| 7111.11. W77 3"”? 8‘5. I‘|‘1 1' "11‘; | 11""121‘31111‘ "mu
': Wk“; 31- ’ 4.114.," U4, ’1 £12: 2 151.1151”
“J11; “a . vhfih‘fii“11111111111114.1111”11111111'

‘ 1;|";1||r' 1. 4| 1: o-|":|‘.' N31“, Ii. 3 11:51:15:
Lam .3 V . $12.1 m|~ I11“ “1111113111111. 3‘11“? 111] ‘4in "11114 . ‘13. ”1.3,”
KW“ M W '8‘ 11f] :m31R‘ng

       
   

 

r75 .-

        
 
   
 
 
     
  
    
 
 
 
   
 
 
    

1“ 32,-,“ i
NEW“

1111111'|11111111é{1;,,1111|1| “11.111411: 1 1’ m‘h'fl "v||‘|l]9‘|| 1|! 11'f',}‘1~m|111||1 1111117!‘

m
. .. ~vo
.N‘.

A'} 4.1.;H 1&5.

 

¥‘>~11'1|113}11 {13111:‘111
"1 \
WW: ' ;5"'| a” Milk-"y

' 5i; L—‘m|'|.» ‘ti 3

u v“: "r;-

|
' : 5‘ (131,251
i‘$%$|¢%fi§*ul 133'“;%fi: $||
“ * fifififiz afi"
3223:; ,I
r1311?
I

:r |

Mnrnflfi 3:11!“ 1111; 3L: ”#1311111

9.:11‘: . I“ , .1
"‘5’ {WW

”.4 ‘
5:36.

       
 

  
 

1" .1

   

    

  
 

   

      
 
 

M £1 1111.111" 1|| 2‘” 7n L - T111333 .g."3,|;.; m 1|]

“y €:|'|l:‘| J. ' ] |"|i:‘33|" 111:1“.“3'1'1‘ 33"..1"; 1‘ ‘a??|,.;', 7‘5...” 2‘51 11;"; “S. i :ii‘ 1.: 5.? s

W ' . ‘ ”1" ‘ “3 5’ 3. "9‘ at? '1‘ 1, 1 fiéW’fia‘mg‘
_ |M 1 Mn; Mi. ;£ J. ‘13: |1N|||41|Et| “my, 1115 MW" 3,3,, 1,1,.” 955% _
’. 1'1‘111u1‘il H I}; I on“. M"; -|;. 11501 1‘: ‘ “a? 4 J v._.; ‘3 __
' |§|l§|;i:r|;1|;h§;iz’u -n 1191 my" "11,132, .J 11111113331 '

       
 

1:? 3. fit I}! .1151

y‘."r ;:"U:

1 Wu, ||111111w11hZé§| {‘41
' ’,': $1111 "1 1Q | W11

       

4 . 1
Mi: ‘ ,
'1 mm, ~:I-cl,' :1?"
O'. 5'”,

13th" L \. n11 , 13131; 1:1 J l CPL}: M"! Mg.

5111111131: “11,1? 111111 11311111 11111111 ,9 E11 11111161.: u
in“ 2‘|- ‘1 11.1.11 1’” P; i.

   

i "‘31
~:.mwT-.”m' L
z. - L' 913:
‘ 11 ”1‘ ‘U‘i’ “F‘dflld‘flfl 1“er 1"!“ 11;;1'1
'h .g l“ - ”1““.
, ?¢;k ;.‘ .1!

     

gum—c

  
     
 
      
    

"11"“

mm

 
    

       

     

   
   
    

11" 1" 3"” "2

   

    

 

 

  
    
   
 
  
 
 
 
 

    

  
   
 
 

  
 

     
 
  
    
 
 

‘ .1132:
.e £|
.3” “pzlhld . \ 1132‘ 1., $119.1! d iii:
hf?” 1‘9 '1m‘r‘r- ‘1.- §§|Eu§§1$|f ‘ «'1 1'3
1,-11?‘ y|![:§ -‘ |;§'g
”191;; "311‘11211111 :11: -!

  

D
|
l ,_
' :53
“1'; '0".

',K'
:‘5'1411'

“11,3“; [ _‘,

 
  

   

      
       
    
    
   
        
   
   
   
      

   

 
 
   

    

   

 

    

 

 
 
 

   

           
 
   
  
 
  
   

   
  
 
   

 

    
  

  

    
  
  
  
  
  

  
    
 
  
  
    

    
 

   
     

  
  
  
  
  

  
  

      
         
    
   

    
     
   

 

. .- If‘
“133‘ M; '1] aim";- n11.
{1‘1"}:‘2’113;
1411311: 1 1 ,1
.. .‘ ' 1
I 1': . . ‘M ' fa; ‘
g ‘i: ‘ fl‘ifi’, t " “"11“. . .
:3; '|""1‘1$1'.'-'e r‘.
. :3-‘1:1|,| .11 9 _ ._ : ,, 2,;3
- .l . ' ' ;_ ' _.‘ , ‘
5|“? 11‘J.|1,u.1 e 115! 1.1: 1 1.
11133 $11111" 5'11 M315 1 M 1' '
'L , '3‘ ‘ .| H '3
I. ., .- .
h. I ’1; g? '
€r"1 1 ' 1.4;:
ef'.54 sasagar
.171.1-. 1‘ | 4"; .2” itin. _
’7 ..";‘.‘| .|,‘” “i; _ .1 I‘
. 'I 4' . . ‘|' 1 I. . .1 .:u ‘
1 . . H. , '1. C > ‘.._ ,,.
. 1'11. 2.:tll H.“ M,‘ 1
.1 , v. ,, 1.5:
j g, ,..| .91} r . ~l’”
|' ‘ 2v: ';:‘1"-" ,
- is ‘ m. .3233: V 9‘11! 1}!
II .I. ’ ;' 1‘I'1'E115111'1; . v|- hi.“
1|]; 111‘” v :11 ii'.||ii||i I F ii.‘ 1 .' "'1" 1 '
- 2,2 , 5;! :1! [1' 1r}: '11'1:||1 13mg! 1: {3| _-= 'i" - 1 .
F C 1 I|I 1111"; 11H 1711,? |§'§|§14.':|'| 1;! -. , ‘ ‘
1 ‘ . :"sW f‘r'fi” : 03 Li ‘6' J , .- a .
|r| HIM ‘41 . ‘0 . Q . I
7 24 I H. ,.;|‘9|\.11||s: 'E: |'- We? 4435‘: . '1
||Jflh"‘ ”E IN! ng 1,5 yw@x 1g%;*v
| | | ' 2 , (1-. :1 :J . . .i ,;
I i ‘7. | . , ‘ ] "511$ ‘.‘ 3""
,1 . 1414.4 '10-"
M 'r..; ,ng , 7.5m"; ‘
2' , 114211 <1 - 35;}. ,,
5 E u h:;¥;';1.:J'-r--’ @333"
«1[i§€¢'i"."' +.
. ,4 ‘1" .

 

” iiiiiijiiiWiiiiiri

01686 3361

This is to certify that the

dissertation entitled

THE ROLES OF RHIZOBIAL SURFACE AND
MEMBRANE COMPONENTS IN SYMBLOSIS

presented by
Yi n. To. n3

has been accepted towards fulfillment
of the requirements for

PhflD degreein BiOChemiStY-H

(idol; 1:Mc/Zs¢nvk‘
\ MapMofessql
Date Ofllln 1} '

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

 

LIBRARY

Michigan State
University

 

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

we chIRC/DflDuopGS—p.“

THE ROLES OF RHIZOBIAL SURFACE AND MEMBRANE
COMPONENTS IN SYMBIOSIS

By

Yin Tang

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Biochemistry

1997

ABSTRACT

THE ROLES OF RHIZOBIAL SURFACE AND MEMBRANE COMPONENTS
IN SYMBIOSIS

By

Yin Tang

A variety of rhizobial cell surface and membrane carbohydrate-containing molecules
have been indicated to play some important roles in rhizobium/legume symbiosis. The
functional roles of these molecules in symbiosis are closely related to the ability of bacterium
to change the relative quantities or structures of these molecules and achieve the optimal
bacterial surface property in order to survive in a changing environment. There is a large
degree of functional and biosynthetic overlap among various rhizobial cell surface and
membrane carbohydrate-containing molecules. A comprehensive picture of the rhizobial cell
surface needs to be obtained in order to understand the roles played by the rhizobial surface
or membrane components in symbiosis.

Studies on the membrane lipids of free living and the bacteroid form of
Bradyrhizobium japonicum indicated that the later has high levels of digalactosyl
diacylglycerol, a glycolipid which has rarely been found outside of the plant kingdom and
photosynthetic bacteria. This result follows our earlier finding that the other predominant
plant glycolipid, sulfoquinivosyl diacylglycerol, is a membrane component of fast growing
Rhizobia and is found even when cells are cultivated in free culture. The near absence of
these lipids in the membranes of bacteria outside of this special group of organisms and

photosynthetic bacteria suggests that the trait could have been passed on through gene

transfer from plants to the bacteria at some point during the development of their symbiotic
relationship. In the case of digalactosyl diacyl glycerol, there is also the possibility that some
common biosynthetic intermediates are used by both the plant and the bacteria. This is a
striking parallel with some host-pathogen interactions.

Culturing Bradyrhizobium japonicum in the presence of glucose as a carbon source
in a rich medium known to stimulate differentiation into bacteroid forms led to the synthesis
of a family of di- and trisaccharide-containing glycerol lipids as the main membrane lipid
components. Like digalactosyl diacylglycerol, these glycolipids are also rarely found in other
bacteria. When Bradyrhizobium japonicum was grown in low-oxygen condition which
mimics the condition inside root nodues. There was a dramatic switch of membrane lipid
chemistry. Phosphatidyl inositol, a typical plant lipid rarely found in bacteria, was also
synthesized. Lowering pH in free-living Bradyrhizobium japonicum did not result in the
change of unsaturation state in the fatty acyl groups of the lipids.

Studies on the interrelationship of the biosynthetic events leading to various
carbohydrate-containing molecules indicated that , when l4C labeled 3-O-methyl-D-glucose
was used as a probe, it was incorporated into membrane glycolipids but not into other major

carbohydrate-containing molecules, including EPS, LPS, Cyclic B-glucans. It is likely that
the biosynthesis of EPS, LPS and cyclic B—glucans does not use the same metabolite pool as

utilized by glycolipids. It is also possible that the substrate specificity of the enzymes

involved in synthesizing glycolipids is different.

To My Dear Mother and Father

iv

ACKNOWLEDGMENTS

My Ph.D. work would not be finished without the help from many people around me
during my five years of being here at MSU.

I am indebted to my thesis advisor, Dr. Rawle I. Hollingsworth. I thank you for your
talents from which I benefit so much. I am not just talking about your talent of being
scientifically insightful and profound. Far more than that, I’m talking about your gift to be
a good person. You are the most busy one, but you never keep others out. You have lots of
pressure, but you never pass it on to your students. You are not always successful in doing
experiments, but you are never short of generating new ideas. From you, I see how open-
minded and tenacious a scientist should be; from you, I see how straightforward and friendly
a colleague better be; also from you, I see how caring and proud a good father is. I thank you,
Rawle, for the scientific guidance and the opportunity for me to learn from you.

I thank my guidance committee members, Dr. Zachary Burton, Dr. Thomas L. Deitz,
Dr. Paul Kinde], Dr. Kenneth Keegstra and Dr. Barbara Sears. I appreciate the time and effort
you spent on reading my reports and thesis, thinking about my projects and attending my
meetings. Especially, I thank you for giving the valuable scientific criticism.

I can’t think of how boring my life would be without many fellow students and

coworkers in the lab. Ben is the best colleague I can imagine in the world. I am fortunate to
be your bench neighbor. You are not only the go-to person when I need scientific critique or
need to operate a new instrument but also the one to consult about home improvement.
Hussein, YW, Luc, Steve and Ben were the ones I often had lunches with ( unfortunately the
restaurants in the area were not enough for these people to go, 80% of the time we ended up
in Thai Kitchen ). I really enjoyed the time with you guys. I know I will miss this. Gang, J ie,
Guijun, Guangfei were the elite organic chemists to ask for help when I needed to do some
organic synthesis occasionally. It was especially nice to talk with you in my mother tongue,
Chinese. Jim is our computer guru. Whenever I got stuck with a PC, I needed to find you. It
is too hard to describe five years with one page of words. I am thankful to many other people,
Rob, Vladmir, Carol, Du, Jung, Kim, Jeongrim, Gabriela, Birgit and J imin. I appreciate the
fellowship with all of you.

Many thanks to MSU NMR facility, especially Dr. Long Li, who is always patient
and willing to help. Thanks also to the Mass facility where I obtained my GC/MS data from
a JEOL instrument.

I’d like to extend my thanks to Department of Biochemistry for giving me the
opportunity to study here and providing me the necessary equipments to finish my work.
Thanks also to the MSU International Center, the Graduate School, MSU Library, College
of Natural Science and the all MSU community and the cities of E. Lansing and Lansing. I
had an very enjoyable time here. I deeply appreciate all the assistance I received from you.

Finally, I am grateful to my husband, II, for your support both in and out of the lab.

I would never leave out my little sweetheart, Lanson, who just turned one and half years old

vi

around the time of my defense, for the unspeakable joy you brought to me and your
unbearable cry to push me work hard.
Above all, behind all of the above-mentioned is the Almighty One, who made

everything happened happened. I thank you, my Lord and Savior for everything!

Wholeheartedly,

Yin

vii

TABLE OF CONTENTS

LIST OF FIGURES .......................................................................................................... xii
LIST OF ABBREVIATIONS .......................................................................................... xiv
CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW ........................................................... 1
Introduction ............................................................................................................ 2
Literature Review .................................................................................................... 5
I. Nitrogen Fixation and Legumes/Rhizobia Symbiosis ........................... 5
II. The Process of Nodulation .................................................................... 8
III. Rhizobial Cell Surface and Surface Carbohydrate-Containing
Molecules ............................................................................................ 13
IV. The Dynamic Property of Rhizobial Surface ...................................... 27
V. Roles Played by Various Rhizobial Surface Carbohydrate-
Containing Molecules At Different Stages of Symbiosis .................... 29
VI. Functional and Biosynthetic Overlap Among Various Rhizobial
Cell Surface Carbohydrate-Containing Molecules ............................. 33
Conclusion ............................................................................................................ 36
Bibliography ........................................................................................................... 38
CHAPTER 2

THE LIPID CHEMISTRY OF BACTEROIDS ISOLATED FROM
PLANT NODULES IS DISTINCTLY DIFFERENT FROM THAT

OF FREE-LIVING RHIZOBIA ........................................................................................ 42
Abstract .................................................................................................................. 43
Introduction ........................................................................................................... 45
Materials and Methods .......................................................................................... 47
Results ................................................................................................................... 51
Discussion ............................................................................................................. 75
Acknowledgments ................................................................................................. 78

viii

Bibliography .......................................................................................................... 79

CHAPTER 3

THE LIPID CHEMISTRY OF FREE—LIVING BACTEROIDS OF RHIZOBIA ............ 81
Abstract .................................................................................................................. 82
Introduction ........................................................................................................... 83
Materials and Methods .......................................................................................... 86
Results ................................................................................................................... 89
Discussion ........................................................................................................... 109
Acknowledgments ............................................................................................... 1 1 1
Bibliography ........................................................................................................ 1 12

CHAPTER 4

THE LIPID CHEMISTRY OF RHIZOBIA GROWN UNDER

LOW OXYGEN AND LOW PH CONDITIONS ........................................................... 113
Abstract ................................................................................................................ 114
Introduction ......................................................................................................... 1 15
Materials and Methods ........................................................................................ 116
Results ................................................................................................................. 119
Discussion ........................................................................................................... 131
Acknowledgments ............................................................................................... 1 32
Bibliography ........................................................................................................ l 33

CHAPTER 5

AN INVESTIGATION ON THE INTERRELATIONSHIP OF RHIZOBIAL

POLYSACCHARIDES AND GLYCOLIPIDS BIOSYNTHESIS .................................. 134
Abstract ................................................................................................................ 135
Introduction ......................................................................................................... 1 36
Materials and Methods ........................................................................................ 141
Results ................................................................................................................. 144
Discussion ........................................................................................................... 162
Bibliography ........................................................................................................ 163

ix

CHAPTER 6

SUMMARY AND PERSPECTIVE ............................................................................... 166
Summary and Perspective .................................................................................... 166
Bibliography ......................................................................................................... 172

Figure 1.1:

Figure 1.2:

Figure 1.3:

Figure 1.4:

Figure 1.5:

Figure 1.6A:

Figure 1.6B:

Figure 1.7:
Figure 1.8:
Figure 1.9:

Figure 2.1:

Figure 2.2:

Figure 2.3:

Figure 2.4:

LIST OF FIGURES

The biological nitrogen cycle ...................................................................... 7
The invasion of the root hair by Rhizobium ............................................... 10
The process of nodulation ......................................................................... 12
Carbohydrate-containing molecules on the surface of Rhizobium ............ 15

General structure of the two major EPS structures found in R. meliloti... 1 8
Basic structure of typical enteric bacterial LPS ........................................ 2O
Structures of LPS lipid A and O-antigen of R. leguminosarum biovars..22
The structures of the phospholipids and glycolipids found in rhizobia ....24
Proposed structure for cyclic [3-(1,6)-B-(1,3)-g1ucan of B. japonicum ...... 25
Structure of chitolipooligosaccharides ....................................................... 26

Scanning electron micrograph showing a typical field in the
bacteroid preparation from soya bean root nodules .................................. 52

Southern hybridization .............................................................................. 54

Proton NMR spectrum of lipids from bacteroids and from extract of
the supernatant which contained plant lipid vesicles ............................... 56

Gas chromatogram of alditol acetate derivatives of the headgroup

components from the total membrane lipid preparation of
Bradyrhizobium japonicum bacteroids ...................................................... 59

xi

Figure 2.5:

Figure 2.6:

Figure 2.7:

Figure. 2.8:

Figure 2.9:

Figure 2.10:

Figure 3.1A:

Figure 3.1B:

Figure 3.2:

Figure 3.3:

Figure 3.4:

Figure 3.5:

Figure 3.6:

Figure 3.7:

Gas chromatogram of the alditol acetate derivatives of the residue
and the supernatant .................................................................................... 61

Gas chromatogram of the chloroform-soluble fraction after acid-
catalyzed methanolysis of the residue, supernatant and the
bacteroid preparations ............................................................................... 63

Negative ion fast atom bombardment mass spectrum of the total
membrane lipid extracts from Bradyrhizobium japonicum

USDA 110 bacteroids and untyped bacteroids .......................................... 65
Orcinol-sprayed thin layer chromatography plate of membrane

lipid extracts ............................................................................................... 68
The structure of digalactosyl diacylglycerol ............................................. 70

Proton nuclear magnetic resonance spectrum of
the isolated digalactosyl diacylglycerol .................................................... 73

Bradyrhizobium japonicum USDA 110 grown in B-III medium .............. 90

Bradyrhizobiumjaponicum USDA 110 grown
in bacteroid-inducing medium ................................................................... 91

Thin layer chromatograms of lipid extract from cell
grown under nutrient enrichment conditions ............................................ 93

Gas chromatography spectrum of alditol acetate derivatives
of the total lipid extract from USDA 110 grown under
nutrients enriched conditions ..................................................................... 96

Proton NMR spectroscopy analysis of the faster moving
component in Panel A of Figure 3.1 ....................................................... 98

The connectivity of H1 glyceryl proton appeared at 4.20 ppm
overlapping with some other signals could be traced
through the DQF-COSY spectra .............................................................. 100

Negative ion electrospray mass spectrometry of the fast moving
component in Panel A of Figure 3.1 ....................................................... 102

Negative ion electrospray mass spectrum of the slow moving
component in Panel A of Figure 3.1 ........................................................ 106

xii

Figure 3.8:

Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Figure 5.1:
Figure 5.2A:
Figure 5.2B:
Figure 5.3:
Figure 5.4:

Figure 5.5:

Figure 5.6:

Figure 5.7:

Figure 5.8A:

Figure 5.8B:

The structures of di- and trisaccharide-containing lipids ........................ 108

Orcinol-sprayed TLC plate of membrane lipids extracts from
B. Japonicum USDA 110 grown in normal and low
oxygen conditions ..................................................................................... 121

Proton nuclear magnetic resonance spectrum of the extracted
membrane lipids of B. Japonicum USDA 110 grown under
free-living and low oxygen conditions ..................................................... 123

Gas chromatograph of alditol acetate derivatives of lipid extract
from B. Japonicum USDA 110 grown under normal
and low oxygen conditions ...................................................................... 126

GC/MS spectrum of alditol acetate derivatives of standard myo-
inositol and lipid extract from B. Japonicum USDA 110 ........................ 128

Proton nuclear magnetic resonance spectrum of the lipids extracted
from the membranes of B. Japonicum USDA 110 grown

under normal and low pH conditions ....................................................... 130

Oxidative demethylation and elimination of 3-O-methyl-D-glucose ...... 139

Phenol-sulfuric acid assay ....................................................................... 145
Scintillation counting ............................................................................... 146
TLC analysis of two fractions with the most radioactivity ...................... 148
TLC analysis of the pooled and concentrated radioactive fractions ........ 151

TLC analysis of the total lipid extract of the bacteria grown with
3-O-methyl glucose as carbon source ...................................................... 153

TLC analysis and the autoradiograph of the lipid extracts from
the bacteria grown on three different media ............................................ 155

The mechanism of periodate oxidative degradation of glucosyl

rings to form dialdehydes ........................................................................ 157
TLC analysis of three lipid extracts before and after oxidation .............. 159
Autoradiograph of the corresponding TLC plate in Figure 5.8A ............. 161

xiii

BIII

CPS
DGDG
DQF-COSY
EPS
ESI-MS
FAB-MS
GC
GC/MS
LPS

N MR
PC

PE

PG

PI

PS

SEM

LIST OF ABBREVIATIONS

modified Bergensen’s media

capsular polysaccharide

digalactosyl diacyl glycerol

double quantum filtered J -correlated spectroscopy
extracellular polysaccharide

electrosray ionization mass spectrometry
fast atom bombardment mass spectrometry
gas chromatography

gas chromatography / mass spectrometry
lipopolysaccharides

nuclear magnetic resonance
phosphatidylcholine
phophatidylethanolamine
phosphatidylglycerol

phosphatidylinositol

phosphatidylserine

scanning electron microscopy

xiv

SQDG sulfoquinovosyl diacylglycerol
TEM transmission electron microscopy

TLC thin-layer chromatography

XV

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

INTRODUCTION

Biological nitrogen fixation has commanded the attention of scientists for more than
100 years. The ability to fix nitrogen is exhibited by a broad spectrum of microorganisms
such as cyanobacteria, archaebacteria and actinomycetes. However, nitrogen fixation with
the highest rates, also with the greatest importance, occurs in symbiotic associations between
bacteria and plants. The Rhizobium-legume is the best known of the symbiotic systems.
Section I of the literature review covers the importance of this symbiotic system in the
nitrogen cycle and Section II introduces the process by which the symbiotic relationship
between rhizobium and leguminous plants is established.

The interaction between rhizobia and their corresponding hosts is very specific. Most
strains only nodulate plant species belonging to a few genera. This host specificity, as well
as many other aspects of symbiosis such as infection thread development and nodule
formation, is determined by a variety of rhizobial cell surface and membrane carbohydrate-
containing molecules. They include extracellular polysaccharides (EPSs), capsular
polysaccharides (CPSs), lipopolysaccharides (LPSs), chitolipooligosaccharides (Nod

Factors), cyclic B-glucans and glycolipids. Their structures and locations on the rhizobial

surface are covered in Section III of the literature review. The surface chemistry of
rhizobium is highly variable and very sensitive to environmental factors which will change
as the organism leaves the rhizosphere and goes inside of the plant (Section IV). The

functions of various glycocongugates in the establishment Of symbiosis are closely related

3

to the ability of rhizobia to change its surface chemistry in response to a changed
environment.

Although numerous studies have been conducted on the involvement of various
rhizobial surface molecules during various stages of symbiosis (summarized in Section V
of the literature review), the specific roles played by each individual component are still far
from clear. It is the diversity of the surface molecules involved in symbiosis that complicates
such studies. These studies become even more complicated because of the large degree of
coupling and overlap of biosynthetic events leading to the synthesis of the various molecules
(Section VI of literature review ).

Three chapters of this dissertation address the aspect of environmental regulation of
rhizobial membrane chemistry. Chapter 2 presents studies on membrane lipid chemistry of
B. japonicum bacteroids living inside plant nodules. We have found that the predominant
lipid in the membranes of Bradyrhizobium bacteroids is digalactosyl diacylglycerol, a typical
plant lipid not found before in non-photosynthetic bacteria. Chapter 3 presents studies on
membrane lipid chemistry of a free-living, bacteroid-like form of B. japonicum. We have
found that the predominant lipid is also a diglycosyl diacylglycerol. Chapter 4 presents
studies on membrane lipid chemistry of Bradyrhizobium grown under low oxygen. Low
oxygen condition was chosen because it mimics conditions inside root nodules. We found
that the low oxygen condition triggers the biosynthesis of phosphatidylinoisitol (PI), another
lipid common in plants and not usually found in bacteria.

Understanding carbohydrate metabolism is central to understanding the connections

among biosynthetic events leading to the various polysaccharides, glycolipids and

4

glycoconjugates which all appear to be so important to the symbiotic process. The work
presented in Chapter 5 was conducted to help to untangle the complex, interdigitized web
of carbohydrate metabolism and biosynthesis in order to understand this complicated picture.
In these experiments, 3-O-[”C]methyl-D-glucose was used to probe the common
intermediates shared by the biosynthetic pathway to various glycolipids and saccharides. We
have found that carbohydrate components in the membrane glycolipids contain labeled 3-O-
methyl glucose. The results indicate that the biosynthesis of glycolipids utilizes either a
different substrate pool or a glycosyl transferase with different substrate specificity.

The last chapter summarized the results obtained from previous chapters and
addressed the possible roles of these unusual membrane phospholipids and glycolipids in the
adaptation and survival of rhizobia inside plants during symbiosis. It also raises some
questions about the origin of the genes involved in the biosynthesis of these unusual
phospholipids and glycolipids. Further studies on the biosynthesis of various saccharides and

glycoconjugates are proposed.

LITERATURE REVIEW

1. NITROGEN FIXATION AND THE RHIZOBIUM/LEGUME SYMBIOSIS

Living organisms continually recycle carbon, nitrogen, oxygen and sulfur from which
they are composed. From both ecological and economic viewpoints, the nitrogen cycle is
extremely important (1). Figure 1.1 is a simple illustration of this cycle. The lower pathway
shows the synthesis of organic nitrogen compounds (principally proteins and nucleic acids)
from inorganic nitrogen compounds (nitrate, nitrite and ammonium ions) during growth of
plants and subsequent plant consumption by animals. Nitrogen returns to the soil as a result
of decay of plant and animal materials. The upper pathway shows the loss of nitrogen to the
atmosphere from nitrates, and its return to the cycle by the process known as nitrogen
fixation. Biological productivity in all agricultural areas of this planet is determined by the
rate at which the cycle turns, i.e. the rate of N2 fixation.

In nature, the ability to fix N2 is restricted to a relatively small number of
prokaryotic organisms, termed diazotrophs. These organisms have the enzyme system,
.nitrogenase, a biochemical machinery for nitrogen fixation (1, 2). In this enzymatic process,
atmospheric dinitrogen is reduced to ammonia. Diazotrophs annually add about 60% of the
earth’s newly fixed nitrogen, while industrially fixed nitrogen contributes only about 25%,
with the remaining 15% produced by lightening, UV radiation, etc.(2). Some nitrogen-fixing

bacteria like Klebsiella pneumoniae, Azotobacter and Rhodopseudomonas are able to reduce

Figure 1.1: The biological nitrogen cycle. Adapted from reference 1.

”.255 v5
”Cm—Q £39.25

«0 .So 2:805

 

//

ZO_._.<o_u_mh_z AMOZ.
mwh<mtz

     

      

ZO_._.<U_"__ZOS_2<

  

ZmOOtha

8
nitrogen into ammonia as free living organisms.

Other bacteria fix nitrogen mainly during symbiotic relationships with plants. Some
well-known symbiotic systems are rhizobium/legume, F rankia/actinorhizal and
cyanobacteria/Gunnera.

The best studied systems of symbiotic nitrogen fixation are the soil bacteria of the
genera Rhizobium, Bradyrhizobium, and Azorhizobium, collectively referred to as rhizobia.
It has been estimated that the rhizobia/legume symbiosis accounts for over 50% of natural
nitrogen fixation (2). The legumes are important in agriculture not only because they fix
nitrogen and thus lower the use of nitrogen fertilizers, but also because of the high protein
content of the plants and seeds. The seeds of crop legumes are important nutritionally for
humans, and the forage legumes (clovers, alfalfa and luceme ) supply a high protein diet to
livestock (3). The most prominent feature of rhizobial symbiosis is the formation of effective
nodules on the roots of leguminous plants (4-8). Inside these nodules the bacteria are able

to fix nitrogen and provide the host plant with ammonia for growth.

II. THE PROCESS OF NODULATION

The first stage of nodulation is bacteria—plant recognition. Rhizobia have a restricted
host range; each strain can only infect and nodulate a limited number of legume species (1,
3). It is thought that this high degree of specificity is due to key interactions between the
molecules on the surface of the partners. These molecules must be sufficiently diverse.
Rhizobia have a wide range of surface saccharides and glycoconjugates that can provide

sufficient diversity. Plants, on the other hand, have lectins. Lectins are proteins that can

9

distinguish different sugars and one theory is that they mediate the recognition process
between legumes and rhizobia.

Once the proper bacteria-plant recognition occurs, the bacteria invade into the root
hair. This starts, in most cases, with the morphological changes on the root hair in the
presence of the bacteria. Figure 1.2 shows this process (7). The root hair curling is caused
by the release of some diffusible signal molecules which now are believed to be bacterial
surface saccharides or glycoconjugates. The bacteria are enclosed in a pocket of the curled
hair. By hydrolyzing the local root hair cell wall, the bacteria enter the epidermal cell and are
then enclosed by a newly formed wall which cuts the bacteria off from the contents of the
host cell. The bacteria divide in the enclosed space and the wall enclosing them extends to
grow as a tube termed the infection thread. The infection thread is actually an intracellular
tunnel bounded by the plant cell wall. The infection thread proceeds to grow beyond the root
hair cell and penetrates into the root cortical cells (Figure 1.3). Simultaneously, the cortical
cells are stimulated to divide and give rise to the nodule primordium. Bacteria are then
released from the infection thread at a point where it is free of host cell wall. They are
released into the cytoplasm of the cortical cells by a process of endocytosis, surrounded by
a plant—derived membrane termed peribacteroid membrane. The bacteria divide frequently
and when the cytoplasm is almost full of bacteria, the bacteria enlarge and change in shape
and differentiate into their endosymbiotic form, bacteroids. The nodule primordium
thereupon develops into a mature nodule. There are also many biochemical changes
accompanying bacteria differentiation to bacteroids. The bacteroids change their many

cellular components to cope with the low oxygen conditions inside nodules and synthesize

10

 

 

 

 

 

 

 

 

 

Figure 1.2: The invasion of the root hair by rhizobium. A, the bacterium becomes enclosed
in the curled root hair, B, the bacterium penetrates the root hair and is surrounded by cell
wall materials; C, initial infection thread growth. The bacteria divide within the infection

thread.

11

Figure 1.3: A, spread of infection thread and division of cortical cells; B, bacteria release;

C, bacteria differentiation into bacteroids; D, mature nodules on plant roots.

12

 

 

 

 

j
bacteroids

 

 

 

 

 

 

 

 

 

g

13

the nitrogenase complex, the enzymatic machinery to fix nitrogen (3). The root nodules are
well adapted for nitrogen fixation in air. The cortex surrounding the central infecting cells
can serve both as a barrier for excess oxygen infusion and as a mediator for the active

metabolite transport within nodules.

III. RHIZOBIAL CELL SURFACE AND SURFACE CARBOHYDRATE-
CONTAINING MOLECULES

The process of the establishment of symbiosis was introduced in the last section.
During this whole process, a variety of rhizobial cell surface and membrane carbohydrate-
containing molecules are believed to be involved. They include EPSs, CPSs, LPSs,

chitolipooligosaccharides (Nod Factors), cyclic B-glucans and glycolipids. Before giving a

detailed discussion of the functions of these various surface carbohydrate-containing
molecules, an introduction of their locations and the structures are needed.

Rhizobium is a Gram-negative bacterium and the envelope of rhizobium consists of
an inner membrane and an outer membrane separated by a periplasmic space. Figure 1.4 is
a schematic diagram of the rhizobial cell surface where the locations of various carbohydrate-
containing molecules are indicated.

EPS and CPS: Like many Gram-negative bacteria, rhizobium produces a discrete
capsule or capsular polysaccharide (CPS) and (or) excretes acidic extracellular
polysaccharide (EPS) as a layer of slime not tightly attached to the bacterial cell surface. The
structures of these two glycoconjugates are generally similar. The most studied EPS is found

in R. meliloti and is often called succinoglycan (EPS I). It belongs to the broad class of acidic

14

Figure 1.4: Carbohydrate—containing molecules on the surface of Rhizobium.

15

 

montmzoommogoogozco IIV mcanEmE .55...

i
j
\ /

 

=8:_m.n.o__o>o IV I \i Emmatmn. «23.2385
)
«29.820 /v i b\
«J: \ i../95959: ESQ

 

m\\\Nll we

ooAlmmm

l6

heteropolysaccharides (9) and is made of repeating octasaccharides containing non-
carbohydrate substitutes (a succinate, pyruvate and O-acyl group). Certain mutant strains of
R. meliloti produce an alternate EPS structure, termed EPS II. The structures of the two R.
meliloti EPSs is presented in Figure 1.5.

LPS: The outer leaflet of the outer membrane of rhizobium is mainly composed of
lipopolysaccharide, molecules unique to the envelope of Gram- negative bacteria. The outer
membrane makes the Gram-negative envelope impermeable to hydrophobic compounds and
higher molecular weight hydrophilic compounds (10). This restrictive permeability properties
of the outer membrane account for the resistance of Gram—negative bacteria to a wide range
of antibiotics. Typical enterobacterial LPS ( Figure 1.6 A) have three covalently linked
domains: membrane embedded lipid A consisting of glucosamine disaccharides linked with
fatty acid chains, an oligosaccharide core and an antigenic polysaccharide chain (O-antigen).
The structure of the LPS Lipid A and O-antigen of R. leguminosarum biovars was shown in
Figure 1.6B (11). This lipid A has carboxylate as the negatively charged groups.

Pbospholipids and Glycolipids: The predominant rhizobial membrane
phospholipids are phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and
phosphatidylcholine (PC). While PE and PG are common phospholipids in the membrane
of Gram-negative bacteria, PC is rarely found in other Gram-negative bacteria but tends to
be predominant in plant cells (12). Besides this uniqueness, rhizobia also synthesize
glycolipids which have not been found in other Gram—negative bacteria. It has been
demonstrated that the membranes of many free-living rhizobia species contain

sulfoquinovosyl diacylglycerol (SQDG), a glycolipid which was thought to occur exclusively

17

Figure 1.5: General structure 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 9.

l8

+G1c M-<31c flak: L” Gai’” n

 

[6
H6 OAc
Glc
I...
Glc
l...
Glc —- OCOCHZCHZCOO ’
134.3
4/O\ /CH3
Glc\ /C\
6 0 COO ’
Succinoglycan

 

+Gal “'13 Glc 13

4/ \6 '6
O\ /O OAc
/C\

COO ' CH3

EPS II

19

in membranes of plant plastids and photosynthetic bacteria (13). One mannose and glucose-
containing diglycosyl diacylglycerol has been reported to accumulate in the membrane of
R. leguminosarum bv. viciae ANU843 when the bacteria were grown under certain culture
conditions (14). The occurrences of these glycolipids make the membrane chemistry of
rhizobium somewhat of an enigma since it is more similar to plant membrane chemistry than
to bacterial membrane chemistry. Figure 1.7 shows the structures of major rhizobial lipid
molecules.

Cyclic B-glucans: Within the periplasmic space is cyclic B-glucans, molecules that
are found almost exclusively in bacteria of the Rhizobiaceae family (15). Reaching
concentrations as high as 5 to 20% of the total cellular dry weight under certain culture
conditions, the cyclic B-glucans are major cellular constituents. In Rhizobium species, these
contains glucose residues linked solely by B-(l,2) glycosidic bonds. In Bradyrhizobium
species, the glucose residues are linked by both B-( l ,3) and B-(1,6) glycosidic bonds (15).
The glucose monomer may become substituted with nonsugar moieties such as anionic
substituents. The degree of polymerization (DP) ranges from 10 to 13 in B. japonicum to 17
to 40 in R. meliloti. The structure of B. japonicum cyclic B-glucans is shown in Figure 1.8.

Chitolipooligosaccharides: Another class of molecules unique to rhizobia is called
chitolipooligosaccharides or Nod factors (16). They are chitin-like oligosaccarides of three
to five 1,4-linked N—acetylglucosaminosyl residues, which are N-acylated on the terminal
non-reducing glucosaminosyl residue. They may be further modified by the addition of

another glycosyl residue, or sulfate, carbamyl, acetyl, and/or glycerol substituents. The

20

 

 

 

 

 

 

 

 

 

 

 

 

 

: A
O-Antigen
J Repeating
Units
‘1
J Core Marker Sugars
Inner
and
Outer
Core
I-leptoseO

 

 

 

Figure 1.6A: Basie structure of typical enteric bacterial LPS. Adapted from Rietschel, E.T.
et a1. 1994. FASEB J., 8, 217-225.

Q-i:0
i

 

21

Figure 1.6B: Structures of LPS lipid A and O-anti gen of R. leguminosarum biovars.

22

HO 0
HO
O
Ilpld A o 0 O o
0 ”$1/ °
0
o 0 o 0 W TKO
0
H0 H0
H0
R1
R1 R1
R2
0 27-OH 28:0
0
in?
, Ot-D-ManNAc
O—Antigen 1
2

-v-3)-a-L-Rha-(1-*3)-a-L-Rha-(l»4)-B-D-GlcNAc-(1+3)-a-L-Rha-(l —)

23

Figure 1.7: The structures of the phospholipids and glycolipids found in rhizobia.

24

WW
OO—$H2
I
HzC—O Jig/R
(3

+
R= —O-CH2-CH2—N(CH3)3 Phosphatidylcholine (PC)
+
—O-CH2-CH2—NH3 Phosphatidylethanolamine (PE)

-O-CH2-CH-CH2-OH Phosphatidylglycerol (PG)
OH

Fatty acid chains vary in length and degree of unsturation

 

Sulfoquinovosyl diacylglycerol (SQDG)
Fatty acid chains vary in length and degree of unsturation

O

«0W0 o
Mannose and glucose-containing diglycosyl diacylglycerol
Fatty acid chains vary in length and degree of unsturation

25

on
on
no “9
HO O OH
H O
"o o o o H
0 no
0 no 0 OH
HO
HO on
HO 0
HO 0
0 OH O OH
HO
O OH
O
HO OH
HO HO 0
o
0
HO 0
OH O
HO

 

Figure 1.8: proposed structure for the unsubstituted cyclic [3—(1,6)-B-(1,3)-glucan of B.

japonicum containing 13 glucose residues. Adapted from reference 15.

26

 

 

 

In n\,
0 o
/ 1‘ ‘ /
cn, o CH20H o CH, o
II.- 0 HO \ HO OH
/N-n, /NH /NH
o o = c\ o =c\
CH CH
_ a J n a
R leguminosarum R. me/i/oti , R Sp. NGR234
bv. wcrae -
n 2 or 3 1, 2. or 3 3
O 018:1 or 018:4 C162 Ci8110rC1620
R1 CHJCO or H CHJCO or H H
R, H SO3H 2-Omethylfucose or
substituted with either
3-O-CH3CO or 4-O-SOJH
Ft3 H H CH3
Ft, H H NH2CO or H
Rs H H NHZCO or H

 

Figure 1.9: Structure of chitolipooligosaccharides. The chitin oligomer and the acyl moiety

(Q) are present in all chitolipooligosaccharides. Adapted from reference 16.

27

Structure of chitolipooligosaccharides is shown in Figure 1.9. Recent studies (17, 18) suggest

that these molecules are membrane-localized.

IV. THE DYNAMIC PROPERTY OF THE RHIZOBIAL SURFACE

In the previous section (section III), various rhizobial cell surface and membrane
carbohydrate-containing molecules are introduced because of their importance in the
establishment of symbioses between rhizobia and legumes. Before the symbiotic roles played
by each class of molecules are summarized in the next section (section V), a very important
property of the rhizobial cell surface needs to be addressed. It is the dynamic property of the
bacterial surface chemistry. The bacterial surface is constantly regulated by environmental
factors including those inside the plant. This dynamic property enables the bacterium to adapt
and survive in a totally new environment such as that inside of the plant cell. The bacterium
achieves its dynamic surface property mainly by changing the relative quantities or structures
of its various surface molecules. The different surface and membrane carbohydrate-
containing molecules present at different stages of symbiosis contributes to this dynamic
property of the bacterial surface. For example, the lipids which constitute the basic structural
elements of membranes are critical to bacterial adaptation (19). They are exquisitely sensitive
to environmental variables, including medium constituents (carbon, nitrogen, inorganic
phosphate, etc.), physical factors (temperature, oxygen tension, pH, etc.) and growth phase
(19-20). The term homeoviscous adaptation (homeostasis) is often used to describe the
ability of membrane to maintain a suitable dynamic state of the bilayer independent of

temperature. The active restructuring of the membrane lipid composition in response to

28
environmental change is essential for an organism to adapt to and survive in an altered
environment.

The dynamic properties of rhizobial surface can be observed by culturing the bacteria
in different media. Studies using monoclonal antibodies as specific submolecular probes for
LPS epitopes suggested that the structure of LPS was regulated by environmental factors
such as ion composition, pH and oxygen tension (21-23). Symbiotic mutations leading to an
impaired LPS structure in R. leguminosarum also affect the normal range of structural
adaptations for LPS macromolecules under different physiological conditions (23).

Rhizobial surface composition changes as bacteria differentiate into bacteroid forms.
The environmental conditions inside root nodules are very different from those experienced
by free-living bacteria (24-25). For example, there is a large decrease in the oxygen
concentration experienced by bacteroids inside nodules (as low as 20 nM) compared with

that experienced by free—living bacterium outside nodules (250 nM). The peribacteroid space

has a lower pH and a different carbon source because of the presence of a proton-ATPase and
a dicarboxylic acid transport system on the peribacteroid membrane (26). Compared with
free-living bacteria, bacteroids inside the nodules have a smaller amount or modified
composition of LPS (27-31) and lower levels of EPS (32). It has even been suggested that
the bacterial outer membrane may be entirely sloughed off during endocytosis by the host
cell (33). Irnmunocytochemical studies monitoring in planta the variations of LPS structure
point to a surprising degree of LPS epitope variations for bacteroids situated in different
regions of the a given nodule (25, 28, 29). These regions represent different physiological

and developmental states within the nodule. The changes in bacteroid LPS or EPS

29

composition can be simulated by growing free-living bacteria under culture conditions that
approximate those inside root nodules. Such conditions include low oxygen (21, 24, 31, 32),
low pH (21, 24, 34), and a combination of low oxygen, low pH and a different carbon source
(25).

It is obvious that the changes in the various surface carbohydrate-containing
molecules are related to the initiation and sustaining of infection. Some mechanisms for the
modes of action of the molecules are proposed. These include roles to facilitate metabolite
transport (34), to evade the plant defense systems (34,35), or to maintain certain physical
properties of the bacterial surface. The next section will be devoted to discussions on the
roles played by various rhizobial surface carbohydrate-containing molecules during

symbiosis.

V. ROLES PLAYED BY VARIOUS RHIZOBIAL SURFACE CARBOHYDRATE-
CONTAINING MOLECULES AT DIFFERENT STAGES OF SYMBIOSIS

The establishment of an effective symbiosis seems to hinge on the multiple
interactions between the plant and various rhizobial cell surface carbohydrate-containing

molecules such as chitolipooligosaccharides, EPSs, LPSs, cyclic B-glucans, and membrane

glycolipids. In the previous sections, the structures and the cellular locations of these
molecules were introduced. The dynamic nature of the biosynthesis of these molecules were
covered because it confers on the organism the capability to adapt to the changing micro-
environment at the different stages of symbiosis. The following is a summary of the studies

on the role(s) played by each class of molecules at different stages of symbiosis.

3O

Chitolipooligosaccharides: Rhizobia have a restricted host range. Each strain is able
to interact with only one or a limited number of legume species. Chitolipooligosaccharides
are considered by many to be the primary host range determinant (16, 36-37) Even though
the mechanism by which they determine host specificity is largely unknown, the production
of chitolipooligosaccharides is found to be specified by rhizobial nod genes which are
located on the symbiotic plasmid. The core oligosaccharide structures of the
chitolipooligosaccharides are specified by nod ABC genes and are common to all rhizobia.
The host-specific nod genes control the modifications of the core molecule with side groups,
which are thought to render chitolipooligosaccharides specific for their particular host.
Chitolipooligosaccharides of R. meliloti have a sulfate group on the reducing sugar, whereas
this substitution is lacking on the chitolipooligosaccharides produced by R. leguminosarum
biovars. Normally the chitolipooligosaccharides of R. meliloti is not active on vetch, a host
plant of R. leguminosarum bv. viciae, but after removal of the sulfate group it gains the
ability to infect vetch and loses this ability on its own host, alfalfa (36). The involvement of
chitolipooligosaccharides substitutions in host specificity is strengthened by studies on R. Sp.
NGR234, which has a broad host range. The bacterium has the unique ability to interact with
more than 70 genera of legumes as well as with Parasponia andersonii, the only non-legume
that can form nodules with Rhizobium. R. Sp. NGR234 produces over 18 different
chitolipooligosaccharides with similar basic structure but different substitutions (38). It
seems that the diversity of chitolipooligosaccharides accounts for the broad host range of this
bacterial strain. In vitro experiments showed that purified chitolipooligosaccharides, when

applied to legume seedlings at concentrations as low as 10‘'2 M, stimulate the differentiation

31

of epidermal cells into root hairs and elicit root hair curling and cortical cell division ( 3, 39-
40).

EPSs: At any stage of symbiosis, EPSs are likely to be chemical signals between the
bacterium and plant because of their locations on the outside of the cell and because of the
diversity of EPSs structures among different bacterial species. EPS are required for the
normal development of root nodules of many rhizobial species (9). Mutants that do not
produce EPSs can be selected based on their lack of fluorescence on Calcofluor medium. By
this method, a cluster of em genes has been found to specify the production of EPSs ( 41).
In R. meliloti, exo mutants consistently failed to nodulate the host alfalfa successfully. In
these exo mutants, nodule invasion appeared to be blocked at the stage when the infection
thread penetrates beyond the epidermal cell layer. This implies that instead of being the
initial signal or the primary host range determinant, EPSs are required for nodule invasion.
It has been shown that low molecular weight form of EPSs can partially suppress the
symbiotic deficiencies of em mutants (41).

LPS: LPSs have also been reported to play some important roles in the symbiotic
infection process. Two mutants of R. leguminosarum bv. Phaseoli with a defective LPS 0-
antigens are symbiotically defective (42). When inoculated onto the host legume, Phaseolus,
these mutants result in the initiation of nodules that become arrested in development because
the infection threads which contain invading rhizobia appear to abort, usually within the root
hair cell. It seems that complete LPS structures, including normal amounts of 0 antigen, are
necessary for the development of the infection thread. In some other studies, LPSs have been

reported to be required for the formation of normal infection thread or the invasion of the

32

root cortical cells (43-49).

Cyclic B-glucans: Cyclic B‘glucans are specified by rhizobial ndvA and nde genes.
The most compelling evidence for a role of cyclic B-glucans during plant infection comes

from studies with ndvA/nde mutants of R. meliloti (15, 50-51). These studies have revealed
that the R. meliloti ndvA and nde mutants form ineffective white pseudo nodules on alfalfa.
Closer examination of the pseudo nodules induced by ndv mutants of R. meliloti reveals a
small number of infection threads that abort at an early stage, and no nitrogen-fixing
bacteroids are observed. The initiation of infection threads by these mutants suggests that the

cyclic B-glucans provide functions during the later stages of alfalfa nodulation. R. meliloti

ndv mutants resemble exo mutants in that they form “empty” (uninfected) nodules and differ
in that they react normally with Calcofuor and curl root hairs normally (26).

Glycolipids: Possible roles for membrane lipids in the symbiotic process have been
proposed in a recent study. One mannose and glucose-containing diglycosyl diacylglycerol
has been reported to accumulate in the membrane of Rhizobium leguminosarum bv. viciae
ANU 843 when the bacteria were grown in the presence of flavonoids (l4). Subnanomolar
concentrations of this glycolipid could also elicit various responses typical of symbiosis such
as root hair deformation and cortical cell division on the host plant. The unusual occurrence
of SQDG in many rhizobial species also suggests a role for these glycolipids during
symbiosis (13).

From the above mentioned studies, it is clear that a diversity of rhizobial surface
carbohydrate-containing molecules are involved at different stages of symbiosis. This

diversity complicates studies aimed at defining roles of each individual components. These

33

studies become even more complicated because of the large degree of coupling or overlap
of biosythetic events leading to the synthesis of various molecules. The next section will be
devoted to this aspect.
VI. FUNCTIONAL AND BIOSYNTHETIC OVERLAP AMONG VARIOUS
RHIZOBIAL CELL SURFACE CARBOHYDRATE-CONTAINING MOLECULES

Most mutant studies on the involvement of certain classes of carbohydrate-containing
molecules during symbiosis are carried out without checking the pleiotropic effects of the
mutation(s) on other classes of molecules. However, it is the whole bacterial surface that
interacts with the plant at each stage of symbiosis. Studies on the functional and biosynthetic
overlap among various rhizobial cell surface components merit special attention.

The best characterized pleiotropic Exo‘ mutants are exoB mutants of R. meliloti.
These mutants are defective in the synthesis of both LPS and EPS. Like other Exo’ mutants
of R. meliloti, the exoB mutants form ineffective nodules. It was found later that these exoB
mutants fail to synthesize active UDP-glucose 4'-epimerase and thus fail to make galactose
(52). Since galactose is a structural element of both EPS and LPS of these strains, it is not
surprising both EPS and LPS are defective in these mutants (52). This raises a question about
the critical factor involved in nodule formation. What accounts for the deficiency in
symbiosis, EPS or LPS?

Some studies on EPS- and LPS-negative mutants of some Rhizobium strain (53, 54)
found that these strains were still able to form nitrogen-fixing nodules, indicating that EPS
or LPS are not required for nodulation. Other studies provided evidence that LPS and EPS

may have the same function in the plant-bacteria interaction in some symbiotic systems

34

(55,56). In R. meliloti, the symbiotic phenotype of em mutants are deficient in EPS and
cannot nodulate alfalfa. Mutants carrying a lpsZ+ gene can restore nodulation. It is found that
lpsZ+ does not restore EPS production but instead alters the composition and structure of
LPS(57). The evidence that EPS and LPS can substitute each other in these strains also lead
to hypotheses on the role of surface polysaccharides in plant-microbe interactions:
irrespective of their exact surface structures, these carbohydrate-containing molecules work
together to provide some signal(s) promoting recognition events (55-57). These signals can
be oligosaccharides, polysaccharides or even the physico-chemical properties of the whole
bacterial surface.

The bacterium achieves the optimum physico—chemical properties for its surface by
constantly changing its composition. The latter is achieved by the biosythetic regulation of
many of its surface molecules. It is not surprising, therefore, that there is a large degree of
coupling of the many biosynthetic events leading to the synthesis of the various surface
components. For instance, both the lipo-oligosaccharides and LPS of Rhizobium meliloti are
known to be sulfated (58). A nod PQ mutant of R.. meliloti which is defective in nodulation
makes a non-sulfated lipo-oligosaccharide. However, it was found later that this mutant also
has an altered LPS (not sulfated) in addition to structural changes in membrane lipids
including SQDG. In rhizobial exo mutants which have an altered EPS, LPS were also
changed (59). R. leguminosarum bv. trifolii 48 has a LPS O-antigen that lacks galactose and
many of the typical glycosyl components found in related strains. It has been recently found
that it synthesizes an alternative glycolipid which contains galactose and the typical 0-

antigen glycosyl components, suggesting that in this strain, the “missing” O-antigen of LPS

35

is transferred to an alternative lipid acceptor (60).

It is obvious that there is functional and biosynthetic overlap among various classes
of rhizobial cell surface components. Without studies on the pleiotropic effects of a particular
mutation, the classical approach of mutating-then—evaluating is not applicable to the studies
on the roles played by each individual bacterial surface component. It is clear that in studies
on the mechanism of plant-microbe interaction, importance should be put on the following
aspects: 1, to obtain a comprehensive picture of the rhizobial cell surface with respect to the
structures of various surface and membrane molecules; 2, to grasp the changes in this picture
when rhizobia live in different environments, especially the environment inside of the plant;
3, to understand the carbohydrate metabolism underlining the biosynthesis of various

carbohydrate-containing molecules which are so central to symbiosis.

36

CONCLUSION

The first two sections of the literature review introduced the importance
rhizobium/legume symbiotic system and the process by which the symbiotic relationship
between rhizobia and leguminous plants is established. At each stages of symbiosis, a
variety of rhizobial cell surface and membrane carbohydrate-containing molecules have been
indicated to play some important roles (section V). The structures and cellular localizations
of these molecules were covered in section III. The functions of various glycoconjugates in
the establishment of symbiosis are closely related to the ability of rhizobia to change its
surface chemistry in response to a changed environment (section IV). There is a large degree
of functional and biosynthetic overlap among various classes of rhizobial cell surface
components (section VI).

It is the whole bacterial surface that interacts with plants. A comprehensive picture
of the rhizobial cell surface is needed in order to understand bacteria-plant interactions.
There are two important parts of the big picture: the environmental regulation of the bacterial
surface and the coupling of the biosynthetic events leading to various surface molecules.

The work presented in this dissertation is based on these two parts. Three chapters
of this dissertation address the aspect of environmental regulation of rhizobial membrane
chemistry. Chapter 2 and 3 respectively presents studies on the membrane lipid chemistry
of B. japonicum living as bacteroids inside plant nodules and living as bacteroid-like form

in bacteroid-inducing media. Chapter 4 presents studies on membrane lipid chemistry of

37

Bradyrhizobium grown under low oxygen and low pH conditions which mimic condition
inside root nodules. Chapter 5 describes an investigation on the interrelationship of the

biosynthetic pathways leading to various rhizobial surface carbohydrate-containing

molecules.

38

BIBLIOGRAPHY

1. Postgate, J. R. 1987. In Nitrogen Fixation (J. R. Postgate, Ed.) pp 1-9, Edward Arnold,
London, Australia and USA.

2. Kim, J. And D. C. Rees. 1994. Biochemistry 33, 389—397.

3. Dixon, R. O. D. and C. T. Wheeler. 1986. In Nitrogen Fixation in Plants (Dixon, R. O.
D. and C. T. Wheeler, Eds.) pp 15-44, Chapman & Hall, New York.

4. Fisher, R. F. and S. R. Long. 1992. Nature 357, 655-660.
5. Long, R. L. 1989. Cell 56, 203-214
6. Van Rhijn, P. and J. Vanderleyden. 1995. Microbiol. Rev. 59, 124-142.

7. D. Werner. 1991. In Biological Nitrogen Fixation ( Stacey, G., Burris, R. H. and H. J.
Evans, Eds. ) pp 399-422, Chapman & Hall, New York and London.

8. Verma, D. P .S. 1992. The Plant Cell 4, 373-383

9. Leigh, J. A. and G. C. Walker. 1994. Trends In Genetics. 10, 63-67.

10. Lugtenbeg, B. And L.V.Alphen. 1983. Biochimica et Biophysica Acta. 737, 51-1 15.
11. Ying Wang. 1995. PhD. Thesis. Dept. of Chemistry, Michigan State University.
12. Goldfine, H., and M. E. Ellis. 1964 . J. Bacteriol. 87, 8-15.

13. Cedergren, R. A., and R. I. Hollingsworth. 1994. J. Lipid Res. 35, 1452-1461.

14. Orgarnbide, G. G., S. Philip-Hollingsworth, R. I. Hollingsworth and F. B. Dazzo. 1994.
J. Bacteriol. 176, 4338-4347.

15. Breedveld, M. W. and K. J. Miller. 1994. Microbiol. Rev. 58, 145-161.

16. Denarie, J and J. Cullimore. 1993. Cell 74, 951-954.

39

17. Cedergren, R. A., L. Jeongrim, L. R. Kerry and R. I. Hollingsworth. 1995. Biochemistry
34, 4467-4477.

18. Orgarnbide, G., Lee, J ., Hollingsworth, R. I., and F. B. Dazzo. 1995. Biochemistry 34,
3832-3840.

19. Hazel, J. R. and E. E. Williams. 1990. Prag. Lipid Res. 29, 167-227.

20. Rose, A. H. 1989. In Microbial Lipids. (Ratledge, C and Wilkinson, S. G. eds.) pp255-
277. Academic Press. New York.

21. Tao, H., N. J. Brewin, and K. D. Noel. 1992. J. Bacterial. 174, 2222-2229.
22. Kannenberg, E. L. and N. J. Brewin. 1989. J. Bacterial. 171, 4543-4548.

23. Kannenberg, E. L., E. A. Rathbun, and N. J. Brewin. 1992. Mol. Microbiol. 6, 2477-
2487.

24. Kannenberg, E. L., S. Perotto, V. Bianciotto, E. A. Rathbun, and N. J. Brewin. 1994. J.
Bacterial. 176, 2021-2032.

25. Sindhu, S. S. , N. J. Brewin, and E. L. Kannenberg. 1990. J. Bacteriol. 172, 1804-1813.

26. Gresshoff, RM. 1990. In The Molecularbialogy of symbiotic nitrogen fixation. (P.M.
Gresshoff, ed). CRC Press, Inc. Florida.

27. de Maagd, R. A., A. S. Rao, 1. H. M. Mulders, and B. J. J. Lugtenberg. 1989. J. Bacterial.
171, 1136-1142.

28. Brussel, A. A. V. van, K. Planque, and A. Quispel. 1977. J. General Microbial. 101,
51-56.

29. Brewin, N. J ., E. A. Wood, A. P. larkins, G. Galfre and G. W. Butcher. 1986. J.General
Microbial. 132, 1959-1968.

30. Planque, K., J. J. van Nierop, A. Burguers, and S. G. Wilkinson. 1979. J. Gen. Microbial.
110,151-159.

31. VandenBosch, K. A., N. J. Brewin, and E. L. Kannenberg. 1989. J. Bacteriol. 171, 4537-
4542.

32. Tully, R. E. and Terry, M. E. 1985. Plant physial. 79, 445-450.

40

33. Bal. A. K. and P. P. Wong. 1982. Can. J. Microbial. 28, 890-896.
34. Bhat, U. R. and R. W. Carlson. 1992. 174, 2230-2235.
35. Parniske, M., B. Ahlbom, and D. Werner. 1991. J. Bacteriol. 173, 3432-3439.

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

37. Carlson, R. W., N. P. J. Price, and G. Stacey. 1994. Mol. plant-microbe Interact. 7, 684-
695.

38. N. P. Price et al. 1992. Mol. Microbial. 6, 3575-3577.

39. H. P. Spaink. 1992. Plant Mol. Biol. 20, 977-983.

40. J. Denarie et al. 1992. Annu. Rev. Microbial. 46, 497-51 1.

41. Reuber, T. L. and G. C. Walker. 1988. Cell 74, 269-280.

42. Noel, K. D., K. A. VandenBosch, and B. Kulpaca. 1986. J. Bacteriol. 168, 1392-1401.
43 Brink, B. A., J. Miller, R. W. Carlson and K. D. Noel. 1990. J. Bacterial. 172, 548-555.

44. Carlson, R. W., S. Kalembasa, D. Turowski, P. Pachori and K. D. Noel. 1987.J.Bacterial.
169, 4923-4928.

45. Noel, K. D., K. A. Vandenbosch and B. Kulpaca. 1986. J. Bacteriol. 168, 1392-1401.

46. Cava, J. R., P. M. Elias, D. A. Turowski , and K. D. Noel. 1989. J. Bacterial. 171, 8-
15.

47. de Maagd, R. A., A. S. Rao, 1. H. M. Mulders, L. G. de Roo, M. C. M. van Loosdrecht,
C. A. Wijffelman, and B. J. J. Lugtenberg. 1989. J. Bacteriol. 171, 1143-1150.

48. Priefer, U. B. 1989. J. Bacteriol. 171, 6161-6168.

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

50. Dylan, T., P. Nagpal, D. R. Helinski, and G. S. Ditta. 1990. J. Bacterial. 172, 1409-1417.

41

51. Dylan, T., P., L. Ielpi, S. Stanfield, L. Kashyap, C. Douglas, M. Yanofski, E. Nester, D.
R. Helinski, and G. S. Ditta. 1986. Proc. Natl. Acad. Sci. USA 83, 4403-4407.

52. Cremers, H.J., M. Batley, J. W. Redmond, L. Eydems, M.W. Breedveld, L.T.M.
Zevenhuizen, E. Pees, C.A. Wijffelman, And B. J. Lugtenberg. 1990. J. Biol. Chem. 265,
21122-21127.

53. Lopez-lara, I. M., G. Orgarnbide, F. B. Dazzo, J. Olivares, and N. Toro. 1993.].Bacterial.
175, 2826-2823.

54. Clover, R. H., J. Kieber, and E. R. Signer. 1989. J. Bacterial. 171, 3961-3967.

55. Putnoky, P., G. Petrovics, A. Kereszt, E. Grosskopf, D. Cam Ha, Z. Banfalvi and
A.Kondorosi. 1990. J. Bacterial. 172, 5450-5458.

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

57. Williams, M. N. V., R. I. Hollingsworth, S. Klein and E. R. Signer. 1990. J. Bacteriol.
172, 2622-2632.

58. Cedergren, R. A., J. Lee, K. L. Ross, and R. I. Holligsworth. 1995. Biochemistry 34,
4467-4477.

59. Diebold, R. and K. D. Noel. 1989. J. Bacterial. 171, 4821-4830.

60. Cedergren, Y. Wang, and R. I. Holligsworth. 1996. J. Bacterial 178, 5529-5532.

CHAPTER 2
THE LIPID CHEMISTRY OF BACTEROIDS

ISOLATED FROM PLANT NODULES IS DISTINCTLY DIFFERENT
FROM THAT OF FREE-LIVING RHIZOBIA

42

43

ABSTRACT

The membrane lipids of free-living and bacteroid forms of Bradyrhizobium
japonicum, obtained from nodules occupied by both typed and untyped bacteria, were
isolated and characterized by a combination of NMR spectrometry, mass spectrometry and
other chemical and physical methods. These studies indicated that both the free-living and
bacteroid forms of this organism contain glycolipids almost exclusively of the type found in
plant cells. In the bacteroid forms, there was a dramatic shift towards the synthesis of
digalactosyl diacylglycerol as the major lipid. This glycolipid has rarely been found outside
of the plant kingdom and photosynthetic bacteria and its occurrence in the bacteroid form
of a plant symbiont is therefore remarkable. The presence of significant plant cell and
organelle contamination in the bacteroid preparations was ruled out by scanning electron
microscopy, southern blot analysis for plant DNA using specific gene probes and chemical
analysis for plant marker steroids, steroid glycosides and glycosyl ceramides. DGDG is not
found in the plasma membrane of plant cells (of which the peribacteroid membrane is an
extension) but is thought to be restricted to plastids. This result follows our earlier finding
that the other predominant plant glycolipid, sulfoquinivosyl diacylglycerol, is a membrane
component of fast growing Rhizobia and is found even when cells are cultivated in free
culture where there is no question of plant contamination. The absence of these lipids in the
membranes of bacteria outside of this special group of organisms and photosynthetic

bacteria suggests that the trait could have been passed on through gene transfer from plants

44

to the bacteria at some point during the development of their symbiotic relationship. In the
case of digalactosyl diacylglycerol, there is also the possibility that some common
biosynthetic intermediates are used by both the plant and the bacteria. This is a striking

parallel with some host-pathogen interactions.

45

INTRODUCTION

The chemistry of bacterial cell surfaces is critical to the physiology and hence the
adaptability of these organisms. When adaptation involves a close symbiotic relationship
with another organism, the relative chemistry of the surfaces of the two species becomes a
major issue. Among the biological systems in which two vastly different species become
involved in a symbiotic relationship which benefits both partners, none (perhaps) has
received as much attention as the Rhizobia-legume symbiosis. Membrane chemistry is very
closely tied to surface chemistry. This is because many of the predominant cellular surface
antigens of both eucaryotes and procaryotes cells are, in fact, often structural components of
the membrane. For instance, lipopolysaccharides which contain the O-somatic antigen are
actually integral bacterial outer membrane components. Various rhizobial surface and
membrane carbohydarate-containing molecules have been implicated to function in the
process of symbiosis as covered in Chapter 1.

In an earlier study (1) we demonstrated that the membranes of members of the family
Rhizobiaceae contained SQDG, a glycolipid which was thought to occur almost exclusively
in membranes of plant plastids and photosynthetic bacteria although its presence has been
demonstrated in a few other bacterial species (2). This unusual occurrence prompted
questions about the developmental and evolutionary relationship between Rhizobium and
plants. Since the membrane is a critical structural feature that determines the ability of an

organism to occupy a particular ecological niche, it is not very surprising that two organisms

46

whose survival revolves around shared ecological pressures of temperature, pH, and water
and nutrient availability and whose development and metabolism must be synchronized
during the symbiotic period, have common membrane components. SQDG is not the only
common component between the membrane lipids of Rhizobium and legumes. One of the
predominant rhizobial membrane lipids is PC, a lipid that is not common in bacteria but
typically occurs in plant cells (3).

In previous studies, we examined vegetative, free-living organisms. It was therefore
important to examine in situ bacteroid forms of Rhizobium to see whether they contained
predominantly large amounts of SQDG, DGDG or any of the common plant glycolipids.
Bacteroids are the in planta nitrogen-fixing form of Rhizobium. For this study, we chose to
investigate the Bradyrhizabium / soya bean system since we had detected small, but
measurable, quantities of SQDG in the membranes of this species and could, therefore,
readily determine whether substantially higher levels were present in their bacteroids. Soya
bean nodules of two types, one inoculated with a known typed strain (USDA 110) and the
other one infected by some untyped indigenous strain from the soil, were used. Soya bean
nodules were used because of their large size which would ensure a high yield of bacteroids

to allow fairly rigorous chemical characterization.

47

MATERIALS AND METHODS

Cultures of Free-living Bacteria and Preparations of Bacteroids

Soya bean nodules were provided by Dr. Thomas Deits, Michigan State University.
Rhizobiumjaponicum USDA 110 was grown at 30°C in B-III broth medium and harvested
as previously described (4). The pellet was collected and used for membrane lipid extraction.
Two different sets of soybean root nodules were used in the experiments. One was obtained
by inoculating the host plant (soya bean) with free-living Bradyrhizobium japanicum USDA
110. Prior to inoculation, plants were grown in sterilized soil and were maintained under
green house conditions. The other set of nodules was obtained from field grown plants in
which indigenous, untyped rhizobium strain(s) resided. Nodules were crushed in 0.3 M
sucrose at 4°C and the bacteroids were separated from plant debris by filtering through glass
wool filter. The filtrate was centrifuged at 100g for 30 min. The supernatant was then
centrifuged at 5000g for 1 h. The two pellets were examined separately by scanning electron
microscopy, found to be the same and therefore combined and used for DNA isolation and
membrane lipid extraction. The plant debris and the final supernatant fraction were also used
for lipid extraction.
Electron Microscopy

Scanning SEM was performed on a JEOL JMS-35CF SEM instrument. Samples
were fixed at 4°C for 1-2 hours in 4% glutaraldehyde and dehydrated in an ethanol series

(25%, 50%, 75%, 95%, 100%). After this, samples were coated with gold (20 nm thickness)

48

in an Emscope Sputter Coater model SC 500, purged with argon gas and then examined by
SEM.
Isolation and Electrophoresis of DNA and Southern Hybridization

Roots and leaves were excised from soyabean plants grown in the fields and treated
separately. One gram of material was cut into small pieces and ground in liquid nitrogen. A
pellet of B. japanicum USDA 110 bacteroids was also ground in liquid nitrogen. 3 mL of
cetyltrimethylammonium bromide buffer (pH 8.0, 20 gram/mL) was added to each sample.
The mixtures were incubated at 60°C for one hour, cooled down to room temperature,
extracted three times with equal volume of chloroform and centrifuged at 8500g for 10
minutes. The final aqueous extractions were transferred to sterile plastic tubes , 5M NaCl
was added to each tube to achieve a final concentration of 2M and then 0.75 volume
isopropanol was added to precipitate the DNA at -20°C for 2 h. The precipitated DNA was
recovered by centrifugation at 6000g for 5 min, redissolved in water, precipitated again at
-20°C in 3M sodium acetate and two volume of ethanol, washed with 70% ethanol and

resuspended in 500 uL triethanolamine buffer at pH 8.0. Comparable amount of DNA

samples were loaded in 1% nondenaturing agarose gel. Electrophoresis was conducted,
bands visualized (ethidium bromide) and DNA transferred to nitrocellulose membranes and
hybridized to probes according to standard procedures (5). The DNA probes used were from
clone pOj116 containing the chloroplast pet A gene and from clone pRS2 containing the
rhizobial nif genes. These were obtained from Dr. Barb Sears and Dr. Frans de Bruijn,

respectively.

49

Lipid Extraction and Thin Layer Chromatography

Standard PS, PC and DGDG were purchased from Sigma Chemical Co. (St Louis,
Missouri). Standard PG was purchased from Avanti Polar Lipids (Alabaster, Alabama).
Lipids were extracted, analyzed by TLC on silica plates using a mixture of chloroform,
acetone, methanol, acetic acid and water in the ratio of 10:4:2: 1:1 (13) and the components
located as previously described (1).
Gas chromatography and Mass Spectrometry

Fast atom bombardment mass spectrometry (FAB-MS) and gas chromatography /
mass spectrometry (GC-MS) were performed and recorded as previously described (1) except
that, for FAB-MS, glycerol was used as the matrix and the ionization voltage was 10 kV. Gas
chromatographic separations for fatty acids were accomplished on a 30 meter DB-l capillary
column ( J & W Scientific, Folsom, California). The program used was: 100° to 150°C at
25°C/minute and 150° to 300°C at 3°C/min with a 10—min hold at 300°C. For the separation
of alditol acetate derivatives, a J & W Scientific DB-225 column was used. The program
used was: 100° to 180°C at 40°C/min and 180° to 220°C at 4°C/min with a 50—min hold at
220°C.
Nuclear Magnetic Resonance Spectroscopy (N MR)

Proton NMR analyses were performed on a Varian VXR-SOO spectrometer (500 MHz
for protons) at 25°C in deuterated chloroform or perdeuterated methanol. Chemical shifts are
referenced relative to the residual chloroform line at 7.24 ppm or the residual methanol line

at 3.30 ppm.

50

Chemical Characterization of Lipids

Fatty acids from lipids of bacteroid, plant debris and supernatant were obtained as
methyl esters as previously described (1) except that the methanolysate was partitioned
between 2 mL of chloroform and lmL of water. The lower chloroform layer was recovered
and subjected to GC and GC/MS. The glycosyl components of the lipids of these three
samples were identified with GC / MS after converting them to alditol acetate derivatives as
described before (6).
Methylation Analysis

The linkage pattern within the carbohydrate moieties of the lipids was determined by
methylation analysis (7) with some modifications. A sample of the total lipid extract (~ 100
pg) was treated with 100 uL of a 0.2 M solution of sodium methylsulfinyl ion at room
temperature for 12 hours. Iodomethane (100 pL ) was then added and the mixture stirred for
another 12 hours and then diluted with 2 mL of water and extracted with 3 mL of
chloroform. The chloroform layer was recovered and the contents of the aqueous layer
adsorbed onto a short C-18 reverse phase cartridge attached to a syringe. The cartridge was
washed with water (2 mL) and the adsorbed material was then released by eluting with
methanol (5 mL). The methanol wash was combined with the first chloroform extract and
the solvent removed under a stream of nitrogen. The residue was hydrolyzed by treatment
with 2 M trifluoroacetic acid for 3 hours at 100°C. The glycosyl components thus released
were then reduced with sodium borohydride and the alditols so fOrmed converted to
peracetates as described earlier (1). The methylated, acetylated products were analyzed by

GC andGC/MS.

51

RESULTS

The method used for the isolation of bacteroids from plant nodules has been widely
used and is known to result in negligible contamination by plant material (8). Most plant
organelles cannot be pelleted with the bacteroids at the centrifugation speeds used, especially
in a medium with the buoyancy of the sucrose solution used. Prior filtration removes plant
debris and associated plant-derived lipids. Several experiments aimed at addressing whether
trace levels of plant contaminants were present were conducted even though gross
contamination could not have led to the levels of glycolipids in the bacteroids. Scanning
electron microscopy examination of the bacteroid suspension was an attractive proposition
because of the large differences in the morphology and sizes of bacteria and bacteroids on
one hand and plant cells and organelles on the other. Plant cells and many organelles are
much larger than bacteroids and vegetative rhizobial cells. The latter cells are readily
recognized by their characteristic cigar shapes. These may be spherical at certain stages of
bacteroid development. SEM micrographs (Figure 2.1) of bacteroids isolated from soybean
nodules in this study showed that there were two cell types, one rod shaped and the other,
though of comparable size, spherical. The cell morphology observed in the SEM were
identical to those in published literature (9). There was no detectable contamination by plant
cells. Southern hybridization using probes for chloroplast marker genes were performed to
look for the presence of plant root plastids. DNA from soybean leaves, soybean roots and

bacteroids of B. japonicum USDA 110 were loaded on to electrophoresis gels. The DNA

52

1/

 

'15KU X4488 ease 1.63 05094

Figure 2.1: Scanning electron micrograph showing a typical field in the bacteroid
preparation from soya bean root nodules. The sharp needle-shaped structures are sucrose

Crystals. Note the typical elongated, cigar and spherical shapes of bacteroid forms of

Bradyrhizobiumjaponicum.

1501;

'110

D.)

Sill

ha

53

isolated from both soybean roots and leaves gave positive signals with chloroplast petA gene
probe ( Figure 2.2, panel A, lane 1 and lane 2 ). In contrast, no signal was detected in the
DNA isolated from USDA 110 bacteroids ( Figure 2.2, panel A, lane 3 ). When DNA
samples were hybridized with the probe for the rhizobial nif gene, only the DNA from
bacteroids gave a positive response ( Figure 2.2, panel B, lane 3 ). Since the pet A gene is
highly conserved in different plastids ( including chloroplasts ) among various plant species,
the conclusion could be made that no intact plastids were present.

To exclude the possibility that the bacteroids were contaminated with plant lipid
vesicles from broken plastids, a proton NMR (Figure 2.3A) spectrum of the lipids was
compared to that (Figure 2.3B) of an extract of the supernatant which contained plant lipid
vesicles that could not be pelleted at the centrifugation speeds used for bacteroid isolation.
Some major difference were observed in the two spectra. One was the presence of several
signals between 6 and 8 ppm (attributable to vinyl protons) in the spectra of the bacteroid
lipid extracts (Figure 2.3A) but were absent from the supernatant extract (Figure 2.38). Other
major differences were observed in the region between 2.8 and 5 ppm. Signals corresponding
to carbohydrate headgroup of lipids in this region were predominant in the bacteroid lipid
spectra. However, some other signals in this region were present in the supernatant extract
but were absent from that of the bacteroid extract. These included (Figure 2.3B) a doublet
at 4. 1 ppm, another doublet at 5.4 ppm and some signals between 3 to 4 ppm. All these results
conclusively ruled out the contamination of plant lipid vesicles which contained lipids of
broken plastids.

The glycosyl components of the lipids of bacteroid samples were identified by GC

54

 

Figure 2.2: In both panels, Lane 1: Soybean leaf DNA; Lane 2: Soybean root DNA; Lane
3: USDA 110 Bacteroid DNA. Panel A: Hybridized with a probe from a clone p0jll6
containing the chloroplast pet A gene which is highly conserved in different plastids (
including chloroplasts ) among various plant species. B: Hybridized with a probe from clone

pRS2 containing rhizobial nif gene.

55

Figure 2.3: Proton NMR spectrum of lipids from bacteroids (A) and from extract of the

supernatant which contained plant lipid vesicles (B).

fi.+__,._

.._,__ _,.

56

1' "—T‘ T’fi—T ‘_7"

 

 

 

 

 

.Fj

 

 

 

W

—t

 

 

 

 

 

 

 

w—T-T-‘w
1 "-

 

57

/ MS after converting them to alditol acetate derivatives. Glycerol and galactose were the
primary components of these lipid preparations (Figure 2.4). Gas chromatography of the
alditol acetate derivatives from chloroform extracts of plant debris and the supernatant
indicated that no glycolipids were recovered from the debris (Figure 2.5A) and that glycerol
and glucose were the primary components in the supernatant (Figure 2.5B). Glucose is a
typical component of glycosyl ceramides, the predominant sphingolipid in plant tissue. These
results demonstrated that the galactose in the bacteroid sample did not come from plant
material and rule out the possibility of contamination of the bacteroid sample by plant lipids.
Gas chromatography of fatty acid methyl ester derivatives from the same 3 samples (Figure
2.6A-C) also supported this conclusion. The chromatogram of the bacteroids sample (Figure
2.6 C) indicated the presence of only fatty acid methyl esters and there were no peaks for any
of the long retention time components (presumably steroids and sphingosine component of
glycosyl ceramides) that the chromatograms from the debris (Figure 2.6A) and supernatant
(Figure 2.6B) contained.

The bacteroid total lipid extracts were analyzed by fast atom bombardment mass
spectrometry (FAB-MS). The spectra of the lipids from the USDA] 10 bacteroids (Figure
2.7A) and untyped bacteroids (Figure 2.78) contained quite intense signals clustered at m/z
941 and 947 respectively. These masses are uncharacteristic of any phospholipids but
corresponded to diglycosyl diacylglycerols containing two octadecanoyl chains in the case
of the 947 ion and a similar lipid but with a total of 3 double bonds between the two fatty
acids in the case of the 941 ion. Since galactose was essentially the only significant hexose

Component of these lipid preparations, it followed that such a glycolipid would have to be

58

Figure 2.4: Gas chromatogram of alditol acetate derivatives of the headgroup components
from the total membrane lipid preparation of Bradyrhizobium japonicum bacteroids. The
peak labeled A is from glycerol and the one labeled B is from galactose. Note that glucose,
a typical component of glycosyl ceramides ( predominant sphigolipid in plant tissue ) is

absent. The elution position of glucose is indicated by arrow.

59

 

 

 

 

 

omcoaaom .2023

 

I
15

Retention time (mins.)

60

Figure 2.5: Gas chromatogram of the alditol acetate derivatives of the residue (A) and of the
supernatant from the bacteroid extraction procedure. It was found that glycerol and glucose

were the primary components in the supernatant.

61

 

uncommom ~33qu

 

 

 

 

 

 

62

Figure 2.6: Gas chromatogram of the chloroform-soluble fraction (A) after acid-catalyzed
methanolysis of the residue, the supernatant (B) and the bacteroid preparations (C). Note that
the bacteroids do not contain any of the long retention time components (presumably steroids
and sphingosine component of glycosyl ceramides) that the residue and supernatant contain.
Only fatty acid methyl esters were present in the bacteroid sample. The major fatty acids

were labeled in all three figures.

Response

Detector

63

- 16:0

18:1

18:2

 

 

 

 

 

16:0
18:1

 

 

 

 

 

16:0
18 1

18:0

 

 

Retention time (mlns.)

64

Figure 2.7: Negative-ion FAB-MS of the total membrane lipid extracts from
Bradyrhizabiumjaponicum USDA 1 10 bacteroids (A) (matrix: glycerol), untyped bacteroids

(B) ( matrix: Glycerol) and untyped bacteroids C ( matrix: triethanolamine).

100

80

60.

20

O

100

80

40

zol

65

 

 

 

 

800

 

lOOOM/Z

 

 

200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

66

a DGDG. Signals for phospholipids were completely suppressed in the MS spectrum of the
untyped bacteroid lipids but could be observed in those from the Bradyrhizobium japanicum
USDA 110 bacteroid lipids. In both spectra, signals for SQDG were not very prevalent in the
positive ion FAB spectra but were major ions in negative ion spectra when triethanolamine
was used as the matrix (Figure 2.7C). Previous studies (10-11) demonstrated that the
predominant membrane phospholipids of free-living Bradyrhizabium japonicum were PC,
PE and PG. TLC analysis of the total lipid extracts from two bacteroids (untyped and USDA
110) and the vegetative, free living bacteria showed that there were major compositional
differences among the three extracts (Figure 2.8). Spots corresponding to PC, PE and PG are
observed in all three extracts. The striking difference between the two bacteroid extracts and
free-living bacterial extract was the presence of a very slow migrating component in the
extracts from the bacteroids. The Rf values were very similar to that of a commercial
standard of DGDG (Figure 2.9). There was considerable streaking indicating a fair degree
of heterogeneity in the fatty acid chains. This was also noticed for the commercial standard.
In both bacteroid extracts, there was a fast moving component (which also appeared as a
streak in USDAl 10 bacteroid extracts) which was identified as poly 3-hydroxybutyric acid
by NMR spectroscopy. The slow moving component on TLC of the untyped bacteroid
sample was isolated by using both preparative silica gel TLC and reverse phase adsorption
chromatography employing a step gradient. The FAB mass spectra of the isolated component
contained the cluster of signals at m/z 947 which was assigned to diglycosyl diacylglycerol.
Compositional analysis confirmed the presence of glycerol and galactose as the sole polar

components.

67

Figure. 2.8: Orcinol-sprayed TLC plate of membrane lipids extracts. DGDG is the authentic
standard of digalactosyl diacylglycerol. The last three lanes contain membrane extracts from

untyped bacteroids, USDA 110 bacteroids and free living USDA 1 10, respectively.

 

68

 

 

m -. .__A-...A. g“

 

 

 

 

 

 

 

 

DGDG P3 P5 pc Untyped USDA‘I‘IO VEG 110

 

69

Figure 2.9: The structure of isolated DGDG.

70

OH

71

The linkage of the two galactosyl residues in the galactose-containing glycolipids was
determined by methylation analysis according to published procedures (7). This analysis
clearly indicated the presence of one non-reducing terminal and one 6-linked galactose
residue. This is the usual linkage of galactose in plant-derived DGDG. To confirm the
anomericity (or vs [3) of the glycosyl linkages and the site of acylation, a proton NMR
spectrum (Figure 2.10) of the isolated DGDG from strain 110 was obtained. One signal, a
doublet at 4.80 ppm with J ~ 2 Hz, was assignable to an a anomeric proton. Another doublet
at 4.11 ppm with J = 8 Hz was assignable to the B anomeric proton. These were in
reasonable agreement with the positions for these signals in a standard spectrum of DGDG
in which the a proton signal was at 4.85 ppm and the B proton signal was at 4.23 ppm. The
fatty acid compositions of the standard and the isolated sample are very different with the
former being rich in C1823 fatty acids. This might give rise to slightly different
conformations of the head group and to slightly different chemical shifts of the anomeric
protons. This result, however supportive of the notion, is not a definitive proof of which
galactose unit has the a configuration and which has the 8. Such a study would necessitate
degradative (chemical) or NMR spectroscopy n.O.e studies. These were precluded because
of the very limited amount of material that can be isolated and the very small amounts that
remain after purification. The information we can extract from the data and TLC
comparisons with a standard, however, strongly indicates that the configurations are the same
as that found in the typical plant glycolipids.

Since all of the lipids present in the membrane of Bradyrhizabium contain one

molecule of glycerol (except for phosphatidyl glycerol which has two) it was possible to

72

Figure. 2.10: Proton NMR spectrum of the isolated digalactosyl diacylglycerol.

73

 

 

 

—H9=H3— '14:)—

—— o—‘Ho—

 

 

 

 

DDT“

74
make an estimate of the proportion of DGDG that was present from the relative amounts of
galactose and glycerol (Figure 2.4). This comparison gave a result of 25-30% DGDG. This
is a minimum value. The actual value should be even higher since one molecule of PG

contains two molecules of glycerol.

75

DISCUSSION

A dramatic change in chemistry in the transition from vegetative cells to bacteroid
forms has been suggested in several chemical and immunochemical studies. In one study,
attempts to isolate LPS from the bacteroid forms of cells were unsuccessful, indications
being that only low molecular weight glycolipids were present (12). Several groups have
now defined antigenic changes in rhizobial cell surface chemistry concurrent with the
transformation to bacteroid forms (13-18). The finding that SQDG and DGDG become major
membrane components in bacteroid forms of Bradyrhizobium (and probably other rhizobia)
is very consistent with the apparently seamless interfacing of plant and bacterial physiology
during the Rhizobium / legume symbiosis. Important questions now arise as to how critical
to symbiosis is the ability of Rhizobium to synthesize these glycolipids. Questions also arise
as to the degree of coupling (common pathways) between the biosynthetic routes for these
molecules and other bacterial carbohydrate-containing molecules such as EPS and LPS. It
should be pointed out that, barring gene redundancy, mutants such as an R. meliloti exo B
mutant (19) and some Rhizobium LPS mutants (20-21), which appear to be impaired in
synthesizing galactose, should also be defective in synthesizing DGDG. Which then is the
critical deficiency, the impairment of polysaccharide synthesis or of DGDG synthesis?

DGDG and SQDG are characteristic plastid lipids. Plastids, according to
endosymbiotic theory, are originated from a symbiotic photosynthetic bacterium. Rhizobium

has similar membrane chemistry to plant organelles and might actually be a (usually

76

extracellular) nitrogen-fixing counterpart to the carbon dioxide fixing chloroplast. The lipid
composition of bacteroids might be a necessary prerequisite for reaching or maintaining a
symbiotic state without triggering plant defense responses. These findings might also explain
why some rhizobial symbiotic mutants which are impaired in LPS or EPS biosynthesis (and
then maybe also DGDG or SQDG) are incapable of being released into nodules to become
bacteroids.

The results of this study are also directly relevant to certain evolutionary questions.
There has been much debate on whether horizontal (lateral) gene transfer occurs between
distinctly different organisms (22). Rhizobium is a case in point since it was discovered that
the sequence of one of the glutamine synthase genes of Bradyrhizabium japonicum was
consistent with it being transferred directly from the plant to the bacterium at some stage
during their co-development (23). This possibility has been discussed from several
perspectives but the case for gene acquisition by horizontal gene transfer, is still quite strong
(16). The ability of Rhizobium to somehow synthesize the plant lipids described here would
seem to represent another very strong case for invoking horizontal gene transfer. This should
not be too surprising considering how ecologically and developmentally close the two
organisms in question are.

There is one other hypothesis that can be advanced to explain the presence of DGDG
in bacteroid membranes. One can invoke parallels with well studied host - pathogen systems
in which there is direct transfer of lipids from the host to the pathogen or parasite. This
phenomenon has been demonstrated in several host-pathogen (parasite) systems such as rat -

Hymenolepis diminuta (24), human-Plasmadium falciparum (25), human-Schistosama

77
mansani (26) and snake-Spirametra erinacei (27). In the case of H. diminuta, the abundant
lipids of the adult parasite contain predominantly unsaturated fatty acids which were obtained
from the host intestinal cells. In this case, the parasite has, apparently, lost its capacity for
synthesizing these unsaturated fatty acids, transferring these lipids from its host is essential
for the biochemical adaptations of the adult parasite to its anaerobic existence in the rat
intestine. In the human-Plasmadium falciparum interaction (malarial infections), the
invading pathogen is incapable of synthesizing its own fatty acids and cholesterol and relies
on the host for these. This requires that phospholipase activities in the cell of the invading
organism be capable of degrading the host cell erythrocyte membrane making the lipid
components available. Another human parasite, S. erinacei, must undergo both
morphological and biochemical changes to adapt to the host environment (c.f. bacteroid
formation in Rhizobium). Developmental stage- dependent utilization (uptake, processing
and incorporating) of host lipids by these parasites have been indicated to play a role in their
evasion of host immunity and to be a requirement for their survival in the new environment.
In this scenario the only troubling aspect is that DGDG is not found in the plasma membrane
of plant cells (of which the peribacteroid membrane is an extension) but is thought to be
restricted to plastids. In view of the large proportion of DGDG present, (over 25% of total
lipids) direct lipid transfer from the plant peribacteroid membrane to the bacteria is not very
likely but cannot be ruled out. The acquisition of glycosyl transferases by the bacterium from

the plant is unlikely since they would have to be transported into the bacterial cell to the

inner membrane surfaces where the activated sugars and lipid carrier are.

78

ACKNOWLEDGEMENTS

We would like to thank Dr. Barb Sears for her advice and her technician, Gabriele
Schewe, for technical assistance with the Southern hybridization experiments. 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-S10-RR04750, NSF grant #CHE-88-00770, and NSF grant #CHE-92-
13241. This work was supported by grant #DE-FG02-89ER 14029 from the US.

Department of Energy to RIH.

8.

9.

79

BIBLIOGRAPHY

. Cedergren, R. A., and Hollingsworth R. I. 1994. J. Lipid Res. 35, 1452-1461.
. Kate, M. 1990. In Handbook of lipid research. Vol. 6. (Morris Kates, ed). Plenum Press.
. Goldfine, H., and Ellis, ME. 1964. J. Bacteriol. 87, 8-15.

. Dazzo, F. B. 1982. In Experimental Microbial Ecology. (R. G. Burns and J. H. Slater Eds).

pp 431-446 Blackwell Scientific Publications, Oxford.

. Sambrook, J ., Fritch, E. F. and Maniatis, T. 1989. In Molecular Cloning, A Laboratory

Manual 9.32 - 9.57.

. Abe, M., Sherwood, J. E., Hollingsworth, R. I. and Dazzo, EB. 1984. J. Bacterial. 160,

517 - 520.

. Pazur, J. H. 1986. In Carbohydrate Analysis: A practical Approach. (M. F. Chaplin, and

J. F. Kennedy eds). pp 67-70.
Brussel, A. V. van, Planque, K. and Quispel, A. 1977. J. General Microbial. 101, 51-56.

Ching, T. M., Hedtke, S. and Newcomb, W. 1977. Plant Physiol. 60, 771-774.

10. Bunn, C. R. and Elkan, G. H. 1970. Can. J. Microbiol. 17, 291-295.

11. Miller, K. J ., Shon, B. C., Gore, R. S. and Hunt, W. P. 1990. Current Microbial. 21, 205-

210.

12. Bolobova, A. V. and Andreev, L. V. 1985. Biakhimiya, 50, 1749-1756.

13. Bhat, U. R. and Carlson, R. W. 1992. J. Bacterial. 174, 2230-2235.

14. Kannenberg, E. L. and Brewin, N. J. 1989. J. Bacterial. 171, 4543-4548.

15. Sindhu, S. S. , Brewin, N. J. and Kannenberg. E. L. 1990. J. Bacteriol. 172, 1804-1813.

l6.Tao, H., Brewin, N. J. and Noel, K. D. 1992. J. Bacterial. 174, 2222-2229.

80

17. VandenBosch, K. A., Brewin, N. J. and Kannenberg, E. L. 1989. J. Bacteriol. 171,
4537-4542.

18.Wood, E. A., Butcher, G. W. N., Brewin, J. and Kannenberg, E. L. 1989. J. Bacterial.
171, 4549-4555.

19. Long, S., Reed, J. W., Himawan, J. and Walker, G. C. 1988. 170, 4239-4248.
20. Cava, J. R., Elias, P. M., Turowski, DA. and Noel, K. D. 1989. J. Bacterial. 171, 8-15.

21. Zhang, Y., Hollingsworth, R. 1., and Priefer, U. B. 1992. Carbohydrate Res. 231, 261-
271.

22. Amabile-cuevas, C. F. and Chicurel, M. E. 1993. American Scientist 81, 332-341.

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

24. Jacobsen, N. S. and Fairbaim, D. 1967. 53, 355-361.

25. Zidovetzki, R., Sherman, I. W., Prudhomme, J. and Crawford, J. 1994. Parasitology.
108, 249-255.

26. Furlong, S. T., Thibault. K. S. Morbelli, L. M., Quinn, L. L. and Rogers, R. A. 1995.
J. Lipid Res. 36, 1-12.

27. Fukushima, T., Abe, K. Nakagawa, A. and Yamane, Y. 1988. Int. J. Parasial. 18, 27-31.

CHAPTER 3
THE LIPID CHEMISTRY OF FREE-LIVING AND
BACTEROID-LIKE FORMS OF RHIZOBIA

81

82

ABSTRACT

The membrane chemistry of vegetative and artificially-induced bacteroid forms of
Bradyrhizabium japanicum was studied by NMR spectroscopy and mass spectrometry.
Culturing Bradyrhizobium japanicum in the presence of glucose as a carbon source in a rich
medium known to stimulate differentiation into bacteroid forms led to the synthesis of a
family of di- and trisaccharide-containing glycerolipids as the main membrane lipid
components. This high level of glycolipid synthesis in induced bacteroid-like forms of free-
living Bradyrhizabium is also of interest since glycolipids are the predominant membrane
lipids in plants. These results underline the close developmental tie between these bacteria
and their plant host and suggest a real possibility of gene transfer or common gene usage

between these plants and their bacterial symbiont at some stage of their co-development.

83

INTRODUCTION

Establishment of an effective symbiosis between Rhizobium and legume plants is
critical to the available nitrogen balance for living systems. It is a process of cellular
interaction leading to compatibility and shared resources between these organisms. The
development of a new plant organ (nodule) and maturation of bacteria to bacteroids inside
nodules seem to be regulated at all stages by a diversity of bacterial factors. These include
bacterial surface glycoconjugates (including integral membrane components such as LPS and
glycolipids), EPS and nod factors (Chapter 1).

The cell surface of Rhizobium is a dynamic entity whose chemistry and antigenicity
is dependent on environmental factors such as oxygen tension and pH. This has been
demonstrated by both chemical and immunochemical analysis (1,2). The environmental
conditions inside root nodules are very different from those experienced by free-living
bacteria (3,4). The process of differentiation from bacteria to nitrogen-fixing bacteroids is
associated with many morphological, biochemical and presumably, cell surface
compositional changes. Bacteroids are more sensitive to detergents and lysozymes, have a
smaller amount and/or a changed composition of LPS (5, 6-8) and usually have lower levels
of EPS (9). Some of the changes, in LPS and EPS, for example, can be simulated by
growing free-living bacteria under appropriate culture conditions, such as low oxygen , low
pH, a special carbon source, or combinations of these conditions (Chapter 1). The rationale

for such experiments is that these conditions are very similar to those experienced by

84

bacteroids inside the nodule although the latter environment is sure to be much more
complicated.

Results from studies described in Chapter 2 showed that true bacteroids of rhizobia,
isolated from plant root nodules, contained DGDG as the predominant membrane lipids. This
raised an important question about the common links between membrane chemistry of
Rhizobium in its infective or its bacteroid stages and that of its plant host. Since bacteroids
are found intracellularly in the plant and since these two symbiotic organisms share
ecological pressures such as oxygen availability, pH, temperature, water and nutrient
availability, it is not surprising that they share some common membrane and surface features
in order for the bacteria to survive the host defense mechanisms. Characterizing the cell
surface and membrane chemistry of bacteroids will enable us to obtain a better understanding
of what biosynthetic functions are necessary for bacteroid transformation and to make a big
step forward in understanding this important and critical cellular differentiation event.
Bradyrhizabium japanicum is a good system to study because we have investigated the
membrane lipid chemistry of the true bacteroids isolated from nodules and Bradyrhizabium
japanicum can be transformed to bacteroid-like forms by changing environmental conditions
(10,11).

At least three previous studies (12-14) have been conducted on the membrane lipid
composition of vegetative, free-living Bradyrhizabium japanicum. Although there was
agreement on the presence of PC and PE, there were still some conflicts with respect to the
presence of PG, PI and PS. No glycolipids were reported in these studies. Our studies

described in Chapter 2 confirmed the existence of PG, PE and PC. Since SQDG has been

85

found in the membrane of free-living rhizobia and DGDG has been found to be predominant
in the membrane of symbiotic bacteroids, we propose that the proportion of membrane
glycolipids in bacteroids is much higher than that in vegetative cells. This makes the
membrane chemistry of bacteroids a mimicry of that of the host plant and may be necessary
for the bacteroid’s transformation and survival. In the experiments described here, we will
address this hypothesis by studying the membrane lipid chemistry of induced, bacteroid-like

forms of rhizobia.

86

MATERIALS AND METHODS

Bacterial Culture

B. japanicum USDA 110 was grown at 30°C in modified Bergensen's medium as
previously described (15). Type I nutrient enrichment medium was created by supplementing
the normal medium with a mixture of 0.1% Casamino Acids, 0.45% sodium succinate, 0.1%
yeast extract and 0.4% D-glucose during exponential growth as described earlier (10,1 1) .
Type II enriched medium was created by supplementing the normal medium with all the
nutrients when starting the culture. All cultures were harvested by centrifugation at late
exponential phase of growth. The pellets were collected and used for membrane lipid
extraction.
Lipid Extraction

Lipids were extracted from cells by stirring the pellets with a mixture of chloroform,
methanol, butanol and water in the ratio of 2: 1 :1 :4 at room temperature for 24 h. The cell
debris was removed by centrifugation and the supernatant was partitioned into two layers.
The lower organic layer was recovered and the cell debris was re-extracted with a mixture
of chloroform, methanol and butanol in the ration of 2: 1: l. The organic layer was again
recovered and combined with the one from the first extraction. The combined organic layers
were concentrated to dryness on a rotary evaporator at a bath temperature of ~40°C and then

dissolved in 1 mL of a 1:1 mixture of chloroform and methanol.

87

Thin Layer Chromatography

Standard PC and PE were purchased from SIGMA Chemical Co. (St. Louis, MO).
The extracted lipids were analyzed by TLC on silica plates using a mixture of chloroform,
acetone, methanol, acetic acid and water in the ratio of 10:4:2: 1 : 1. After development, the
plates were sprayed with orcinol and heated at 120°C. For the detection of glycolipids, the
orcinol-treated plates were heated for only 3-4 min over which time only the spots
corresponding to glycolipids were visible. These appeared as bright purple against a white
background. Prolonged heating for 15 min led to the development of brown spots for the
phospholipids. Components that moved too slowly on straight-phase chromatography to be
eluted efficiently in preparative separations were isolated by preparative C-18 reverse-phase
chromatography using HZO/methanol (4:1) as an eluent.
Nuclear Magnetic Resonance Spectroscopy

NMR analyses were performed on a Varian VXR-SOO spectrometer (500 MHZ for
protons) at 25°C in perdeuterated methanol. Chemical shifts are referenced relative to the
residual methanol line at 3.30 ppm. Double quantum filtered J-correlated spectroscopy
(DQF—COSY) spectrum was recorded using a spectral width of 4748.9. Generally, 32 scans
were accumulated in the f2 dimension and a total of 512 spectra transformed in the f 1
dimension.
Electrospray Mass Spectrometry

ES-MS spectra was recorded on a Fisons Platform mass spectrometer. The spectra
were recorded in the negative ion mode using a capillary voltage of 3 kv. Samples were

infused into the instrument in methanol.

88

Chemical Characterization

The glycosyl components of the lipids were identified by GC after converting them
to alditol acetate derivatives. Samples (300 pg) were hydrolyzed with 2M TFA at 100°C for
2 h. After removing the free fatty acids by extraction with chloroform, the aqueous acid was
removed under a stream of nitrogen and the free glycosyl components were reduced with
sodium borohydride. The alditols thus formed were peracetylated with 0.1 mL of a 1:1
mixture of acetic anhydride and pyridine at room temperature for 24 h. The acetylating
reagents were removed under a stream of nitrogen and the residue partitioned between 1 mL
of water and 1 mL of chloroform. The lower chloroform layer was recovered and subjected
to GC and GC/MS analysis.
GC and GC/MS

Gas chromatography was performed on a Hewlett Packard 5890 gas chromatograph
using helium as a carrier and a flame ionization detector. For the separation of alditol
acetate derivatives, gas chromatography was performed on a J and W Scientific DB-225
column. The initial temperature was 180°C, the temperature was raised at the rate of
2°C/min to a final temperature of 220°C. The column was held at 220°C for 50 min.
GC/MS was performed on a J EOL 505 mass spectrometer using an accelerating voltage of

70 eV.

89

RESULTS

In our studies, we used bacteroid-inducing media as demonstrated by others (10, l 1).
Electron nricrographs of Bradyrhizabium japanicum USDA 110 grown in regular B-III
media (Figure 3.1A) and in bacteroid-inducing media (Figure 3. 18) showed that under the
same magnification of the microscope, bacteroid-like forms were longer than those of the
control. Pleomorphism was also observed in bacteroid-like forms. These are the
characteristic morphological changes the bacteria undergo in the bacteroid-inducing media
(10,11).

Culturing B. japanicum USDA l 10 in bacteroid-inducing media had a dramatic effect
on the spectrum of membrane lipids synthesized. Figure 3.2 is a TLC chromatogram of the
total lipids from the various bacterial cultures. In panel A are the lipids from the type I and
II bacteroid inducing media. Spots with the characteristic purple color of glycolipids were
visible after heating the plate for 30 seconds after spraying with orcinol. Bacteria grown in
both type I and type H bacteroid-inducing media synthesized the same two new components.
Lanes 1 and 2 of panel B had the same material as in panel A but the plate was heated for a
longer period so that all of the lipid components (including phospholipids) were detected.
Lane 3 was a lipid preparation from vegetative cells. Lanes 4-5 are phospholipid standards.
The spots indicated by arrows were due to glycolipids. The lower one was barely visible
when the plate was heated for 30 seconds (panel A). These glycolipids are not synthesized

in un-induced cells but are major lipids in the bacteroid forms.

9O

 

Figure 3.1A: Bradyrhizabiumjaponicum USDA 110 grown in regular BIII medium. The

sample was negatively stained with 1.5% potassium phosphotungstate. Bar=1.8 um.

91

 

Figure 3.13: Braabzrhizobiumjaponicum USDA 110 grown in bacteroid-inducing medium.

The sample was negatively stained with 2% ammonium molybdate. Bar=1.8 um.

92

Figure 3.2: Thin layer chromatograms of lipid extract from cell grown under nutrient
enrichment conditions. In panel A the plate was heated for 2 min to reveal only glycolipids.
In panel B the plate was heated for 20 min. Panel A lanes 1 and 2 shows the membrane
lipids extracts from Bradyrhizabium japanicum USDA 110 grown in type I and II enriched
media respectively. The first two lanes in panel B correspond to panel A. Arrows indicate
the glycolipids. The next lane is a membrane preparation from un-induced cells and the last

two lanes are standards.

93

On.

Mn. EEOC

__ BE

_ 8%
seesaw 888m

__ m5
889m sauna

.8:

 

 

 

 

 

 

 

.. .
o
n
. A
n. . c
o A
p
u
a v
F
a
.
g

0 .
..
.
at...
..

Q
.o
.-
. o.
to
.. o
s r

 

 

 

 

 

' .

 

111]

 

 

94

Analysis of the glycosyl components of the total crude lipids extracts from bacteria grown
in type I and type H media using alditol acetate derivatives showed that glucose was a major
polar component of the total lipids (Fig. 3.3). There was also a peak due to the alditol
acetate of ribose indicating the presence of a ribosylated glycolipid. The faster-moving
glycolipid components in lanes 1 and 2 of panel A were isolated by preparative thin layer
chromatography on silica. Gas chromatography and GC/MS analysis of alditol acetate
derivatives revealed that glucose and glycerol were the only polar components present.
Proton NMR spectroscopy analysis (Fig. 3.4) indicated that the component was the same for
both lipid preparations. Two signals (doublets at 4.30 and 4.38 ppm with 8 Hz coupling
constants) were readily assignable to B-anomeric protons. Also readily assignable were
signals due to a glyceryl moiety. These were a multiplet at 5.28 ppm due to H2 and a doublet
of doublet at 4.43 ppm, with 14 Hz and 3 Hz coupling, due to one of the two H1 glyceryl
protons. The other H1 glyceryl proton appeared at 4.20 ppm overlapping with some other
signals but its connectivity could be traced through the DQF-COSY spectra (Fig. 3.5). The
other signals could also be readily assigned from the IH-NMR DQF-COSY spectra. The
presence of two signals for B-anomeric protons in the NMR spectrum and glucose and
glycerol as the sole polar components of this lipid (by GC-MS analysis) indicated that it was
a diglucosyl diacyl glycerol. This was readily confirmed by negative ion electrospray mass
spectrometry (Fig. 3.6). In the high mass region of the spectrum, a cluster of ions with
individual signals at m/z 887, 901, 915 and 929 corresponding to M-H parent species was
detected. These differed by 14 mass units indicating that they are a homologous series of ions

and varied only in the lengths of the fatty acyl groups in the different species. The ions

95

Figure 3.3: Gas chromatography spectrum of alditol acetate derivatives of the total lipid
extract from Bradyrhizabium japanicum USDA 110 grown under nutrients enriched

conditions. Peak A, Glycerol; Peak B, Ribose; Peak C, Glucose.

Response

 

96

 

 

 

 

 

Retention Time

97

Figure 3.4: Proton NMR spectroscopy analysis of the fast moving component in Panel A
of Figure 3.1 indicated that two signals (doublets with 8 Hz coupling constants at 4.30 and
4.38 ppm) were readily assignable to B-anomeric protons. The multiplet at 5.28 ppm was due
to H2 and a doublet of doublet at 4.43 ppm, with coupling constants of 14 Hz and 3 Hz, was

due to one of the two H1 glyceryl protons.

98

_L

or"

hLL FL

 

 

1P

-

 

 

 

b

n.n

FF—

 

P

b

 

F

rid
» p

P

P

h

b

0.0

 

99

Figure 3.5: The connectivity of H1 glyceryl proton appeared at 4.20 ppm overlapping with
some other signals could be traced through the DQF—COSY spectra. Some other signals

could also be assigned from this spectra.

100

and. E

o.N m.N o.m

bbPhphhbb-y—Phpb—hb»P—bbrhwbbe—r~

o.v
- b Ln p

m.v

p tr — _ u r- h

 

all

-.Io.’.!‘..l.‘ O. 00.06-8500 .-

 

 

i.. OO.IIIIID)TI..' .I III. .

Ll» I'I I ill DIDl
‘1! ‘I .‘ J I

.I-’.O

   

30.0..

.3
.8

I

at:

 

 

 

be 3

E»

101

Figure 3.6: Negative ion ES-MS confirmed the fast moving component in Panel A of Figure
l was diglucosyl diacyl glycerol. In the high mass region of the spectrum, a cluster of ions
with individual signals at m/z 887, 901, 915 and 929 corresponding to M-H parent species
was detected. The ions corresponded to fatty acid chains ranging from 14 to 18 carbons with
a maximum of two double bond equivalents due to unsaturation or cyclopropyl groups. In
the middle mass region, another cluster of ions with peaks at m/z 680, 694, 708 and 722
were due to loss of acyloxy groups from the parent species. Low intensity fragment ions at
m/z 721 and 749, corresponding to loss of one glucose residue from the parent species at
m/z 901 and 929 respectively. Most of the ions appeared as doublets. Signals corresponding
to fatty acyloxy ions were prominent in the low mass region of the spectrum. These appeared
at m/z 227 (14:0), 241 (15:0), 255 (16:0), 253 (16: 1), 251 (16:2 or 16:1 cyclopropyl), 265
(17:2 or 17:1 cyclopropyl), 279 (18:2 or 17:1 cyclopropyl), 281 (18:1) 279 (18:2) and 293

(1821 cyclopropyl).

102

Qua—en— s: 03— as can 2:. as com 3% _ can a.“

. /
\ . Se 8.. Lr :8
a: 2... we. as

 

L2!

61.!
$92

697
, m

 

€62

1.x.

 

992

 

 

L99

emu: reo—

103

corresponded to fatty acid chains ranging from 14 to 18 carbons with a maximum of two
double bond equivalents due to unsaturation or cyclopropyl groups. In the middle mass
region, another cluster of ions with peaks at m/z 680, 694, 708 and 722 was also detected.
These were due to loss of acyloxy groups from the parent species. Low intensity fragment
ions at m/z 721 and 749, corresponding to loss of one glucose residue from the parent
species at m/z 901 and 929 respectively, were also observed. Most of the ions appeared as
doublets since both saturated and unsaturated forms of most fatty acids were present. Signals
corresponding to fatty acyloxy ions were prominent in the low mass region of the spectrum.
These appeared at m/z 227 (14:0), 241 (15:0), 255 (16:0), 253 (16:1), 251 (16:2 or 16:1
cyclopropyl), 265 (17:2 or 17:1 cyclopropyl), 279 (18:2 or 17:1 cyclopropyl), 281 (18: 1) 279
(18:2) and 293 (18:1 cyclopropyl). These were consistent with the fatty acid analysis. The
linkage between the two glycosyl units was determined to be 1,6 from the characteristic
fragments at m/z 251 and 281. These are due to the 03A2 and the cleavage and 0'2A2 cleavages
respectively (16). The positive mode MS (data not shown) also gave confirmation of the
molecular weight.

Both GC-MS analysis and the electrospray mass spectrum (Fig. 3.7) of the slower
moving component (isolated by C-18 reverse phase chromatography and by far the minor
component) indicated that it was a ribosylated version of the diglucosyl diacyl glycerol. In
the high mass region of this glycolipid clusters of ions at 993, 1007, 1021 and 1035 were
detected. This sequence corresponded to the diglucosyl diacyl glycerol series plus a ribosyl
residue. In this series, the typical number of double bond equivalents in the fatty acid chains

was 1. The ribosyl residue was attached to the diglucosyl core. This was apparent since,

104

except for the parent ions, the other fragment ions corresponded to those from the diglucosyl
compound indicating that the 5-carbon sugar was the first to be lost leaving the diglucosyl
compound which then further fragments by the known route. For instance, in the middle
mass region, the same sequence of ions at m/z 680, 694, 708 and 722 seen in the
fragmentation of the diglucosyl compound was also detected. The quantities of this
triglycosyl diacylglycerol that were obtained precluded a thorough analysis of the linkage

position. The structures of the two glycolipid components are shown as _1_ and 2 in Figure 3.8.

105

Figure 3.7: Negative ion EI-MS of the slower moving component in Panel A of Figure 3.1
confirmed GC-MS analysis that it was a ribosylated version of the diglucosyl diacylglycerol.
In the high mass region, a cluster of ions at 993, 1007, 1021 and 1035 were detected,
corresponding to the diglucosyl diacylglycerol series plus a ribosyl residue. In the middle
mass region, the same sequence of ions at m/z 680, 694, 708 and 722 seen in the

fragmentation of the diglucosyl compound was also detected.

 

 

106

 

7.3:: .234. .Su:;:.8__.~: _ $.92: :53..._:.8m.:_:.8+:;:.3»: Ifi
<<i 1 II<

LII». p »— p».—> rub—b» .-
N}: <1

:8
«Nb

 

6mg 1x. I

 

vs

 

 

 

r2:

 

mm:—

107

Figure 3.8: The structures of di- and trisaccharide-containing lipids.

108

OH

HO

 

109

DISCUSSION

The bacteroid models we use here resembled true bacteroids found in nodules since
they share certain undefined carbohydrate cell surface antigens with the latter (4). They have
similar morphology as true bacteroids and are even capable of fixing nitrogen in their free-
living state. The switch by Bradyrhizabium japanicum to synthesizing glycolipids during
bacteroid differentiation is quite noteworthy especially since the disaccharide is the gluco-
analog of the common galactose-containing plant glycolipid DGDG and these glycolipids
are rarely found outside the plant kingdom (Chapter 1)

The results become even more interesting when considering the finding (described
in Chapter 2) that DGDG has also been found to be the predominant membrane lipid of true
bacteroids isolated from nodules. It seems that the synthesis of these plant glycolipids are
critical to the formation of bacteroids and ex-planta bacteroid-like forms of rhizobia. Why
the bacteroid-like forms synthesize gluca- containing glycolipids instead of galactose-
containing ones as synthesized in true bacteroids? One possible explanation was that the
glycosyl transferases typically do not have absolute specificity. If UDP-galactose were
provided (say from a plant) it is quite possible that this organism could synthesize the
galacta-lipid instead.

Questions now arise as to whether it is a coincidence that Bradyrhizabiumjapanicum
synthesizes large amounts of plant-like lipids in bacteroid forms. Several groups have now

defined antigenic changes in rhizobial cell surface chemistry concurrent with transformation

110

to bacteroid forms (1-4, 8). There is every indication of shared epitopes between bacteroids
and plant cell surfaces. This notion and the results of this study are consistent with the
apparently seamless interfacing of plant and bacterial physiology during the Rhizobium/

legume symbiosis.

lll

ACKNOWLEDGMENT

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-S10-RR04750, NSF grant #CHE-88-00770, and NSF grant #CHE-92-
13241. This work was supported by grant #DE-FG02-89ER 14029 from the US.

Department of Energy to RIH.

112

BIBLIOGRAPHY

l. Kannenberg, E. L. and N. J. Brewin. 1989. J. Bacterial. 171, 4543-4548.

2 Tao, H., N. J. Brewin, and K. D. Noel. 1992. J. Bacterial. 174, 2222-2229.

3. Kannenberg, E. L., S. Perotto, V. Bianciotto, E. A. Rathbun, and N. Brewin. 1994. J.
Bacteriol. 176, 2021 -2032.

4. Sindhu, S. S. , N. J. Brewin, and E. L. Kannenberg. 1990. J. Bacteriol. 172, 1804-1813.

5. Lopez-lara, I. M., G. Orgarnbide, F. B. Dazzo, J. Olivares, and N. Toro. 1993. J.Bacteriol.
175, 2826-2823.

6. Brewin, N. J., E. A. Wood, A. P. Larkins, G. Galfre, and G. W. Butcher, 1986. J. General
Microbial. 132, 1959-1968.

7. Brussel, A. V., van. K. Planque and A. Quispel. 1977. J. General Microbial. 101, 51-56.

8. de Maagd, R. A., A. S. Rao, 1. H. M. Mulders, and B. J. J. Lugtenberg. 1989. J. Bacterial.
171, 1136-1142.

9. Tully, R. E. and M. E. Terry. 1985. Plant Physiol. 79, 445-450.

10. Urban, J. E. 1979. Appl. Environ. Microbial. 38, 1173-1178.

11. Urban, J. E. 1988. Can. J. Microbial. 34, 910-913.

12. Bunn, C. R. and G. H. Elkan. 1970. Can. J. Microbial. 17, 291-295.

13. Gerson, T. and J. J. Patel. 1975. Appl. Microbiol. 30, 193-198.

14. Miller, K. J., B.C. Shon, and W. P. Hunt. 1990. Current Microbial. 21, 205-210.

15. Dazzo, F. B. 1982. In Experimental Microbial Ecology. (R. G. Burns and J. H. Slater,
eds.) pp 431-446. Blackwell Scientific Publications, Oxford.

16. Orgarnbide, G. G., R. I. Hollingsworth , and F. B. Dazzo. 1992. Carbohydr. Res. 233,
151-159.

CHAPTER 4
THE LIPID CHEMISTRY OF RHIZOBIA
UNDER LOW OXYGEN AND LOW pH CONDITIONS

113

114

ABSTRACT

Lowering oxygen tension in free-living Bradyrhizabium japanicum resulted in a
dramatic switch in membrane chemistry in which PC, the predominant lipid in aerated
cultures, was no longer synthesized and PE became the major lipid. Besides this change, PI,
a typical plant lipid rarely found in bacteria, was also synthesized. Lowering pH in free-living
Bradyrhizabium japanicum did not result in any change in the lipid headgroups but resulted

in a change of the unsaturation state of the fatty acyl groups of the lipids.

115

INTRODUCTION

Bacteroids live in a microoxic environment. The oxygen diffusion barrier of nodules,
in conjunction with the plant leghemoglobin and active bacteroid respiration, creates a low
oxygen concentration zone in the infected zone of the nodule. Compared with dissolved

oxygen concentration at 250 um outside the nodule, the oxygen concentration inside nodules

is below 30 nm. This microoxic environment is critical for the operations of nitrogen
reducing enzyme nitrogenase (1). Besides, the bacteroids live in an acidic environment
because of the acidity of peribacteroid fluid (Chapter 1).

Like any other environmental factor, oxygen tension and pH regulate membrane and
cell surface chemistry. Both chemical and immunochemical analyses support this (2, 3). In
Chapter 2 and Chapter 3, I described the effects of two environments, one inside root nodules
and the other bacteroid-inducing medium, on membrane lipid chemistry of rhizobium.
Experiments described here were conducted to study the effects of oxygen tension and pH
on glycolipid biosynthesis in Rhizobium. These are factors a bacterium experiences once it
enters the plant and are critical to the adaptability and survival of the bacterium once it is
inside of the plant. Low oxygen concentrations and low pH in the media are known to trigger
bacteroid differentiation in free-living cultures ( 2, 3 and Chapter 1). Bacteroids are the

nitrogen-fixing form of the bacterium when it resides inside of the plant cell.

116

MATERIALS AND METHODS

Bacterial Culture

Rhizobium japanicum USDA 1 10 was grown at 30°C in Modified Bergensen's
medium as previously described ( 4 ). Low oxygen conditions were obtained by stoppering
the four liter Erlenmeyer flasks containing two liter culture media with butyl rubber stoppers
at the time of inoculation to restrict oxygen exchange with the atmosphere. The concentration
at the time of harvesting the cells was typically between 500 and 1000 ppm as measured by
an O2 electrode. The cultures were harvested by centrifugation at late exponential phase of
growth. The pellets were collected and used for membrane lipid extraction. .
Lipid Extraction

Lipids were extracted from cells by stirring the pellets with 160 mL of a mixture of
chloroform, methanol, 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 was partitioned
into two layers. The lower organic layer was recovered and the cell debris was re-extracted
with 240 mL of a mixture of chloroform, methanol and butanol in the ratio of 2: 1:1. The
organic layer was again recovered and combined with the one from the first extraction. The
combined organic layers were concentrated to dryness on a rotary evaporator at a bath
temperature of 40°C and then dissolved in 1mL of 1:1 mixture of chloroform and methanol.
Thin Layer Chromatography

Standard PC and PE were purchased from SIGMA Chemical Co. (St Louis,

117

Missouri). Standard PI and PG were purchased from AVANTI Polar Lipids (Alabaster,
Alabama). The extracted lipids were analyzed by TLC on silica plates using a mixture of
chloroform, acetone, methanol, acetic acid and water in the ratio of 10:4:2: 1: 1. After
development, the plates were sprayed with orcinol or phosphomolybdic acid reagent and
heated at 100°C for 15 minutes.
Nuclear Magnetic Resonance Spectroscopy

Proton NMR analyses were performed on a Varian VXR-SOO spectrometer (500 MHz
for protons) at 25°C in perdeuterated methanol. Chemical shifts are referenced relative to the
residual methanol line at 3.30 ppm.
Chemical Characterization

The glycosyl components of the lipids were identified by GC after converting them
to alditol acetate derivatives. Samples (300;: g) were hydrolyzed with 2M TFA at 100°C for
two hours. After removing the free fatty acids by extraction with chloroform, the aqueous
acid was removed under a stream of nitrogen and the free glycosyl components were reduced
with sodium borohydride. The alditols thus formed were peracetylated with 0.1 mL of a 1:1
mixture of acetic anhydride and pyridine at room temperature for 24 h. The acetylating
reagents were removed under a stream of nitrogen and the residue partitioned between 1 mL
of water and 1 mL of chloroform. The lower chloroform layer was recovered and subjected
to GC analysis.
GC and GC/MS

For the separation of alditol acetate derivatives, gas chromatography was performed

on a J and W Scientific DB-225 column. The initial temperature was 180°C, the temperature

118

was raised at the rate of 2° C/min to a final temperature of 220 °C. The column was held at
220°C for 50 min. GC/MS was performed on a JEOL 505 mass spectrometer interfaced with
a Hewlett Packard 5890 gas chromatograph. Spectra were obtained in the positive mode by

using electron impact ionization at an acceleration voltage of 70 eV.

119

RESULTS

Fig. 4.1 was the TLC analyses of the lipids from Bradyrhizabium japanicum USDA
1 10 grown under low oxygen (microanaerobic) conditions and grown under normal aeration.
There were major compositional differences between the two. Under the low oxygen
conditions, there was a large increase in PG and PE and the amount of PC synthesized was
greatly reduced. The most striking difference between the TLC analysis of the two different
lipid extracts was the presence of a slow moving component in the extracts from the bacteria
grown under microanaerobic conditions. The retention time was very similar to that of the
PI standard.

Proton NMR spectroscopy of the total lipid extract from bacteria grown under normal
aeration was shown in Fig. 4.2A. It indicated that the major lipid species under normal
aeration was PC. This was clear from the presence of strong signals at 3.23 ppm (the N-
methyl groups). Also present in the spectrum were signals for PE and the other N-methyl
derivatives of PE. Signals for these other N -methyl groups appeared as sharp singlets
between 2.7 and 3.3 ppm. The proton NMR spectrum of the total lipid extract from the
bacteria grown under microanaerobic conditions was quite different (Fig. 4.2B). Changes
observed in the latter were: disappearance of signals corresponding to PC, a large decrease
in the peak intensity for the N-methyl groups of mono and dimethyl PE, a large increase in
peaks corresponding to PE at 3.18 and 4.07 ppm and appearance of new signals between 3.5

and 4.0 ppm. There were new signals between 3.5 to 4.0 ppm characteristic of a carbohydrate

120

Figure 4.1: Orcinol-sprayed TLC plate of membrane lipids extracts from the lipid extract
of USDA 110 grown both in normal and low oxygen conditions. Lane 1, PG; Lane 2, PC;
Lane 3, lipid extract of USDA 110 grown in normal conditions; Lane 4, lipid extract of B.
japanicum USDA 110 grown in low oxygen conditions. Lane 5, PI; Lane 6, PE. Note that
under the low oxygen conditions, PC was no longer synthesized, but there was a significant
increase in the amount of PE. The most striking difference between the two different lipid
extracts was the presence of slow moving component in the extracts from the bacteria grown
under low oxygen conditions. The retention time of this slow moving component was very

similar to that of an authentic standard PI.

121

 

 

 

 

 

 

 

 

 

 

 

PC

Control

Law Oxygen

PI

PE

 

122

Figure 4.2: Proton NMR spectrum of the extracted membrane lipids of free living B.
japanicum USDA 110 (A) and of the membrane lipids of free-living B. japanicum USDA
110 grown under low oxygen conditions (B). In A, signals between 2.8 and 4.8 ppm are due
to carbohydrates and glyceryl components. The signals between 0.5 and 2.4 ppm are due to
resonances in the fatty acids. The strong resonances at 3.30 and 4.86 ppm are from residual
CHDZOD and OH signals, respectively, of the solvent. In B, the signals at 3.23 ppm and 4.27
ppm which correspond to PC are gone. There is also a decrease in the peak intensity at 2.7
ppm and 3.3 ppm (other N -methyl groups) and an increase in peak intensity at 3.18 ppm and

4.07 ppm (PE). Note the appearance of new signals between 3.5 and 4.0 ppm.

123

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

W."j‘VYIVY—v—YIfjvvfiiY‘Y—‘ijv‘Y1IVV'11"‘VI'YfifiIVYY'IT‘V‘I

5.3 5.0 4.3 4.0 3.5 3.0 2.5 2.0 1.5 1.0 pp

124

residue; however, there was no signal that was assignable to an anomeric proton indicating
that an inositol group was present as indicated by the TLC analysis.

Inositol was conclusively identified by GC and GC-MS analysis of the lipid
headgroup after hydrolysis and peracetylation. Gas chromatography analysis showed that the
predominant glycosyl component of membrane lipid from bacteria grown under
microanaerobic conditions (Figure 4.33) had the same retention time as the peracetate of
standard inositol (Figure 4.3A). The mass spectrum of this component (Figure 4.4B) showed
the same fragmentation pathway as that of the standard PI (Figure 4.4A).

Compared with that of bacteria grown under normal conditions, the NMR spectrum
of the total lipid extract of Bradyrhizabium japanicum USDA 110 grown under low pH
conditions showed no change in the region below 6.0 ppm. This indicated that there were no
changes in the head groups of the lipids. Some changes in spectrum from 6.0 to 8.0 ppm
were observed for a total lipid extract of Bradyrhizabiumjaponicum USDA l 10 grown under
low pH conditions (Figure 4.58). Signals corresponding to the unsaturation state of the fatty
acyl groups in the lipids became stronger compared with those in the spectrum of bacteria

grown under normal conditions (Figure 4.5A).

125

Figure 4.3: Gas chromatogram of alditol acetate derivatives of lipid extract from B.
Japanicum USDA 110 grown under normal conditions (B) and under low oxygen conditions

(B). The predominant peak in B has the same retention time as that of standard myo-inositol

in A.

127

Figure 4.4: GC/MS spectrum of alditol acetate derivatives of standard myo-inositol (A) and
of lipid extract from USDA 110 (B). Note that the predominant component in GC (Figure

4.3B) has the same fragmentation pathway as standard inositol derivatives.

128

 

 

 

 

 

 

 

 

 

 

 

n.
.mn
all L
4
.m
mIL .4
.3 L
. l
3
a
0
3.1113 .
all.”
3 T. .m
#3
o E
Z
2
SI.
21L
I1
0
2 9 M
.
aI Z
1 f
.e i.
t W
H S
4
1‘ v
6 .
t V
rfl.m
9 1
n
38
7
m m w m m
A Relative abundance

 

 

 

 

 

 

 

 

 

 

 

 

I F I b h z
m m
T L
e
m j g...“
e If
I I1
I
a«JIIII 1
3 0
MIL m
e fiw
3
A
mllr.
z
:1 L
”.1,
a Z W
a o
mlwrié
m .. .
I 7
1
no-
5 H
I. 3
.. m
1
3
7
7
5.
U ‘ fl 1 1‘ o
M“ m m w a

Rcllesivc abundlncc

129

Figure 4.5: Proton NMR spectrum of the extracted membrane lipids of free-living B.
japanicum USDA 110 grown under normal conditions (A) and low pH conditions (B).

Changes were observed in signals corresponding to the unsaturation in the fatty acyl groups

of the lipids.

130

IE ed a.» o... u.» v... o... n... . o.-
~V-HP-nbppp—ppthE-hbphpppbrrPhpppbhEbpppphpppr—up.»-pthp-pprPLpp—PppprP-p—ppplb

II,

 

1 1 I

Jr 4

 

 

 

 

RE u.» «6 cg. «.1. v... w... u.» o.-
pb—uph--urn—pp-p-plrpn—ppb-ppprr—ppphprrph—p>-pup-hp—pF-hpbbhh—ppppbPrPhr—r-PIPCPPbb-n

1 i 141:

F D

 

 

131

DISCUSSION

At least three previous studies (5,6,7) have been conducted on the membrane lipid
composition of free-living Bradyrhizabium japanicum. Although there was much agreement
on the presence of PC and PE , there were still some conflicts with respect to the presence
or absence of PG, PI and PS. These discrepancies could, however, arise from differences
in growth medium (nutrients) and/or aeration (oxygen tension) and therefore could be
reflecting the sensitivity of the bacteria to these. In liquid and unsparged culture the extent
of aeration is determined by the geometry of the culture flask and the speed of shaking,
conditions that may not be controlled by laboratory routines. N -Methyl derivatives of PE and
PG are not easily resolved on TLC although they can be easily identified by NMR or MS.
This may explain why in the previous studies (5,6) PG or N-methyl derivatives of PE were
not detected.

It is of interest that Bradyrhizabium japanicium synthesize large amounts of inositol-
containing lipids in low-oxygen growth media because PI is rarely found in Gram negative
bacteria (8). It is one of the major phospholipids of plant membranes and may comprise up
to 19% of the total phospholipid of the plasma membrane (9). Its presence here is certainly
noteworthy since it further underlines the uniqueness of rhizobial membrane chemistry

among bacteria and its close similarity to plant membrane chemistry.

132

ACKNOWLEDGMENT

This work was supported by grant #DE-FG02-89ER 14029 from the US. Department
of Energy to RIH. 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-RR04750, NSF grant #CHE-88-00770, and NSF

grant #CHE-92-13241.

133

BIBLIOGRAPHY

l. Batut J. And P. Boistard. 1994. Antonie van leeuwenhaek. 66, 129-150.
2. Kannenberg, E. L. and N. J. Brewin. 1989. J. Bacterial. 171, 4543-4548.
3.Tao,H., N. J. Brewin, and K. D. Noel. 1992. J. Bacteriol. 174, 2222-2229.

4. Dazzo, F. B. 1982. In Experimental Microbial Ecology. (R. G. Burns and J. H. Slater eds.)
pp 431-446. Blackwell Scientific Publications, Oxford.

5. Bunn, C. R. and G. H. Elkan. 1970. Can. J. Microbial. 17, 291-295.
6. Gerson, T. and J. J. Patel. 1975. Appl. Microbial. 30, 193-198.

7. Miller, K. J ., B. C. Shon, R. S. Gore, and W. P. Hunt. 1990. Current Microbiol. 21, 205-
210.

8. Goldfine, H. 1982. In Current topics in membranes and transport, (Razin S, S. Rottem
eds.). Academic Press, New York. 17, 2-44.

9. Hetherington, A. M. and B. K. Drobak. 1992. Prag. Lipid Res. 21, 53-63.

CHAPTER 5
AN INVESTIGATION ON THE INTERRELATIONSHIP
OF RHIZOBIAL POLYSACCHARIDE AND
GLYCOLIPID BIOSYNTHESIS

134

135

ABSTRACT

There is functional and biosynthetic overlap among the various classes of rhizobial
cell surface components. Understanding carbohydrate metabolism is central to understanding
the connections among the biosynthetic events leading to the various polysaccharides,
glycolipids and glycoconjugates. 3-O-[”C]methyl-D-glucose was used as a probe to see

whether it can be taken up, intact, into carbohydrate-containing molecules. 3-O-methyl-D-
glucose is used extensively in glucose uptake studies. Our results clearly demonstrated that
3-O-[”C]-methyl-D-glucose was incorporated into membrane glycolipids but not into other
major carbohydrate-containing molecules, including EPS, LPS and cyclic B-glucans. It is
likely that the biosynthesis of EPS, LPS and cyclic B-glucans do not use a common pool of
monosaccharide with membrane glycolipids but rather utilizes locally generated pools.
Alternatively, the substrate specificity of the glycosyltransferase(s) synthesizing glycolipids
may be different from enzymes synthesizing EPSs, LPSs, and cyclic B-glucans. Biosynthesis

of glycolipids in this strain was shown to be independent of those synthesizing other

carbohydrate-containing molecules.

136

INTRODUCTION

In section III of Chapter 1 we introduced a variety of rhizobial cell surface

carbohydrate-containing molecules such as EPS, CPS, LPS, lipooligosaccharides, cyclic B—

glucans and glycolipids (1-6). In section V of Chapter 1 we discussed that all these molecules
are believed to function in the establishment of symbiosis between rhizobia and legumes (7-
20)

As described in section VI of Chapter 1, the diversity of bacterial cell surface
carbohydrates involved in symbiosis make it difficult for studies aiming to define the roles
of each individual component. The existence of a large degree of coupling of many
biosynthetic events leading to the synthesis of various surface components further
complicates these studies. Examples for such coupling go from the common structural (and
genetic) links between chitolipooligosaccharides, LPSs and general membrane lipids (21),
the transfer of the “missing” 0 antigen of LPS to an alternative glycolipid (22), to structural
links between EPSs and LPSs (12-25). In addition to this, there also exist overlaps in the
symbiotic functions of these molecules (26-29).

Understanding carbohydrate metabolism is central to understanding the connections
among the biosynthetic events leading to the various polysaccharides, glycolipids and
glycoconjugates. We are concerned with probing the biosynthesis of various molecules by
using a monosaccharide analog to see whether the analog can be taken up, intact, into a

carbohydrate-containing molecule. This can give us information about the common

137

intermediates among the biosynthetic pathways leading to different molecules.

In experiments described here, 3-O-[”C]-methyl-D-glucose was used a probe. 3-O-
methyl-D-glucose is used extensively in glucose uptake studies (30—32). It is generally known

to be taken up via the same facilitated diffusion uptake pathway as glucose (33). It has been
shown that 3-O-methy1-glucose can be phosphorylated both in yeast and in mammalian
systems (34-37). In our experiments, 3-O-methyl-D-glucose was chosen because it is a close
analogue of glucose, a component of almost all of the rhizobial cell surface carbohydrate—
containing molecules. The l"C label will be lost if demethylation occurs to give glucose.
Oxidative demethylation will yield labeled formaldehyde of forrnate (Figure 5.1A).
Elimination gives labeled methanol (Figure 5.1B). Both are C 1 units that do not enter into
normal metabolism. Hence, any radiolabeled polysaccharides or glycolipids would contain
3-O-methyl-glucose.

There are other aspects of biosynthesis that such experiments will allow us to test.
If, for example, the biosynthesis of a particular polysaccharide or glycolipid does not use a
common pool of the monosaccharide but utilizes instead a locally generated pool, then we
might expect the polysaccharide or glycolipid not to contain the label. This will address a
very important point. The biosynthesis of that particular glycolipid or polysaccharide will not
be affected by mutations affecting the common pool. We therefore can determine to what
extent the biosynthesis of certain carbohydrate-containing molecules utilizes common

pathways.

138

Figure 5.1: A, Oxidative demethylation of 3-O-methyl-D-glucose yields formaldehyde of

forrnate; B, Elimination of 3-O-methyl-D-g1ucose gives methanol.

139

OH
A o
HO

OH

3-O-methyl-glucose

\IOI
OH
O
H

OH
OH
O 4
HO
Hm OH

OH HCHO or HCOOH
Glucose
B

OH OH

0 o
014CH3 H H

HO HO __

H on 14CH3OH H/ on

3-O-methyl glucose
Enol

140

In a series of experiments described in this chapter, Rhizobium leguminosarum bv.
viciae ANU843 was grown in a medium with 3-O-['4C]-methyl-D-glucose as the only
carbohydrate component. We reasoned that if it were used as the sole carbohydrate
component in a culture medium, it would perhaps be activated for glycosyl transfer and
compete with glucose at several levels of biosynthesis of polysaccharides, oligosaccharides
and glycolipids. This will allow us to separate the carbohydrate-containing molecules that
are synthesized from the common pool of available glucose from those that need locally-

generated pools of de novo synthesized or activated glucose.

141

MATERIALS AND METHODS

Cell Culture

3-O-methyl-glucose was purchased from SIGMA Chemical CO. (St Louis, Missouri)
and 3-O-[”C]-methyl-D-glucose was purchased from Amersham International plc
(Buckinghamshire, England). Rhizobium leguminosarum bv. viciae ANU843 was grown at
30°C in modified Bergensen's medium with mannitol omitted and 3-O-methyl glucose

(10mM) added. 15 uCi 3-O-["’C]methyl-D-glucose was added to two liters of medium at

the time of inoculation. The bacteria grown on regular modified Bergensen's medium was
used as control. To cultures which were used for tracing S-containing molecules, 0.3 mCi
of 35S-sulfate was added. The cultures were harvested by centrifugation at late exponential
phase of growth. The supernatant was used for EPS extraction and the pellet was used for
the extraction of membrane lipids, LPSs, cyclic B-glucans and cell metabolites.
EPS Extraction

The supernatant from the bacterial culture was concentrated to one-forth of its
original volume on a rotary evaporator under reduced pressure at a bath temperature of ~
40°C. Equal volume of ethanol was added and the mixture was left at room temperature for
half an hour. EPS was recovered by centrifugation at 6,000g for 20 min and dialyzed against
distilled water. After five to six water change, EPS solutes were lyophilized to dryness. A

small amount was redissoved in water by sonication and subjected to scintillation counting.

142

Lipid Extraction

Lipids were extracted from cells by stirring the pellets of bacteria with 160 mL of
2: 1: 1:4 chloroform: methanol: butanol: water at room temperature for 24 h. The bacterial
cell debris was removed by centrifugation at 6,000g for 20 min at room temperature. The
supernatant was separated into two layers. The lower organic layer was recovered after
centrifugation was concentrated to dryness on a rotary evaporator under reduced pressure at
a bath temperature of ~ 40°C. The lipids were redissolved in 1 mL of 1:1 chloroform:
methanol.

Separation of LPS and cyclic B—glucans

The upper aqueous layer from the lipid extract mixture was centrifuged and the
supernatant was concentrated to a syrup and chromatographed on a Sepharose-4B column
( 76 cm x 6 cm) in 0.05 M formic acid (pH%5.5). Fractions of 1 mL were analyzed for
carbohydrate by using the phenol-sulfuric acid assay (21) and for radioactivity by using
scintillation counting.

Periodate Oxidative Cleavage of Lipid

The lipid extraction was concentrated to dryness and suspended in 0.1 mL of
methanol. To this suspension 0.1mL of 0.1% sodium periodate was added. The reaction
mixture was sonicated , left at room temperature overnight and concentrated to dryness. A
solvent of water: chloroform: methanol in the ratio of 6:5:1 was added to the reaction

mixture and the lipid was recovered in the lower organic layer.

143
Thin Layer Chromatography
Standard PC was purchased from SIGMA Chemical Co. (St Louis, Missouri). The

lipids extracted were analyzed by TLC on silica plates by using a mixture of chloroform,
acetone, methanol, acetic acid and water in the ratio of 10:4:2: 1: l. The cyclic B-glucans were
analyzed by TLC on silica plates by using a mixture of propanol, ammonium hydroxide and
water in the ratio of 60:40:10. After development, the plates were sprayed with orcinol

reagent, heated at 100°C for 15 min and exposed to a phosphoimager.

144

RESULTS

EPSs

EPSs extracted from cells grown in the 3-O-[”C]methyl-D-glucose media were
analyzed for radiolabel by scintillation counting. No label was detected.
LPSs

The upper aqueous layer from the lipid extraction mixture of bacteria grown on 3-O-
[14C]methyl-D-glucose was chromatographed on Sephrose-4B column. Phenol-sulfuric acid
assay revealed a few carbohydrate-containing peaks (Figure 5.2A). The first three small
peaks (a,b,c) corresponded to LPSs. The large, late-eluting peak ((1) corresponded to the
normal elution volume of cyclic B-glucans. Scintillation counting (Figure 5.2B) revealed that
the label was located only in this latter peak. There was no detectable radioactivity associated
with the first three LPSs peaks, indicating that 3-O-[”C]-methyl-D-glucose was not
incorporated into LPSs.
Cyclic B—glucans

To separate cyclic b-glucans from free 3-O-methyl glucose, the late eluting
carbohydrate peak was subjected to thin layer chromatography. Figure 5.3 was the photocopy
of the orcinol-sprayed plate. 10 [IL of sample from the two fractions with the highest
radioactivity were applied onto the TLC plates and each was resolved into two components.
The fast-moving component was free 3-O-methyl glucose as shown by the same retention

time as the standard whereas the one at the origin was cyclic B-glucans. All the fractions in

145

 

Absorbonco at 4%

 

AAAAAAAAAAAAAA

 

 

 

vvvvvvvvvvvvvvv

 

 

Figure 5.2A: Phenol-sulfuric acid assay revealed a few carbohydrate-containing peaks. The
first three small peaks (a,b,c) corresponded to LPSs. Tire large late-eluting peak ((1)

corresponded to the normal elution volume of cyclic b-glucans and free 3-O-methyl glucose.

146

 

140

120

 

 

 

 

 

 

 

100 >
.1
0
E 30
0
60
1
40
0
0 :
O 10 20 30 40 50 60 70 80 90 100

Figure 5.2 B, Scintillation counting revealed that the radioactivity was associated with the

latter peak (d) in 5.2A.

147

Figure 5.3: TLC analysis of two fractions with the highest radioactivity (in peak (1 from
Figure 2). Each was resolved into two components respectively. The fast-moving component
was free 3-O-methyl glucose as shown by the same retention time as the standard whereas

the one at the origin was cyclic B-glucans.

148

 

 

 

4m-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

~._.
' v
‘. r
0 ,~
- --4—“ |'
I _ a]
. ‘ t l

3-O-methyl rum Fr. so Glucose
glucose

 

r

149

Membrane Lipids

Total lipid extract of bacteria grown on 3-O-[14C]methyl-D-glucose was analyzed
by TLC (Figure 5.5A) and the chromatogram analysed by autoradiography (5.5B). Two
groups of lipids were radiolabeled, indicating that they were glycolipids containing 3-0-
methyl glucose. There was a possibility that 3-O-methyl group was incorporated into SQDG,
a common glycolipid in rhizobial species containing sulfoquinovose as the head group (38).
The biosynthetic pathway of SQDG is largely unknown. One hypothesis about the
biosynthesis of SQDG was that SQDG came directly from glucose. If there was the case, 3-
O-methyl glucose could be changed to 3-O-methyl sulfoquinovose and 1"C labeled 3-O-
methyl SQDG could be detected on the TLC plate. To investigate the possibility that radio-
labeled glycolipids are SQDG, the same strain of bacteria was grown on both regular media
and media with non-radioactive3-O-methyl glucose as the only carbohydrate source. 35S
labeled sulfate was added to both media to trace SQDG (38) by analyzing the membrane lipid
extract on TLC. Figure 5.6A was the photocopy of the ocinol sprayed plate showing the lipid
extracts from bacteria grown on three different media: CM: 3-O-methyl glucose with 3-0-
[”C] methyl-D-glucose ; SM: 3-O-methyl glucose with [358] sulfate; SR: regular media with
[35S] sulfate. Figure 5.6 B was the autoradiograph of the same plate. Two major 35S-

containing lipids were found respectively in the lipid extracts of bacteria grown on two

150

Figure 5.4: A. TLC analysis of the pooled, concentrated radioactive fractions (Peak d from
Figure 2). B. Autoradiograph of the same plate. All of the radioactivity was associated with

the fast-moving carbohydrate component corresponding to free 3-O-methyl glucose.

151

 

 

 

tax... 4 .l ... ... . . .. .3. .. . . . . .
“44.5.5100... Win"... aw 4.1%.. 1.. W: 4.4... v . . 4 a1 . . .3 . . ...7.... .41.. m.
Quinn Iris 48.11845. .

Outawvrvtlsm 405.71.”..‘411 rtlnum.

a. . I
5.. ..\. .. 4 0.. 4.: n\ ‘4. .
Eggf+rh if. 4.4193. «tell. I... ..... 11..” . . .3... .. 5. ..
314491.14. Jnselilcis. .341: via Stalin-.7! .
u

       
    

   
    

  

4.44.... w” 4.5.... .54.. 5:4 a5... .......4; fl
away“; . .5. .

«MYTH EJHY-‘utt awe?

M-Glc Glucan Glc

Glucan Glc

M-Glc

152

Figure 5.5: A. TLC analysis of the total lipid extract of bacteria grown on 3-O-methyl
glucose; B. Autoradiograph of the same plate. Two groups of lipids contained 3-O-methyl

glucose.

153

 

 

. . .
v . ..«, \‘ 3 . .. . ... v
. . . y. \. .. . u /r\ .. . ‘ .
. . .s - . . . v.3, a \ . x ,
._ ..I .. Z .. . o . — .. . x
. I v . . .. 5 \r . p . .
. . . . c . .s a I. 1 . «I \ fl \ ,
. .. . .1 » . I . I A . . .
. (....r. ‘. . . .r. . . .f‘... I
. t .. I“ \‘ o . . I
. . ‘ J. x . . .. . . . .
‘ J I c . u . . ~ . .
.. . . . . ~
.\ 1.. , o l '. u .u. . y
. .u u . . . .Io . . ‘ .. . , .
. . ». . ‘. \ 6n.“ 4...C .n/ .f ‘A. w .-.l l .9 .P 2. . . i.. Iv . wc‘.) .
. . ~ .. . g . . .
n ‘ -
u . ~ D . . I . . _
. . . .. .. In x .u\. I .. . u r »
. . .\ I, . r \ u
. I 1 - . . v
. a. I 1‘ . y u .\ . .A u
I o I n i 1v
- n u .
. .. . . .
.
. .4 \
. . o v
4 l o \ - §
..,. .
A ~ .
v ‘
.v\
. . . I
I ...v.fl .. . I: Va I .
. 3 IV
. . . .rviut,gw .. orb .v
. . . . .I,
. .0 c (I y.'- IJ
. x , . y
c .
. .1. . .
. .
- V « ~ I I I; )
¢ I . 1
. . . . ~ .
. .l r . I
\ ~ 4., . I .p K}. a}... . f
. . « . 4
{V n’ » .oq . TI‘.. .4 .1 i. 4‘ .-
. .\ \s. ’ . ..
.. .51 _ I .» o .
s . g a
p . . ~
I .w » .. .. ,
a . . ‘
_ Z t N c . .,
. _ .v . _ u. . I ,
. q f .. fl.' '1 u ).
I .. .- 4 . t
l .
.

.w . III?! IIv

 

. I Ill.

1. . . . sffi, . .‘... l'.

I‘

\

!|' . .ol..all|l ‘III

& 4

154

Figure 5.6: A.TLC analysis of the lipid extracts from bacteria grown on three different
media: CM: MC labeled 3-O-methyl glucose medium; SM: 35S labeled 3-O-methyl glucose

medium; SR: 35S labeled regular B-HI medium; B. Autoradiograph of the same plate.

155

 

t. .

 

 

“at . . .§ ....

O 04......

‘.;I.

CM SM

SR

SM SR

CM

156

kinds of 35S- labled media (SM and SR). These ”S- labeled lipids had the same retention
time as that of SQDG ( 2) and their retention time was different from those of glycolipids
containing 3-O-methyl glucose.
Periodate Oxidative Cleavage of Lipid

To confirm the glycolipids containing 3-O-methyl glucose was not the same as 3-0-
methyl SQDG, periodate oxidative degradation was used. The mechanism of the reaction is
shown in Figure 5.7. In this reaction, vicinal glycols are cleaved to yield a dialdehyde and
iodate via a cyclic intermediate. This reaction has tremendous analytical significance since
it can be used to ascertain what proportion of glucosyl residues might be blocked at the 0-2
or 0-3 position. Substitution of either renders the ring insensitive to periodate oxidation. The
three aforementioned lipid extracts before and after oxidation were analyzed by TLC.
Changes were observed in all three lipid extracts (SR, CM, SM) after oxidation (Figure
5.8A). Figure 5.8 B was the corresponding autoradiograph. The major 35S labeled lipids
corresponding to SQDG in both SR and SM disappeared after oxidation, indicating that both
were 35S labeled SQDG instead of 3-O-methyl SQDG. The glycolipids containing 3-0-

methyl glucose in CM remained after oxidation.

157

HO HIO4 0

OR

OH
O OH
/ O
“0 H°%
+—
0 0 OR 0 0

/ OR
+ \ IO4H3

HIO3

Figure 5.7 : The mechanism of periodate oxidative degradation of glucosyl rings to form
dialdehydes. In the reaction, vicinal glycols are cleaved to yield a dialdehyde and iodate via

a cyclic intermediate.

158

Figure 5.8A: TLC analysis of three lipid extracts before (SR, CM, SM) and after oxidation

(SRO, CMO, SMO).

159

 

PC

SMO

SRO CM CMO SM

SR

160

Figure 5.8B: Autoradiograph of the same plate. Note that the major 35S labeled lipids
corresponding to SQDG in both SR and SM disappeared after oxidation while the glycolipid

containing 3—O-methyl glucose in CM remained after oxidation.

161

I.
t

s
I.
‘

.J.g..u.n3.a...f. .

.. . a‘f. ~u ..
Hun). £101 .
. o.‘ \tclh

f
V.

 

PC

SMO

CMO SM

SRO CM

SR

162

DISCUSSION

Our results clearly indicated that 3-O-['4C]-methyl-D-glucose was incorporated into
membrane glycolipids but not into other major carbohydrate—containing molecules, including

EPS, LPS and cyclic B-glucans. This gives us some information about the biosynthetic
pathways for the synthesis of these molecules. It is likely that EPS, LPS and cyclic B-glucans

are not generated from a common pool of the monosaccharides with membrane glycolipids
but utilizes locally generated pools. In this local pool, normal polysaccharide molecules may
be synthesized from glucose synthesized de novo. We can conclude that the biosynthesis of
glycolipids in this strain was independent, to some degree, from those synthesizing other
carbohydrate-containing molecules. There was another possible explanation to our results.
Previous studies (1) have indicated the similarities of glycosyltransferases involved in the
biosynthesis of EPS, glycogen, cellulose and the O-antigen of LPS. Our results may indicate
the existence of a substrate specificity difference between the glycosyltransferase

synthesizing glycolipids and those synthesizing EPSs, LPSs, and cyclic B-glucans.

In these experiments, we did not further characterize the structure of the 3-O-methyl-
D-glucose containing glycolipid. In the membrane of same bacterial strain,, one mannose-
and glucose-containing diglycosyl diacylglycerols (Figure 1.7 in Chapter 1) has been
identified ( 12). These molecules had a 1,3 glycosidic linkage between the two sugars. It is
very likely that the glycolipids containing 3-O-methyl glucose was the same as DGDG except
for the terminal sugar which is 3-O-methyl glucose instead of glucose. The 1,3 linkage

between the two sugars will render the whole molecule resistant to periodate oxidation.

163

BIBLIOGRAPHY

l. Leigh, J. A. and G. C. Walker. 1994. Trends Genetics. 10, 63-67.

2. Cedergren, R. A., and R. I. Hollingsworth. 1994. J. Lipid Res. 35, 1452-1461.
3. Breedveld, M. W. and K. J. Miller. 1994. Microbial. Rev. 58, 145-161.

4. Denarie, J and J. Cullimore. 1993. Cell 74, 951-954.

5. Orgarnbide, G. G., S. Philip-Hollingsworth, R. I. Hollingsworth and F. B. Dazzo.
1994. J. Bacteriol. 176, 4338-4347.

6. Breedveld, M. W. and K. J. Miller. 1994. Microbial. Rev. 58, 145-161.

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

8. Carlson, R. W., N. P. J. Price, and G. Stacey. 1994. Mal. plant-microbe Interact. 7,
684-695.

9. N. P. Price et al. 1992. Mal. Microbial. 6, 3575-3577.
10. H. P. Spaink. 1992. Plant Mol. Biol. 20, 977-983.
1 l. J. Denarie et al. 1992. Annu. Rev. Microbiol. 46, 497-511.

12. Brink, B. A., J. Miller, R. W. Carlson and K. D. Noel. 1990. J. Bacterial. 172, 548-
555.

13. Carlson, R. W., S. Kalembasa, D. Turowski, P. Pachori and K. D. Noel. 1987. J.
Bacterial. 169, 4923-4928.

14. Noel, K. D., K. A. Vandenbosch and B. Kulpaca. 1986. J. Bacterial. 168, 1392-1401.

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

16. de Maagd, R. A., A. S. Rao, 1. H. M. Mulders, L. G. de Roo, M. C. M. van Loosdrecht,

164

C. A. Wijffelman, and B. J. J. Lugtenberg. 1989. J. Bacterial. 171, 1143-1150.

17.

18.

19.
20.
. R. Helinski, and G. S. Ditta. 1986. Proc. Natl. Acad. Sci. USA 83, 4403-4407.

21.

22.
23.

24.
. B. Perry. 1992. J. Bacterial. 174, 2172-2177.

25.

26.

27.

28.

29.

30.

31.

Priefer, U. B. 1989. J. Bacterial. 171, 6161-6168.

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

Dylan, T., P. Nagpal, D. R. Helinski, and G. S. Ditta. 1990. J. Bacterial. 172, 1409-
1417.

Dylan, T., P., L. Ielpi, S. Stanfield, L. Kashyap, C. Douglas, M. Yanofski, E. Nester, D
Dubois, M., K. A. Gilles, J. K. Hamilton, P. A. Rebers, and F. Smith. 1956. Anal. Chem.
28, 350-356.

Cedergren, Y. Wang, and R. I. Hollingsworth. 1996. J. Bacterial. 178, 5529-5532.
Aragon, V., R. Diaz, E. Moreno, and I. Moriyon. 1996. J. Bacterial. 178, 1070-1079.

Beynon, L. M., A. J. Dumanski, R. J. C. Mclean, L. L. maclean, J. C. Richard, and M

Kogan, G., B. Jann. And K. Jann. 1992. FEMS Microbial. Lett. 91, 135-140.

Lopez-lara, I. M., G. Orgarnbide, F. B. Dazzo, J. Olivares, and N. Toro. 1993. J.
Bacterial. 175: 2826-2823.

Putnoky, P., G. Petrovics, A. Kereszt, E. Grosskopf, D. Cam Ha, Z. Banfalvi and A.
Kondorosi. 1990. J. Bacterial. 172, 5450-5458.

Williams, M. N. V., R. I. Hollingsworth, P. M. Brzoska and E. R. Signer. 1990. J.
acterial. 1172, 6596-6598.

Diebold, R. and K. D. Noel. 1989. J. Bacterial. 171, 4821-4830.

Erdman, S. H., M. H. Hart, J. H. Park and J. A. Vanderhoof. 1991. J. Pediatr.
Gastraenteral Nutr. 14, 360-366.

Gould, G. W., H. M. Thomas, T. J. Jess and G. 1. Bell. 1991. Biochemistry 30, 5139-
5145.

165

32. Saiti, Y. and D. F. Silbert. 1979. J. Biol. Chem. 254, 1102-1 107.

33. Furler, S. M., A. B. Jenkins, L. H. Storlien, and E. W. Kraegen. 1991. Am. J. Physiol.
261, 337-347.

34. Malaisse-Lagae, F., M. H. Giroix, A. Sener and W. J. Malaisse. 1986. FEBS-Lett. 198,
292-294.

35. Jay, T. M., G. A. Dienel, N. F. Cruz, K. Mori, T. Nelson and L. Sokoloff. 1990. J.
Neurachem. 55, 989-1000.

36. Gancedo, C. And J. M. Gancedo. 1985. Eur. J. Biochem. 148, 593-597.

37. Gatley, S. J ., J. E. Holden, J. R. Halama, T. R. DeGrado, D. R. Bernstein and C. K. N g.
1984. Biochem. Biophys Res Commun. 119, 1008-1014.

CHAPTER 6
SUMMARY AND PERSPECTIVE

166

167

SUMMARY AND PERSPECTIVE

At each stages of rhizobium/legume symbiosis, a variety of rhizobial cell surface and
membrane carbohydrate-containing molecules have been indicated to play some important
roles. The functions of these molecules in the establishment of symbiosis are closely related
to the ability of rhizobia to change its surface chemistry in response to a changed
environment. It is the whole bacterial surface that interacts with plants and there is a large
degree of functional and biosynthetic overlap among various rhizobial cell surface and
membrane carbohydrate-containing molecules. A comprehensive picture of the rhizobial cell
surface needs to be obtained in order to understand the roles played by the rhizobial surface
or membrane components in symbiosis. There are two important parts of the big picture: the
environmental regulation of the bacterial surface and the coupling of the biosynthetic events
leading to various surface and membrane molecules.

Chapter 2 to Chapter 4 of this dissertation address the aspect of environmental
regulation of rhizobial membrane chemistry. In the studies described in Chapter 2, the
membrane lipids of free living and bacteroid forms of Bradyrhizabium japanicum, obtained
from nodules occupied by both typed and untyped bacteria, were isolated and characterized
by a combination of NMR spectroscopy, mass spectrometry and other chemical and physical
methods. These studies indicated that in bacteroid forms of Bradyrhizabiumjapanicum, there
was a dramatic shift towards the synthesis of DGDG, the glycolipid which has rarely been

found outside of the plant kingdom and photosynthetic bacteria. This result follows our

168

earlier finding that the other predominant plant glycolipid, SQDG, is a membrane component
of fast growing Rhizobia and is found even when cells are cultivated in free culture. In the
studies described in Chapter 3, culturing Bradyrhizabium japanicum in the presence of
glucose as a carbon source in a rich medium known to stimulate differentiation into bacteroid
forms led to the synthesis of a family of di- and trisaccharide-containing glycerolipids as the
main membrane lipid components. Like DGDG, these glycolipid are also rarely found in
other bacteria. In the studies described in Chapter 4, Bradyrhizabium japanicum was grown
in a low-oxygen condition which mimics the condition inside root nodues. There was a
dramatic switch of membrane chemistry in which PC, the predominant lipid in aerated
cultures, was no longer synthesized and PE became the major lipid. Besides this change, PI,
a typical plant lipid rarely found in bacteria was also synthesized. Lowering pH in free living
Bradyrhizabium japanicum did not result in any change in the lipid headgroups but resulted
in the change of unsaturation state in the fatty acyl groups of the lipids.

The possible functional role played by these unusual lipid molecules in symbiosis
deserves some further discussion. When the bacterium is living inside of the plant, the plant
cell in which the bacterium is growing provides the growth medium for the bacterium. The
bacteria have to use many intermediates in this growth medium to modify its own membrane
and surface chemistry in order to support its division and differentiation. If the cellular
chemistry of the bacterium is modified to such a degree that it approximates that of its host,
the change in the cellular chemistry of the host will be minimal and the defense system of
the host may be avoided. There is another way to look at this. Membrane is a critical

structural feature that determines the ability of an organism to occupy a particular ecological

169

niche. It is, therefore, not too surprising that rhizobium and its plant host have developed
similar membrane chemistry because the two organisms share ecological pressures of
temperature, pH, water and nutrient availability and their metabolism have to be
synchronized. This is a striking parallel with some host-pathogen interactions.

The results presented in this dissertation not only suggest a close developmental tie
between rhizobia and its host plant but also raise questions about the origin of the genes
involved in the biosynthesis of the membrane lipids of rhizobium. Were these genes
transferred from the plant or were they developed in the bacterium by parallel evolution? It
is very possible that these genes be passed on through gene transfer from plants to the
bacteria at some point during their long symbiotic relationship. Although it is still a
conjecture, the close relationship between the two organisms could provide many
opportunities for gene transfer..

Future work can be conducted on the characterization of various membrane and
surface saccharides from different species to establish the generality of our findings. Besides,
the changes in the bacterial surface and membrane chemistry in different environments
should also be investigated.

There is functional and biosynthetic overlap among various classes of rhizobial cell
surface components. In the studies described in Chapter 5, l“C labeled 3-O-methyl-D-
glucose was used as a probe to investigate the interrelationship of the biosynthetic events
leading to various carbohydrate-containing molecules which are important to symbiosis. Our
results clearly indicated that 14C labeled 3-O-methyl-D-glucose was incorporated into

membrane glycolipids but not into other major carbohydrate-containing molecules, including

170

EPS, LPS and cyclic B-glucans. It is likely that the biosynthesis of EPS, LPS and cyclic B-

glucans do not use the same metabolite pool as glycolipids. It is also possible that the
substrate specificity of the glycosyltransferase(s) synthesizing glycolipids is (are) different.
Our conclusion is that the biosynthesis of glycolipids in this strain was independent, to some
degree, from those synthesizing other carbohydrate-containing molecules. Future studies on
the interconnections among the biosynthetic pathways leading to various saccharides can be
conducted by using both radiolabeled and functional labeled analogs of proposed substrates
or intermediates, following their incorporation into the final products, and seeing that if they
are incorporated into more than one saccharide.

The most important question that can be answered in this dissertation is that by what
mechanism rhizobia can survive the plant defense systems. Can rhizobia live inside of the
plant as an organelle? The unusual occurrence of SQDG, PC and 27-hydroxyl-C28-fatty acid
in the rhizobial species has strongly revealed similarities in the composition of membrane
lipids among rhizobia, chloroplast, and cyanobacterium( a photosynthetic bacterium ).
Cyanobacterium is also a functional counterpart of rhizobium. According to endosymbiotic
theory, chloroplast was originated through an endosymbiotic invasion of the plant cell by
cyanobacterium (1, 2). Modern cyanobacteria are often found in symbiotic relationships
with other organisms and live inside of these organisms as functional chloroplasts. In this
sense, cyanobacterium is a C02- fixing symbiont and rhizobium is a Nz-fixing symbiont,
both associated with plants. The similarities in both the function and the membrane
composition suggest that rhizobium can live inside of the plant as an organelle. The work

presented in this thesis further strengthened this hypothesis. The ability of these rhizobial

171

species to make the unusual plant-type lipids, such as SQDG, DGDG, PI, PC and di- and tri-
saccharides-containing glycolipids, strongly suggests that the membrane surface of rhizobium
mimics that of the plant (including that of plastids). This enables rhizobium to live inside of
the plant successfully after the invasion and establish an effective symbiosis. The results
obtained in Chapter 5 addressed this ability of rhizobia from another perspective. It is likely
that the glycosyl transferases involved in the biosynthesis of these glycolipids have a
relatively low substrate specificity. These enzymes can even incorporate monosaccharide
analogue such as 3—O-methyl-glucose into membrane glycolipids. It is therefore not
surprising that they can pick up different sugars in different growth conditions and make
different glycolipids. In this way, rhizobium can quickly modify its membrane structure and

adapt to a new environment.

172

BIBLIOGRAPHY

1. Whatley, J. M. 1993. International Review of Cytology. 144, 259-297.

2. Raver, P. H. 1970. Science 169, 641-646.