I l I I I I I I I I I l I I I
4 5 4 O 3.5 3 0

 

Figure 4.6: 1I-i-NMR spectrum of purified 358—labelled lipid

component (Rf=0.22). Complete proton assignments for

sulfoquinovosyl diacylglycerol are shown. Primed numbers

refer to the carbohydrate protons.

142

Figure 4.7: Negative ion fast atom bombardment mass
spectrum of purified lipid component. The cluster of ions
between 700 and 90011 are (M-H)‘ pseudo-molecular ions for
the different species. The ion at m / z 793 is due to the molecule
containing two hexadecanoyl residues (Structure 4.1). This is
the predominant component. The other ions are due to
heterogeneity in the fatty acyl components. The assignments
are made in the text. Signals between 250 and 300 are due to

carboxylate ions of the fatty acyl components formed by

primary cleavage. The assignments arezlm/z 255 (C16) the
predominant component, 28] (C185), 295 (C18 with one

cyclopropyl group) and 279 (C18:2).

143

asawxz

 

05m

 

onw mmw

 

a;

 

m2.

8&0

58v

GEN

 

‘1‘»11—‘1-10 4111 l 4 i A—Wdi 1141..

 

mom 5N

nmm

.1. 1

man

 

 

 

Figure 4.7

144
((M-H)’=793u) was assigned to sulfoquinovosyl dihexadecanoyl glycerol

(Structure 4.1). The next major ion at m/z 819 corresponded to a form in
which the fatty acids were a hexadecanoyl component and an
octadecenoyl component. The next major peak at m/z 833 corresponded
to the presence of a methyleneoctadecenoyl and a hexadecanoyl fatty acid
residue. The next most intense peak appeared at m/z 859 and
corresponded to the presence of octadecenoic acid and a
methyleneoctadecenoic acid. Other minor peaks corresponding to
different combinations of the fatty acids present were also observed.
Signals for the carboxylate ions of the fatty acids appear between 250 and
300u. The peaks corresponding to C16 and C183 are the most abundant
(m/z 255 and 281, respectively). The peak at m/z 225 is most likely due to
the dehydrosulfoglycosyl anion corresponding to a minor sulfoglycosyl ion
at m/z 243. The stereochemistry of the glycerol to carbohydrate linkage
was confirmed by removing the fatty acid by base treatment and
subjecting the carbohydrate component to polarimetry. Not enough
material was obtained to allow an accurate measurement of the optical
rotation of the glycoside but the sign could be determined. The rotation of
the glyceryl glycoside in water was small and positive, consistent with the

proposed D-configuration (180).

145

Structure 4.1: Structure of the major component in the family

of sulfoquinovosyl diacylglycerols elaborated by R. meliloti

2011. The most prominent fatty acyl species are C16, C11“ and

C18 cyclopropyl species. The l-position of glycerol is acylated.

146
Occurrence of sulfoquinovosyl diacylglycerol (SQDG) in symbiotic species

of the Rhizobiaceae.

Organic extracts of several strains belonging to Rhizobium,
Bradyrhizobium and Agrobacterium were screened for the presence of
SQDG by TLC followed by visualization with orcinol. The results are
shown in Figure 4.8. Glycosyl-containing bands (Rf=O.22 and 0.18) with
similar mobilities to the standard SQDG bands seemed to be present in all
members of the rhizobia species tested. No such bands occurred in the
Escherichia coli control. The A. tumefaciens sample showed what
appeared to be a faint band in this region also. However, when the TLC
plates were subjected to autoradiography (Figure 4.9), the Rhizobium
bands were found to be radiolabelled while the same region of the A.
tumefaciens lane was not. These results were verified by scraping the
bands from the plate, extracting with a chloroform:methanol mixture (1:1,
vzv), and checking radioactivity by scintillation counting. In addition, a
third faint radioactive band with slower mobility than SQDG appeared in
the autoradiogram in the lanes of the rhizobia species. This band has not
yet been identified. The Bradyrhizobium strain tested also showed a band
with a similar mobility to the standard and when extracted showed about
twice baseline radioactivity (results not shown). In a previous study, a
TLC of the total lipid extract from R. meliloti failed to show the presence

of SQDG (181). This was most likely due to the different methods of

147

Figures 4.8: Orcinol stained TLC plate of membrane extracts
from members of the Rhizobiaccae (Strains are identified in
text). Two bands can be seen with similar mobilities to the
two SQDG standards in all strains except the E. coli control
and the A. tumefaciens sample. Some major lipids are
identified as a reference in the Rhizobium wild type (2011)
strain. They are phosphatidylcholine (PC),
phosphatidylglycerol (PG), phosphatidylethanolamine (PE),

and diphosphatidylglycerol (DPG).

148

 

Figure 4.8

149

Figure 4.9: Autoradiogram corresponding to TLC presented
in Figure 4.8 (not same scale) of membrane extracts from
members of the Rhizobiaceae (Strains are identified in text).
Two radiolabelled bands can be seen with similar mobilities to
the SQDG standard in all strains except the E. coli control
and the A. tumefaciens sample. The minor radiolabelled
component (X) has yet to be identified. The standard was not

radiolabelled.

150

 

Figure 4.9

151

extraction. The other major bands, however, had similar mobilities to
those observed here. Extracted radioactive TLC bands were also
analyzed by Fourier transform infrared spectroscopy (Figure 4.10). All

samples as well as the standard showed broad absorption bands for the

hydroxyl groups at about 3400cm‘1 along with strong bands for CH2 and
CH3 (2800-2950cm'1), for acyl ester groups (1750cm'1) and overlapping

bands in the region for sulfonic acid groups (1000-1400cm’1). These values

agree with previously published data (179).

152

Figure 4.10: Fourier transform infrared spectra of (A) SQDG
standard and (B) extract from putative SQDG TLC band of
the R. meliloti 2011 sample (spectra from the other samples
were identical). The major absorption bands from the
standard can also be seen in the sample and are identified in
the text. The bands at 1400 and 1600 cm" of the sample

spectrum are due to a contaminant.

STnonamtttlnoo

”TFInIM1999009

93.193

99.190 97.972

94.997

91.703

90.211

91.691 93.093 94.576 99.097 97.659

90.099

153

 

 

 

 

l
—I

3333.2 3310.3 3222.4 2334.4 2443.3 2033.3 i37o.3 1232.7 324.7 303.33
Havonuabor (ca-1)

 

 

 

 

 

 

 

 

 

3233.2 33:0.3 3222.4 2334.4 2443.3 2033.3 1370.3 1232.7 334.77 303 33
u3v3nu-bnr (cm-1)

Figure 4.10

154

DISCUSSION

It has been shown in these studies that sulfoquinovosyl
diacylglycerol (SQDG) is a membrane lipid in certain cultured rhizobial
species. The occurrence of SQDG outside of plants or photosynthetic
bacteria has virtually never been reported before. A strict exclusivity of
this lipid to photosynthetic systems has generally been assumed. It is,
therefore, remarkable that this lipid component is found in a Gram-
negative bacteria. It is at once puzzling and understandable that it is
found in a plant symbiont. Several questions immediately arise in
connection with this finding. The first one is: Do these lipids have any
functional role in symbiosis? It is tempting to propose that, perhaps, the
presence of this lipid might be necessary in order to ensure some
compatibility between the plant and bacterial membranes during
symbiosis. It has been known for many years that another typical plant
lipid, phosphatidylcholine (PC), exists in the membrane of members of the
Rhizobiaceae. However, a role for PC in symbiosis and/ or pathogenesis
has not yet been elucidated. The second question is: What is the origin of
the genes responsible for biosynthesis of SQDG; were they transferred
from the plant or were they developed in the bacterium? Based on zero

evidence in either direction, the former scenario seems much more

155

palatable since the rhizobia and legumes have shared a long association
and there has been sufficient chance for this to have occurred. Another
alternative is that these bacteria were, at one stage, plant organelles.
With an aim of determining the generality of SQDG amongst
legume bacterial symbionts, included in this study were a representative
sampling of the family Rhizobiaceae. R. leguminosarum bv. trifolii
ANU843 and R. sp. strain NGR234 are wild type strains that represent a
variety in the range of Rhizobium host specificity. The finding of SQDG
in these organisms increases the consequence of the previous finding and
strengthens the possibility of the widespread importance of this lipid in
symbiosis. R. leguminosarum bv. trifolii ANU845, which lacks the
"symbiotic plasmid", was studied in an effort to determine whether the
pSym genes are involved in the biosynthesis of SQDG thus indicating a
possible role in nodulation as opposed to a later stage in infection. The
presence of SQDG in this mutant suggests that this lipid may have a role
(if any) during or after bacteroid formation, possibly by ensuring
membrane compatibility between the bacterium and the plant thus
allowing for release from the infection thread. More studies will need to
be performed on other mutant strains that lack this ability. B. japonicum
was studied because of its genetic dissimilarity to R. meliloti yet similarity
in the function of symbiosis. The lower amounts of SQDG found in B.

japonicum does not disallow the possibility of its importance to symbiosis

156

in this organism. The possibility exists that SQDG is a major component
only of the bacteroid form and that the levels of expression observed here
are background. Further studies on the presence of this lipid in bacteroid
membranes are ongoing. Finally, A. tumefaciens was investigated to
probe the possibility that SQDG is a key determinant in the distinction of
symbiosis from pathogenesis. Although there appeared to be an orcinol
positive band corresponding to the faster migrating SQDG component in
the TLC analysis for the Agrobacterium species, this band did not show
any detectable levels of radioactivity either by autoradiography or
scintillation counting. Despite the obvious difference in function,
Agrobacterium and Rhizobium species have been shown to be
indistinguishable phenotypically (182) and genetically by DNA-rRNA
hybridization experiments (183) and 16S rRN A sequence analysis (184).
Since this species is genetically quite similar to R. meliloti, the lack of
SQDG suggests that this lipid plays a role in symbiosis. Although this
study was not exhaustive in the strains tested, it does reinforce the
possibilities of a role for sulfoquinovosyl diacylglycerol in the

Rhizobium / legume relationship.

157

ACKNOWLEDGMENTS

Nuclear magnetic resonance studies related to the characterization
of sulfoquinovosyl diacylglycerol were performed by Rawle I.
Hollingsworth. This work was supported by a grant (DE-F602-
89ER14029) from the US. Department of Energy. The NMR data were
obtained on instrumentation that was purchased in part with funds from
NIH grant #1-SlO-RRO4750, NSF grant #CHE-88-00770, and NSF grant

#CHE-92-13241.

CHAPTER 5

INVESTIGATION OF THE BIOSYNTHESIS OF

SULFOQUINOVOSYL DIACYLG LYCEROL IN

RHIZOBILUVI MELILO'H

158

159

ABSTRACT

Investigations into the biosynthetic pathway of sulfoquinovosyl
diacylglycerol in Rhizobium meliloti were performed. One and two-
dimensional chromatography were used to analyze 35S-labelled extracts
of R. meliloti and other related organisms. These experiments revealed
major differences in the labelled metabolites of these various strains
attributable to the variety of their sulfur-containing glycolipids (i.e.,
lipopolysaccharide, chitolipooligosaccharide and sulfoquinovosyl diacyl-
glycerol). The identity of these components in R. meliloti was probed by
qualitative tests on the extract followed by one-dimensional
chromatography. Two compounds were identified that showed resistance
to acid and base hydrolyses, cleavage by periodate treatment, oxidation by
a bromine/ water solution and no affinity for anion or cation exchange
resins. Based on results of these tests coupled with comparisons to
standards, these two compounds were proposed to be thiolactaldehyde
and thioquinovose. Although these molecules have not previously been
identified in relation to sulfoquinovosyl diacylglycerol synthesis, it is
entirely possible that they would fit into one of the many proposed
pathways. Attempts were made to verify the identity of these compounds

by comparison to synthetic standards but proved unsuccessful.

160

INTRODUCTION

Using 3F’s-labelled extracts of some photosynthetic microorganisms
and higher plants, Benson and coworkers in 1959 were able to isolate
significant amounts of a sulfur-containing lipid and characterize it as 1'-
O-(6'-deoxy—a-D-glucosyl-6'-su1fonic acid)-3-O-diacyl glycerol (Structure
4.1) (185). This lipid, sulfoquinovosyl diacylglycerol (SQDG), has since
been detected in all plants, various algae (including green, blue-green, red
and brown algae), purple bacteria, some photosynthetic bacteria and a
non-photosynthetic marine diatom (186, 187). Although SQDG has been
found to be mainly associated with photosynthetic membranes, it has also
been shown to occur in such diverse non-photosynthetic tissues such as
roots (188), potato tubers (189) and apples (190). Significant amounts have
also been found in etiolated tissue (191). Sulfoquinovosyl diacylglycerol
represents a major portion of the ether-extractable lipids in the organisms
in which it has been identified. For example, it can account for up to 40%
of marine red algae glycolipids (206). The amount of SQDG in the leaves
of higher plants is generally only about 5%, however, the lipid is confined
to the chloroplast lamellar membranes. This confinement of SQDG to
plastids has led to the assumption of a role for the lipid in photosynthesis

(for review see 207). Various functions of the lipid that have been

161
proposed include chlorophyll binding, involvement in electron transport

and association with certain enzymatic activities (207).

Since the discovery of the plant sulfolipid, relatively little progress
has been made towards the elucidation of its biosynthesis. Numerous
theories have been proposed based on analogy to classical pathways but
have not been supported by conclusive evidence. A compilation of some of
these various hypotheses is represented in Figure 5.1. An early study by
Shibuya et al., used radiochromatography and radioelectrophoresis to
analyze 35S-labelled extracts of the green algae Chlorella for the presence
of possible intermediates in SQDG metabolism (192). These workers
found over 50 sulfur-containing compounds, of which at least twenty
contained a sulfonic acid group. Most notable of the identified
compounds were sulfolactate, sulfolactaldehyde, sulfopropanediol,
sulfoquinovose and a nucleoside diphosphate sulfoquinovose. These
results led Benson to propose that the activated sulfoquinovose might
react with diacylglycerol (DAG) to give SQDG (193). It was also
suggested that the sulfoquinovose precursor was synthesized in a
"sulfoglycolytic" pathway; i.e., the aldol condensation of a 3-carbon
sulfonic acid-containing moiety with dihydroxyacetone phosphate
(DHAP). This idea was expanded and outlined in more detail by Davies
and coworkers (194). These researchers found that cysteic acid could act

as a precursor for sulfolipid biosynthesis. They also suggested the possible

Figure 5.1: Compilation of some of the proposed pathways
for the biosynthesis of sulfoquinovosyl diacylglycerol.
Enzymes are indicated where the particular reactions have
been demonstrated. Compounds that are underlined have

been used as precursors. Adapted from ref. 207.

163

am 2..me

 

 

 

3.333303%
30265300230
22......
80m
BannosmA0modbé . 20.80.33.636
2x00058300 . 0530a 13003-0
ImmeIO \ 0<OO~IM_V
row: will: 23... N12%...
_ . :6me N
10%: 33./3.30:3 m _ I 00
:00 Imommzo 12w:
_ I_ X. 1.00 .53. 5......
O u 0 O I 0 -Nmom 343206

sodoazw Imom IIN 1/ \

 

82: .3023.
0443095250 mm... .F<
0 a 0 a a 0 3.. 39030334“ 83>???—

p a E U _ a m -0nmmosm-m -_0:00£mm0£m
1.0210 :6me N5
.230 _ _ a __
IO~I0 IOWI ‘11 u IOwIAm. .mOdOan - Odom—u
Hm 03:0 :«00 :«00 1.00

30.0.10

164
involvement of ATP-sulfurylase due to an inhibitory effect of molybdate

on 358042" incorporation. This implicated either adenosine-5'-

phosphosulfate (APS) or 3'-phospho-adenosine-5'-phosphosulfate (PAPS)
as intermediates. Accordingly, the proposal of Davies et al. begins with a
transfer of the sulfur from the high energy intermediate, PAPS, to
phosphoenolpyruvate followed by various modifications of the 3-carbon
moiety and finally the aldolase reaction suggested by Benson. As
mentioned earlier, however, this theory is supported by very little concrete
evidence.

A second possible pathway in the biosynthesis of 6-sulfoquinovose
centers around the controversial involvement of cysteic acid. The
conclusion of the aforementioned study by Davies et al. was that
incorporation of cysteic acid into SQDG was due to a side reaction or to
degradation products (194). In subsequent work, Haines suggested a more
direct involvement of cysteic acid as an important intermediate and
modified the existing theory (Figure 5.1) (186). This pathway involved the
displacement of an O-acetyl group from O-acetyl serine by a sulfite
moiety made by reduction of the activated sulfate. This theory was based
on 1) results from Hodgson et al. which showed that PAPS was directly
involved in the formation of a sulfite intermediate in Chlorella (195) and 2)
a study showing the availability of O-acetyl serine as an intermediate in

cysteine biosynthesis in Salmonella typhimarium (196). Once again, no

165
direct evidence supported this pathway other than the possible association

of cysteic acid which itself has been controversial. One subsequent study
showed incorporation of labeled cysteic acid into SQDG in alfalfa (197)
while another found no evidence for incorporation in spinach (198).

A novel proposed pathway for the biosynthesis of 6-sulfoquinovose
involves the addition of sulfite to a double bond to yield the sulfonic acid
group (Figure 5.2). Lehmann and Benson were able to synthesize 6-
sulfoquinovose from sulfite and methyl-a-D-glucoseenide in 5 minutes at
room temperature in aqueous solution at pH 6.4 to 7.0 and suggested a
similar reaction could be possible in viva (199, 200). Recent studies by
Benning and Sommerville have identified four genes in the SQDG
biosynthetic pathway, one of which displayed sequence similarity to a
UDP-glucose epimerase (203, 204). The presence of this enzyme is
consistent with the proposed scheme in Figure 5.2. A recent study has also
supplied new biochemical data to support this pathway (205). The
evidence included stimulation of SQDG production by UTP and the lack of
inhibition by compounds likely to repress the "sulfoglycolytic" pathway. In
addition, these authors demonstrated that UDP-[14C]-glucose could be
incorporated into SQDG and that this incorporation was augmented by
the addition of methyl-a-D-glucoseenide. As referred to above, several

studies have shown direct evidence for the involvement of sulfite in

166

Figure 5.2: Proposed pathway for the biosynthesis of
sulfoquinovosyl diacylglycerol. These workers suggested that
the sulfite group could be added either to UDP-4-ketoglucose-
5-ene or to UDP-glucose-5-ene This figure was taken from

reference 205.

W
011.0"

01.0" o o Ho o
OH O‘UDP _= OH O‘UDP OH duo?
on OH OH
UDP-Glc ’ ”DP-G"

I"
'I
'I
I
I
’I
I
I
I
I
CH.OH it
I" 4‘30
o ’
OH Wop I
I
OH 'I
’I
I”
-H.° " [so
"I
CH. -
o ‘80.”
0 o” -_. O
0110? OH 0110?
OH OH
1+2" 1+2”
CH. 0;
O .
o I,
OH 0110? H0 on 0110?
OH OH
Umw'phoqul

l
l
j .
”('3 I
o‘UDP
0H
-. . sulphollpld

. on
WM " of
Figure 5'2

168
various systems (197, 201, 202). However, the role of sulfite was called

into question in a study by Hoppe and Schwenn on cell-free extracts of
Chlamydomonas reinhardtii. While studying the incorporation of possible
precursors into SQDG, these workers found PAPS to follow a normal
substrate saturation curve while sulfite was not saturable and therefore
possibly due to a non-enzymatic artifact (201).

The only steps in the pathway for the biosynthesis of SQDG that
have sufficient evidence to support them are the first one, activation of
sulfate to either APS or PAPS, and the last one, reaction between the
sulfoquinovose nucleoside diphosphate and diacylglycerol. Several
results from various labs have supported the first step and these are
mentioned in detail above. A significant experiment by Heinz et al. seems
to have also definitively proven the final step. Using 14C-labelled
diacylglycerol, these authors found that addition of UDP-sulfoquinovose
increased the amount of label in SQDG while the GDP derivative had
only half the activity and the ADP and CDP derivatives were inactive
(208).

The experiments presented in Chapter 4 proved sulfoquinovosyl
diacylglycerol to be a major component of various rhizobial strains. In
addition, Agrobacterium tumefaciens, a genetically closely related plant
pathogen, was found to be devoid of the lipid. The significance of these

discoveries lies not only in the implications of a role for this lipid in

169
symbiosis, but in the possibilities for clarifying its biosynthetic pathway.

Since it was clear that various rhizobia (and A. tumefaciens) synthesized
different sulfur—containing membrane components, it followed that
intermediates unique to one molecular species may be isolable. In other
words, the results presented in Chapters 2 and 4 show that Rhizobium
meliloti makes SQDG and sulfated lipopolysaccharide and
chitolipooligosaccharide, R. trifolii synthesizes only SQDG while A.
tumefaciens and E. coli have not been found to contain these sulfated
species. It may be possible, therefore, to gain some insight into the
biosynthesis of SQDG by comparing the sulfur-containing metabolites in
R. meliloti and A. tumefaciens, two otherwise closely related bacteria. In
addition, since this problem has generally only been studied in plants, it
may be easier to analyze in a less complex system.

The aforementioned study by Shibuya, et al. used two-dimensional
paper chromatography to identify sulfur-containing biosynthetic inter-
mediates in algae (192). It was supposed that this type of analysis would
yield a "fingerprint" for each organism that could then be compared to the
others. The compounds that differed between the strains could also be
identified by cutting the bands, eluting the sample from the paper and
identifying the component. The information gleaned from these studies
would possibly resolve some of the uncertainties related to the biosyn-

thesis of SQDG and either confirm or reject the theorized pathways.

170

MATERIALS AND METHODS

Strains and bacterial cultures.

The various bacterial strains which were used are as follows:
Rhizobium meliloti 2011 (from Ethan Signer, Massachusetts Institute of
Technology), R. leguminosarum bv. trifolii ANU843 (from Barry Rolfe,
Australian National University), Agrobacterium tumefaciens C58 (from
Frans DeBruijn, Michigan State University), and Escherichia coli TG-l
(from Gregory Zeikus, Michigan State University). All strains of bacteria
(with the exception of E. coli) were grown at 30°C in liquid cultures
containing modified Bergensen's (BIII) media as previously described (142)
except that in addition, 35S-labelled sulfate at a level of IOOuCi/ liter of
culture was added in the form of carrier free sulfuric acid (New England
Nuclear). The E. coli strain was grown at 37°C in MSX4 media with 353-
labelled sulfate added as above. Alfalfa seeds were from the laboratory of

Frank Dazzo, Michigan State University.

One and two dimensional chromatographic analyses of cell extracts.
Cells were lysed with a combination of freeze/ thaw cycles, probe
sonication and lysozyme treatment and then extracted twice each with hot

80% ethanol, absolute ethanol and 50% ethanol. Extracts were condensed

171
to dryness and redissolved in 50% ethanol for chromatography. Alfalfa

was grown on a Fahraeus agar medium (142) containing 358-labelled
sulfate at a level of 200 uCi/ liter of culture. Plants were harvested after
about six days growth. Roots were separated from stems and both
ground and extracted as above. Two-dimensional paper chromatography
was performed using a modified method of Shibuya, et al. (192).
Whatman #1 filter paper (20 x 20 cm) was developed in the first direction
with phenol / water (5:2 w/ w) and in the second direction with butanol /
propionic acid / water (142:71:100 w/w). Radioactivity was visualized
using a Molecular Dynamics Phosphorimager after exposure of the
phosphor screen for approximately ten days. One-dimensional paper

chromatography was run using the mobile phase containing butanol.

Qualitative analyses of cell extracts.

The ethanol extract mentioned above was subjected to various
treatments for qualitative analysis by one-dimensional TLC. Acid and
base hydrolyses were performed using 1% HCl and 0.1% NaOH,
respectively, and heating the sample for 20 minutes at 70°C. Periodate
oxidation was achieved using a 0.1% sodium meta-periodate solution at
room temperature for 15 minutes. The bromine / water reactionwas
performed using a 5% solution in the dark at room temperature for 15

minutes. Anion and cation exchange chromatography were performed

172
using TEAE cellulose and CM cellulose, respectively. Samples were

eluted with water and the flow-through collected. Trifluoroacetic acid
(TFA) hydrolysis was accomplished with a 2M acid solution that was
heated for two hours at 120 °C. Radiolabelled SQDG was acquired from
cells grown in 358-1abelled culture. The cells were extracted with
chloroform / methanol / l-butanol / water (2:1:1:4) and the organic layer,
containing the sulfolipid, was condensed. Sulfoquinovose was prepared
by hydrolyzing a portion of this SQDG-containing organic fraction with
TFA as above. Radiolabelled sulfate was obtained carrier free from
DuPont (Wilmington, DE). Thin-layer chromatography was performed
using ethyl reversed phase layers with a mobile phase consisting of 2-
propanol / ammonium hydroxide (3:2). Carbohydrate spots were
visualized using an orcinol / sulfuric acid solution. Methionine and

cysteine were visualized using a 0.25% ninhydrin in butanol spray.

173

RESULTS AND DISCUSSION

Radiolabelled cultures of Rhizobium meliloti 2011 and R. trifolii
ANU843 were analyzed using a combination of treatments intended to
lyse the cells followed by extraction of the cell debris with ethanol. This
extraction method has previously been shown to be effective in allowing
an analysis of 35S—labelled components by two-dimensional paper
chromatography (192). Figures 5.3 and 5.4 show the phosphorimager
scans of paper chromatograms of the R. meliloti and R. trifolii extracts,
respectively. The chromatogram of the R. meliloti extract is fairly
complex and appears to consist of at least a dozen distinct radiolabelled
components. This is not surprising since R. meliloti would be expected to
have various sets of biosynthetic pathways and intermediates for its
diverse sulfur-containing components (i.e., LPS, CLOS and SQDG).

In contrast to the R. meliloti results, the chromatograph of the R.
trifolii extract shows relatively few distinct labelled components. It is
entirely reasonable that the pattern of sulfur-containing components is
simpler in the R. trifolii extract due to absence of sulfate in the LPS and
CLOS of this organism. Regarding SQDG synthesis, however, these
results may be indicative of two possibilities; either the pathway involves

relatively few sulfur-containing intermediates or the steady-state

174

 

Figure 5.3: Phosphorimager scan of paper chromatogram of

R. meliloti extract. Paper was developed first in the y-
direction with the phenol mobile phase. 0 = origin, F =

solvent front.

$.23 \_‘..'.“l.1\;::/'

42.; 4;.
. , ,_ If" . l
"7"- .5 l.

 

Figure 5.4: Phosphorimager scan of paper chromatogram of
R. trifolii extract. Paper was developed first in the y-direction

with the phenol mobile phase. 0 = origin, F = solvent front.

176
concentrations of these intermediates are below the level of detection.

This latter point illustrates an important caveat about these types of
experiments. Although these results show few radiolabelled components,
the bacteria obviously makes many other compounds bearing sulfur in
order to survive (i.e., cysteine, methionine, coenzyme A, etc.). The steady-
state concentrations of such molecules or their intermediates, however,
must be sufficiently low as to not be detected by the methods employed
here. Accordingly, we must always keep in mind that the intermediates of
SQDG may also be present at levels which are not readily observable.

In order to verify these results, the entire experiment was repeated.
Suprisingly, the pattern of radiolabelled components on the
chromatogram of the R. meliloti extract was vastly different from that
seen in the first experiment. The chromatogram of the R. trifolii extract
appeared again, however, to be virtually identical to the that in the
original. Since this method proved to be too irreproducible for a direct
comparison of the various extracts, a one-dimensional system was
employed. This method was more reproducible and allowed a direct
comparison of the various bacterial and plant extracts.

An example of a one-dimensional paper chromatographic analysis
is shown in Figure 5.5. The extracts from various organisms including R.
meliloti, R. trifolii, Agrobacterizun tumefaciens, Escherichia coli, and

alfalfa were chromatographed concurrently on paper using the l-butanol/

177

 

E. coli R.trifolii R.meliloti Agrobact. Alfalfa root & leaf

Figure 5.5: Phosphorimage of one-dimensional paper

chromatogram of ethanol extracts of various organisms.

178
propionic acid / water (142:71:100) mobile phases from the previous

experiment. As seen in the two-dimensional study, the extract from R.
meliloti contained a significant number of radiolabelled components while
that from R. trifolii had relatively fewer. Several of the bands between
these samples seemed to have similar relative retention times and
therefore may be common intermediates. There were also a couple of
commonalties between R. meliloti and the plant pathogen A. tumefaciens.
It has been shown in several phylogenetic studies that this bacterial
species is genetically very closely related to R. meliloti (182-184). A.
tumefaciens was shown in Chapter 4 to be devoid of SQDG and, perhaps,
as a result displays a much simpler pattern of radiolabelled components.
It is quite possible, however, that this strain could contain some of the
same intermediates if not part of the pathway for making SQDG as R.
meliloti, but does not produce the sulfolipid itself.

The E. coli strain was tested in order to compare this "typical"
Gram-negative bacterium to the rhizobia. The chromatogram from this
sample showed a single broad band near the origin that is most likely
composed of free sulfate and perhaps other components with similar
mobilities. This seems to agree with results from Chapter 4 that
demonstrated the absence of sulfur-containing lipids in this strain. As

mentioned above, however, these results may not reflect the true level of

179
sulfur-containing components due to the transience of biosynthetic

intermediates.

The alfalfa was tested in order to see how related the pattern of
radiolabelled components in its extract compared to that of its symbiont,
R. meliloti. The root and leaf were segregated to determine what
difference, if any, existed between them especially since the latter would
be expected to harbor most of the SQDG. Each extract, however, showed
a relatively simple pattern of bands and had some overlap to those in the
R. meliloti sample. These results are encouraging since one possibility for
the appearance of SQDG in rhizobia, as discussed in Chapter 4, could be
because of horizontal gene transfer. If this were true, then the same
intermediates would exist in both the bacteria and its host.

In order to begin to characterize some of these sulfur-containing
metabolites, qualitative analytical tests were performed on the extract of
R. meliloti. The treatments, to which the 355-1abelled extract was
subjected, included mild acid and base hydrolyses, periodate cleavage,
bromine oxidation, anion and cation exchange chromatography, and
triflouroacetic acid hydrolysis. The mild acid and base hydrolyses were
intended to reveal the presence of any molecules containing sulfate groups
since they would be expected to be cleaved under these conditions. The
resulting disappearance of a band in a particular lane of the paper

chromatograph compared to the untreated extract would indicate that

180
that component contained a sulfate group. The periodate oxidation

reaction was used to reveal the possible presence of carbohydrates or
polyalcohols. The outcome of this reaction is the cleavage of the bond
between carbon atoms bearing hydroxyl groups and subsequent oxidation
of the hydroxyls to aldehydes. Any molecule containing vicinal hydroxyls
will be cleaved and therefore have its chromatographic properties altered.
A bromine solution will oxidize aldoses nearly quantitatively by
converting them to the corresponding lactone while leaving ketoses
unaffected. This test was meant to provide some insight into what
particular types of radiolabelled sugars were present in the extract. Small
amounts of the extract were also passed through anion and cation
exchange columns in order to detect what types of charged molecules
might be present. This would help define the oxidation state of the sulfur
in the radiolabelled bands. Most of the sulfur-bearing components
involved in SQDG biosynthesis would likely remain on the anion
exchange column since they would be expected to be in the form of the
initial state, sulfate, or the final one, sulfonic acid. A compound that
passed through the anion exchange would either contain an uncharged
sulfur-containing group, such as a thiol or thioether, or have an overall
positive charge. The triflouroacetic acid treatment, in addition to
indicating the presence of sulfate groups as with the mild hydrolysis,

would be expected to cleave any glycosidic linkages that might be present.

181

Figure 5.6: Phosphorimage of one-dimensional paper
chromatogram of ethanol extract of R. meliloti after various
treatments (See Materials and Methods for complete
description). The glucose and methionine were run in a
different experiment than the samples using the identical

method.

 

3..

an
AV.

34 fl. ...

134%.; . n.“

_ we»... ..
1&fi‘ufi .9:fo

.

.7'.

 

30430553

«78000

0.00 «3:08

mcmoafiuogmm

mOUO

Hm». 5.9.0qu

8005 883.3%

030:. «wag—mm
9.330
00100.05
200m
300.3an

mg

5.6

igure

F

183
The products of each treatment were chromatographed on paper

one-dimensionally as above along with standards. The standards used
included 35S—labelled SQDG, sulfoquinovose, and free sulfate as well as
unlabelled methionine, cysteine, and glucose. An example of the paper
chromatography is shown in Figure 5.6. As expected, the phosphorimage
of the mild hydrolysis lanes appeared similar, however, the resolution of
the various bands was generally poor. It appears as though all of the
bands in these sample lanes lost intensity compared to the untreated
sample, except, of course, for the band corresponding to the free sulfate.
Suprisingly, the results were similar for the samples treated with
periodate and bromine which were intended to show the presence of
vicinal hydroxyl groups and reducing sugars, respectively. Although these
tests would not be expected to produce the same effect on the extract, the
results were generally reproducible.

More interesting results were obtained for the samples subjected to
ion exchange chromatography. In the case of the extract passed through
the anion exchange column, only two distinct bands appeared in the
phosphorimage of the paper chromatogram. Interestingly, these two
bands were also detected, albeit slightly less resolved, in the sample
exposed to the cation exchange column. This would imply that these
components are uncharged and that the sulfur might be in the form of a

thiol group. These two compounds were also found to be stable to

184
triflouroacetic acid hydrolysis which would also be consistent with the

presence of thiol moieties. Neither of the components had relative
retention times matching the standard SQDG, sulfoquinovose, sulfate,
methonine or cysteine. The slower moving compound, however, did align
fairly well with the glucose standard. Based on this it seems that a likely
candidate for the slower moving component could be thioquinovose
(Figure 5.7). This compound would definitely have a very similar mobility
on the chromatogram as glucose and fit very well with the additional data
presented above (i.e., resistant to acid hydrolysis, flow-through ion
exchange columns, and possible susceptibility to periodate cleavage and
bromine oxidation). Furthermore, based on an analogy to the proposed
"sulfoglycolytic" pathway (Figure 5.1) it can then hypothesized that the
other molecular species in question could be thiolactaldehyde. This
compound should combine easily with dihydroxyacetone phosphate to
form thioquinovose (Figure 5.7). In order to verify this, attempts were
directed at synthesizing these compounds in order to compare them
directly on the chromatogram but these efforts proved unsuccessful. In
order to increase resolution and make sample retrieval easier, a thin layer
chromatography (TLC) system was developed. A fresh bacterial culture
was grown and extracted identically as above. The TLC analysis,
however, showed vastly different results than before in that only two

radiolabelled bands were detected. The slower moving component tested

185

?H20
H O H OH
CH OPO -
$0: | 2 3 Ho H
H + = ——"’ —” H OH
CHZSH CHZOH H OH
thiolactaldehyde DHAP (f
CHZSH
thioquinovose

Figure 5.7: Proposed scheme of thioquinovose formation

from thiolactaldehyde condensation with DHAP.

positive for the presence of carbohydrate by staining pink after reaction
with orcinol. It was unresolved, however, from other carbohydrate
components with similar mobilities and therefore not easily isolable. The
faster moving radiolabelled component was most likely SQDG due to its
mobility (very near the solvent front) along with its reaction to orcinol

(stained a more brownish color indicative of a glycolipid).

CONCLUSIONS

Based on the data obtained in this study a hypothesis was made that
thiolactaldehyde and thioquinovose could be possible intermediates in
SQDG synthesis. This not only leaves the question of SQDG biosynthesis

unresolved, but it may even increase the confusion since these types of

186
intermediates (i.e., containing the reduced form of sulfur) have not yet

been proposed. It is, however, entirely possible for incorporation of sulfur
into the pathway at the thiol level followed by a final oxidation step from
thioquinovose to sulfoquinovose. In fact, unpublished results from the
Hollingsworth lab have indicated the incorporation of 35S into SQDG
using radiolabelled cysteine as the only form of sulfur in the media. This
would be consistent with a pathway analogous to the one presented in
Figure 5.1 with cysteine replacing the cysteic acid. The possibility also
remains that there exists more than one pathway for the biosynthesis of
SQDG. Given the evidence for such disparate scenarios as those shown
in Figures 5.1 and 5.2, this concept seems more than likely.

The results presented in this chapter are perhaps reflective of why
the question of SQDG biosynthesis has remained a mystery for so long;
the transience of biosynthetic intermediates makes them very difficult to
isolate or reproduce from culture to culture using wild type strains. This is
demonstrated in the results from the extract of the E. coli strain which
show the lack of any significant amounts of sulfur-containing compounds.
Generally, strains carrying mutations that allow intermediates to
accumulate are used in these types of studies. Perhaps the recent
discovery of four genes related to SQDG biosynthesis (203, 204) will lead

to such mutants and a solution to this long-studied question.

187

ACKNOWLEDGMENT

This work was supported by a grant (DE-FG02-89ER14029) from

the US. Department of Energy.

CHAPTER 6

RETROSI’ECT

188

189

Over the years, the cell-surface carbohydrates of Rhizobium have
been the central focus of our attempts to understand the important
symbiotic interactions of these bacteria with their host leguminous plants.
Previous ideas regarding the roles of rhizobial polysaccharides and
glycolipids had these molecules each providing a specific function to the
infection process. For instance, it has been generally believed that the role
of the extracellular polysaccharides in infection is to assure bacterial
release from the infection thread and nodule invasion. It has also been
suggested that the role of the capsular polysaccharides is to provide a
recognition component to the lectins which reside on the root surface. A
belief even more passionately held than these is that the
chitolipooligosaccharides are the sole determinants of host specificity and
that the genes involved in their synthesis are exclusive to them. The
studies presented in this dissertation begin to dispel these ideas and
provide the basis for a comprehensive model in which the key determinant
of host specificity and infection is the entire rhizobial membrane
distribution and not individual molecular species.

Since their discovery, the chitolipooligosaccharides (i.e., CLOS or
Nod factors) have garnered a large proportion of the attention in the area
of rhizobial research. These molecules are oligomers of N-

acetylglucosamine with a fatty acid in place of the acetyl group on the

190
non-reducing terminus. Various modifications to this basic structure,

including sulfation and fatty acid unsaturations, are presumed to be the
absolute determinants of the bacteria’s ability to infect the proper host.
This assertion was made on the lone observation that exogenous addition
. of CLOS to plant roots causes morphological changes that mimic those
seen in the early stage of an authentic infection. This hardly reflects the
actual situation of a bacteria/ root interaction since it is well known that
the bacterium will bind to a root hair before it actually begins to curl (209).
Why, then, would rhizobia need to synthesize a ”diffusable metabolite”
such as CLOS if it is already in close contact with the root? This fact,
coupled with the basic structure of CLOS and other points raised in
Chapter 2, would intuitively lead one to believe that these molecules are
membrane localized. In these studies, we verified this by isolating ten-fold
more Nod factor from the membrane of a wild-type rhizobial strain than
other researchers have been able to extract from the culture medium of
genetically engineered induced bacteria. The CLOS from these latter
results was undoubtedly the result of normal blebbing of the membrane
during cell divisions. Small amounts of both inner and outer membrane
components are extractable from culture supernatants due to this
phenomenon (2] 4).

The exclusivity of the nod genes and Nod factors to host range

selection has always been a questionable hypothesis in our minds. The

191
definitive reason for our skepticism is the inability of anyone to show that

a non-infective Nod" mutant could be ”rescued” by the addition of purified
Nod factors. The only explanation for the constant failure of this
experiment is that the process of host selection is more complex than is
generally thought. Either CLOS is not the sole determinant of host range
or the nod genes are not responsible for only CLOS synthesis. Results
presented here prove that the former scenario is highly probable and the
latter is definite. The CLOS produced by Rhizobium meliloti has a sulfate
group on the reducing sugar and it is this substituent that is believed to be
responsible for the ability of this bacteria to infect its proper host, alfalfa.
Strains of this bacteria with mutations in the nod genes responsible for the
attachment of this sulfate group to the Nod factor have a concomitant loss
in their ability to nodulate alfalfa. It has been shown here that these
strains also have the identical impairment in their ability to attach a
sulfate group to their lipopolysaccharide (LPS). This is irrefutable
evidence that directly links structure, function and genetic control of the
chitolipooligosaccharides with another family of cell-surface molecules.
An interesting, albeit challenging avenue to pursue would be the
total characterization of the O-antigen of R. meliloti 2011. This is not a
trivial undertaking as reflected by the fact that only one other rhizobial
LPS has been characterized. The challenge is magnified by the lability of

the sulfate linkage. It would be extremely difficult to pinpoint its location

192
in the molecule due to losses during the isolation process. Nevertheless,

interesting questions remain regarding the location of the sulfate ester.
For instance, does the O-antigen repeat contain a reducing terminal
glucosamine residue and if so, is the sulfate attached as in the Nod factor?
Also, is the sulfate attached to the diglucosamine in the lipid A and if this is
the case, what functionality provides the negative charge to take its place
in the case of the 11on and nodH mutants?

To prove the universality of the concept of membrane structural
overlap in rhizobia, it was demonstrated that another strain that
synthesizes a sulfated CLOS, R. sp. strain NGR234, also makes LPS that
contains sulfate. Furthermore, the sulfate group is not the only
functionality involved in the structural overlap. It was also shown that
there is an interrelation in the fatty acid distribution and carbohydrate
substituents between the CLOS and other membrane components in these
strains. These types of commonalties in the distribution of fatty acids
between lipid components of the same membrane is a well documented
phenomenon (146, 147). It is ridiculous, therefore, to propose that specific
fatty acids of one molecular species, the chitolipooligosaccharides, are
responsible for directing host range.

In order to probe further this flexibility in fatty acid distribution, R.
trifolii 48 cells were grown in a culture medium containing parinaric acid.

This fatty acid is a conjugated polyene which could be detected easily due

193
to its spectroscopic properties. It was thought that the cells would take up

the fatty acid and incorporate it into its membrane components as was
shown in an earlier study using E. coli (215). Unfortunately, using several
different methodologies, chitolipooligosaccharides were not successfully
isolated from this strain. Perhaps in the future other strains from which
we have already isolated CLOS, e.g. R. meliloti 2011 or R. trifolii
ANU843, might be a better choice for such a study.

The results presented in Chapter 2 paint a picture that depicts the
membrane as a highly integrated web of biosynthesis. Altering one
specific gene will not necessarily result in the perturbation of one
molecular species but will have a cascade effect throughout the
membrane. This hypothesis was tested by comparing the total membrane
extracts of the various non mutants that were used in Chapter 2 with
the wild type strain. The putative function of the 71on gene is to control
sulfation of the chitolipooligosaccharide by making the activated sulfate
transfer molecule. When the total membrane extracts of these mutant
strains were compared, however, important differences seemingly
unrelated to sulfation were observed (Figure 6.1)‘. Primarily, the changes
seen were in the relative intensities of the various bands, although at least

one component was detected in the mutants that was not seen in the wild

 

1 For information regarding the strains used in this experiment, see Chapter 2. For the extraction procedure
and chromatographic conditions, see Chapter 4.

194

 

 

 

 

 

10-102901!
-zOrou
-mpou

.75
3

SE.
'6‘
'2‘.
N
O
D—‘
t-J

Figure 6.1: Thin layer chromatogram of membrane extracts

of three 11on mutant strains along with the parent wild type
strain. Note the differences in relative intensities of various
bands and presence of a novel band in the mutants (arrow).
Weak bands which did not show up when the plate was copied

are indicated. Bands were visualized with orcinol.

195
type bacteria. The effects of these mutations, therefore, are much more

far-reaching than is generally thought. Not only are the levels of the
various lipids affected, but the machinery to make molecules undetected in
the wild type is suddenly turned on. It is rather absurd, then, to propose
that a single gene and the associated CLOS structural modification are
solely responsible for host specificity and not take into account the
dynamic nature of the membrane. Studies such as this may be expanded in
the future in order to possibly identify the novel components in the
mutants and quantitate more carefully the changes in the various lipid
species produced.

Another example of structural overlap of membrane components
important to rhizobia is presented in Chapter 3. These studies focused on
an unusual. strain, R. trifolii 48, that has a membrane composition very
distinct from closely related bacteria which have the same host range. In
particular, this strain lacks one of the two typical LPS core regions
produced in its species. It also produces an LPS O-antigen and
extracellular polysaccharide that are very unique. Questions therefore
arose as to the molecular basis for the infectability of this strain. The
isolation and partial characterization of a novel glycolipid is presented
here which may answer these questions. The molecule contains a
polysaccharide with a composition that mimics the ”typical” LPS O-

antigen of this species. This polysaccharide appears to be linked to the

196
membrane by an assortment of fatty acids usually found in various

membrane components. It therefore seems that although the membrane
has an inconsistency in its ”normal” cell-surface carbohydrate
biosynthesis, it has found a way to circumvent this problem and make a
new glycolipid that apparently corrects for the putative defect. This
example demonstrates the power that the membrane affords the bacteria
in its ability to adapt. It also suggests again the importance of general
membrane distribution versus the specific structure of any one component.

A further important aspect of the rhizobia/ legume symbiosis,
besides host specificity, is the ability of the invading organism to avoid the
plant’ 5 defense mechanisms. In some way the bacterium must be able to
disguise itself in order for the plant to disregard it as a hostile agent. A
substantial step was taken in this thesis regarding a possible molecular
mechanism to explain this phenomenon. Sulfoquinovosyl diacylglycerol
(SQDG) was shown to be a membrane component of several rhizobial
species. The significance of these findings lies in the fact that this lipid was
previously thought to have been restricted to photosynthetic organisms.
The occurrence of a plant lipid in Rhizobium may play a critical role in
ensuring an avoidance of plant defenses. These implications are
magnified when considering the recent discovery in the Hollingsworth lab
of other classes of plant lipids in a rhizobial species; the mono- and

digalactosyl diacylglycerols as well as phosphatidyl inositol. In addition,

197
another typical plant lipid, phosphatidylcholine, has been known to exist

in rhizobia for years. Consequently, the rhizobial membrane may appear
to the host plant to look more like one of its own organelles than an
invading bacterium. It is interesting to note that in the case of the mono-
and digalactosyl diacylglycerols and phosphatidyl inositol, the synthesis of
these lipids are augmented once the bacteria are subjected to low oxygen
or other conditions that would simulate the environment of the bacteroid
state. This is an important detail since it is at this point, after the bacteria
have been released from the infection thread into the cell, in which they
should be most vulnerable to host defenses. We see once again, therefore,
the adaptability of the membrane and the criticality of the entire spectrum
of lipids in the rhizobial membrane in determining a viable symbiotic
interaction.

The final experiments in this dissertation were directed at
elucidating the unresolved biosynthetic pathway of the plant sulfolipid,
SQDG. Based on these studies, two compounds, thioquinovose and
thiolactaldehyde, were proposed as possible intermediates in a variation
of the long-standing ”sulfoglycolytic” pathway. The intention of these
experiments were to delineate the pathway in order to eventually surmise
the origin of the genes involved. The complexity of SQDG biosynthesis
dictates that a simple scenario of molecular mimicry would not be

sufficient. Multiple genes in the bacterium most likely would have had to

198
be transferred from the plant at some point in their evolutionary

coexistence. The likelihood of horizontal gene transfer has, in fact, been
demonstrated in the Rhizobiaceae (210) as well as in other systems where
there is close contact between eukaryotes and prokaryotes (211-213). The
possibility also exists that rhizobia were at one point plant organelles and
subsequently developed the ability to become independent of the plant.
This concept is becoming increasingly plausible as we are discovering
more about the rhizobial membrane and its similarities to that of the
plant. These possibilities can only be proven by the isolation and
comparison of plant and bacterial genes responsible for common links
such as SQDG. Therefore, future studies need to be aimed at

investigating plant lipid biosynthesis as well as the rhizobial system.

BIBLIOGRAPHY

10.

11.

12.

199

BIBLIOGRAPHY

Dixon, R. O. D. and Wheeler, C. T. (1986) in Nitrogen Fixation in
Plants, p. 9. (Chapman and Hall, New York).

Yao, P. Y. and Vincent, J. M. (1969) Aust. I. Biol Sci., 22, 413-423.

Dazzo, F. B. and Gardiol, A. (1984) In Genes involved in microbe
plant interactions, pp. 3—31. (Springer Publishers, New York).

Libbenga, K. R. and Harkes, P. A. A. (1973) Planta, 114, 17-28.

Caetano-Anolles, G., Crist-Estes, D. K. and Bauer, W. D. (1988) I.
Bacteriol., 170, 3164-3169.

Bergman, K., Gulash-Hofee, l\/I., Hovestadt, R. E., Larosiliere, R.
C., Ronco, P. G. and Su, L. (1988) I. Bacteriol., 170, 3249-3254.

Callaham, D. A. and Torrey, I. G. (1981) Can. I. Bot., 59, 1647-1664.

Robertson, I. G., Lyttleton, P., Bullivant, S. and Grayston, G. F.
(1978) I. Cell Sci., 30, 129-149.

Dixon, R. O. D. (1967) Arch. Microbiol., 56, 156-166.

Djordjevic, M. A. and Weinman, I. I. (1991) Aust. I. Plant. Physiol,
18, 543-557.

Hollingsworth, R. I. and Carlson, R. W. (1989) I. Biol. Chem, 264,
9300—9303.

Bhat, U. R., Mayer, I-l., Yokota, A., Hollingsworth, R. I. and
Carlson, R. W. (1991) I. Bacteriol, 173, 2155—2159.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

200

Gil-Serrano, A. M., Gonzalez-Jimenez, I., Tejero-Mateo, P.,
Megias, M. and Romero-Vazquez, M. I. (1994) I. Bacteriol., 176,
2454-2457.

Hollingsworth, R. I. and Lill-Elghanian, D. A. (1989) I. Biol. Chem,
264, 14039-14042.

Hollingsworth, R. I. and Lill-Elghanian, D. A. (1990) in Cellular and
molecular aspects of endotoxin reactions, pp. 73-84. (Nowotny, A.,
Spitzer, I. I. and Ziegler, E. I., eds.) Elsevier, The Netherlands.

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

Bhat, U. R., Forsberg, L. S. and Carlson, R. W. (1994) I. Biol. Chem,
269, 14402-14410.

Carlson, R. W., Hollingsworth, R. I. and Dazzo, F. B. (1988)
Carbohydr. Res., 176, 127-135.

Hollingsworth, R. I., Carlson, R. W., Garcia, F. and Gage, D. (1989)
1. Biol. Chem, 264, 9294-9299.

Hollingsworth, R. 1., Carlson, R. W., Garcia, F. and Gage, D. (1989)
I. Biol. Chem, 265, 12752.

Bhat, U. R., Bhagyalakshmi, S. K. and Carlson, R. W. (1991)
Carbohydr. Res., 220, 219-227.

Zhang, Y., Hollingsworth, R. I. and Priefer, U. B. (1992) Carbohydr.
Res., 231, 261-271.

Wang, Y. (1995) Analysis of the molecular basis for the
supramolecular structure of lipopolysaccharides. Doctoral Thesis,
Michigan State University, Department of Chemistry.

Peterson, A. A. and McGroarty, E. I. (1985) I. Bacteriol., 162, 738-
745.

Peterson, A. A., I-Iaug, A. and McGroarty, E. I. (1986) I. Bacteriol.,
165, 116-122.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

201

Cava, I. R., Elias, P. M., Turowski, D. A. and Noel, K. D. (1989) I.
Bacteriol., 171, 8-15.

De Maagd, R. A., Rao, A. S., Mulders, I. H. M., Goosen-De R00, L.,
Van Loosdrecht, M. C. M., Wijffelman, C. A. and Lugtenberg, B. I. I.
(1989) I. Bacteriol., 171, 1143-1150.

Clover, R. H., Kieber, I. and Signer, E. R. (1989) I. Bacteriol., 171,
3961-3967.

Priefer, U. B. (1989) I. Bacteriol., 171, 6161-6168.

Carlson, R. W., Kalembasa, S., Turowski, D., Pachori, P. and Noel,
K. D. (1987) I. Bacteriol., 169, 4923-4928.

Breedveld, M. W., Canter Cremers, H. C. J., Batley, M., Posthumus,
M. A., Zevenhuizen, L. P. T. M., Wijffelman, C. A. and Zehnder, A. I.
B. (1993) I. Bacteriol., 175, 750-757.

Wang, Y. and Hollingsworth, R. I. (1994) Carbohydr. Res., 260, 305-
317.

Leigh, ]. A. and Walker, G. C. (1994) Trends in Genetics, 10, 63-67.
Glazebrook, I. and Walker, G. C. (1989) Cell, 56, 661-672.

Zhan, H., Levery, S. B., Lee, C. C. and Leigh, I. A. (1989) Proc. Natl.
Acad. Sci. USA, 86, 3055-3059.

Philip-Hollingsworth, S., Hollingsworth, R. I. and Dazzo, F. B.
(1989) l- Biol. Chem, 264, 1461-1466.

Hollingsworth, R. I., Abe, M., Dazzo, F. B. and Hallenga, K. (1984)
Carbohydr. Res., 134, C7-C11.

Hollingsworth, R. I., Abe, M., Sherwood, I. E. and Dazzo, F. B.
(1984) I. Bacteriol., 160, 510-516.

Abe, M., Amemura, A. and Higashi, S. (1982) Plant Soil, 64, 315-320.

Miller, K. 1., Kennedy, E. P. and Reinhold, V. N. (1986) Science, 231,
48-51.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55

202

Dylan, T., Helinski, D. R. and Ditta, G. S. (1990) I. Bacteriol., 172,
1400-1408.

Reuhs, B. L., Carlson, R. W. and Kim, I. S. (1993) I. Bacteriol., 175,
3570-3580.

Reuhs, B. L., Carlson, R. W., Kim, I. S. and Williams, M. N. V. (1993)
14th North American Conference on Symbiotic Nitrogen Fixation,
University of Minnesota.

Perotto, S, Brewin, N. I. and Kannenberg, E. L. (1994) Mol. Plant-
Microbe Interact., 7, 99-112.

Battisti, L., Lara, I. C. and Leigh, I. A. (1992) Proc. Natl. Acad. Sci.
USA, 89, 5625-5629.

Hamblin, I. and Kent S. P. (1973) Nat. New Biol., 245, 28.
Bohlool, B. B. and Schmidt, E. L. (1974) Science, 185, 269-271.
Dazzo, F. B. and Hubbell, D. H. (1975) Appl. Microbiol., 30, 172-177.

Dazzo, F. B. and Hubbell, D. H. (1975) Appl. Microbiol., 30, 1017-
1033.

Dazzo, F. B. and Brill, W. I. (1977) Appl. Environ. Microbiol., 33,
132-136.

Dazzo, F. B., Napoli, C. A. and Hubbell, D. H. (1976) Appl. Environ.
Microbiol 32, 166-171.

Dazzo, F. B., Yanke, W. E. and Brill, W. I. (1978) Biochim. Biophys.
Acta, 539, 276-286.

Kato, G., Maruama, Y. and Nakamura, M. (1980) Agric. Biol.
Chem, 44, 2843-2855.

Hughes, T. A., Leece, I. G. and Elkan, G. H. (1979) Appl. Environ.
Microbiol., 37, 1243-1244.

Hughes, T. A. and Elkan, G. H. (1981) Plant Soil 61,87-91.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

203

Wolpert, I. S and Albersheim P. (1976) Biochem. Biophs. Res.
Commun., 70, 729-737.

Hrabak, E. M., Urbano, M. R. and Dazzo, F. B. (1981) .I. Bacteriol.,
148, 697-711 .

Philip-Hollingsworth, S., Hollingsworth, R. I., Dazzo, F. B.,
Djordjevic, M. A. and Rolfe, B. G. (1989) I. Biol. Chem, 264, 5710-
5714.

McNeil, M., Darvill, I., Darvill, A. G., Albersheim, P., van Veen, R.,
Hooykaas, P., Schilperoort, R. and Dell, A. (1986) Carbohydr. Res.,
146, 307-326.

Canter Cremers, H. C. I., Batley, M., Redmond, I. W., Wijfles, A. H.
M., Lugtenberg, B. I. I. and Wijffelman, C. A. (1991) I. Biol. Chem,
266, 9556-9564.

O'Neill, M. A., Darvill, A. G. and Albersheim, P. (1991) I. Biol.
Chem, 266, 9549-9555.

Hollingsworth, R. I., Dazzo, F. B., Hallenga, K. and Musselman, B.
(1988) Carbohydr. Res., 172, 97-112.

Abe, M., Sherwood, I. E., Hollingsworth, R. I. and Dazzo F. B.
(1984) I. Bacteriol., 160, 517-520.

Gardiol, A. E., Hollingsworth, R. I. and Dazzo, F. B. (1987) I.
Bacteriol., 169, 1161-1167.

Bhuvaneswari, T. V., Mills, K. K., Crist, D. K., Evans, W. R. and
Bauer, W. D. (1983) I. Bacteriol., 153, 443-451.

Dazzo, F. B., Truchet, G. L., Sherwood, I. E., Hrabak, E. M., Abe,
M. and Pankratz, S. H. (1984) Appl. Environ. Microbiol., 48, 1140-
1150.

Sherwood, I. E., Vasse, I. M., Dazzo, F. B. and Truchet, G. L. (1984)
I. Bacteriol., 159, 145-152.

Diaz, C. L., Melchers, L. S., Hooykaas, P. J. I., Lugtenberg, B. I. I.
and Kijne, I. W. (1989) Nature, 338, 579-581.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

204

Bishop, P. E., Dazzo, F. B., Applebaum, E. R., Maier, R. I. and Brill,
W. I. (1977) Science, 198, 938-939.

Dazzo, F. B. and Brill, W. I. (1979) I. Bacteriol. 137, 1362-1373.
Kamberger, W. (1979) FEMS Microbiol. Lett., 6, 361-365.

Paau, A. S., Leps, W. T.and Brill, W. I. (1981) Science, 213, 1513-1515.
Maier, R. I. and Brill, W. I. (1978) I. Bacteriol., 133, 1295-1299.

Russa, R., Urbanik, T., Kowalczuk, E. and Lorkiewicz, Z. (1982)
FEMS Lett., 13, 161-165.

Puvanesarajah, V., Schell F. M., Gerhold, D. and Stacey, G. (1987)
I. Bacteriol., 169, 137-141.

Lagares, A., Caetano-Anolles, G., Niehaus, K., Lorenzen, I.,
Ljunggren, H. D., Puhler, A. and Favelukes, G. (1992) I. Bacteriol.,
174, 5941-5952.

Dazzo, F. B., Truchet, G. L., Hollingsworth, R. I., Hrabak, E. M.,
Pankratz, H. S., Philip-Hollingsworth, 8., Salzwedel, I. L.,
Chapman, K., Appenzeller, L., Squartini, A., Gerhold, D. and
Orgambide, G. (1991) I. Bacteriol., 173, 5371-5384.

Hollingsworth, R. 1., Zhang, Y. and Priefer, U. B. (1994) Carbohydr.
Res., 264, 271-280.

Leigh, I. A., Reed, I. W., Hanks, I. F., Hirsch, A. M. and Walker, G.
C. (1987) Cell, 51, 579-587.

Urzainqui, A., Ahmann, D. and Walker, G. C. (1992) I. Bacteriol.,
174, 3403-3406.

Djordjevic, S. P., Chen, H., Batley, M., Redmond, I. W. and Rolfe, B.
G. (1987) I. Bacteriol., 169, 53-60.

Ozga, D. A., Lara, I. C. and Leigh, I. A. (1994) Mol. Plant-Microbe
Interact., 7, 758-765.

Kosch, K., Iacobi, A., Parniske, M., Werner, D. and Muller, P. (1994)
Z. Naturforsch., 49c, 727-736.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

205

Reed, I. W. and Walker, G. C. (1991) Genes and Dev., 5, 2274-2287.

Kondorosi, E., Banfalvi, Z. and Kondorosi, A. (1984) Mol. Gen.
Genet., 193, 445-452.

Fisher, R. F., Tu, I. K. and Long, S. R. (1985) Appl. Environ.
Microbiol., 49, 1432-1435.

Marvel, D. I., Kuldau, G., Hirsch, A., Richards, E., Torrey, I. G. and
Ausubel, F. M. (1985) Proc. Natl. Acad. Sci. LISA, 82, 5841-5845.

Djordjevic, M. A., Schofield, P. R., Ridge, R. W., Morrison, N. A.,
Bassam, B. I., Plazinski, I., Watson, I. M. and Rolfe, B. G. (1985)
Plant Mol. Biol., 4, 147-160.

Dudley, M. E., Iacobs, T. W. and Long, S. R. (1987) Planta, 171, 289-
301.

Rossen, L., Johnston, A. W. B. and Downie, I. A. (1984) Nucl. Acids
Res., 12, 9497-9508.

Debellé, F., Rosenberg, C., Vasse, I., Maillet, F., Martinez, E.,
Dénarié, I. and Truchet, G. (1986) I. Bacteriol., 168, 1075—1086.

Djordjevic, M. A., Schofield, P. R. and Rolfe, B. G. (1985) Mol. Gen.
Genet., 200, 463-471.

Cervantes, E., Sharma, S. B., Maillet, F., Vasse, I., Truchet, G. and
Rosenberg, c. (1989) Molec. Microbiol., 3, 745-755.

Schwedock, I. and Long, S. R. (1989) Molec. Plant-Microbe Interac.,
2, 181-194.

Spaink, H. P., Sheeley, D. M., van Brussel, A. A. N., Glushka, I.,
York, W. S., Tak, T., Geiger, 0., Kennedy, E. P., Reinhold, V. N. and
Lugtenberg, B. I. I. (1991) Nature, 354, 125-130.

Banfalvi, Z., Sakanyan, V., Konzc, C., Kiss, A., Dusha, I. and
Kondorosi, A. (1981) Mol. Gen. Genet., 184, 334-339.

Rosenberg, C., Boistard, P., Dénarié, I. and Casse-Delbart, F.
(1981) Mol. Gen. Genet., 184, 326-333.

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

110.

111.

112.

206

Burkardt, B. and Burkardt, H. I. (1984) I. Mol. Biol., 175, 213-218.

Mulligan, I. T. and Long, S. R. (1985) Proc. Natl. Acad. Sci. USA, 82,
6609-6613.

Innes, R. W., Kuempel, P. L., Plazinski, I., Canter Cremers, H.,
Rolfe, B. G. and Djordjevic, M. A. (1985) Mol. Gen. Genet., 201,
426-432.

Rossen, L., Shearman, C. A., Iohnston, A. W. B. and Downie, I. A.
(1985) EMBO I., 4, 3369-3373.

Horvath, B., Bachem, C. W. B., Schell, I. and Kondorosi, A. (1987)
EMBO I., 6, 841-848.

Peters, N. K., Frost, I. W. and Long, S. R. (1986) Science, 233, 977-
980.

Ceatano-Anolles, G., Crist-Estes, D. K. and Bauer, W. D. (1988) I.
Bacteriol., 170, 3164-3169.

Kosslak, R. M., Bookland, R., Barkei, I., Paaren, H. E. and
Appelbaum, E. R. (1987) Proc. Natl. Acad. Sci. USA, 84, 7428-7432.

Djordjevic, M. A., Redmond, I. W., Batley, M. and Rolfe, B. G.
(1987) EMBO I., 6, 1173-1179.

Thornton, H. and Nicol, H. (1936) Nature, 137, 494-495.

Truchet, G., Michel, M. and Dénarié, I. (1980) I. Differentiation, 16,
163-173.

Finan, T. M., Hirsch, A. M., Leigh, I. A., Iohansen, E., Kuldau, G. A.,
Deegan, S., Walker, G. C. and Signer, E. R. (1985) Cell, 40, 869-877.

Faucher, C., Maillet, F., Vasse, I., Rosenberg, C., van Brussel, A. A.
N ., Truchet, G. and Dénarié, I. (1988) I. Bacteriol., 170, 5489-5499.

Banfalvi, Z. and Kondorosi, A. (1989) Plant Molec. Biol., 13, 1-12.

Faucher, C., Camut, S., Dénarié, I. and Truchet, G. (1989) Molec.
Plant-Microbe Interac., 2, 291-300.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

123.

124.

125.

126.

127.

207

Lerouge, P., Roche, P., Faucher, C., Maillet, F., Truchet, G., Promé,
I. C. and Dénarié, I. (1990) Nature, 344, 781-784.

Roche, P., Lerouge, P, Ponthus, C. and Promé, I.-C. (1991) I. Biol.
Chem, 266, 10933-10940.

Truchet, G., Roche, P., Lerouge, P., Vasse, I., Camut, S., de Billy, F.,
Promé, I.-C. and Dénarié, I. (1991) Nature, 351, 670-673.

Schultze, M., Quiclet-Sire, B., Kondorosi, E., Virelizier, H.,
Glushka, I. N., Endre, G., Gero, S. D. and Kondorosi, A. (1992)
Proc. Natl. Acad. Sci. USA, 89, 192-196.

Iournet, E. P., Pinchon, M., Dedieu, A., de Billy, F., Truchet, G. and
Barker, D. G. (1994) Plant I., 6, 241-249.

Schwedock I. and Long S. R. (1990) Nature, 348, 644-647.

Leyh, T. S., Taylor, I. C. and Markham, G. D. (1988) I. Biol. Chem,
263, 2409-2416.

Roche, P., Debellé, F., Maillet, F., Lerouge, P., Faucher, C., Truchet,
G., Dénarié, I. and Promé, I.-C. (1991) Cell, 67, 1131-1143.

Ervin, S. E. and Hubbell, D. H. (1985) Appl. Environ Microbiol., 49,
61-68.

Hubbell, D. H. (1970) Bot. Gaz., 131, 337-342.
Solheim, B. and Raa, I. (1973) I. Gen. Microbiol., 77, 241-247.

Bhuvaneswari, T. (1984) In Advances in nitrogen fixation research,
pp. 389-396. (Martinus Nijhoff/ Dr. W. Iunk Publishers, The Hague).

Bhuvaneswari, T. and Solheim, B. (1985) Physiol. Plant, 63, 25-34.

Atkinson, E. M. and Long, S. R. (1992) Mol. Plant/Microbe
Interact., 5, 439-442.

Debelle, F., Rosenberg, C. and Dénarié, I. (1992) Mol.
Plant/Microbe Interact., 5, 443-446.

128.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

139.

140.

208

Geremia, R. A., Mergaert, P., Geelen, D., van Montagu, M. and
Holsters, M. (1994) Proc. Natl. Acad. Sci. USA, 91, 2669-2673.

John, M., Schmidt, I., Wieneke, U., Krussmann, H.-D. and Schell, I.
(1988) EMBO I., 7, 583-588.

Rohrig, H., Schmidt, I., Wieneke, U., Kondorosi, E., Barlier, I.,
Schell, I. and John, M. (1994) Proc. Natl. Acad. Sci. USA, 91, 3122—
3126.

Zaat, S. A. I., Wijffelman, C. A., Spaink, H. P., Van Brussel, A. A. N.,
Okker, R. I. H. and Lugtenberg, G. I. I. (1987) I. Bacteriol., 169, 198-
204.

Redmond, I. W., Batley, M., Djordjevic, M. A., Innes, R. W.,
Kuempel, P. L. and Rolfe, B. G. (1986) Nature, 323, 632-634.

Bibb, M. I., Biro, S., Motamedi, H., Collins, J. F. and Hutchinson, C.
R. (1989) EMBO I ., 8, 2727-2738.

Shearman, C. A., Rossen, L., Johnston, A. W. B. and Downie, I. A.
(1986) EMBO I ., 5, 647-652.

Downie, I. A. (1989) Molec. Microbiol., 3, 1649-1651.

Baev, N., Endre, G., Petrovics, G., Banfalvi, Z. and Kondorosi, A.
(1991) Molec. Gen. Genet., 228, 113-124.

Rietschel, E. T., Kirakae, T., Schade, F. U., Mamat, U., Schmidt, G.,
Loppnow, H., Ulmer, A. I., Zahringer, U., Seydel, U., Di Padova, F.,
Schreier, M. and Brade, H. (1994) FASEB I., 8, 217-225.

Price, N. P. I., Relic, B., Talmont, F., Lewin, A., Promé, D., Pueppke,
S. G., Maillet, F., Dénarié, I., Promé, J.-C., 8: Broughton, W. J.
(1992) Molecular Microbiology 6, 3575-3584.

Sanjuan, I., Carlson, R. W., Spaink, H. P., Bhat, U. R., Barbour, W.
M., Glushka, J. and Stacey, G. (1992) Proc. Natl. Acad. Sci. USA, 89,
8789-8793.

Carlson, R. W., Sanjuan I., Bhat, U. R., Glushka, I., Spaink, H. P.,
Wijfjes A. H. M., Van Brussell, A. A. N., Stokkermans, T. J. W.,
Peters, K. and Stacey, G. (1993) I. Biol. Chem, 268, 18372-18381.

141.

142.

143.

144.

145.

146.

147.

148.

149.

150.

151.

152.

153.

154.

155.

209

Schwedock, I. and Long, SR. (1992) Genetics, 132, 899-909.

Dazzo, F. B. (1982) Leguminous root nodules, pp. 431-446. In R.
Burns and T. Slater (eds.), Experimental Microbial Ecology.
Blackwell Scientific Publications, Oxford, U. K.

Williams, M.N.V., Hollingsworth, R.I., Klein, S. 8: Signer, ER.
(1990) I. Bacteriol., 172, 2622-2632.

Cedergren, RA. and Hollingsworth, RI. (1994) I. Lipid Res., 35,
1452-1461.

Geiger, O., Thomas-Oates, I. E., Glushka, I., Spaink, H. P., 8:
Lugtenberg, B.I.J. (1994) I. Biol. Chem, 269, 11090-11097.

Hazel, J. R., and Williams, E. E. (1990) Prag. Lipid Res., 29, 167-227.
Shibuya, I. (1992) Prog. Lipid Res., 31, 245-299.
Weissbach, A. and Hurwitz, I. (1958) I. Biol. Chem, 234, 705-709.

Dubois, M., Gilles, K. A., Hamilton, I. K., Rebers, P. A. and Smith, F.
(1956) Anal. Chem, 28, 350-356.

Leigh, I. A., Signer, E. R. and Walker, G. C. (1985) Proc. Natl. Acad.
Sci. USA., 82, 6231-6235.

Stacey, G., So, I.-S., Roth, L. E., Bhagya Lakshmi, S. K. and
Carlson, R. W. (1991) Mol. Plant-Microbe Interact., 4, 332-340.

Relic, B., Perret, X., Estrada-Garcia, M. T., Kopcinska, I.,
Golinowski, W., Krishnan, H. B., Pueppke, S. G. and Broughton, W.
I. (1994) Mol. Microbiol., 13, 171-178.

Spaink, H. P. (1992) Plant Mol. Biol., 20, 977-986.

Breedveld, M. W. and Miller, K. I. (1994) Microbiol. Rev., 58, 145-
161.

Orgambide, G. G., Hollingsworth, R. I. and Dazzo, F. B. (1992)
Carbohydr. Res., 233, 151-159.

156.

157.

158.

159.

160.

161.

162.

163.

1 64.

165.

166.

1 67.

168.

169.

170.

210

Orgambide, G. G., Philip-Hollingsworth, S., Hollingsworth, R. I.,
and Dazzo, F. B. (1994) I. Bacteriol., 176, 4338-4347.

Reuhs, B. L., Williams, M. N. V., Kim, I. 8., Carlson, R. W. and Coté,
F. (1995) I. Bacteriol., 177, 4289-4296.

Aragén, V., Diaz, R., Moreno, E. and Moriyon, I. (1996) I.
Bacteriol., 178, 1070-1079.

Beynon, L. M., Dumanski, A. I., McLean, R. J. C., MacLean, L. L.,
Richards, I. C. and Perry, M. B. (1992) I. Bacteriol., 174, 2172-2177.

Kogan, G., Jann, B. and Jann, K. (1992) FEMS Microbiol. Lett., 91,
135-140.

Hollingsworth, R. I., Dazzo, F. B. and Mort, A. I. (1987) I.
Bacteriol., 169, 3369-3371.

Amemura, A., Harada, T., Abe, M. and Higashi, S. (1983)
Carbohydr. Res., 115, 165-174.

Carlson, R. W., Sanders, R. E., Napoli, C. and Albersheim, P. (1978)
Plant Physiol., 62, 912-917.

Carlson, R. W. (1984) I. Bacteriol., 158, 1012-1017.

Zevenhuizen, L. P. T. M., Scholten-Koerselman, I. and Posthumus,
M. A. (1980) Arch. Microbiol., 125, 1-8.

Carlson, R. W., Garcia, R, Noel, D. and Hollingsworth, R. I. (1989)
Carbohydr. Res., 195, 101-110.

Weissbach, A. and Hurwitz, I. (1958) I. Biol. Chem, 234, 705-709.

Dell, A., York, W. S., McNeil, M., Darvill, A. G. and Albersheim, P.
(1983) Carbohydr. Res., 117, 185-200.

Williams, M. N. V., Hollingsworth, R. I., Brzoska, P. M. and Signer,
E. R. (1990) I. Bacteriol., 172, 6596-6598.

Goldfine, H and Ellis, ME. (1964) I. Bacterial, 87, 8-15.

171.

172.

173.

174.

175.

176.

177.

178.

179.

180.

181.

182.

183.

184.

211

Thompson, E.A., Kaufman, A.E., Johnston, NC, and Goldfine, H.
(1983) Lipids, 18, 602-606.

Wilkinson, S.G. (1988) Gram-negative bacteria, pp. 299-488. In C.
Ratledge, and S.G. Wilkinson, (eds.), Microbial lipids. Vol. 1.
Academic Press, San Diego, California.

Vijn, I., das Neves L., van Kammen, A., Franssen, H., and Bisseling,
T. (1993) Science, 260, 1764-1765.

Harwood,I.L. (1980) Sulfolipids. pp. 301-320, In P.K. Stumpf, and
EB. Conn, (eds.), The biochemistry of plants, a comprehensive
treatise. Vol. 4. Lipids: structure and function. Academic Press,
New York.

Doddrell, D.M., Pegg, D.T., and Bendall, MR. (1982) I. Magn.
Reson, 48, 323-327.

Ernst, R.R., Bodenhansen, G. and Wokaun, A. (1987) Principles of
nuclear magnetic resonance in one and two dimensions. pp. 431-
440. Oxford University Press, Oxford.

Bax, A. and Davis, D. G. (1985) I. Magn. Reson, 65, 355-360.

Bax, A. and Subramanian, S. (1986) I. Magn. Reson, 67, 565-569.

Araki, S., Sakurai, T., Oohusa, T., Kayama, M., and Sato, N. (1989)
Plant Cell Physiol, 30, 775-781.

LePage, M., Daniel, H., and Benson, A. A. (1961) The plant
sulfolipid. II. Isolation and properties of sulfoglycosyl glycerol. I.
Amer. Chem. Soc, 83, 157-159.

Hubac, C., Guerrier, D., Ferran, I., Tremolieres, A., and Kondorosi,
A. (1992) I. General Microbiology, 138, 1973-1983.

White, L. O. (1972) J. Gen. Microbiol., 72, 565-574.
De Smedt, I. and De Ley, I. (1977) Int. I. Syst. Bacteriol., 27, 222-240.

Yanagi, M. and Yamasato, K. (1993) FEMS Microbiol. Lett., 107,
115-120.

185.

186.

187.

188.

189.

190.

191.

192.

193.

194.

195.

196.

197.

198.

212

Benson, A. A., Daniel, H. and Wiser, R. (1959) Proc. Natl. Acad. Sci.
USA, 45, 1582-1587.

Haines, T. H. (1973) Sulfolipids and halosulfolipids, pp. 197-232. In
J. A. Erwin (ed.), Lipids and biomembranes of eukaryotic
microorganisms. Academic Press, New York.

Anderson, R., Kates, M. and Volcani, B. E. (1978) Biochim. Biophys.
Acta., 528, 89-106.

Stuiver, C. E. E., Kuiper, P. J. C. and Marschner, H. (1978) Physiol.
Plant, 42, 124-128.

Galliard, T. (1968) Phytochem., 7, 1907-1914.
Galliard, T. (1968) Phytochem., 7, 1915-1922.

Tevini, M. (1977) Light, function and lipids during plastid
development, pp. 121-145. In M. Tevini and H. K. Lichtenthaler
(eds.), Lipid and lipid polymers in higher plants. Springer Verlag,
Berlin.

Shibuya, I., Yagi, T. and Benson, A. A. (1963) Sulfonic acids in
algae, pp. 627-636. In Jpn. Soc. Plant Physiol. (ed.), Microalgae and
photosynthetic bacteria. Univ. of Tokyo Press, Tokyo.

Benson, A. A. (1963) Adv. Lipid Res., 1, 387-394.

Davies, W. H., Mercer, E. I. and Goodwin, T. W. (1966) Biochem. I.,
98, 369-373.

Hodgson, R. C., Schiff, J. A. and Mather, J. P. (1971) Plant Physiol.,
47, 306-311.

Kredich, N. M. (1971) I. Biol. Chem, 246, 3474-3484.

Harwood, I. L. (1975) Biochim. Biophys. Acta., 398, 224-230.

Mudd, J. B., Dezacks, R. and Smith, J. (1980) Studies on the
biosynthesis of sulfoquinovosyldiacylglycerol in higher plants, pp.

57-66. In P. Mazliak, P. Beneviste, C. Costes and R. Douce (eds.),
Biogenesis and function of plant lipids. Elsevier, Amsterdam.

199.

200.

201.

202.

203.

204.

205.

206.

207.

208.

209.

210.

211.

212.

213.

214.

213

Lehmann, I. and Benson, A. A. (1964) I. Amer. Chem. Soc., 86, 4469-
4472.

Lehmann, I. and Benson,'A. A. (1964) Proc. Int. Congr. Biochem., 6,
66.

Hoppe, W. and Schwenn, I. D. (1981) Z. Naturforsch., 36c, 820-826.

Saidha, T. and Schiff, I. A. (1989) Biochim. Biophys. Acta, 1001, 268-
273.

Benning, C. and Sommerville, C. R. (1992) I. Bacteriol., 174, 2352-
2360.

Benning, C. and Sommerville, C. R. (1992) I. Bacteriol., 174, 6479-
6487.

Pugh, C. E., Roy, A. B., Hawkes, T. and Harwood, I. L. (1995)
Biochem. I., 309, 513-519.

Dembitsky, V. M., Pechenkina-Shubina, E. E. and Rozentsvet, O. A.
(1991) Phytochemistry, 30, 2279-2283.

Harwood, I. L. (1980) Sulfolipids, pp. 301-320. In P. K. Stumpf
(ed.), The Biochemistry of Plants, Vol. 4. Academic Press, New York.

Heinz, E., Schmidt, H., Hoch, M., Jung, K. H., Binder, H. and
Schmidt, R. R. (1989) Eur. I. Biochem., 184, 445-453.

Dazzo, F. B. and Truchet, G. L. (1983) I. Membrane Biol., 73, 1-16.
Carlson, T. A. and Chelm, B. K. (1986) Nature, 322, 568-570.

Lamour, V., Quevillon, S., Diriong, S., N’Guyen, V. C., Lipinski, M.
and Mirande, M. (1994) Proc. Natl. Acad. Sci. USA, 91, 8670-8674.

Doolittle, R. F., Feng, D. R, Anderson, K. L. and Alberro, M. R.
(1990) I. Mol. Evol., 31, 383-388.

Lemassu, A. and Daffe, M. (1994) Biochem. I ., 297, 351-357.

Kadurugamuwa, I. L. and Beveridge, T. I. (1995) I. Bacteriol., 177,
3998-4008.

214

215. Tecoma, E. S., Sklar, L. A., Simoni, R. D. and Hudson, B. S. (1977)
Biochemistry, 16, 829-835.

HIGQN STATE UNI V.

“II 2IIIIII III I IIIIIIIIIIIIIIIIIS