EXPERIMENTAL AND EPIDEMIOLOGICAL APPROACHES TO CAMPYLOBACTER
JEJUNI-ASSOCIATED GUILLAIN BARRÉ SYNDROME
By
Jessica L. St. Charles

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Comparative Medicine and Integrative Biology – Doctor of Philosophy
2013

ABSTRACT
EXPERIMENTAL AND EPIDEMIOLOGICAL APPROACHES TO CAMPYLOBACTER
JEJUNI-ASSOCIATED GUILLAIN BARRÉ SYNDROME
By
Jessica L. St. Charles
Guillain Barré Syndrome (GBS), an immune-mediated acute polyneuropathy of
the peripheral nerves, is the leading cause of acute autoimmune neuromuscular
paralysis in the Western world. Approximately two-thirds of patients report a prior
respiratory or gastrointestinal infection; Campylobacter jejuni is the most commonly
reported antecedent infection. C. jejuni is the leading cause of foodborne bacterial
gastrointestinal infection; transmission is by ingestion of contaminated water, milk, and
poultry. The proposed mechanism of GBS is similarity of the lipo-oligosaccharide (LOS)
on the outer surface of some C. jejuni strains to gangliosides, glycolipid structures found
in the nervous system. This molecular mimicry results in an immune response directed
at both the LOS and gangliosides. Current data surrounding molecular mimicry are
conflicting, therefore suggesting that other surface structures of C. jejuni are involved in
GBS pathogenesis. Epidemiological studies on isolates collected from patients that
presented with either enteritis or neuropathy to define potential genetic relationships of
GBS to the highly variable C. jejuni flagellar and major outer membrane proteins have
given inconclusive and contradictory results. Therefore additional work is needed to
improve molecular typing of C. jejuni isolates.

The work reported here took two complementary directions: development of a
murine model of GBS using C. jejuni strains of an LOS type known to be associated
with GBS in humans and an epidemiological study of C. jejuni isolates from the U.S.
with a focus in Michigan, in which the LOS biosynthetic locus, the flagellar protein, and
the major outer membrane protein were all characterized in addition to standard
multilocus sequence typing.
To understand the role of the genetic factors of C. jejuni and potential host
factors in the pathogenesis of GBS more in vivo studies are needed. However, current
animal models to study GBS employ immunization of rabbits with purified LOS or mice
with bovine myelin to initiate an immune response similar to that in GBS patients. These
are not natural models of GBS following C. jejuni infection. In a more natural model in
chickens, birds given GBS-associated strains of C. jejuni develop both clinical signs and
immunological responses similar to those seen in GBS patients. However, chickens are
anatomically and physiologically different from humans. The goal of the first half of my
studies was development of a murine model secondary to C. jejuni infection; we found a
strain of mouse that has the potential to be such a model, but more work must be done.
The goal of the second half of my studies was to examine possible
epidemiological relationships between other variable surface components of C. jejuni
and LOS types associated with GBS. In epidemiological studies using molecular typing
of both variable and conserved genes, we found that both human C. jejuni isolates from
a limited collection from the U.S. with focus in Michigan and from calves on a Michigan
dairy farm that had a C. jejuni outbreak had LOS biosynthetic loci characteristic of GBSassociated strains.

Copyright by
JESSICA LYNN ST. CHARLES
2013

Dedicated in Loving Memory to Neil T. St. Charles

v

ACKNOWLEDGMENTS

First and foremost I would like to thank my family and loved ones for their love
and endless support throughout my time here at MSU. A very special thanks to my
better half, Ben Dunn, for the love, unconditional support, and encouragement he has
provided me continuously day and night as I worked to complete my degree.
I’d like to thank my wonderful advisor, Dr. Linda S. Mansfield, for her support,
knowledge, and guidance throughout my studies. She was my advisor as an
undergraduate and is the reason I considered graduate school in the first place, so
without her I wouldn’t be here today and for that I am grateful. I would also like to
sincerely thank my committee members Dr. Bill Atchison, Dr. Katheryn Meek, Dr. John
Linz, and Dr. Jon Patterson for their guidance and support.
In addition, I thank Dr. Julia Bell for her support and guidance; she taught me all
the basic laboratory skills a scientist needs to succeed when I was an undergraduate in
the lab. I would also like to thank Dr. Shannon Manning for the knowledge and support
she shared with me.
Many thanks to the members of the Comparative Enteric Disease Laboratory
who created a wonderful laboratory environment to work in. I had the pleasure to work
with Andy Flies, JP Jerome, Ankit Malik, Barbie Gadsden, and Phillip Brooks who I will
miss and wish them the best of luck with their studies. I have also had the pleasure to
work with past members of the lab that include Alice Murphy, Vijay Rathinam, Eric
Smith, Melisa Bailey, Anthony Brooks, Allison Zaleski, Laura Krzykowski, Jenna

vi

Gettings, Jamie Kopper, and Amanda Stanton. I also thank Rebekah Mosci, Samantha
Wengert, and Amber “Cody” Springman from Dr. Manning’s lab.
Thank you to Dr. Vilma Yuzbasiyan-Gurkan, Victoria Hoelzer-Maddox, and the
Comparative Medicine and Integrative Biology program for support and encouragement
throughout my studies. And thank you to the Microbiology Research Unit (MRU) and
Enteric Research Investigative Network (ERIN) both funded by the National Institute of
Health for the financial support of my research.
I would also like to thank the great people at the MSU University Research
Containment Facility that helped with maintenance of my mice. I want to thank the great
people at the MSU Investigative Histopathology Laboratory and the MSU Genomic
Research Technology Facility for their help and knowledge in the analysis of data.

vii

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................xiii
LIST OF FIGURES ..................................................................................................xiv
CHAPTER 1: Introduction .......................................................................................1
Guillain Barré Syndrome ..............................................................................2
Subtypes of GBS ..........................................................................................7
Acute Inflammatory Demyelinating Polyneuropathy ...........................7
Acute Motor Axonal Neuropathy (and Acute Sensory Motor Axonal
Neuropathy .........................................................................................8
Miller Fisher Syndrome .......................................................................9
Campylobcter jejuni ......................................................................................11
Lipo-oligosaccharides (LOS) ..............................................................12
Molecular Typing of C. jejuni ..............................................................15
Animal Models ..............................................................................................19
Rationale ......................................................................................................23
CHAPTER 2: Non-Obese Diabetic (NOD) mice as working models of Guillain
Barré Syndrome following Campylobacter jejuni infection.......................................25
Abstract ........................................................................................................26
Introduction ...................................................................................................28
Materials and Methods .................................................................................34
Animals ...............................................................................................34
Campylobacter jejuni ..........................................................................35
Experimental Design...........................................................................35
Experiment 1: Screening mice of various NOD genotypes for
GBS following C. jejuni infection ...............................................36
Experiment 2: Screening larger group sizes of NOD WT and
-/NOD B7-2 mice for GBS ........................................................36
-/-

Experiment 3: Examination of NOD IL-10 mice using
antibiotics to treat enteritis ........................................................37
Neurological Phenotype Testing .........................................................37
Open field testing ......................................................................38
Hang testing .............................................................................38
Rotarod testing .........................................................................39
Footprinting ...............................................................................39
DigiGait™ testing.....................................................................39
Enzyme-Linked Immunosorbent Assays (ELISAs) .............................41

viii

Assessment of Clinical Signs..............................................................42
Necropsy Methods and Assessment of Gross Pathology ...................42
Culture for Presence of C. jejuni, DNA Extraction and PCR Assays ..43
Histopathological Scoring of the Gastrointestinal Tract ......................43
Histopathological Scoring of the Sciatic Nerve ...................................43
Nerve Histology Criteria ......................................................................45
Immunohistochemical Staining (IHC) .................................................46
Statistical Analyses .............................................................................47
Results .........................................................................................................49
Selection of Mice for Analysis .............................................................49
Experiment 1: Screening mice of various NOD genotypes for GBS
following C. jejuni infection .................................................................50
Experimental screen for GI disease with GBS-associated
C. jejuni strains .........................................................................50
Detection of anti-ganglioside antibodies ...................................51
Neurological phenotype testing ................................................53
NOD WT mice.................................................................54
-/NOD IL-10 mice ...........................................................55
-/-

NOD B7-2 mice............................................................56
Experiment 2: Screening larger groups sizes of NOD WT and NOD
-/B7-2 mice for GBS ...........................................................................57
Experimental screen for GI disease with GBS-associated
C. jejuni strains .........................................................................57
Detection of anti-ganglioside antibodies ...................................58
Neurological phenotype testing ................................................59
NOD WT mice.................................................................60
-/NOD B7-2 mice............................................................61
Presence of nerve lesions and detection of cellular infiltrates ..62
-/Experiment 3: Examination of NOD IL-10 mice using antibiotics to
treat enteritis .......................................................................................64
Experimental screen for GI disease with GBS-associated
C. jejuni 260.94 .........................................................................64
Detection of anti-ganglioside antibodies ...................................65
Neurological phenotype testing ................................................67
Rotarod ...........................................................................68
DigiGait™ ......................................................................68
Stride Length and Hind Base Width ......................68
Clinical Analysis ....................................................69
Presence of nerve lesions and detection of cellular infiltrates ..70
Detection of macrophages in nerves ........................................70
Discussion ....................................................................................................72
Comparison of NOD mouse genotypes ..............................................73
Difficulties encountered and their resolutions .....................................76
Future work.........................................................................................77

ix

Acknowledgements ......................................................................................79
Tables ...........................................................................................................80
Figures .........................................................................................................83
CHAPTER 3: Examination of Miller Fisher Syndrome in NOD WT and C57BL/6
WT mice subsequent to Campylobacter jejuni CF93-6 infection .............................103
Abstract ........................................................................................................104
Introduction ...................................................................................................106
Materials and Methods .................................................................................110
Animals ...............................................................................................110
Campylobacter jejuni ..........................................................................111
Experimental Design...........................................................................111
Neurological Testing ...........................................................................112
Open field testing ......................................................................112
Rotarod testing .........................................................................113
Footprinting and tracking ..........................................................113
Enzyme-Linked Immunosorbent Assays (ELISAs) .............................114
Assessment of Clinical Signs..............................................................115
Necropsy Procedure and Assessment of Gross Pathological
Features .............................................................................................115
Assessment of Enteric Disease ..........................................................116
Histopathological Scoring for the GI Tract and Sciatic Nerve .............117
Gastrointestinal tract .................................................................117
Sciatic nerve .............................................................................117
Statistical Methods .............................................................................118
Results .........................................................................................................119
Screening for Enteric Disease with GBS-associated and
MFS-associated C. jejuni Strains........................................................119
Detection of Anti-Ganglioside Antibodies with ELISA Assays.............120
Neurological Testing ...........................................................................121
Rotarod .....................................................................................122
Open Field Test ........................................................................122
Footprint Measurements ...........................................................123
Presence of Nerve Lesions.................................................................124
Discussion ....................................................................................................126
Table ............................................................................................................129
Figures .........................................................................................................130
CHAPTER 4: Multiple molecular typing schemes applied to define genetic
relationships among C. jejuni isolates that share the same LOS classification .......139
Abstract ........................................................................................................140
Introduction ...................................................................................................142
Materials and Methods .................................................................................147
Campylobacter jejuni Strains and Growth Conditions.........................147
Preparation of Genomic DNA .............................................................148

x

Determination of Campylobacter spp. ................................................149
Classification of LOS Classes by PCR ...............................................149
DNA Sequencing and Allele Assignment ............................................150
Multi-locus sequence typing (MLST).........................................150
flaA SVR ...................................................................................150
porA ..........................................................................................150
Sequence Analysis .............................................................................151
Statistical Analysis ..............................................................................152
Results .........................................................................................................153
LOS Classification ..............................................................................153
MLST Analysis ....................................................................................154
porA Typing ........................................................................................155
flaA SVR Sequence Types .................................................................156
Comparison of Isolates with All Molecular Typing Methods ................156
Discussion ....................................................................................................158
Table ............................................................................................................163
Figures .........................................................................................................168
CHAPTER 5: Campylobacter jejuni isolates from calves have A, B, and C lipooligosaccharide (LOS) biosynthetic locus classes similar to human Guillain Barré
Syndrome associated strains ..................................................................................173
Abstract ........................................................................................................174
Introduction ...................................................................................................176
Materials and Methods .................................................................................179
Case Study .........................................................................................179
Campylobacter jejuni Isolation ............................................................179
Preparation of Genomic DNA and Confirmation of Campylobacter
spp. .....................................................................................................180
Classification of LOS Biosynthesis Loci ..............................................181
Multi-locus Sequence Typing (MLST) .................................................181
porA Allele Typing and flaA SVR Sequence Typing ...........................182
Sequence Analysis .............................................................................183
Statistical Analysis ..............................................................................183
Results .........................................................................................................185
Verification of C. jejuni ........................................................................185
LOS Classification ..............................................................................185
MLST Analysis ....................................................................................185
porA Allele Typing and flaA SVR Sequence Typing ...........................186
Composite Profiles .............................................................................188
Discussion ....................................................................................................189
Zoonotic Transmission Between the Calves and Family Members ....189
Examination of Genetic Relationships Among Calf Isolates with
LOS Locus Classes A–C ....................................................................190
Table ............................................................................................................192
Figures .........................................................................................................193

xi

CHAPTER 6: Conclusions.......................................................................................196
Summary ......................................................................................................197
Chapter 2 Summary: Non-Obese Diabetic (NOD) mice as working
models of Guillain Barre Syndrome following Campylobacter jejuni
infection ..............................................................................................200
Chapter 3 Summary: Examination of Miller Fisher Syndrome in
NOD WT and C57BL/6 WT mice subsequent to Campylobacter
jejuni CF93-6 infection ........................................................................204
Chapter 4 Summary: Multiple molecular typing schemes applied to
define genetic relationships among C. jejuni isolates that share the
same LOS classification .....................................................................205
Chapter 5 Summary: Campylobacter jejuni isolates from calves have
A, B, and C lipo-oligosaccharide (LOS) biosynthetic locus classes
similar to human Guillain Barre Syndrome associated strains............207
Future Directions ..........................................................................................210
Defining the NOD WT mouse as a natural model of GBS ..................210
-/Potential for the NOD IL-10 mouse as a natural model of GBS .......211
Examination of other factors contributing to the pathogenesis of
GBS ....................................................................................................213

APPENDICES .........................................................................................................216
Appendix A: Chapter 2 Experiment 1 ...........................................................217
Appendix B: Chapter 2 Experiment 2 ...........................................................230
Appendix C: Chapter 2 Experiment 3 ...........................................................243
Appendix D: Chapter 2 Rotarod Controls .....................................................250
REFERENCES ........................................................................................................252

xii

LIST OF TABLES

Table 2.1: Clinical Assessment for DigiGait™ videos ............................................80
Table 2.2: Statistical results for change within treatment groups over the course
of the experimental time. .......................................................................81
Table 2.3: P values for ELISA assays for ELISA assays for all 3 experiments .......82
Table 3.1: P values for ELISA assays.....................................................................129
Table 4.1: Summary of C. jejuni Isolates and their assigned allele types ...............163
Table 5.1: Summary of Molecular Typing ...............................................................192

xiii

LIST OF FIGURES

Figure 2.1: Experiment 1 Data Summary ...............................................................83
Figure 2.2: Experiment 2 Data Overview................................................................86
Figure 2.3: Examination of Peripheral Nerves. .......................................................91
Figure 2.4: Immunohistochemical Data for Experiment 2. ......................................92
Figure 2.5: Macrophage infiltration of NOD WT mice. ............................................93
Figure 2.6: Experiment 3 Data Summary ...............................................................95
Figure 2.7: Heat Map of Clinical Neurological Assessment ....................................99
Figure 2.8: Infiltration of macrophages in the nerve root ........................................101
Figure 3.1: Examination of GI disease ...................................................................130
Figure 3.2: ELISA data ...........................................................................................132
Figure 3.3: Assessment of Clinical Neurological Signs with Phenotype Tests .......134
Figure 4.1: Dendogram of the LOS biosynthetic loci ..............................................168
Figure 4.2: Comparison of LOS classes to MLST Clonal Complexes ....................170
Figure 4.3: Phylogenic tree of the porA based on DNA sequences .......................171
Figure 5.1: SplitsTree diagram based on MLST alleles ..........................................193

xiv

Figure 5.2: Dendograms of porA and flaA SVR DNA sequences ...........................194
Figure 5.3: SplitsTree analysis of combined allelic profiles of 7 MLST loci, porA
alleles, and flaA SVR sequence types. .................................................195
Figure A-1: Examination of enteric disease from C. jejuni infection in
Experiment 1. ......................................................................................218
Figure A-2: Experiment 1 ELISA assay data ..........................................................219
Figure A-3: Phenotypic data for Experiment 1 .......................................................225
Figure B-1: Examination of enteric disease caused by C. jejuni infection in
Experiment 2. ......................................................................................231
Figure B-2: ELISA assay data for days 0 and 14 P.I. .............................................232
Figure B-3: Experiment 2 Phenotype Data.............................................................238
Figure B-4: Presence of nerve lesions in Experiment 2 .........................................242
Figure C-1: Evidence of enteric disease in Experiment 3 .......................................244
Figure C-2: ELISA assay data for Experiment 3 .....................................................246
Figure D-1: Rotarod Control Experiment with C57/BL6 mice .................................251

xv

CHAPTER 1

Introduction

1

GUILLAIN BARRÉ SYNDROME

Guillain Barré Syndrome (GBS) was first described in 1916 during World War I,
when G. Guillain, J. A. Barré, and A. Strohl examined two soldiers who had symptoms
of a rapid and progressive motor neuropathy that involved loss of reflexes and
spontaneously recovered (7, 153). In these patients they also described an elevated
protein level in the cerebrospinal fluid (CSF) without an increase in white blood cells (9,
82, 153). Today, the disease is clinically characterized as progressive weakness with a
symmetrical ascending flaccid paralysis, areflexia, and increased CSF protein level (9,
67, 129).
The annual incidence rate of GBS worldwide is approximately 0.6–4.0 per
100,000 population (7, 55, 63, 72, 116, 141, 157). In Europe, incidence rates are1.2–1.9
annually per 100,000 persons (116). In Western countries, the annual incidence rate
ranges from 0.89–1.89 cases per 100,000 persons (163). With the near eradication of
polio, GBS is the leading cause of acute flaccid paralysis in the Western world (7, 63,
67, 140, 150, 157, 160, 162, 165). Incidence rates are lower in children than adults and
are highest in the elderly (72). The occurrence of GBS can increase with age, a rate of
1.0 per 100,000 persons below the age of 30 increases to 4.0 cases per 100,000
persons in people older than 75 years of age (116). Men are more commonly affected
than women with a rate of 1.25–1.5 men to 1.0 women, which is an unusual occurrence
in autoimmune diseases (72, 116, 148, 157).
GBS is typically a monophasic disease with symptoms most severe at weeks 2–4
and then plateauing. However, 7% of GBS patients have had 2 or more occurrences (7,

2

72, 163). Symptomatic patients present with overall weakness that frequently involves
muscles served by the motor, sensory, and autonomic nerves associated with the limbs
and may involve the respiratory muscles and muscles served by facial and ocular motor
nerves as well (7, 72, 116). Patients report the first onset of symptoms to include
numbness, tingling sensation, weakness, and pain in the limbs (116, 163). The average
stay in the hospital is 7 days; nearly 25% of patients require mechanical ventilation (72,
148, 163). Ten to twenty percent of patients have lifelong severe disability and mortality
rates range from 3–10% (7, 72, 148, 163).
Two-thirds of GBS patients report a preceding infection prior to developing GBS
(17, 63, 147, 148, 160, 162, 163). Prior infections include Campylobacter jejuni,
Mycoplasma pneumonia, Haemophilus influenza, cytomegalovirus, and Epstein-Barr
virus (EBV) (7, 163). C. jejuni is the most common antecedent infection in GBS patients
(63, 140, 160, 162, 163). It is estimated that 0.25–0.65 per 1000 cases of C. jejuni
infections go on to develop GBS (163). A study by Kuroki et al. (1993) determined that
41% of GBS patients were infected with C. jejuni prior to development of GBS (92).
Another study, which was performed in England between April, 2000, and March, 2001,
was performed to identify the number of GBS cases reported that were associated with
C. jejuni infection. The results indicated that 15% of all GBS cases in England during
this time period were the result of C. jejuni infection (140). A separate and larger study
of 229 GBS patients showed that 23% (53/229) of GBS patients had a prior C. jejuni
infection (58). Eight percent and 2% of patients had cytomegalovirus and EBV
respectively (58).

3

Other less common causes of GBS include surgery, head injury, and
vaccinations (26, 148). One study demonstrated that a patient presented with clinical
features of GBS one week after a serious head injury. Pathological studies showed
peripheral nerve demyelination, confirming that the patient had GBS (142). Other
reports of GBS following head injury have been cited in the medical literature; patients
commonly present with increased levels of antibodies reactive to myelin proteins (37,
95, 142). GBS has also been shown to develop after administration of vaccines (147). A
meta-analysis was performed for the 2009 H1N1 flu vaccination and association of
developing GBS. There was a 1.6 per 1 million rate of vaccinated individuals developing
GBS (129).
The mechanism of GBS is thought to be molecular mimicry between antibodies
that are produced during a preceding infection and neural epitopes (7, 63, 148).
Specifically, it is thought that the lipo-oligosaccharides (LOS) on the surface of C. jejuni
cross react with gangliosides localized in the nervous tissue (7, 77). The terminal glycan
portion of the C. jejuni LOS is highly variable and is thought to be the region of the
bacterium that mimics gangliosides (63). Gangliosides are sialylated glycosphingolipids
highly enriched in the nervous tissue (7, 15, 16, 84, 162). They are made up of a
ceramide tail inserted into the phospholipid bilayer and a highly variable sialylated
oligosaccharide tail that is exposed extracellularly (7, 84, 160). Over 100 different
gangliosides have been identified to date; they play many roles in cell growth and
differentiation and maintenance of stability of paranodal junctions (7, 160).
Association of anti-ganglioside antibodies with GBS was first reported in 1988 by
Ilyas et al. in 5/26 patients who presented with anti-ganglioside antibodies (74). Anti-

4

ganglioside antibodies are highly associated with GBS in patients; approximately 60%
of patients present with anti-ganglioside antibodies during the acute phase of the
disease (84, 157, 168). Antibodies reactive to gangliosides GM1 and GD1a have been
shown to disrupt the axons, lipid raft, and ion channels along the peripheral nerves
(160). It has also been shown in multiple in vivo studies that anti-GQ1b antibody binds
to the neuromuscular junction in patients resulting in complement activation (117, 151,
152). One study examined the association of the complement system and its role in
damage to the nerves by exposing neuromuscular junctions (NMJ) of the diaphragm to
anti-disialoside antibody, antibody that binds to portions of gangliosides GQ1b and
GT1a. The study saw deposits of antibody on the NMJ that resulted in injury that was
observed histologically (62). They also observed a correlation with nerve damage and
the presence of the membrane attack complex of the complement system, but showed
no nerve damage occurred in mice that lacked the complement system therefore
reinforcing its role in nerve damage (62).
A set of criteria has been established to verify that molecular mimicry is the
mechanism producing an autoimmune disease (7). Thus, the criteria for molecular
mimicry in GBS include: 1) establishment of an epidemiological association between the
immune-mediated disease and an infectious agent, 2) identification of T cells or
antibodies directed against host tissue, 3) identification of a component of the infectious
agent mimicking the target antigen, and 4) reproduction of the disease in an animal
model (7, 160). It has been shown by Ang et al. (2004) that GBS fulfills these criteria
and therefore occurs as a result of molecular mimicry (160).

5

Patients can spontaneously recover from GBS or may need treatment; treatment
options include plasma exchange (PE), intravenous immune globulin (IVIG), and
complement inhibitors (7, 72). PE was first used as a treatment in 1978 on a GBS
patient and was successful in alleviating the symptoms. In 1985 PE became the primary
treatment method for GBS (134). The success of PE is thought be to a result of
removing anti-ganglioside antibodies that attack the nerves from the plasma of patients
(116, 134). IVIG has also been successful in the treatment of GBS patients. It is thought
that IVIG neutralizes the IgG autoantibodies that are produced during the disease (116,
134). Complement inhibitors have also been examined as a possible treatment. It has
been demonstrated that the membrane attack complex (MAC) formed during
complement cascade activation plays a role in the damage seen in nerve tissue (116).
Markers for the MAC have been identified in the sera of GBS patients and have been
found in the affected neurons with its associated Schwann cells (116). Inhibitors of
components of the complement cascade may become a treatment option in the future.
Steroids have also been examined as a treatment option but were found to be
unsuccessful (116).

6

SUBTYPES OF GBS

Acute Inflammatory Demyelinating Polyneuropathy
Acute inflammatory demyelinating polyneuropathy (AIDP) is the most common
subtype of GBS in the Western world; Europe and North America have 69–90% of AIDP
cases (17, 72, 134, 147, 157, 163). AIDP patients present clinically with flaccid paralysis
and areflexia that progress for 2–4 weeks with mild sensory changes (134, 147, 148,
157). Patients with AIDP more commonly have autonomic nerves affected, experience
facial weakness (71% of patients), and need mechanical ventilation (134, 147, 148).
Anti-ganglioside antibodies are not commonly associated with AIDP, but antibodies to
ganglioside GD2 have been linked with this syndrome (116, 153). The pathogenesis of
this subtype includes inflammatory infiltrates in nervous tissue, primarily macrophages
and T cells associated with peripheral nerve demyelination. In AIDP, axons are
generally not affected (17, 116, 147, 150, 163). Complement activation has also been
shown to play a role (116, 163). Histopathological studies using human specimens have
shown that complement activation in the Schwann cell membrane is important in the
development of GBS (60). Three patients that suffered from this subtype of GBS were
autopsied, nerves removed, and immunohistochemistry was performed. It was observed
in all autopsies that markers for the complement cascade were localized along the outer
surface of the Schwann cells, therefore suggesting complement is activated by antibody
bound to the nerve and results in damage to the myelin (60). Nerve conduction studies
showed that demyelination occurs but re-myelination can occur as well as the disease

7

wanes. Furthermore, Schwann cells have been observed to migrate to the site of the
lesions to initiate re-myelination of neurons (134, 150).

Acute Motor Axonal Neuropathy (and Acute Motor Sensory Axonal Neuropathy)
Acute motor axonal neuropathy (AMAN) is a subtype of GBS prevalent in China,
Japan, South and Central America, and India; 38–65% of GBS cases in these regions
are AMAN and 22–46% are classified as AIDP (134, 147, 157, 163). In North America,
approximately 10–15% of GBS cases are classified as AMAN variants (17, 147, 148,
157). C. jejuni is the leading antecedent infection associated with AMAN (116, 147).
AMAN presents clinically with abrupt motor weakness that can progress. Recovery
begins within 2–3 weeks and complete recovery is often achieved. Also, nerve
conduction studies showed that recovery from AMAN is quicker than recovery from
AIDP (17, 134). Antibodies to gangliosides GM1, GM1b, GD1a, and GalNAc-GD1a are
all highly associated with AMAN with reactivity to GM1 and GD1a the most commonly
associated antibodies (77, 134, 153). The pathogenesis of this subtype is an immune
attack on the axon of the neuron rather than the Schwann cells and myelin (17, 154,
157). Complement and IgG bind to the axolemma of the motor nerves at the nodes of
Ranvier which results in MAC formation and subsequent axonal damage; demyelination
rarely occurs (163). Using a rabbit model that has been immunized with a ganglioside
mixture including ganglioside GM1, it was shown that immunoglobulins were deposited
on the axons of the anterior roots (167). Moran et al. demonstrated that anti-GM1
antibodies induced from C. jejuni LOS bind to the nodes of Ranvier in humans,
+

+

disrupting Na and K channels and resulting in poor nerve conduction (103).

8

A GBS subtype similar to AMAN is acute motor sensory axonal neuropathy
(AMSAN), in which damage involves both motor and sensory peripheral nerves (17).
Onset is abrupt, as in AMAN, and many patients require mechanical ventilation within
days of onset (17). Incidence of AMSAN is less than 10% of all AMAN cases, making it
the most uncommon subtype of GBS (154, 157).

Miller Fisher Syndrome
Miller Fisher Syndrome (MFS) is commonly found in East Asia, where 20–25% of
cases present as MFS, compared to 1–5% in Western countries (134, 163). MFS was
first described in 1956 by C. Miller Fisher, who examined patients who had symptoms
similar to those seen with GBS but also presented with ophthalmoplegia, ataxia, and
areflexia (153). Today, MFS is clinically characterized by this triad of symptoms:
areflexia, ataxia, and ophthalmoplegia, though not all symptoms may present (17, 93,
134, 137, 157, 162). Unlike GBS, MFS presents as a descending paralysis that affects
both the central and peripheral nervous system (17, 139). The disease usually
progresses for 1 week after the first onset of symptoms, and improvement begins within
2 weeks. The average time for full recovery is 1–3 months in most patients (163).
Reactivity against ganglioside GQ1b is strongly associated with MFS; 83–100% of MFS
patients present with anti-GQ1b antibodies (17, 72, 77, 93, 137, 157). Immunostaining
assays have shown that ganglioside GQ1b is highly enriched in the human cranial
nerves that control the extrinsic eye muscles, including the 3
(trochlear), and 6

th

rd

(oculomotor), 4

th

(abducens) cranial nerves (93, 116, 153, 154). Presence of anti-

9

GQ1b is a good diagnostic marker for MFS because very few of these patients lack
these antibodies (93).

10

CAMPYLOBACTER JEJUNI

In the 1970’s Campylobacter jejuni was identified as a foodborne pathogen (68).
C. jejuni is a gram negative curved rod bacterium with a unipolar or bipolar flagella (7,
75, 91). It is the leading cause of foodborne bacterial gastroenteritis in the developing
world (7, 31, 63, 75, 162). In the United States the annual incidence of C. jejuni infection
is approximately 1–2.4 million infections (44, 68, 91, 131). Symptoms of C. jejuni
infection include fatigue, abdominal cramps, nausea, watery diarrhea and bloody
diarrhea (102). Mortality rates are low for campylobacterosis in the United States; it is
estimated that approximately 124 deaths result annually (44). Studies have shown that
infection of C. jejuni is the highest during early Fall and that all age groups can be
affected (91). In some instances C. jejuni infection can also be associated with postinfectious syndromes such as Guillain Barré Syndrome, Reiter’s Syndrome,
Inflammatory Bowel Disease, and Irritable Bowel Syndrome (44, 68, 75, 87, 125). C.
jejuni can colonize the gastrointestinal tracts of both wild and domesticated animals, but
rarely cause clinical disease in food animals (pigs, cows, chicken, etc) (68). Therefore,
routes of infection with C. jejuni include contaminated meat (mainly poultry),
contaminated water, and unpasteurized milk (75, 112, 156).
It is well recognized that C. jejuni is highly associated with the development of
GBS, specifically the AMAN subtype. However, only select strains of C. jejuni have the
ability to mimic gangliosides (38). The genes involved in LOS biosynthesis have been
identified in many sequenced strains and strains not sequenced (47, 112). The genetic
mechanisms that result in the highly variable LOS have also been identified (47, 112).

11

GBS patients can be co-infected with multiple strains of C. jejuni but only one of the
strains may have the ability to mimic gangliosides (38). Some strains have a higher
frequency of association with the development of GBS as a result of the presence or
absence of certain genes in the LOS biosynthesis loci (7). Other genetic factors such as
the flagella or major outer membrane protein have been examined to identify a genetic
relationship of C. jejuni strains associated with GBS, but have yielded contradicting
results (27, 35, 36).

Lipo-oligosaccharides (LOS)
The LOS is a surface molecule found on some gram negative bacteria in place of
the larger lipopolysaccharide (LPS) structures found on such organisms such as
Salmonella (112, 125). LOS is anchored to the bacterial outer membrane by insertion of
the lipid A tail into the membrane, with the highly variable region—the oligosaccharide
portion—extending into the extracellular space (112). Variability of this outer membrane
component is a result of the diverse monosaccharide components and linkages
between them that make up the structure of the LOS (112). Mechanisms that result in
the high variability of the LOS include: 1) lateral gene transfer, 2) phase variation due to
homopolymeric tracts, and 3) gene inactivation, deletion, duplication, or fusion (31, 51,
125).
The LOS has been classified into 23 different classes (A–W) based on gene
content and organization in the complex LOS biosynthesis locus; in C. jejuni 11168, this
locus occupies the open reading frames (ORFs) from Cj1131 to Cj1151 (46, 47, 63).
Initially the LOS was classified into 3 classes (A–C), class B was interpreted as an

12

evolutionary intermediate between class A and C (47). Classes A and B may also be
further classified by allele type (A1/B1 or A2/B2) based on presence the cgtA (orf5abI)
and cgtB (orf6ab) genes (112). Parker et al. (2005) further organized the LOS
biosynthesis loci, resulting in 8 classes (A–H) and eventually identified an additional 11
classes (A–S) (63, 111, 112). Classes T–W have also been identified recently by
Stanhope and colleagues (125). In the most recent organization, classes A, B, C, M, R,
and V all have the necessary genes needed to synthesize a ganglioside mimicking LOS
(51, 63, 111). These LOS classes all have the genes neuBCA that encode for the
biosynthesis of sialic acid and the cst gene (cstII for classes A, B, M, and R; cstIII for
class C), that encodes sialyltransferase needed to sialylate the LOS (63, 111, 112, 124,
125). All other classes of the LOS lack the genes necessary for an LOS structure that
may mimic gangliosides; they lack the sialic acid biosynthesis genes and a
sialyltransferase.
LOS class A C. jejuni strains are more commonly associated with GBS while C.
jejuni strains with class B LOS are more common in MFS patients (47, 63, 124). A study
by Parker et al. (2005) examined the LOS biosynthesis loci of 16 GBS isolates and
concluded that 14/16 were classified as LOS class A1 and 2/16 as LOS class allele A2
(112). Another study by Godschalk et al. (2004) focused on examining the LOS loci of
17 GBS strains and found that 53% of these C. jejuni GBS associated strains had class
A LOS and 18% belonged to a class unable to mimic gangliosides (49). They postulated
that this observation could be the result of a unique sialyltransferase that allows the
LOS structure to mimic gangliosides or that these GBS patients were co-infected with
multiple strains of C. jejuni, some having the ability to mimic gangliosides and others

13

not, but culture only resulted in the isolation of a single strain (49). Godschalk and
Parker in two different studies determined that 62% and 64%, respectively, of the C.
jejuni strains that only caused enteritis were classified as LOS classes A–C. This result
suggests that presence of ganglioside mimicking epitopes on the LOS of C. jejuni may
be advantageous to the bacterium in gastrointestinal colonization and infection (49,
111). Contradictory to this suggestion, an in vitro study by Guerry et al. (2002) showed
C. jejuni strains that lacked the cgtA gene (N-acetylgalactosaminyl (GalNAc)
transferase) had a higher rate of invasion in intestinal epithelial cells (56).
An example of variation within the LOS is a polymorphism that occurs on the
gene that encodes a sialyltransferase, cstII (134). Of the 291 amino acids in the gene,
amino acid 51 determines its enzymatic activity (134). If the amino acid is threonine
(Thr51) then the enzyme encodes a mono-functional α-2,3, sialyltransferase which
commonly results in gangliosides associated with GBS (90). If the amino acid produced
is asparagine (Asn51) the cstII is bi-functional, encoding an enzyme with both α-2,3,
sialyltransferase and α-2,8, sialyltransferase activity, resulting in a LOS structure that
more commonly mimics the GQ1b ganglioside associated with MFS (63, 90).
Recently, work has been performed with C-Dps, which is a protein produced by
C. jejuni in response to levels of nutritional or oxidative stress (114, 115). Two studies
by Piao et al. (2010 and 2011) have shown that C-Dps binds to sulfatide, a neutrophil
glycosphingolipid that is needed for paranodal junction formation and maintenance of
ion channels along the axons. Once bound it results in demyelination of the myelin
sheath and axonal degeneration (114, 115). Using immunohistochemistry, he was able
to demonstrate both in vitro and in vivo that C-Dps binds to the nodes of Ranvier, outer

14

most portion of the myelin sheath and basement membranes of the peripheral nerves
(115). Using cell culture, he was able to show further binding of C-Dps to the anterior
horn cells in the spinal cord grey matter and to myelin sheath in the white matter (114).
These findings suggest that C-Dps should be analyzed further in its role in the
pathogenesis of GBS.

Molecular Typing of C. jejuni
The most common approach used to classify C. jejuni strains is multi-locus
sequence typing (MLST), which is based on partial sequences of seven housekeeping
genes. MLST has been used to characterize the source of C. jejuni infections (124).
There is an accessible website that allows assignment of sequence types and
examination of the different clonal complexes based on sequence of their MLST alleles
(http://pubmlst.org/campylobacter). Studies have shown that clonal complexes (CC) CC
21 and CC 45 are the most common assigned to C. jejuni isolates (124). A study was
done to examine 335 C. jejuni strains that included human, bovine, and poultry isolates
for correlation of their clonal complex assigned based on MLST and their LOS class (A–
H) (124). Results showed that LOS class A1 was highly associated with CC 22 (27/27),
CC 508 (2/2), CC 1034 (1/1) and CC 1332 (4/4) while LOS class A2 was associated
with CC 61 (7/7). Two of two LOS locus class B1 strains were in the CC 42, while two of
two B2 strains were in the CC 58 and 14/15 B2 strains were in CC 48. In this study, the
LOS class E, which is not sialylated, was completely associated with complexes CC 677
and CC 283 (124). Therefore, these results suggest that the combination of MLST and
LOS classification may be useful approach for predicting disease outcomes.

15

Other typing schemes used to classify and analyze GBS-associated C. jejuni
strains involve examination of the major outer membrane protein, porA, and the major
flagella protein, flaA. It has been demonstrated that a highly variable region termed,
short variable region (SVR) in the flaA of C. jejuni is a molecular typing tool useful to
compare C. jejuni isolates (101, 102). The SVR is only 321 base pairs in length and
therefore provides a quick and reliable typing scheme (102). Comparisons of the SVR in
C. jejuni strains are commonly used to examine the different C. jejuni strains in a
population (102). One study in Curacao performed multiple molecular typing methods to
identify the spread of clonal C. jejuni including MLST and sequencing of the flaA SVR
(36). The study examined 234 isolates of C. jejuni from patients that presented with
gastroenteritis, 2 of which also developed GBS. It was performed from March, 1999, to
March, 2000, and based on the flaA SVR 11 different types were identified; MLST
identified 29 sequence types (ST). The 2 C. jejuni isolates from the GBS patients were
in two different sequence types for both MLST and flaA SVR. There was a poor
correlation of the MLST typing with the flaA SVR typing together; each ST identified with
MLST had flaA SVR sequence with variation, thus supporting the high variability in the
flaA (36). A study by Tsang et al. (2001) proposed that a flaA SVR sequence type was
highly associated with patients that developed GBS (144). They compared their isolates
to GenBank®, and based on their results concluded that isolates of C. jejuni that result
in neuropathy do cluster together based on their flaA SVR sequence types (144). They
also identified that 3 isolates from their study were identical in flaA SVR type and found
it unusual since each isolate was from a different geographical part of the world (Japan,

16

Chile, and America). Based on these data, the authors suggested that this specific flaA
SVR sequence type could be a marker for GBS (144).
Dingle et al. (2001) examined the typing of 25 C. jejuni isolates from both GBS
and MFS patients with MLST and flaA SVR typing; they also wanted to examine the rate
of the proposed flaA SVR sequence type described above in his isolates (35). They
found 20 different sequence types that were able to group into 13 different clonal
complexes, though they did note that the second most common clonal complex of C.
jejuni, CC 45, was not present in any of the 25 isolates suggesting that isolates
associated with CC 45 are less likely to develop neuropathy. The sequencing of the flaA
SVR identified 20 distinct sequences and only 4 of the isolates had the flaA SVR
sequence that was a proposed marker for GBS (35). The lack of the proposed flaA SVR
sequence to the GBS isolates conflicts with the results of Tsang et al. and therefore
more work would need to be done to understand this discrepancy.
The other proposed typing scheme involves using the porin that is the major C.
jejuni outer membrane protein as a target for rapid classification of strains. This porin is
a hypervariable region of C. jejuni (27). Clark et al. (2007) sequenced the porA gene of
105 C. jejuni isolates to examine if any correlation of the porA sequence was present
during development of GBS. Based on their sequence results they identified 3 different
sequence types but no correlation to development of GBS was noted with any of the
sequences (27).
Both the flaA SVR and porA have been isolated and sequenced from C. jejuni
from patients who presented with either enteritis only or enteritis and neuropathy, and
based on findings there is contradicting evidence of whether molecular typing of the flaA

17

SVR or porA show any association with the development of GBS (35, 144). Therefore,
we hypothesized that assigning C. jejuni strains based on MLST, porA sequence, and
flaA SVR sequence together could result in correlations between GBS-associated
strains and C. jejuni associated with enteritis only based on LOS classification.

18

ANIMAL MODELS

Currently there are few animal models used to study the natural development of
GBS. Rabbits provide the current model of the GBS subtype, AMAN. Previously, Yuki et
al. provided proof for a rabbit model by immunizing rabbits subcutaneously with either
2.5 mg of a mixture of gangliosides or 1 mg of GM1 isolated from bovine brain. At 3
week intervals, rabbits were injected intraperitoneally with the same cocktail until limb
weakness was observed; if no limb weakness occurred, rabbits were observed for 6
months and euthanized (167). All of the rabbits given the mixture of gangliosides
developed flaccid paralysis within 35–57 days and antibodies to GM1 were observed 2–
3 weeks after the first injection. Studies of the sciatic nerves from these rabbits showed
mild to severe Wallerian-like degeneration with infiltration of macrophages. Nine of
eleven rabbits given isolated GM1 extracted from bovine brain developed flaccid
paralysis and had presence of anti-GM1 antibodies as determined by enzyme-linked
immunosorbent assays (167). Sciatic nerves from these rabbits showed mild to severe
Wallerian-like degeneration with no presence of lymphocytes (167).
More recently, Yuki et al. immunized two groups of rabbits with C. jejuni LOS
known to mimic ganglioside GM1; this LOS preparation was from a C. jejuni isolate of a
GBS patient that presented with antibodies to ganglioside GM1 (165, 166). Group 1 had
rabbits immunized at 3 week intervals with 2.5 mg of C. jejuni LOS and group 2 rabbits
were immunized at the same time points but received 10 mg of LOS (165). After
immunization, the rabbits in both groups 1 and 2 exhibited clinical, electrophysiological,
and histopathological features that are common in AMAN. The rabbits in group 1 began

19

to develop flaccid paralysis between 133 to 329 days after the first inoculation. Group 2
of rabbits developed tetraparesis 40 to 227 days after the initial inoculation of LOS.
Rabbits in both groups 1 and 2 developed anti-GM1 IgM and anti-GM1 IgG antibodies
2–4 weeks and 4–6 weeks, respectively, after inoculation (165). Histological studies
showed macrophage infiltration in the nerve roots of all rabbits that exhibited any type of
paralysis. In the sciatic nerve, Yuki observed Wallerian-like degeneration with little to no
demyelination. He also observed no T cell infiltration but did see IgG bound to the axons
(165). Thus, despite documenting some new observations, these rabbit models are not
a natural model of the pathogenesis of GBS.
Another model is experimental autoimmune neuritis (EAN), which is a T cellmediated demyelinating disease in the peripheral nervous system that resembles GBS.
Peripheral nerve myelin, purified myelin protein P2 or T cell transfers (T cells are from
stimulated lymph node cells in vitro with P2) are used to immunize SJL/J mice to elicit
an anti-myelin response (20, 126). The clinical and histological presentations in these
mice are very similar to those of GBS (8, 163). Some mice given these preparations
developed lesions of the sciatic nerve that peaked at 40 days with infiltration of T cells in
the nerves observed. The mice display clinical and pathological features that resemble
the GBS subtype, AIDP (52, 126, 163). Lewis rats have also been used as an EAN
model; myelin proteins P0 or P2 along with Freund’s complete adjuvant are used to
immunize Lewis rats which have been shown to have T cell infiltration and macrophagemediated demyelination that is similar to that seen in GBS (20, 61). These EAN models
do not make a good GBS model however, because there is not conclusive evidence to
show that T cells occur in GBS patients (134, 163).

20

The only true model of GBS following C. jejuni infection is the chicken. To initiate
disease, chickens were inoculated by oral gavage with 5 ml of a C. jejuni isolate taken
from a boy who developed GBS (94). After inoculation, chickens were observed for
neurological signs and at endpoint, the sciatic nerves from both control and infected
chickens were removed for pathological analysis that included both histological
examination with hematoxylin and eosin staining for cellular infiltrates and Luxol fast
blue staining for myelin damage. Within 12 days post inoculation, chickens infected with
C. jejuni developed weakness and were not able to stand; some chickens were
paralyzed. Both wing and head droop were also observed. The earliest onset of
weakness was as early as day 5 post inoculation (94). At endpoint only the sciatic nerve
was removed for analysis. Histopathological studies of the sciatic nerve showed
Wallerian-like degeneration and varying degrees of lengthening of the nodes of Ranvier
with rare occurrence of paranodal demyelination. The study notes the most severe
Wallerian-like degeneration of the sciatic nerve was observed in a chicken that became
weak 18 days post inoculation. In some instances cellular infiltration of macrophages
could be detected around the axon of the sciatic nerve (94). The uninfected chickens
showed no Wallerian-like degeneration in any of their sciatic nerves.
Chickens however are very different physiologically and anatomically from
humans and therefore do not make good models of human GBS. For example, when
screening the model for new treatments one would prefer that the animal’s physiology
closely mimics that of humans so that drug distribution and clearance would be
analogous to those in the human. Therefore, we hypothesized that a mouse, which is
physiologically similar to humans, could be a model of GBS secondary to C. jejuni

21

infection. This would offer an animal model that can be used to study the natural course
to the development of GBS.

22

RATIONALE

The role of C. jejuni in the pathogenesis of GBS is still not well understood. More
comparative in vivo studies need to be performed to understand better the role of
molecular mimicry in the development of GBS. However the current animal models
used to study GBS are either not natural models of the course of development of the
disease or are physiologically and anatomically different from humans (94, 158, 167).
Therefore, a murine model that develops GBS secondary to C. jejuni infection is
needed. Chapters 2 and 3 of this dissertation describe progress towards such a model.
Analysis of the role of LOS in molecular mimicry has revealed conflicting data
that suggest that other genetic factors of C. jejuni may contribute to the development of
GBS. Epidemiological studies have been performed to collect isolates from patients with
defined clinical presentations to study the genetic relationship of enteric and
neuropathic strains of C. jejuni. The flagella and major outer membrane protein (MOMP)
have each been examined in these epidemiological studies because both molecules are
located on the outer surface of the C. jejuni cell (27, 35). Studies have associated
development of GBS with particular C. jejuni serotypes (120) or with strain clusters
identified by amplified fragment length polymorphism sequence typing of the SVR of the
flaA gene, or sequence typing of the porA gene encoding the MOMP (27, 144). These
approaches have not yielded definitive molecular markers for strains likely to produce
GBS (27, 35, 36, 144); therefore, additional work to improve molecular typing is needed.
Chapters 4 and 5 of this dissertation describe molecular genetic typing methods
targeting both conserved and variable loci of C. jejuni that were applied to strains

23

collected in two epidemiologic studies. Interestingly, both studies demonstrate that the
LOS loci of many of the C. jejuni isolates demonstrated characteristics similar to those
of C. jejuni strains associated with GBS. These GBS-like strains were found both in the
general U.S. population with focus on the Michigan population (Chapter 4) and
associated with diarrhea in calves on a Michigan dairy farm (Chapter 5).
Therefore, based on the stated hypotheses, I pursued the following specific aims
in this dissertation:
Specific Aim 1: Assess the presence of autoimmune sequelae following infection of
C. jejuni in Non-Obese Diabetic (NOD) mice based on neurological signs and
presence of nerve lesions

Specific Aim 2: Correlate genetic differences between the LOS biosynthesis loci
and other variable outer surface structures of C. jejuni strains with potential to elicit
neuropathy

24

CHAPTER 2

Non-Obese Diabetic (NOD) mice as working models of Guillain Barré Syndrome
following Campylobacter jejuni infection

25

ABSTRACT

Campylobacter jejuni is a leading cause of bacterial gastroenteritis linked to
several serious autoimmune sequelae including Inflammatory Bowel Disease, Irritable
Bowel Syndrome, Reiter’s Arthritis, and Guillain Barré Syndrome (GBS). GBS is an
acute peripheral neuropathy characterized by limb weakness and loss of tendon reflex.
C. jejuni strains from GBS patients have lipo-oligosaccharides (LOS) structures that
mimic gangliosides found on the peripheral nerves. The GM1 and GD1a gangliosides
are most commonly associated with acute motor axonal neuropathy (AMAN), the form
of GBS that follows C. jejuni infection. We hypothesized that GBS can result in NOD
mice or their congenic IL-10 and B7-2 knockouts secondary to C. jejuni infection and
that clinical signs and pathological changes can be detected. To address this
hypothesis mice were gavaged orally with C. jejuni strains HB93-13 and 260.94 from
patients with GBS. Mice were assessed for clinical neurological signs, anti-ganglioside
antibodies, cellular infiltrates, and lesions in peripheral nervous tissue. Infected mice of
the NOD WT and the NOD IL-10

-/-

genotypes produced anti-ganglioside antibodies that

were of the IgG1 isotype directed against a mixture of gangliosides GM1 and GQ1b.
Significant increases in anti-ganglioside antibodies against single gangliosides (GM1
and/or GD1a) occurred in infected mice of all genotypes. Phenotypic tests showed
significant differences between treatment groups of all genotypes. Peripheral nerve
lesions with macrophage infiltrates were significantly different between the shaminoculated mice and C. jejuni infected mice of both the NOD WT and NOD IL-10

26

-/-

genotypes. Based on these parameters, we conclude that NOD WT mice are the best
candidate for a potential mouse model of natural onset GBS.

27

INTRODUCTION

Guillain Barré Syndrome (GBS) is a post-infectious polyradiculoneuropathy
considered to be an autoimmune disease (116). GBS is defined as a group of several
disorders that are differentiated by particular forms of immune attack on the peripheral
nervous system. Subtypes of GBS include acute inflammatory demyelinating
polyneuropathy (AIDP), acute motor axonal neuropathy (AMAN), and Miller Fisher
Syndrome (MFS). AIDP is the most common subtype; occurring in approximately 90%
of GBS cases in North America and Europe (134, 163). AIDP presents as segmental
demyelination affecting mainly the myelinated nerves of the limbs and lower cranial and
sensory nerves. AMAN is a form seen in China, Japan, and Central and South America
with a frequency of 38–65% (134). In patients, AMAN presents as damage to the motor
neurons at the nodes of Ranvier. A related form—acute motor sensory axonal
neuropathy (AMSAN) — has also been recognized where patients are affected in both
motor and sensory peripheral nerve function (108). MFS is commonly seen in eastern
Asia; at a rate of 20–25% (134, 163). MFS is an anatomically localized variant of GBS
characterized by the triad of symptoms areflexia, ataxia, and ophthalmoplegia (17, 137).
In contrast to GBS, MFS causes descending paralysis because different nerve groups
are affected (139).
GBS is described as a monophasic disease in humans, with the most severe
disease occurring 2–4 weeks after first onset of symptoms; however, 7% of GBS
patients have reported reoccurrence (7, 163). Weakness is followed rapidly by
symmetrical ascending paralysis and, if progressive, the muscles of respiration are

28

paralyzed. The mortality rate is 3–10%; 10−20% of patients do not recover completely
and may suffer life-long disability (163). Current treatments of GBS include plasma
exchange and intravenous immunoglobulin (7, 72). Males are more likely to develop
GBS than females with a ratio of 1.25−1.5 males to 1.0 females; most autoimmune
diseases present with a higher ratio of females than males (72, 116).
Two-thirds of GBS patients have a history of prior infection, most commonly
respiratory or gastrointestinal infections (72, 116, 162, 163). Common bacterial and viral
antecedent infections of GBS include Campylobacter jejuni, Mycoplasma pneumoniae,
Epstein-Barr virus, and cytomegalovirus (7, 70, 116, 154, 157, 163). In 1993, Mishu and
Blaser recognized that C. jejuni infection can result in development of GBS (67, 105).
The Gram negative bacterium C. jejuni is the most frequent bacterial etiology for human
gastroenteritis with 1–2.4 million cases annually in the U.S. and is the most common
enteric infection associated with development of the AMAN form of GBS (19, 63, 162,
163, 165). The risk of developing GBS after C. jejuni infection is less than 1 GBS case
per 1000 C. jejuni infections in the U.S. (4, 5, 7). Neurological symptoms of the AMAN
subtype occur 1–3 weeks after the onset of C. jejuni induced enteric disease, although
some GBS patients are colonized without enteritis (157, 163).
In GBS, molecular mimicry is thought to play a role in the production of antiganglioside antibodies, which bind to and trigger attack on host nerve tissue (7, 63, 76,
121, 128, 134, 157, 160, 162, 165). The structure of lipo-oligosaccharides (LOS) of C.
jejuni can mimic nerve gangliosides in the host, eliciting an autoimmune response that
cross-reacts with gangliosides in the peripheral nerves (7, 107). It has been shown both
in humans and in animal models that the immune response to the bacterium results in

29

production of anti-ganglioside antibodies that leads to damage of nerves and Wallerianlike degeneration; such antibodies (“autoantibodies”) have been shown to bind to motor
neurons, nodes of Ranvier, and neuromuscular junctions (5, 71, 107, 145, 159).
Autoantibodies directed against gangliosides GM1 and GD1a have been detected most
commonly in patients with the AMAN variant of GBS, while autoantibodies against
GQ1b are more often associated with MFS (13, 40, 150). The peak titers of antiganglioside antibodies following C. jejuni infection have been associated with onset of
GBS symptoms in patients. An example of correlation between autoantibodies and
disease manifestations occurred during an outbreak of C. jejuni in 2007 when 36 cases
of GBS were identified in northern China in patients that produced high titers of antiGM1 autoantibodies. Bacterial fecal culture was performed, and C. jejuni was identified
as the antecedent infectious agent (169). Another study conducted in southeast
England during 1983–1984 examined serum samples of 95 GBS patients and 88 control
patients and demonstrated that 14/95 patients had high titers of antibodies reactive to
GM1 based on ELISA testing (149).
Despite these advances little progress has been made in understanding the
pathogenesis of GBS and even less in developing effective therapies, largely because
of the lack of accurate, tractable animal models. Currently, rabbits and chickens serve
as the induced animal models for GBS and other peripheral neuropathies. GBS is
induced in the rabbit model by immunizing rabbits with C. jejuni LOS known to mimic
particular gangliosides. Rabbits given a dose of 2.5 mg C. jejuni LOS subcutaneously at
3 week intervals developed flaccid limb weakness between 133 to 329 days after the
initial inoculation, while rabbits treated similarly with 10 mg LOS developed tetraparesis

30

40 to 227 days after the initial inoculation (165). Anti-GM1 IgM and anti-GM1 IgG
antibodies were detected 2–4 weeks and 4–6 weeks, respectively, after the first
injections (135, 165). Although LOS exposure elicits anti-ganglioside reactivity,
immunized rabbits are not a precise model for GBS induced following oral C. jejuni
infection. Alternatively, chickens acquire a form of peripheral neuropathy secondary to
oral inoculation with C. jejuni GBS patient strains. Thirty-three percent of chickens
receiving C. jejuni HB93-13 became paralyzed within 12 days (94). In paralyzed
chickens, early lesions included nodal lengthening and paranodal demyelination that
were later followed by Wallerian-like degeneration and even paranodal re-myelination in
some long-term survivors. Thus, chickens inoculated orally with GBS associated C.
jejuni strains from patients can be considered a naturally occurring model for GBS.
However, chickens are outbred and very different anatomically and physiologically from
humans; these facts detract from their value as a model. Thus, a murine model is
needed to study GBS pathogenesis, including the roles played by both host and
pathogen genetics. Susceptible inbred mice would provide the ideal model where the
interplay of host and pathogen genetics could be explored.
Non-Obese Diabetic (NOD) inbred mice have been documented to develop
autoimmune diabetes mediated by auto-reactive T cells infiltrating the pancreas (66).
They also develop autoimmune diseases of the salivary glands and thymus (13).
Additionally, NOD mice deficient for the co-stimulatory molecule B7-2 (NOD-B7-2

-/-

mice) are largely protected from autoimmune diabetes but develop a spontaneous
autoimmune peripheral polyneuropathy that resembles GBS called chronic inflammatory
demyelinating polyradiculoneuropathy (CIDP) (13, 130). By 20 weeks of age, peripheral

31

nerves of neuropathic NOD-B7-2

-/-

+

mice have infiltrates of dendritic cells and CD4 and

+

CD8 positive T cells. Based on knockout of genes in mice of the same genetic
background, these investigators showed that neuropathy developed in the absence of
perforin or fas proteins, suggesting that classic cytotoxicity pathways for nerve damage
were not operating. Stronger evidence for the autoimmune basis of this neuropathy was
provided by experiments which showed that the neuropathy was transferable to naïve
+

NOD.SCID mice by CD4 T cells isolated from neuropathic animals (13, 130). Neither of
these mouse models has been used to explore the effects of natural infection with C.
jejuni in the initiation of clinical GBS or classical GBS lesions with Wallerian
degeneration of the peripheral nerves.
We hypothesized that NOD mice or their congenic IL-10 and B7-2 knockouts
develop spontaneous GBS secondary to C. jejuni infection and that neurological signs
and pathological lesions can be detected by established methods such as neurological
evaluation, enzyme-linked immunosorbent assays (ELISAs) for anti-ganglioside
antibodies, and histological examination for Wallerian degeneration and inflammatory
infiltrates. The strains of C. jejuni we used, C. jejuni HB93-13 and C. jejuni 260.94, were
isolated from patients who presented with GBS. It has been shown that both strains
mimic the GM1ganglioside; C. jejuni HB93-13 also has been shown to mimic the GD1a
ganglioside (2). Our rationale for using NOD mice and their congenic knockouts is
based on the known propensity of mice of these genotypes to develop autoimmune
diseases. These NOD mice exhibit autoimmune diseases due to defects in the activity
of natural killer cells, cytokine production from macrophages, and regulation of adaptive

32

T cell populations (3). Furthermore, we previously established that NOD WT mice are
colonized by C. jejuni and that NOD IL-10

-/-

mice are colonized and develop enteritis in

a reproducible fashion after challenge with a pathogenic strain of C. jejuni (99). Also,
minimal enteric disease was observed in NOD WT and NOD B7-2
C. jejuni strains tested compared to the NOD IL-10

-/-

-/-

genotypes for both

mice. There were significant

differences between the sham-inoculated mice and the C. jejuni infected mice in the
presence of IgG1 antibody to the ganglioside mixture GM1/GQ1b in the NOD WT and
NOD B7-2

-/-

-/-

mice. NOD WT, NOD IL-10 , and NOD B-7

-/-

mice had significant

differences in plasma levels of anti-GD1a IgG3 antibody. Mice of all genotypes also had
significant differences between the sham-inoculated mice and infected mice in one or
more phenotypic tests. Based on our findings in long term time course studies with
extensive neurological phenotype testing, we conclude that NOD WT mice have the
greatest potential to be a natural murine model for GBS.

33

MATERIALS AND METHODS

Animals
NOD mice, along with their congenic IL-10 and B7-2 knockouts, were originally
purchased from The Jackson Laboratory (Bar Harbor, ME, USA), and a breeding colony
was established as described previously (98). For experimental work, mice were
housed in individual filter top sterile cages at the Michigan State University Research
Containment Facility (URCF). Cages were changed every 14 days, and mice were fed
sterilized mouse diet 7913 (Harlan Teklad) and given autoclaved reverse osmosis
water. Prior to inoculation, fecal samples were taken and tested for presence of
colitogenic bacteria including Campylobacter spp., Helicobacter spp., Citrobacter
rodentium, Enterococcus faecalis, and E. faecium by PCR. A modified PCR assay from
Jackson Laboratories was used to confirm the genotypes of all experimental mice
(http://jaxmice.jax.org/pub-cgi/protocols/protocols.sh?objtype=protocol
&protocol_id=346). DNA extraction from ear punches was performed using DNeasy
Blood and Tissue Kit (Qiagen, Hilden, Germany) or from fecal pellets according to
Murgatroyd et al., with modifications (104). Mice were inoculated with C. jejuni strains
between the ages of 7–10 weeks. This age range was used to enable the experiments
to be performed before the age at which NOD B7-2

-/-

mice begin to develop

spontaneous neuropathy. It is known that B7-2 knockout mice begin to develop mild
clinical signs of spontaneous neuropathy no earlier than 20 weeks of age (Gadsden,
unpublished data) (13, 130). After inoculation, mice were monitored daily for clinical
signs of enteric disease and/or neurological disease. All animal experiments followed

34

NIH guidelines and were approved by the MSU All University Committee on Animal Use
and Care under protocol numbers 04/07-030-00 and 06/09-092-00.

Campylobacter jejuni
C. jejuni strains, C. jejuni 260.94 (ATCC BAA-1234) and C. jejuni HB93-13
(ATCC 700297) used in these studies were purchased from the American Type Culture
Collection (ATCC, Manassas, VA, USA) and were isolated from patients with GBS. The
growth conditions of C. jejuni inocula, confirmation of the spiral morphology and motility,
and mouse inoculation methods were previously described (98).

Experimental Design
Three experiments were conducted to address our hypothesis that C. jejuni
patient strains can elicit peripheral neuropathy in mice of the NOD genotype. In all three
experiments, post inoculation testing was similar, but improvements were made with
each experiment. Mice were subjected to a series of weekly phenotypic tests to detect
any signs of neurological disorder as explained in further detail below. In Experiments 1
and 2 all mice were bled from the saphenous vein prior to inoculation to collect plasma
to test for anti-ganglioside antibodies; mice were also bled 14 days after inoculation and
at the endpoint. Mice were not bled prior to inoculation or at day 14 after inoculation in
Experiment 3 based on the non-significance of the data of Experiments 1 and 2 at these
time points.

35

Experiment 1: Screening mice of various NOD genotypes for GBS following C. jejuni
infection.
The first experiment screened 3 groups of NOD mice of different genotypes—
-/-

-/-

NOD wild type (WT), NOD IL-10 , and NOD B7-2 —with 30 mice in each group. Ten
mice from each group were infected with the GBS-associated C. jejuni strain HB93-13,
10 were infected with the GBS-associated C. jejuni strain 260.94, and the remaining 10
were sham-inoculated with tryptose soya broth (TSB; negative controls). At day 12 post
inoculation, fecal pellets were collected and cultured for the presence of C. jejuni to
confirm colonization. Mice were euthanized and necropsied 56 days post inoculation.
Samples of blood and sections of the GI tract were taken and examined for presence of
anti-ganglioside antibodies and C. jejuni colonization respectively; detailed methods are
given below.

Experiment 2: Screening larger group sizes of NOD WT and NOD B7-2

-/-

mice for GBS.

Sample size calculation based on experiment 1 results prompted screening of
larger sample sizes for further optimization. Experiment 2 was performed using only
NOD WT and NOD B7-2

-/-

mice and only C. jejuni 260.94. C. jejuni 260.94 was used
-/-

because NOD WT, NOD IL-10 , and NOD B7-2

-/-

mice infected with this strain in

Experiment 1 showed high colonization rates with little or no enteric disease. Each
mouse genotype had a group size of 36 with 18 mice in each group orally inoculated
with C. jejuni 260.94 and the remaining 18 mice in each group sham-inoculated with
TSB. Group size for this experiment was determined based on numbers of mice in

36

Experiment 1 showing neurological signs. The sample size (15–18 mice per treatment
group) was calculated based on a dichotomous study with two proportions
(independent, prospective) and an uncorrected chi-square test design with alpha=0.05
and power=0.80. Mice were euthanized and necropsied at an endpoint of 80 days post
inoculation. At necropsy, samples of the GI tract, sciatic nerve, and blood were taken to
determine C. jejuni colonization, nerve lesions, and presence of anti-ganglioside
antibodies respectively; detailed methods given below. A longer experimental endpoint
was used to see if more mice developed clinical neurological signs.

Experiment 3: Examination of NOD IL-10

-/-

mice using antibiotics to treat enteritis.

Experiment 3 included 4 treatment groups, all NOD IL-10

-/-

mice. Thirty mice

were given C. jejuni 260.94 alone, 20 mice were given C. jejuni 260.94 and then treated
with the antibiotic chloramphenicol (2.5 mg/ml in drinking water for 26 days beginning at
day 12 post infection), 20 mice were given TSB, and the remaining 20 mice were given
TSB and then treated with chloramphenicol as above. Mice were euthanized and
necropsied at 84 days post inoculation. Sections of the GI tract, blood, and the sciatic
nerve from ankle to spinal nerve root were removed to determine C. jejuni colonization,
presence of anti-ganglioside antibodies, and nerve lesions with presence of cellular
infiltrates respectively; detailed methods are given below.

Neurological Phenotype Testing
Neurological tests were performed on a weekly basis to detect any signs of
neurological damage in mice infected with C. jejuni strains compared to sham

37

inoculated controls. Mice were introduced to the tests prior to bacterial inoculation to
remove any effects of a learning curve (22). Tests used included the open field test
(OFT), hang test, rotarod test, foot printing and tracking, and DigiGait™ analyses.
Experiment 1 included the use of the OFT, hang test, and foot printing. Experiment 2
included those tests in experiment 1 with the addition of the rotarod. Experiment 3 only
utilized the rotarod and DigiGait™.

Open field testing
The open field test, which allows a trained observer who is blinded to detect
abnormal gait and ataxia, was performed using a large clear polycarbonate box with 4
quadrants marked on the bottom. Mice were placed at the center and observed for 30
seconds. Parameters recorded included the number of quadrants crossed, rearing, and
unusual gait. This test was repeated three times and results were averaged (14, 24).

Hang testing
The hang test, which can detect loss of motor strength, consisted of placing a
mouse on a wire grate surface suspended from a ring stand; the surface was then
flipped over so that the mouse was upside down. The period of time the mouse could
remain hanging upside down for up to 2 minutes was recorded (29). This test was
repeated 3 times, with15 minute rest periods between trials, and the longest time on the
grate was recorded as the test value. The hang test was only used in Experiments 1
and 2 (22, 29).

38

Rotarod testing
Rotarod (IITC Rotarod for Rats and Mice, IITC Inc. Life Science, Woodland Hills,
CA) was used to detect loss of balance and motor function; this test was used in
Experiments 2 and 3. Mice were placed on the rotarod for 60 seconds at a speed of 3
RPM with an acceleration of 0.10 RPM per second to reach 30 RPM for a maximum
speed. The time that the mouse stayed on the rotating rod was recorded. The test was
repeated 3 times, with 15 minute rest periods between trials, and the longest time the
mouse achieved was used for analysis (14, 22).

Footprinting
In Experiments 1 and 2, footprinting was used to detect foot drag, abnormal gait,
or abnormal foot placement as previously described by Crawley and others (29). Briefly,
the fore and hind paws of the mouse were painted with different colored non-toxic paint.
Each mouse was placed at the end of a tube, with a slot cut out of the top to monitor
mice. Mice were allowed to walk along the tube on a strip of filter paper to record the
footprints. Footprints from three strides were then analyzed for stride length, hind base
width, stride variability, toe width, and print length. This was done by identifying the
midpoint of each paw and then measuring the distances between paws of the forelimbs
and hind limbs for strides and stances. Length of the foot and width of the toe spread
were measured and analyzed as well (14, 22, 29).

DigiGait™ testing

39

In Experiment 3, a DigiGait™ apparatus (Mouse Specifics, Quincy, MA, USA),
was used to measure stride variability, stance width, gait symmetry, and stride length.
This machine is designed specifically to measure a rodent’s gait using a clear motorized
treadmill belt that can be set at various speeds. The mice are placed into the chamber
on the belt and forced to run by the movement of the belt as a sensitive video camera
records their movement from below. The video is then processed with specifically
designed proprietary software to analyze the gait. Mice were trained on the DigiGait™
for 2 weeks prior to inoculation at speeds of 10 cm/sec, 15 cm/sec, and 20c m/sec. For
the experiment mice were run at both 15 cm/sec and 20 cm/sec for a minimum capture
of 300 frames. Videos were then acquired and analyzed using DigiGait™ Imaging
Software (Mouse Specifics, Quincy, MA, USA). Stride variability, stance width, gait
symmetry, and stride length were all analyzed. Mice that were behaviorally
uncooperative throughout the entire experiment, producing poor videos, were removed
from the analysis.
We also used the weekly videos to provide clinical observations of the gait,
stance, and toe positions of the mice. A clinical observation score sheet was developed
that allowed for the quantification of clinical signs, including splayed hind legs, hind limb
dragging, knuckling, foot drag, paddling, galloping, sidestepping, crossing over, and
bunny hopping. Each of these clinical signs were given one point for presence and two
points if it occurred at a high frequency; splayed hind legs and foot drags were given
one point for present on one limb and two points if present on 2 limbs. Toes were also
scored for being splayed, curled, or flaccid; points were assigned for each (1, 2 and 3

40

points respectively). The stance of the mouse was also examined for a shorter and/or
wider stance; presence or absence was given one or no points, respectively (Table 2.1).
Neurological scores for a mouse each week were then added to give a cumulative
clinical phenotype score for each mouse. Three trained operators, one of whom was a
veterinarian trained in neurological examination, watched the DigiGait™ videos
simultaneously to assign the scores; they were blinded and unbiased. In the rare
instances that the three disagreed, the videos were rescored to resolve the
discrepancies.

Enzyme-Linked Immunosorbent Assays (ELISAs)
Mice were bled from the saphenous vein prior to inoculation and at day 14 post
inoculation, and from the heart post euthanasia at the time of necropsy. ELISA assays
for anti-C. jejuni antibody detection were performed as previously described (98). For
detection of autoantibodies against gangliosides GD1a, GM1, and GQ1b, ELISA assays
were performed using antigens GD1a (Sigma Aldrich, St. Louis, MO, USA at 20 μg/ml),
GM1 (US Biological, Swampscott, MA, USA at 2.0 μg/ml), and GQ1b (Calbiochem/EMD
Millipore, Gibbstown, NJ, USA at 0.2 μg/ml). Positive control monoclonal antibodies
used for GD1a and GM1 assays were purchased from Chemicon/EMD Millipore
(Gibbstown, NJ, USA); anti-GQ1b was purchased from Associates of Cape Cod (East
Falmouth, MA, USA). Detection of autoantibodies to a mixture of two gangliosides was
also performed using gangliosides GM1 and GD1a with a mixture of 1:1. A 1:1 mixture
of GM1 and GQ1b was also tested. Biotinylated goat antibodies included anti-mouse
IgG1, IgG2a, IgG2b, and IgG3 (Jackson ImmunoResearch, West Grove, PA, USA) and

41

biotinylated goat anti-mouse IgM (Sigma Aldrich, St. Louis, MO, USA). Extravidin
peroxidase (Sigma Aldrich, St. Louis, MO, USA) and TMB substrate (Rockland
Immunochemicals Inc., Gilbertsville, PA, USA) were used for detection.

Assessment of Clinical Signs
Mice were monitored once a day and then twice a day after the first onset of
illness. Clinical signs for enteritis assessed included hunched posture, rough coat,
diarrhea, inactivity, and dehydration. Any mice that exhibited clinical signs of enteric
disease, based on a clinical score sheet (98), prior to end of experimental time points,
were euthanized immediately to prevent any discomfort or distress. Neurological clinical
signs were examined during the weekly phenotype tests.

Necropsy Methods and Assessment of Gross Pathology
At the humane endpoint or at the end of each experiment, all mice were
humanely euthanized using an overdose of CO2 (1). After euthanasia, mice were
weighed and a blood sample taken directly from the heart, followed by puncture of the
diaphragm to ensure death. The entire GI tract was then removed from stomach to
rectum, and upon examination any gross pathology features were noted, as previously
described (98). Mice were then skinned from the back and the spinal cord was exposed.
The sciatic nerve was also exposed on both hind limbs. The mouse and legs were
placed in a specimen cup and fixed in 10% buffered formalin. The brain was then
removed and placed in 10% buffered formalin.

42

Culture for Presence of C. jejuni, DNA Extraction and PCR Assays
TSA-CVA agar plates (TSA supplemented with 5% sheep blood, 20 µg/mL
cefoperazone, 10 µg/mL vancomycin, and 2 µg/mL amphotericin B) on which GI tissues
removed from mice were streaked were scored for presence of C. jejuni (98). Plates
with C. jejuni growth were swabbed up and whole cell PCR was performed to verify that
isolates were C. jejuni as previously described (98). DNA was extracted from cecal
tissue of all control mice and PCR was performed to verify mice were C. jejuni negative
(98).

Histopathological Scoring of the Gastrointestinal Tract
At the time of necropsy, the ileocecocolic junction was removed and fixed in 10%
phosphate-buffered formalin; the junctions were paraffin embedded and stained with
hematoxylin and eosin by the Investigative Histopathology Laboratory, Division of
Human Pathology, Department of Physiology, Michigan State University. Sections were
scored on a scale of 0 to 44 for lesions present in the lumen, epithelium, lamina propria,
and the submucosa as previously described (98). Slides were evaluated by a single
investigator (LSM) who was blinded to their identity.

Histopathological Scoring of the Sciatic Nerve
For Experiment 2, at the time of necropsy, the sciatic nerves were removed as
follows. The legs of each mouse were removed, skinned, and muscles were dissected
to expose the nerves. The nerves were exposed from the pelvis to the tibial nerve.

43

One leg (left) was fixed in 10% buffered formalin for 6 hours and then placed in
60% ethanol for paraffin embedment for both longitudinal and cross sections at the
Investigative Histopathology Laboratory, Division of Human Pathology, Department of
Physiology, Michigan State University. One section was stained with hematoxylin and
eosin and the other with Luxol fast blue. Both longitudinal and cross sections of nerves
were then examined and photographed using a Nikon Eclipse E600 microscope with a
SPOT camera with Windows TM version 4.09 software (RTSlider Diagnostic
Instruments, Inc., Sterling Heights, MI). Nerves stained with hematoxylin and eosin were
examined for cellular infiltrates and axonal degeneration. Nerves stained with Luxol fast
blue were examined for myelin damage.
The other leg (right) with the exposed nerve was placed in 30% sucrose for up to
one week and then placed in a plastic cassette, covered with OCT compound, and
frozen at -80°C to be saved for immunohistochemistry.
For Experiment 3, the dissection process was extended to remove the entire
sciatic nerve up to and including the nerve root (L3–L5); the brain and the spinal cord
were dissected and preserved separately. After removal of internal organs, the skin, and
muscle along the sciatic nerve on both legs were removed to expose the sciatic nerve.
The sciatic nerve was followed up to the spinal column and the vertebral bone removed
to expose the nerve root. The dorsal portion of the vertebral column over the spinal cord
was snipped off with Student Vannas Spring scissors (Fine Science Tools, Foster City,
CA, USA) to expose the cord. The brain was removed by removing the dorsal surface of
the skull and transsecting the cranial nerves at the base. Then the mouse carcass with
exposed nerves, spinal cord, and brain was placed in 10% buffered formalin for 24

44

hours and then transferred to 60% ethanol until further nerve preparation could be
performed. After fixation, the dorsal root ganglia were exposed and removed, ensuring
that the L4 nerve root was included. The sciatic nerve was then removed starting from
the spinal cord all the way to anterior tibial nerve at the tarsus of the mouse. Finally the
brachial plexus nerves at the foreleg were removed in their entirety as a group. All
nerves were placed in a cassette and kept in 60% ethanol until paraffin embedment and
analysis as described above for Experiment 2. During embedment, nerves were placed
in longitudinal orientation to facilitate analysis. Sciatic nerves, dorsal root ganglia, and
the brachial plexus sections were stained with either Luxol fast blue or hematoxylin and
eosin. Sections were observed and imaged with a Nikon Eclipse E600 microscope with
a SPOT camera (RTSlider Diagnostic Instruments, Inc., Sterling Heights, MI).

Nerve Histology Criteria
The criterion for scoring nerve lesions were developed by 2 veterinary
pathologists (BJG, JSP) and critiqued by a human pathologist (HTC). Specific features
were evaluated on hematoxylin and eosin (H&E) stained sections and Luxol fast blue
stained sections as follows. The H&E stained nerve root sections were evaluated for
inflammatory infiltrate severity (0=normal, 1=mild infiltrate, 2=moderate infiltrate, and
3=marked infiltrate), distribution (scattered, clustered), and composition (lymphocytes,
plasma cells, macrophages, neutrophils, eosinophils, and mast cells). The H&E stained
sciatic nerve and brachial plexus nerve sections were evaluated in a similar manner
except that the distribution of cells was scored as perivenular, endoneurial, perineurial,
epineurial, and intramuscular. The H&E stained sciatic nerve sections were evaluated

45

for myelin sheath loss (0=not present, 1=mild in 0–30% of section, 2=moderate in 30–
60% of the section and 3=severe in greater than 60% of the section). All histological
sections were scored by a single board certified veterinary pathologist (BJG) who was
blinded to the identities and experimental groups of the individual mice. Blinding of
slides and data analyses were performed by a second investigator (JLS).

Immunohistochemical Staining (IHC)
We examined the sciatic nerves for infiltration of T cells and macrophages using
IHC. Nerves were removed, fixed, embedded, and sectioned as previously described
above. Serial sections 3–5 μm thick were cut, attached to silanized glass slides and
stained for either CD3 (polyclonal, ab5690, abcam Cambridge, MA) or F4/80 (BM8,
eBiosciences San Diego, CA). Sections were deparaffinized in 2 changes of xylene and
rehydrated in a graded alcohol series. Sections were then incubated in 1% hydrogen
peroxide in tris-buffered saline (TBS; pH 7.4) to block endogenous peroxidase.
Subsequently the sections for CD3 were boiled for 15 minutes in citrate buffer (Vector
Laboratories, Burlingame, CA); the F4/80 sections were incubated at 37°C for 15
minutes in Proteinase K antigen retrieval solution. The sections were blocked for 1–3
hours at room temperature with 1% bovine serum albumin (Sigma Aldrich, St. Louis,
MO) +1.5% goat serum (Vector Laboratories, Burlingame, CA) in TBS 0.025% Triton X100 (Sigma Aldrich, St. Louis, MO) and then incubated with the primary antibody at 4°C
overnight. Sections were then washed twice and stained with the Vectastain ABC kit
(Vector Laboratories, Burlingame, CA) for CD3 or rat on mouse HRP polymer kit
(Biocare Medical, Concord, CA) for F4/80. The sections were developed with Impress

46

Nova Red kit (Vector Laboratories, Burlingame, CA), counterstained, differentiated,
dehydrated and mounted with Permount™ (Sigma Aldrich, St. Louis, MO, USA).
Sections were observed and imaged using a Nikon Eclipse E600 microscope with a
SPOT camera (RTSlider Diagnostic Instruments, Inc., Sterling Heights, MI).

Statistical Analyses
Statistical analyses were performed using GraphPad Prism version 6.01
(GraphPad Software Inc., La Jolla, CA, USA). Analyses of results of the open field test,
hang test, and rotarod phenotype tests in all experiments were done using two-way
ANOVA with repeated measures followed by the Holm-Sidak multiple comparisons test
if significance was obtained in the two-way ANOVA. If the assumption of normality was
not met for the two-way ANOVA with repeated measures, data were transformed using
the square root function, which resulted in the normality assumption being met. The
assumption of equal variance was met as well. In Experiment 1, the footprint
measurements stride length and hind base width were not analyzed with a two-way
ANOVA with repeated measures because there were too many missing data points
caused by individual mice refusing to perform the test at one or more time points during
the experimental time span. These measurements were evaluated using a one-way
ANOVA at each individual time point; if assumptions were not met, then a nonparametric Kruskal-Wallis ANOVA on ranks was used. In Experiment 2, few mice had
missing data points during the experimental time span. To allow the repeated measures
ANOVA to be performed, mice that did not have a complete set of measurements
during the entire experiment were removed from the data set; the removal of mice

47

resulted in slightly smaller group sizes. This test was followed with the Holm-Sidak test
if significance was obtained.
For evaluations that did not have time as a factor (histological scoring, ELISA,
IHC), one-way ANOVA was used followed by the Holm-Sidak post hoc test. This
approach was applied in Experiments 1 and 3. If normality and equal variance
assumptions for one-way ANOVA were not met, then the non-parametric Kruskal-Wallis
ANOVA on ranks was used followed by Dunn’s post hoc multiple comparison test. Data
for Experiment 2 were analyzed with the non-parametric Mann-Whitney U test,
comparing uninfected mice to infected mice within each genotype. The Fisher Exact test
was used to examine any cell types present in the nerve.
Stride length and hind base width measurements taken digitally by the
DigiGait™ apparatus in Experiment 3 were not analyzed because there were too many
missing data points. A clinical observation scoring system was developed for the raw
DigiGait™ videos (Table 2.1), and results were analyzed using two-way ANOVA with
repeated measures to examine only mice that were videotaped at every time point for
the entire length of the experiment. Spearman’s rank/ordinal correlation was used to
examine the correlation of the clinical assessment of the DigiGait™ videos to the GI
histological scores of the mice using the software, PAST version 1.89 (Øyvind Hammer,
Oslo, Norway).

48

RESULTS

Selection of Mice for Analysis
-/-

NOD WT, NOD IL-10 , and NOD B7-2

-/-

mice were bred in a specific-pathogen

free facility and tested by PCR to confirm genotype and to be free of the known
colitogenic bacteria, Campylobacter spp., Helicobacter spp., Citrobacter rodentium,
Enterococcus faecalis, and E. faecium as previously described (98). Preliminary and
published results have shown that NOD B7-2

-/-

mice do not develop spontaneous

neuropathy prior to 20 weeks of age (Gadsden, unpublished data) (13, 130); in all
experiments, mice were used prior to 20 weeks of age. Mice that developed diabetes,
prolapsed rectum, or died of respiratory failure after accidental inhalation of the
inoculum during oral gavage in any experiment were removed from analysis. In the
three experiments combined, there were 2/28 and 2/28 NOD WT mice that developed
diabetes or respiratory disease respectively. Of the NOD IL-10

-/-

mice, 11/100 and

2/100 mice developed prolapsed rectums or respiratory failure respectively, both of
-/-

which resulted in early euthanasia. Two mice (1/28 NOD WT and 1/28 NOD B7-2 )
were removed from analysis due to failure to detect colonization by C. jejuni HB93-13
by either culture or PCR.
Body weight was also analyzed at the end of all experiments; we found no
significant differences between treatment groups within each genotype using KruskalWallis ANOVA on ranks test for Experiments 1 and 3 (data not shown) and Mann-

49

Whitney U for Experiment 2 (data not shown). Therefore we concluded body weight
played no role in the examination of footprints.

Experiment 1: Screening mice of various NOD genotypes for GBS following C.
jejuni infection

Experimental screen for GI disease with GBS-associated C. jejuni strains
Colonization was determined by culture of GI tract tissue and/or C. jejuni specific
PCR. As seen in Figure 2.1, panel A, all of the sham-inoculated control mice were
negative by cecum culture and negative for C. jejuni specific PCR on DNA extracted
-/-

from cecum tissue. Four of nine NOD WT, 7/10 NOD IL-10 , and 4/9 NOD B7-2
infected with C. jejuni HB93-13 were colonized. All NOD WT and NOD IL-10

-/-

-/-

mice

mice

infected with C. jejuni 260.94 were colonized as detected by culture and/or PCR. Nine
of ten NOD B7-2

-/-

mice were colonized in the cecum by C. jejuni 260.94.

Three mice inoculated with C. jejuni HB93-13, one mouse of each genotype, had
to be euthanized early due to high scores for clinical GI disease. Two NOD IL-10

-/-

mice, both in the C. jejuni 260.94 infection group, had to be euthanized early for high
clinical scores of GI disease.
At necropsy, mice were examined for the following gross pathological changes:
enlarged ileocecocolic lymph nodes and thickened wall of gastrointestinal tract (Figure
A-1, panel A). Four of nineteen C. jejuni infected NOD WT mice, 16/18 NOD IL-10
mice, and 4/19 NOD B7-2

-/-

-/-

mice exhibited gross pathological changes in the GI tract.

50

In addition, 4 sham-inoculated NOD IL-10

-/-

mice exhibited spontaneous colitis, as

expected in IL-10 deficient mice. These 4 mice were not colonized with either strain of
C. jejuni as confirmed by culture and PCR. No sham-inoculated NOD WT mice or NOD
B7-2

-/-

mice exhibited gross pathological changes in the GI tract.
Most NOD IL-10

-/-

mice infected with both C. jejuni HB93-13 and C. jejuni 260.94

had pronounced gross pathological changes in the GI tract. More NOD B7-2

-/-

mice in

the C. jejuni HB93-13 infection group had gross pathological changes than B7-2

-/-

mice

in the C. jejuni 260.94 infection group. Mice in the NOD WT C. jejuni 260.94 and C.
jejuni HB93-13 infection groups had similar levels of gross pathological changes in the
GI tract (Figure A-1, panel A).
At necropsy the ileocecocolic junction was removed, fixed, stained, and scored
as previously described (98). Kruskal-Wallis ANOVA on ranks was used to analyze the
histology scores; no significant differences were found between treatment groups within
any of the three genotypes (Figure A-1, panel B). As with the gross pathological
changes, the NOD IL-10

-/-

mice exhibited the highest GI histology scores, as did the C.

jejuni HB93-13 infected NOD WT mouse that was euthanized early due to GI disease.

Detection of anti-ganglioside antibodies
ELISA assay results were statistically tested using one-way ANOVA followed by
the Holm-Sidak post hoc test if a significant result was obtained in the ANOVA. When

51

normality and equal variance assumptions were not met for one-way ANOVA, the nonparametric Kruskal-Wallis ANOVA on ranks was used followed by Dunn’s post hoc test.
Plasma samples obtained on days 0 and 14 were analyzed for the presence of
anti-ganglioside GM1 and GD1a antibodies using ELISA assays. Only small amounts of
blood can be taken from mice at these time points, so two ELISA assays were done.
Mouse IgM was tested for anti-GM1 activity, and mouse IgG1 was tested for anti-GD1a
activity. At days 0 and 14, no significant differences were found between the control and
infected mice in levels of plasma IgM reactive to ganglioside GM1 and IgG1 reactive to
-/-

ganglioside GD1a in the NOD WT, NOD IL-10 , and NOD B7-2

-/-

genotypes (Figure

A-2, panels A and B). ELISA assays were performed at endpoint for the presence of
anti-C. jejuni IgG2b antibody to verify the development of adaptive immune responses
to C. jejuni. ELISA assays were also performed at endpoint for the presence of (1) antiGD1a IgG1, IgG2a, and IgG3 antibodies, (2) anti-GM1 IgG1, IgG2a, IgG3, and IgM
antibodies, and (3) anti-GQ1b IgG1, IgG2a, and IgG3 antibodies (Figure A-2, panels C
and D; GQ1b data not shown). Further ELISA assays were performed at endpoint for
the presence of IgG1 antibodies reactive with two mixtures of gangliosides: a
GM1/GQ1b mixture and a GM1/GD1a mixture.
Significant differences between the treatment groups within all mouse genotypes
at endpoint were found for IgG2b reactive with crude C. jejuni antigen (Kruskal-Wallis;
-/-

-/-

NOD WT: p=0.0004, NOD IL-10 : p=0.0001 and NOD B7-2 : p<0.0001). Dunn’s
multiple comparison test demonstrated significant differences between the shaminoculated and both the C. jejuni HB93-13 and C. jejuni 260.94 inoculated groups in all
three genotypes as seen in Figure 2.1, panel B-1. No significant differences were found

52

for any of the gangliosides tested individually. A significant difference was found
between IgG1 antibody levels in NOD WT uninfected mice and mice infected with C.
jejuni 260.94 for the mixture of gangliosides GM1/GQ1b (one-way ANOVA; p=0.0100).
The Holm-Sidak post hoc test was used and a significant difference in plasma levels of
IgG1 antibody to the ganglioside mixture GM1/GQ1b was found between the shaminoculated and the C. jejuni 260.94 infected NOD WT mice (p ≤ 0.01) as noted in Figure
2.1, panel B-2. All other mouse genotypes had no significant differences in plasma
levels of IgG1 antibody to the mixture of gangliosides GM1/GQ1b; NOD WT mice
infected with C. jejuni HB93-13 also displayed no significant differences in levels of
IgG1 antibody reactive with ganglioside mixture GM1/GQ1b. Levels of IgG1 antibody
reactive with ganglioside mixture GM1/GD1a in sham-inoculated and C. jejuni infected
mice were not significantly different in any of the three genotypes (Figure A-2 panel E).

Neurological phenotype testing
Statistical analyses for the open field test (examination of abnormal gait) and
hang test (muscle strength) were performed with two-way ANOVA with repeated
measures followed by the Holm-Sidak post hoc test for multiple comparisons. If the
normality assumption was not met, data were transformed using the square root
function. In the two-way ANOVA with repeated measures we tested for both (1)
differences between treatment groups of the same genotype (discussed below) and (2)
changes in the genotype over time (Table 2.2). The occurrence of significant differences
within the treatment groups occurring over the course of the experimental time period
was a general phenomenon in all phenotypic tests and will not be discussed further.

53

The stride length and hind base width footprint data sets (detection of gait
abnormalities including foot dragging) for all mouse genotypes had many missing data
points and therefore one-way ANOVA or Kruskal-Wallis ANOVA on ranks at each time
point was performed.
NOD WT mice. NOD WT infected mice had significant differences from the
sham-inoculated control mice in the open field test, hang test, and hind base width.
Analysis of the open field test data (Figure 2.1, panel C) indicated significant
differences between the treatment groups of the NOD WT mice (two-way ANOVA RM;
p=0.0048). Post hoc pairwise Holm-Sidak multiple comparison tests indicated significant
differences between NOD WT sham-inoculated mice and C. jejuni HB93-13 infected
mice at the 2-, 3-, and 6- week time points (wk2 p≤0.01, wk3 p≤0.05, wk6 p≤0.01).
Holm-Sidak post hoc tests also indicated significant differences between shaminoculated mice and C. jejuni 260.94 infected mice at weeks 2, 3, and 6 (wk2 p≤0.05,
wk3 p≤0.05, wk6 p≤0.01).
Analysis of the hang test data (Figure 2.1, panel D) detected significantly
different hang test times between the uninfected NOD WT mice and both the C. jejuni
HB93-13 and C. jejuni 260.94 infected NOD WT mice (two-way ANOVA RM; p=0.0083).
Sham-inoculated mice had significant differences from those infected with C. jejuni
260.94 at weeks 1 and 3 of the experiment (Holm-Sidak post hoc pairwise multiple
comparisons test; wk1p p≤0.05, wk3 p≤0.01). Qualitative examination of the data shows
that there was a decrease in hang time of most treatment groups at week 1; this
observation could indicate that a longer training time was needed for this test. It was
also observed that the NOD WT C. jejuni 260.94 infected mice followed a similar pattern

54

over time as the sham-inoculated mice and C. jejuni HB93-13 infected mice, but never
had hang times as high as these treatment groups.
Footprint measurements of stride length and hind base width were analyzed with
one-way ANOVA at each weekly time point; Holm-Sidak post hoc tests were used if
significant results were indicated in the ANOVA (Figure A-3, panel B). No significant
changes between the treatment groups were observed in the stride length data at any
time point. Hind base width data showed significant changes between all the treatment
groups at week 6 (one-way ANOVA; p=0.0494). The post hoc test, Dunn’s, indicated no
significant changes between any 2 treatment groups.
NOD IL-10

-/-

mice. Significant differences between the treatment groups were

found in stride length measurements. No other test indicated any significant changes in
treatment groups.
The open field test data (Figure A-3, panel A-1) showed no significant differences
between the treatment groups. Analyzing the hang test time of the NOD IL-10

-/-

mice

revealed no significant differences between the treatment groups. The sham-inoculated
mice and C. jejuni HB93-13 infected mice both were qualitatively observed to have a
decrease in hang time at week 1; this phenomenon did not occur in the C. jejuni 260.94
infected mice (Figure 2.1 panel D). This occurrence was also noted in the NOD WT
above.
Footprint patterns (Figure A-3, panel C) were measured for stride length and hind
base width and one-way ANOVA was performed at each time point. Significant
differences in stride length were found between the treatment groups at weeks 4, 5, 6,
and 8 (one-way ANOVA; wk4 p=0.0363, wk5 p=0.0298, wk6 p=0.0054, wk8 p=0.0042).

55

Pairwise analysis indicated significant differences between the sham-inoculated mice
and the C. jejuni 260.94 infected mice at weeks 5, 6, and 8 (Holm-Sidak; wk5 p≤0.05,
wk6 p≤0.01, wk8 p≤0.01). Mice infected with C. jejuni HB93-13 and sham-inoculated
mice had significant differences in stride length at weeks 6 and 8 (Holm-Sidak; wk6
p≤0.05 and wk8 p≤0.01). The hind base width data analysis showed no significant
differences between the infected mice and the uninfected mice.
NOD B7-2

-/-

mice. Significant differences between the treatment groups were

observed in the hang test, stride length, and hind base width.
Two-way ANOVA with repeated measures analysis of the number of quadrants
crossed in the open field test indicated no significant changes between the treatment
groups of this genotype (Figure A-3, panel A-2). Qualitative examination of the data
shows that mice infected with both C. jejuni HB93-13 and C. jejuni 260.94 exhibited an
increase from a small number of quadrants crossed (weeks 2–3) to a higher number
crossed (weeks 4–6) and back to a smaller number of quadrants crossed (weeks 7–8)
during the course of the experiment.
Analysis of hang test data (Figure 2.1 panel D) with two-way ANOVA with
repeated measures indicated significant differences between the treatment groups of
the NOD B7-2

-/-

mice (two-way ANOVA RM; p=0.0353). Post hoc pairwise analysis

indicated that significant differences between sham-inoculated mice and C. jejuni
260.94 infected mice occurred at week 1 of the experiment (Holm-Sidak; p≤0.05).
Qualitative examination of these data showed that there was a decrease in hang time
occurring at week 1 in all treatment groups. This was noted previously in the other 2
genotypes and therefore suggests a longer training time was needed for this test.

56

Stride length and hind base width measurements were analyzed using one-way
ANOVA at each time point (Figure A-3, panel D). Significant differences in stride length
occurred between treatment groups at the 1- and 3- week time points (one-way
ANOVA; p=0.0248 and p=0.0085 respectively). Pairwise analysis indicated significant
differences between the sham-inoculated mice and the C. jejuni HB93-13 infected mice
at week 3 (Holm-Sidak; p≤0.01). Analysis of the hind base width measurements
indicated significant changes in hind base width between the treatment groups at week
1 (one-way ANOVA; p=0.0157). Holm-Sidak post hoc tests indicated significant
differences between the sham-inoculated mice and the C. jejuni HB93-13 infected mice
(p≤0.05) and between the sham-inoculated mice and C. jejuni 260.94 infected mice
(p≤0.05).

Experiment 2: Screening larger group sizes of NOD WT and NOD B7-2

-/-

mice for

GBS

Experimental screen for GI disease with GBS-associated C. jejuni strains
Colonization of mice with C. jejuni 260.94 was determined by culture of GI tract
tissue fragments and/or PCR (Figure 2.2, panel A-1). All C. jejuni 260.94 infected mice
of both NOD WT and NOD B7-2
infected NOD B7-2

-/-

-/-

genotypes were colonized. One C. jejuni 260.94

mouse was negative by culture, but positive by PCR. All sham-

inoculated mice of both NOD WT and NOD B7-2
jejuni 260.94 by culture and/or PCR.

57

-/-

genotypes were negative for C.

At necropsy, gross pathological changes in the GI tract were recorded Figure 2.2,
panel A-2; these changes were thickened wall of the GI tract and enlarged lymph
nodes. Gross pathological features in C. jejuni 260.94 infected mice were minimal; only
4/18 NOD WT mice and 1/18 NOD B7-2

-/-

mice exhibited a single gross pathology

feature. All other mice had no gross pathological changes in the GI tract.
One mouse from the NOD WT C. jejuni 260.94 group had to be euthanized early
due to a high clinical sign score for GI disease; this mouse still only presented a single
gross pathological feature on necropsy.
At the end of the experiment (80 days), the ileocecocolic junction was examined
histologically and scored by a single blinded investigator (LSM) (100). Non-parametric
Kruskal-Wallis ANOVA on ranks indicated no significant difference between histological
scores of infected and uninfected mice of either genotype (Figure B-1).

Detection of anti-ganglioside antibodies
ELISA data were analyzed using the non-parametric Mann-Whitney U test to
compare uninfected mice to C. jejuni 260.94 infected mice within each genotype. ELISA
assays were performed on plasma samples obtained at days 0 and 14 to test for the
presence of IgM antibodies reactive with ganglioside GM1 and IgG1 antibodies reactive
with ganglioside GD1a. On day 0, NOD WT mice that were going to be sham-inoculated
had significantly lower levels of anti-GM1 IgM than NOD WT mice that were going to be
inoculated with C. jejuni 260.94 (p=0.0229). This significant difference observed at day
0 was the result of a few low values (3 of 17 mice). When these mice were removed
from analysis the significant differences were no longer observed. All other ELISA tests

58

at both day 0 and day 14 showed no significant differences between groups (Figure B-2,
panels A and B).
ELISA assays for anti-C. jejuni, -GM1, -GD1a, -GQ1b, and ganglioside mixtures
GM1/GD1a and GM1/GD1a were performed at the endpoint of experiment (80 days) as
in Experiment 1. Significant differences were found in levels of anti-C. jejuni IgG2b
antibody in both NOD WT and NOD B7-2

-/-

mice when infected mice were compared to

congenic uninfected mice (Figure B-2, panel C; Mann-Whitney U; p<0.0001 for both
genotypes). Significantly different levels of IgG3 anti-ganglioside GD1a between
infected and uninfected mice were found for both mouse genotypes (Figure 2.2, panel
-/-

B-1; NOD WT: p=0.0296; NOD B7-2 : p=0.0395). Analyses of data from ELISA assay
for antibodies reactive with the other individual gangliosides indicated no significant
differences (Figure B-2, panels D-E; ganglioside GQ1b not shown). C. jejuni 260.94
infected NOD WT mice had significantly higher levels of IgG1 reactive with the mixture
of gangliosides GM1 and GQ1b (Figure 2.2, panel B-2; Mann-Whitney U; p=0.0351);
NOD B7-2

-/-

mice did not. No significant differences were detected in levels of IgG1

antibody reactive with the mixture of gangliosides GM1 and GD1a in either mouse
genotype (Figure B-2, panel F).

Neurological phenotype testing
Statistical analyses for the open field test, hang test, and rotarod were performed
using two-way ANOVA with repeated measures; if significant differences between the
uninfected mice and infected mice were observed, the ANOVA was followed with the

59

post hoc Holm-Sidak test. The footprint measurements of stride length and hind base
width were analyzed with two-way ANOVA with repeated measures with the removal of
any mice that did not have a measurement for a characteristic for every time point
during the experiment. Only 4 mice were removed from the stride length measurement;
none were removed for hind base width. Any significant changes within the treatment
groups that occurred over the course of the experimental time are shown in Table 2.2
and will not be discussed any further as it was a general phenomenon.
NOD WT mice. No significant changes between the uninfected mice and C.
jejuni 260.94 infected mice were observed in any of the phenotypic tests below.
Data from the open field test, which is used to determine the presence of gait
abnormalities and overall well-being of mice, had no significant differences in the
number of quadrants crossed between treatment groups. As can be seen in Figure 2.2,
panel C-1, there was high variability in the NOD WT C. jejuni 260.94 mice which
resulted from a “circling” mouse in that group. NOD mice often exhibit an obsessively
circling phenotype. The same mouse caused the high variability throughout the entire
experiment. Removal of this mouse from the analysis did not affect the statistical nonsignificance of changes over time seen in the NOD WT mice.
The hang test (Figure B-3, panel A-1) was used to assess muscle strength and
gripping ability of the mice. No significant differences were found in the NOD WT mice
between the treatment groups. Nevertheless, most NOD WT mice infected with C. jejuni
260.94 had shorter hang times than uninfected NOD WT mice at week 4, slightly later
than seen in Experiment 1.

60

Only one NOD WT sham-inoculated mouse had to be removed from the stride
length measurements from lack of footprint patterns for the entire course of the
experiment (Figure B-3, panel B). There were no significant changes in the stride length
between treatment groups. There were also no significant changes in hind base width
observed between treatment groups; no mice had to be removed for this analysis.
Use of a rotarod was introduced in this experiment; this apparatus is used to
detect changes in balance and motor capability of mice. NOD WT sham-inoculated and
C. jejuni 260.94 mice showed no significant change in the time spent on the rotarod
between treatment groups (Figure 2.2, panel D-1). There was a statistically nonsignificant but qualitatively noticeable trend occurring within the C. jejuni 260.94 infected
mice; most of these mice remained on the rotarod for shorter lengths of time than
uninfected mice at 7 of 11 time points after inoculation.
NOD B7-2

-/-

mice. Data analyses indicated significant differences between the

uninfected mice and the C. jejuni 260.94 infected mice in the open field test and hind
base width measurements.
NOD B7-2

-/-

mice had significant differences over time between the treatment

groups of number of quadrants crossed (Figure 2.2, panel C-2; p=0.0299). However, no
significant differences were noted at any particular time point using a pairwise test.
Analysis of the hang test data indicated no significant differences between the
treatment groups of the NOD B7-2

-/-

mice. Data shown in Figure B-3, panel A-2.

To assess the stride length measurements of the NOD B7-2

-/-

mice, 1 sham-

inoculated mouse and 2 C. jejuni 260.94 infected mice had to be removed from the data

61

set because they lacked complete footprint patterns for the course of the experiment.
No significant changes in stride length occurred between the treatment groups (Figure
B-3, panel C-1). Analysis of hind base width data indicated significant differences
between the sham-inoculated mice and the mice infected with C. jejuni 260.94 (two-way
ANOVA RM; p=0.0498); pairwise tests indicated significant differences between the
infected and uninfected mice occurring at week 11 of the experiment (Figure B-3, panel
C-2; Holm-Sidak; p≤0.001).
Analysis of rotarod data indicated no significant changes in the time spent on the
cylinder between the treatment groups of NOD B7-2

-/-

mice. As seen in the NOD WT C.

jejuni 260.94 mice, there was a qualitative trend toward shorter times spent on the
rotarod at 10 of 11 time points after inoculation in the NOD B7-2

-/-

infected mice (Figure

2.2 panel D-2).
To determine whether this phenomenon of a decrease in time on the rotarod was
limited to mice of the NOD genetic background, a control experiment with uninfected
C57/BL6 WT and C57BL/6 IL-10

-/-

mice was performed for the same length of time.

Both genotypes of C57/BL6 mice experienced a decrease in time spent on the rotarod
over the course of 84 days (Figure D-1). This suggests that decrease of time spent on
the rotarod over the experimental period is not limited to mice of the NOD genetic
background. Most importantly, neurological disease can still be examined by
comparison of the treatment groups.

Presence of nerve lesions and detection of cellular infiltrates

62

At endpoint the sciatic nerve was removed from the mouse to be fixed, stained
and examined histologically. The nerves were stained with both H & E to examine for
cellular infiltrates and Luxol fast blue (LFB) to examine for myelin damage. The nerves
were scored blinded and unbiased by a board-certified pathologist (BJG). Cellular
infiltrates, most commonly mast cells and lymphocytes, were observed in the nerve
sections of some mice. However, no significant differences in the presence of mast cells
or lymphocytes between the infected mice and the uninfected mice in either the NOD
WT or the NOD B7-2

-/-

mice were detected by the Fisher Exact test. Histological scores

(see Materials and Methods) were analyzed with non-parametric Mann-Whitney U test
to identify any significant changes between treatment groups within each genotype. No
significant results were observed between treatment groups in either the NOD WT mice
or NOD B7-2

-/-

mice (Figure B-4). One C. jejuni 260.94 infected NOD B7-2

-/-

mouse did

exhibit cellular infiltrates (mast cells, lymphocytes, neutrophils, and plasma cells) as well
as myelin damage as seen in Figure 2.3.
Immunohistochemical staining was performed on sections of the sciatic nerve to
examine them for the presence of T cells and macrophages. Since the nerve is so small
and frail, only a subset of each treatment group could be analyzed depending on
whether or not any nerve tissue remained in the paraffin block after sectioning for
hematoxylin and eosin and Luxol fast blue staining. There were no significant
differences in the presence of T cells between infected and uninfected mice of either
genotype; however, the difference between infected and uninfected NOD B7-2

-/-

mice

approached significance (Mann-Whitney U; p=0.0698). Staining for macrophages

63

showed significant differences between infected and uninfected NOD WT mice (MannWhitney U; p=0.0289) but not between infected and uninfected NOD B7-2

-/-

mice as

seen in Figure 2.4. Infiltration of macrophages in the NOD WT C. jejuni infected mice is
shown in Figure 2.5.

Experiment 3: Examination of NOD IL-10

-/-

mice using antibiotics to treat enteritis

Experimental screen for GI disease with GBS-associated C. jejuni 260.94
Colonization was determined at the end of the 84 day experiment by detection of
C. jejuni in the GI tract of the mouse by culture and/or PCR (Figure C-1, panel A).
Twenty six mice were colonized with C. jejuni 260.94 and two of twenty-eight mice
infected with C. jejuni and not treated with chloramphenicol (CMP) were found to be
negative for the bacteria. In the C. jejuni + CMP group, 13/17 mice were not colonized
at the conclusion of the experiment. Fourteen of sixteen mice in the sham-inoculated
group and 19/19 mice in the sham-inoculated and CMP-treated group were negative by
culture and/or PCR. Two mice in the sham-inoculated group thought to be culture
positive were determined to be negative for C. jejuni 260.94 based on PCR confirming
that the colonies observed by culture were not campylobacters.
Two gross pathological features in the GI tract were noted: thickened GI wall and
presence of enlarged ileocecal lymph nodes (Figure 2.6, panel A). Six of sixteen shaminoculated mice, 22/28 C. jejuni 260.94 inoculated mice, 10/19 sham-inoculated and
CMP-treated mice, and 14/17 C. jejuni 260.94 inoculated and CMP-treated mice
presented with pathological features. A subset of mice that presented with gross

64

pathological features in all treatment groups presented with both thickened GI wall and
enlarged lymph nodes (3/6 sham-inoculated, 19/22 C. jejuni 260.94 inoculated, 4/10
sham-inoculated and CMP-treated, and 9/14 C. jejuni 260.94 inoculated and CMPtreated mice). C. jejuni inoculated and CMP-treated mice had levels of gross
pathological change similar to those mice that received C. jejuni only. Some shaminoculated mice and sham-inoculated and CMP-treated mice presented with gross
pathological features from spontaneous colitis that is common in IL-10 deficient mice
(132).
Two of twenty-eight C. jejuni 260.94 infected mice were euthanized early for high
clinical scores for enteric disease. No other mice were euthanized early due to high
clinical scores for GI disease.
Analysis of histological scores of GI disease showed no significant differences
between groups using Kruskal-Wallis ANOVA on ranks; mice in all treatment groups
exhibited high histology scores (Figure C-1, panel B). Spontaneous colitis is expected in
NOD IL-10

-/-

mice and accounts for the marked gross pathological changes and high GI

histological scores seen in both sham-inoculated and the sham-inoculated and CMPtreated mice.

Detection of anti-ganglioside antibodies
Data from ELISA assays were analyzed using one-way ANOVA; if significant
differences were obtained, the post hoc Holm-Sidak test was used. If assumptions of
normality and equal variance were not met, Kruskal-Wallis ANOVA on ranks was used
with post hoc Dunn’s tests if significant differences were noted.

65

Levels of antibodies reactive with single and mixed gangliosides (GM1, GD1a
and GQ1b; GM1/GQ1b and GM1/GD1a) were determined at the endpoint of the
experiment (84 days). Anti-C. jejuni IgG2b levels were also tested and significant
differences were obtained using Kruskal-Wallis ANOVA on ranks (Figure C-2, panel A;
p<0.0001). Dunn’s post hoc tests indicated significant differences in anti-C. jejuni IgG2b
levels between the following groups: (1) the sham-inoculated mice and mice infected
with C. jejuni 260.94 only (p≤0.0001), (2) mice infected with C. jejuni 260.94 only and
sham-inoculated mice treated with CMP (p≤0.0001), and (3) sham-inoculated mice
treated with CMP and mice inoculated with C. jejuni 260.94 and treated with CMP
(p≤0.01).
Levels of anti-GD1a IgG1and IgG3 antibodies were observed to be significantly
different (Figure C-2, panels C-1 and C-3; Kruskal Wallis, p=0.0030 and p=0.0421
respectively). Dunn’s pairwise post hoc tests indicated significant differences in levels of
both IgG1 and IgG3 antibody reactive with ganglioside GD1a between the shaminoculated mice and mice infected with C. jejuni 260.94 not treated with CMP (IgG1:
p≤0.01 IgG3: p≤0.05). Plasma levels of anti-GQ1b IgG2a antibodies were significantly
different (Figure C-2, panel D; Kruskal-Wallis, p=0.0007). Dunn's pairwise post hoc tests
indicated significant differences in the anti-GQ1b levels of the C. jejuni 260.94 infected
mice treated with CMP and (1) mice that were sham-inoculated (p≤0.001), (2) C. jejuni
260.94 infected mice (not treated with CMP)( p≤0.001), and (3) sham-inoculated mice
treated with CMP (p≤0.001). Significant differences in plasma levels of anti-GM1 IgG3
antibody were observed (Figure C-2, panel B-4; Kruskal-Wallis; p=0.0217); Dunn's

66

pairwise post hoc tests found significant differences in antibody level between the
sham-inoculated mice and the sham-inoculated mice treated with CMP (p≤0.05).
Levels of IgG1 antibodies reactive with the mixture of gangliosides GM1 and
GQ1b were significantly different between the treatment groups (Figure 2.6, panel C-1;
Kruskal-Wallis; p=0.0017). Dunn's post hoc tests indicated significant differences in antiGM1/GQ1b IgG1 antibody levels between the sham-inoculated mice and mice given
both C. jejuni 260.94 and CMP (p≤0.001). IgG1 antibodies reactive with the mixture of
anti-GM1/GD1a were found to be significantly different between the treatment groups
(Figure 2.6, panel C-2; Kruskal-Wallis; p=0.0332). All other ELISA assay data are
shown in Figure C-2, panels B-C; GQ1b data are only shown for antibody IgG2a.

Neurological phenotype testing
Two-way ANOVA with repeated measures was used to analyze any significant
changes occurring in mouse performance on the rotarod. Stride length and hind base
width measurements were computed using the DigiGait™ software, but could not be
analyzed using two-way ANOVA with repeated measures because of missing
measurements. One-way ANOVA was not performed as in Experiment 1 because
missing measurements made group sizes too small. Visual clinical assessment of
DigiGait™ videos was performed by blinded and unbiased scientists (LSM and JAB).
Each clinical feature noted in videos was given a score as described above in Materials
and Methods (Table 2.1); scores for the individual clinical features were then summed
for each mouse at each week. Results for mice that had videos for every time point
during the experimental time were analyzed using two-way ANOVA with repeated

67

measures. Cumulative clinical assessment scores for the entire length of the experiment
were correlated with the GI tract histological scores using Spearman’s rank/ordinal
correlation test to assess the possible influence of enteritis on the assessment of
neurological signs.
Rotarod. This test was used to examine balance and motor function in mice. No
significant differences in time spent on the rotarod wheel were observed between
treatment groups. Both antibiotic treated and untreated mice infected with C. jejuni
260.94 were observed to have shorter times in weeks 5 and 6. As seen in Experiment 2,
there was also a decrease in time on the rotarod throughout the entire experiment in
mice of all four treatment groups (Figure 2.6 panel B).
DigiGait™. This phenotypic test was introduced in this experiment to replace
the footprint analysis used in the previous two experiments. Mice were videotaped from
underneath while walking on a clear treadmill belt, and gait analysis was then performed
using proprietary software supplied with the apparatus. We were able to apply a visual
clinical evaluation of neurological changes by watching the videos of the mice and
scoring them as shown in Table 2.1.
Stride Length and Hind Base Width. These measurements could not be obtained
for mice that either: (1) were uncooperative and unable to be assessed in a given week
or (2) whose videos were recorded but were unable to be analyzed by the software. The
large number of missing measurements resulted in failure of all statistical tests and
therefore these data were not analyzed (data not shown). Mice that were uncooperative
never developed neurological disease based on visual observation and rotarod data.

68

Clinical Analysis. Comparison of both the cumulative clinical scores given to mice
based on any neurological phenotype noted in the videos captured by the DigiGait™
apparatus and the GI histological assessment resulted in no statistical association
between the enteritis and neurological phenotypes either when all mice were
considered together (Spearman's rank/ordinal correlation; p=0.87) or when treatment
groups were analyzed separately (p=0.77 for C. jejuni-infected mice, p=0.41 for C.
jejuni-infected and CMP-treated mice; p=0.20 for sham-inoculated mice, and p=0.42 for
sham-inoculated and CMP-treated mice) (Figure 2.6 panel D). The scores of each week
for mice that had a video for the entire length of the experiment were analyzed with a
two-way ANOVA with repeated measures and no significant changes within treatment
groups were observed.
The heat map in Figure 2.7 is based on weekly clinical neurological scores of
individual mice based on DigiGait™ videos using a reduced scale; only mice for which
videos were obtained in all weeks are included. To construct the heat map, the features
examined in the videos were simplified to stance abnormality, gait abnormality, and
other (toe abnormality and general weakness). The additional feature of balance
(rotarod time<40 secs) was also included in the heat map. A mouse could present in
one of 5 features: (1) no features, (2) 1 feature, (3) 2 features, (4) 3 features, or (5) all
features, which are demonstrated by intensity of color in the heat map. The heat map
thus summarizes the clinical histories of the individual mice. While mice of all treatment
groups displayed neurological signs, these signs tended to be more severe and to
persist longer in the three groups of treated mice than in sham-inoculated mice. Also,
included in the heat map is a summary of mice positive or negative for anti-ganglioside

69

antibodies. Mice positive or negative for anti-ganglioside antibodies were determined by
using two standard deviations from the mean of the control mice. As seen in the map
two thirds of the mice infected with C. jejuni 260.94 and no treatment were positive for
presence of anti-ganglioside antibodies. All other treatment groups had less than two
thirds of mice positive for anti-ganglioside antibodies.

Presence of nerve lesions and detection of cellular infiltrates
At necropsy (84 days), the entire sciatic nerve, including the nerve root(s), was
removed, fixed, and stained with hematoxylin and eosin for histologic examination of
cellular infiltrates. The brachial plexus was also removed, fixed, and examined. Blinded
slides were read by a single pathologist (BJG). In some of the samples, hematoxylin
and eosin staining revealed rare macrophages with small, round, clear intracytoplasmic
vacuoles; these will be confirmed with IHC staining of F4/80 for macrophages. The
Fisher Exact probability test indicated no significant differences in the presence of
vacuoles between treatment groups. None of the vacuoles were stained with Luxol fast
blue, indicating that they did not contain myelin. In addition, Luxol fast blue staining did
not indicate myelin loss. No significant differences were found between groups in scores
derived from hematoxylin and eosin staining of nerve roots, sciatic nerve and brachial
plexus (data not shown).

Detection of macrophages in nerves
The hematoxylin and eosin staining of the sciatic nerves did not indicate the
presence of cells with T cell morphology in nerves of mice any of the treatment groups

70

and therefore immunohistochemical staining for T cells was not performed.
Macrophages in nerves were examined in mice from the TSB with no treatment group
and the C. jejuni with no treatment group. Only a subset of mice from each group was
analyzed because the nerve sections were very small and not all mice had nerve
remaining on the paraffin block after the histological stains were performed. No
significant differences in the infiltration of macrophages were seen in the sciatic nerve or
brachial plexus. As seen in Figure 2.8, the nerve root did have significant differences of
infiltration of macrophages between the uninfected mice and the infected mice (MannWhitney U; p= 0.0436). Also, based on our results we were able to conclude that the
vacuoles observed with hematoxylin and eosin were fat vacuoles and not macrophages.

71

DISCUSSION

Small animal models are needed to explore pathogenesis, treatment, and
prevention of GBS. C. jejuni-associated GBS in humans is characterized by production
of anti-ganglioside antibodies, clinically evident impairment of the peripheral nervous
system, and histological evidence of damage to peripheral nerves (7, 38, 72). We
addressed our hypothesis by screening three inbred strains of NOD mice for
susceptibility to these three autoimmune manifestations following a single oral
inoculation with GBS-associated strains of C. jejuni. These C. jejuni strains were
isolates from patients who developed GBS, thereby increasing the likelihood of mice
developing neuropathy. The outcomes were assessed using ELISA, clinical scoring and
neurologic assessment tools developed previously (22), and histopathological
examination of nerve tissue.
Based on our findings we have concluded that NOD mice, specifically NOD WT
mice, have the greatest potential to be a murine model of GBS disease following C.
jejuni infection. All NOD mice developed anti-ganglioside antibodies and exhibited
clinical neurological signs; however, the anti-ganglioside antibodies and their isotypes
varied between mouse genotypes. There was also variation between mouse genotypes
in the neurological clinical signs exhibited and it was difficult to demonstrate nerve
damage in histological analysis of sections of nerves in all mouse genotypes.
Macrophage infiltrates in the nerve root of infected NOD WT and IL-10

-/-

mice were

notable, but cellular infiltrates other than macrophages were not found to be significant
in any mice infected with C. jejuni GBS isolates. It is clear that because of their small

72

size it is difficult to demonstrate nerve lesions in peripheral nerves of mice without
preliminary studies to allow specialized techniques to be employed on the relevant
structures.

Comparison of NOD mouse genotypes
-/-

In our studies we were able to conclude that NOD WT, NOD IL-10 , and NOD
B7-2

-/-

mice can all be colonized with either C. jejuni HB93-13 or C. jejuni 260.94. Mice

infected with C. jejuni HB93-13 displayed severe gross pathological changes in the GI
tract in the first experiment. Such changes could mask any neurological signs that were
present. Severe gross pathological changes in the GI tract were also associated with
enteric disease which resulted in early euthanasia of affected mice. For these reasons
C. jejuni HB93-13 was discontinued in Experiments 2 and 3.
NOD WT mice are all colonized by C. jejuni but experience minimal enteritis,
produce anti-ganglioside antibodies, and have a significant infiltration of macrophages
in nervous tissue after infection with C. jejuni when compared to sham-inoculated mice.
C. jejuni infected mice developed a strong reaction to C. jejuni bulk antigen reactive with
IgG2b. These mice also had significant differences of IgG3 and IgG1 antibody reactive
to ganglioside GD1a and ganglioside mixture GM1/GQ1b respectively, when compared
to their uninfected counterparts. Significant differences between the uninfected mice
and C. jejuni infected mice were detected in stride length and hind base width. The C.
jejuni 260.94 infected NOD WT mice had significant differences in infiltration of the
nervous tissue by macrophages, but did not exhibit significant differences in myelin
damage in these nerves at the light microscopic level. One disadvantage of the NOD

73

WT mice is that they begin to develop diabetes with age; diabetes could result in
decreased sample sizes in long-term experiments.
NOD IL-10

-/-

mice developed a strong immune response to C. jejuni and

developed anti-ganglioside antibodies to individual gangliosides GM1 and GD1a. They
also developed antibodies to a mixture of gangliosides GM1 and GQ1b and a mixture of
GM1 and GD1a. Stride length and hind base width measurements also varied
significantly between the sham-inoculated mice and those infected with either C. jejuni
HB93-13 or C. jejuni 260.94. The NOD IL-10

-/-

mice also had a significant difference

between the uninfected mice and infected mice in the infiltration of macrophages in the
nerve root.
C. jejuni infected NOD IL-10

-/-

mice developed enteric disease in Experiment 1

and Experiment 3. Therefore in Experiment 3, a group of C. jejuni-infected mice was
treated with chloramphenicol (CMP) in an attempt to prevent enteric disease from
occurring and interfering with neurological assessments. Mice that received C. jejuni
with CMP still exhibited severe gross pathological changes and high histological scores,
similar to those seen in the mice infected with C. jejuni only; suggesting that the CMP
may have changed the microbiota of the gut and caused GI irritation. Another
explanation is that these IL-10 deficient mice exhibited the high GI histological scores
and gross pathological features observed in the experiments because they were unable
to down-regulate the inflammation initiated by C. jejuni infection. The high histological
scores at the ileocecocolic junction and high levels of gross pathological changes noted
in the NOD IL-10

-/-

mice also demonstrate that an increase in damage to the gut does

74

not necessarily result in an increase in neurological clinical signs. The NOD IL-10

-/-

mice did exhibit significant differences between uninfected and infected mice with the
presence of anti-ganglioside antibodies. However, to continue work with NOD IL-10

-/-

mice as a GBS model it would be advisable to use GBS strains that are less likely to
elicit spontaneous colitis in order to keep enteric disease to a minimum to maintain the
sample sizes. Alternately, immunizing these mice with C. jejuni LOS that mimics
gangliosides could be performed to develop neurological disease without enteritis.
NOD B7-2

-/-

mice had significant differences in anti-GD1a antibody responses to

IgG3 between the sham-inoculated mice and the C. jejuni 260.94 infected mice. There
were also significant differences between the sham-inoculated mice and the infected
mice in both the hang test and open field test. A potential disadvantage to the use of
NOD B7-2

-/-

mice is that they develop spontaneous autoimmune peripheral neuropathy

that resembles human GBS past 20 weeks of age. An extensive study was performed in
our laboratory to examine the onset of neurological disease in our NOD B7-2

-/-

mice

because the phenotype was not well described. Results indicated that NOD B7-2

-/-

mice

do not develop neuropathy prior to 20 weeks of age (Gadsden, unpublished). Therefore,
all experiments with these mice were ended prior to 20 weeks of age to prevent any
overlap of the spontaneous neuropathy and neuropathy triggered by C. jejuni infection.
This strategy of using these mice at a younger age must be advocated for future work
on inducible GBS.

75

Difficulties encountered and their resolutions
Some difficulties encountered were identifying the right ELISA assays to use,
determining the proper phenotypic tests to use to assess neurological clinical signs, and
finding the correct location of nerve lesions in the peripheral nervous system. Two
studies by Kaida et al. (2004 and 2007) showed that sera from GBS patients that are
negative for presence of individual anti-ganglioside antibodies can be positive for
antibodies reactive with ganglioside mixtures (85, 86). As seen in Table 2.3, we saw low
levels of anti-ganglioside antibodies reactive with individual gangliosides, so we
examined presence of antibody reactive to a mixture of gangliosides and detected
significant differences between uninfected and infected mice in antibody levels reactive
with the mixtures of gangliosides.
We concluded that the DigiGait™ was sensitive enough to detect the subtle
neurological changes that occurred; however, the speed (20 cm/sec) that we used in
Experiment 3 was not fast enough to allow for consistent running of all mice. This
resulted in the DigiGait™ software not being able to statistically analyze all videos; in
other experiments we increased the belt speed to 30 or 40 cm/sec and obtained better
results; (Gadsden; unpublished data). Data from our clinical evaluations indicated
significant differences in neurological signs in the uninfected mice and infected mice,
suggesting nerve lesions may be present. Only a small section of the nerve was
removed in Experiment 2 and no lesions were detected, so we improved our technique
in Experiment 3 by removing the sciatic nerve from the tarsus to the spinal nerve root
and removed the brachial plexus; however, no significant differences were detected

76

between the uninfected mice and those infected with C. jejuni; this will be discussed
below.

Future work
In future experiments we will examine the neuromuscular junction (NMJ); it is
highly enriched with gangliosides (157). It also does not have the blood-nerve barrier
present at other sites along the nerves; therefore, antibodies reactive to the
gangliosides have ready access to the NMJ (157).
Another strategy would be staining for components of the complement system
associated with the nerve to detect sites of damage as well as providing information on
the mechanism of damage. Activation of the complement system has been implicated in
damage to the axon of the nerve in GBS patients (53). Other evidence has been
presented showing that IgG antibody and components of the complement system, such
as C5a, are localized on the motor nerve axons, specifically the axolemma, of AMAN
patients (59). Complement components were also shown to bind to Schwann cells of
AIDP patients (60). These models (especially the NOD WT) could serve to confirm
these mechanisms if the correct nerve locations are chosen and the correct fixation and
level of examination (e.g. transmission electron microscopy) are conducted.
As described above, NOD WT mice have the greatest potential to be useful as a
working murine model to study GBS following infection by C. jejuni. NOD WT mice have
been shown to be colonized with GBS-associated C. jejuni with minimal enteric disease,
produce anti-ganglioside antibodies, display clinical neurological signs, and have influx
of activated macrophages present in the peripheral nerves. However, we use the term

77

“working model” because NOD WT mice need to be further assessed before they can
be considered a true animal model of GBS. In particular, a time course experiment with
these mice should be performed to measure peaks of antibody titers to describe the
time of onset of disease in detail. ELISA assays require optimization to provide stronger
evidence that anti-ganglioside antibodies are produced that react with the mixture of
gangliosides. The neurological disease phenotypes of mice need to be analyzed using
the DigiGait™, this apparatus has the potential to be the most sensitive means of
detecting neurological clinical signs. Furthermore, these mice should be examined by
other methods of identifying nerve lesions, immune cell or infiltrates, and complement
binding to nervous tissue by staining for these immune system components. If the
results of these tests show significant differences between mice infected with C. jejuni
and uninfected mice, then validation of murine model to study the natural onset of GBS,
would be complete and provide proof that molecular mimicry is the pathogenic
mechanism behind GBS.

78

Acknowledgements: The authors thank Drs. Jon S. Patterson and Howard T. Chang for
their contribution in the development of a histological criterion to score nerves.

79

Table 2.1: Clinical Assessment for DigiGait™ videos. Scoring method used for
visual assessment of clinical neurological signs in the videos obtained weekly using the
DigiGait™ apparatus. Each sign was assigned a score as shown; scores were added
for each mouse at each week. Cumulative scores were used in statistical tests.

Score Methodology for Clinical Assessment of DigiGait™ videos

Sidestepping
Crossing Over
Bunny Hopping
Paddling
Galloping
Rolling
Shortened stance
Widen stance
Knuckled Over
Weakness

Absent
0
0
0
0
0
0
0
0
0
0

Present
1
1
1
1
1
1
1
1
1
1

Repeated
2
2
2
2
2
2
NA
NA
NA
NA

Splayed Hind Leg
Foot Drag

Absent
0
0

Present in 1 Leg
1
1

Present in both Legs
2
2

Toes

Absent Toes Splayed Apart
0
1

80

Toes Curled
2

Toes Flaccid
3

Table 2.2: Statistical results for change within treatment groups over the
course of the experimental time. The table below shows results from two-way
ANOVA with repeated measures analysis of change in time over the course of the
experiment. No significant result obtained is designated . Abbreviations: NS, not
significant; NA, not applicable.

Experiment 1

Experiment 2

Experiment 3

NOD WT NOD IL-10-/- NOD B7-2-/- NOD WT NOD B7-2-/- NOD IL-10-/OFT
NS
p=0.0447
p<0.0001 p<0.0001 p<0.0001
NA
Hang Test p<0.0001 p<0.0001
p<0.0001
NS
NS
NA
Rotarod
NA
NA
NA
p<0.0001 p<0.0001
p<0.0001
Stride
Length
NA
NA
NA
p<0.0001 p<0.0001
NA
Hind Base
Width
NA
NA
NA
p=0.0013 p=0.0004
NA

81

Table 2.3: P values for ELISA assays for ELISA assays for all 3 experiments. Abbreviations: 1W, one-way
ANOVA; KW, Kruskal-Wallis ANOVA; MW, Mann-Whitney U.
NOD WT

Experiment 1
NOD IL-10-/- NOD B7-2-/-

Experiment 2
Experiment 3
NOD WT
NOD B7-2-/- NOD IL-10-/-

ELISA
GM1 (IgG1)
GM1 (IgG2a)
GM1 (IgG3)

1W; p=0.2933 1W; p=0.6478 KW; p=0.1909 MW; p=0.4275 MW; p=0.5148 KW; p=0.8321
KW; p=0.4515 KW; p=0.7704 1W; p=0.6926 MW; p=0.2634 MW; p=0.7087 KW; p=0.6412
1W; p=0.8760 1W; p=0.2671 1W; p=0.4883 MW; p=0.9232 MW; p=0.7474 KW; p=0.0217

GM1 (IgM)

1W; p=0.8160 1W; p=0.2149 1W; p=0.9403 MW; p=0.6657 MW; p=0.8699 KW; p=0.9319

GD1a (IgG1)

KW; p=0.4051 KW; p=0.8051 1W; p=0.5113 MW; p=0.3232 MW; p=0.5148 KW; p=0.0030

GD1a (IgG2a)
GD1a (IgG3)

1W; p=0.8416 1W; p=0.7702 1W; p=0.1925 MW; p=0.3248 MW; p=0.8392 KW; p=0.2208
1W; p=0.8430 KW; p=0.5369 1W; p=0.2028 MW; p=0.0296 MW; p=0.0395 KW; p=0.0421

GQ1b (IgG2a)

1W; p=0.6931 1W; p=0.5181 1W; p=0.6903 MW; p=0.2686 MW; p=0.3751 KW; p=0.0029

GM1/GD1a (IgG1)

1W; p=0.3624 KW; p=0.1086 1W; p=0.7989 MW; p=0.3795 MW; p=0.7132 KW; p=0.0332

GM1/GQ1b (IgG1)

1W; p=0.0100 1W; p=0.9237 1W; p=0.8887 MW; p=0.0351 MW; p=0.5361 KW; p=0.0017

Cj (IgG2b)

KW; p=0.0004 KW; p=0.0001 KW; p<0.0001 MW; p<0.0001 MW; p<0.0002 KW; p<0.0001

Day 0 GD1a (IgG1) KW; p=0.9020 1W; p=0.2053 1W; p=0.9429 MW; p=0.7498 MW; p=0.9382
Day 14 GD1a (IgG1) 1W; p=0.5481 1W; p=0.1695 1W; p=0.9854 MW; p=0.2292 MW; p=0.4709

NA
NA

Day 0 GM1 (IgM)
Day 14 GM1 (IgM)

NA
NA

1W; p=0.7371 1W; p=0.5059 1W; p=0.9255 MW; p=0.0229 MW; p=0.9545
1W; p=0.7070 1W; p=0.6219 KW; p=0.3752 MW; p=0.2683 MW; p=0.2230

82

Figure 2.1: Experiment 1 Data Summary. Panel A shows the number of NOD WT,
NOD IL-10-/-, and NOD B7-2-/- mice colonized in the cecum with C. jejuni HB93-13
and C. jejuni 260.94. Panel B shows the levels of plasma IgG2b antibody reactive with
C. jejuni in all mice (B-1) and the plasma levels of IgG1 antibody reactive with the
ganglioside mixture GM1/GD1a (B-2). Panel C shows the number of quadrants crossed
in the open field test for only the NOD WT mice. Panel D shows the hang time for mice
of all genetic backgrounds during the course of the experiment with the hang test. The
Holm-Sidak pairwise test found significant differences in hang times between shaminoculated NOD WT mice and C. jejuni 260.94 infected NOD WT mice at week 1
(p≤0.05) and week 3 (p≤0.01). Unless stated otherwise, significance levels in all graphs
are denoted by * (p≤0.05); ** (p≤0.01); *** (p≤0.001); and **** (p≤0.0001).

A

Colonization of Cecum at Endpoint (56 Days)

83

Figure 2.1 (cont’d)
B-1

B-2

84

Figure 2.1 (cont’d)
C

D

*Significant differences in hang time are noted in the legend.

85

Figure 2.2: Experiment 2 Data Overview. Panel A -1 shows the colonization levels of
C. jejuni 260.94 in the ceca of the mice at endpoint of experiment (80 days). Panel A-2
indicates the gross pathological features present in mice at time of necropsy. Gross
pathological features examined were enlarged lymph nodes and thickened wall of the
GI tract. Panel B shows ELISA results. Panel B-1 shows plasma levels of antibody
IgG3 reactive with the ganglioside GD1a and panel B-2 demonstrates the plasma
levels of IgG1 antibody reactive with ganglioside mixture GM1/GQ1b. Note that only
NOD WT mice had significant differences in plasma levels of IgG1 to ganglioside
mixture GM1/GD1a. Panel C shows the number of times a quadrant was crossed over
the course of the experiment; panel C-1displays the NOD WT mice and panel C-2
shows the NOD B7-2-/- mice. Panel D shows the time spent on the rotarod by NOD
WT mice (panel D-1) and NOD B7-2-/- mice (panel D-2). Unless stated otherwise,
significance levels in all graphs are by

* (p≤0.05), ** (p≤0.01), *** (p≤0.001) and

**** (p≤0.0001).

86

Figure 2.2 (cont’d)
A-1

A-2

87

Figure 2.2 (cont’d)
B-1

B-2

88

Figure 2.2 (cont’d)
C-1

C-2

89

Figure 2.2 (cont’d)
D-1

D-2

90

Figure 2.3: Examination of Peripheral Nerves. Panel A shows a Luxol fast blue
stain of a sciatic nerve section from a sham-inoculated NOD B7-2-/- mouse; panel B
shows a Luxol fast blue stain of sciatic nerve section from a C. jejuni 260.94 infected
NOD B7-2-/-mouse. There is myelin loss visible in panel B, but as seen in panel A,
the nerve from the uninfected mouse has intact myelin. Panels C and D show H & E
stained sections of sciatic nerve section from a NOD B7-2-/- sham-inoculated mouse
and a C. jejuni 260.94 infected mouse, respectively. The arrows in panel D indicate
cellular infiltration within the nerve. All panels are shown at the same magnification
(40X). For interpretation of the references to color in this and all other figures, the
reader is referred to the electronic version of this dissertation.

A

B

C

D

91

Figure 2.4: Immunohistochemical Data for Experiment 2. Panel A shows the
number of cells that stained positive for CD3 (T cells). Panel B shows the number of
cells that stained positive for F4/80 (activated macrophages). Significance is noted
with * (p≤0.05).
A

B

92

Figure 2.5: Macrophage infiltration of NOD WT mice. Panels A and C show sciatic
nerve sections from sham-inoculated NOD WT mice stained with F4/80; panels B and
D show a F4/80 stain of sciatic nerve sections from C. jejuni 260.94 infected NOD WT
mice. There is visible infiltration of macrophages in panels B and D, but none is seen
in panels A and C. The black arrows indicate stained macrophages. Panels A–D are
shown at the same magnification (20X). Panels F–I show the same sciatic nerve
sections at 40X magnification; panels F and H are sham-inoculated NOD WT mice and
panels G and I are C. jejuni 260.94 infected NOD WT mice.

A

B

C

D

93

Figure 2.5 (cont’d)

F

G

H

I

94

Figure 2.6: Experiment 3 Data Summary. Panel A shows colonization levels of C.
jejuni 260.94 in the cecum at the end of the experiment (84 days). Panel B shows the
time spent on the rotarod over the course of the experiment. Panel C shows the
ELISA assay data for plasma levels of IgG1 reactive with ganglioside mixture
GM1/GQ1b in panel C-1; ganglioside mixture GM1/GQ1b is in panel C-2. There were
significant differences between the sham-inoculated mice and C. jejuni infected mice
treated with CMP in both assays. Panel D displays the cumulative scores of the
clinical assessment of neurological signs using the videos obtained from the
DigiGait™; only mice that had videos for the entire length of the experiment are
shown. The black line indicates the median of scores for each time point. No
significant differences were observed between treatment groups with a two-way
ANOVA with repeated measures. Unless stated otherwise, significance levels in all
graphs are by * (p≤0.05), ** (p≤0.01), *** (p≤0.001) and **** (p≤0.0001).

95

Figure 2.6 (cont’d)
A

B

96

Figure 2.6 (cont’d)
A

C-1

C-2

97

Figure 2.6 (cont’d)

12

DigiGait Assessment of Clinical Signs

Clinical Score

10
8
6
4
2
0

0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
TSB only
C. jejuni only
TSB + CMP
C. jejuni + CMP
Weeks Post Inoculation

98

Figure 2.7: Heat Map of Clinical Neurological Assessment. This heat map
demonstrates the assessment of clinical neurological signs obtained from DigiGait™
videos and from rotarod testing for each mouse that provided videos for entire
experiment (84 days). To construct this graph, features observed in the primary
assessment of clinical signs with the initial scoring of the video were categorized into
3 main features: 1) gait abnormalities, 2) stance abnormalities, and 3) other (toe
abnormalities and general weakness). Balance was also assessed; a mouse was
classified as having an abnormality in balance if spent <40 secs on the rotarod. Thus,
a mouse could be assigned to one of 5 groups: 1) no features, 2) 1 feature, 3) 2
features, 4) 3 features, or 5) all features. These features are expressed based on
color intensity of the weekly rectangles on the heat map. To the far right of the heat
map the histological scores of the ileocecocolic junction are shown. As seen on the
graph, a high histological score did not interfere with assessment of clinical
neurological signs. Presence of anti-ganglioside antibodies is also shown on the far
right of the heat map. Over two-thirds of mice that had anti-ganglioside antibodies
were C. jejuni infected mice not treated with CMP.

99

Figure 2.7 (cont’d)
Weeks Post Inoculation
0 1 2 3 4 5 6 7 8 9 10 11 12 GI Ab

TSB only

Neurological Signs
0 Features
1 Feature
2 Features
3 Features
All Features

C. jejuni Only

GI Histological Score
0≥10
10≥20
>20

C. jejuni + CMP

TSB + CMP

Anti-ganglioside Antibodies
Positive
Negative

100

Figure 2.8: Infiltration of macrophages in the nerve root. Panels A and B are
nerve root sections stained with F4/80 (activated macrophages). Panel A is a shaminoculated mouse and panel B is a C. jejuni infected mouse. The black arrows
indicated macrophages. Panel C shows graphically the presence of activated
macrophages in the nerve root sections of C. jejuni mice not treated with CMP and
sham-inoculated mice not treated with CMP. Significance levels are noted by *
(p≤0.05).

101

Figure 2.8 (cont’d)

C

102

CHAPTER 3

Examination of Miller Fisher Syndrome in NOD WT and C57BL/6 WT mice subsequent
to Campylobacter jejuni CF93-6 infection

103

ABSTRACT

Acute neuropathies Guillain Barré Syndrome (GBS) and Miller Fisher Syndrome
(MFS) can follow Campylobacter jejuni infection. GBS is an ascending symmetrical
paralysis affecting the peripheral nervous system. MFS is a descending neuropathy
affecting the central nervous system before affecting peripheral nerves. Lipooligosaccharides (LOS) of C. jejuni mimic nerve gangliosides in the peripheral nervous
tissue of the host, resulting in autoimmune attack on host gangliosides. Antibodies to
gangliosides GM1 and GD1a are associated with GBS and antibodies to GQ1b with
MFS. We hypothesized that NOD WT and C57BL/6 WT mice infected with MFSassociated C. jejuni CF93-6 would exhibit neurological signs, autoantibodies, and
lesions consistent with MFS; whereas NOD WT mice infected with GBS-associated C.
jejuni 260.94 would exhibit features of GBS disease. Orally infected mice of both the
NOD and C57BL/6 genetic backgrounds were tested weekly for neurological
phenotypes; GI tract, nerves, and blood were taken to determine C. jejuni colonization,
nerve lesions, and presence of anti-ganglioside antibodies at endpoint. ELISA assays
showed significant differences in the presence of IgG1 antibody reactive to ganglioside
mixture GM1 and GQ1b between sham-inoculated NOD WT mice and C. jejuni 260.94
infected NOD WT mice. No other significant differences in levels of IgG1 or any other
IgG isotype tested were observed in either the NOD WT C. jejuni CF93-6 infected mice
or between the uninfected mice and infected mice of the C57BL/6 WT mice. For all
phenotypic tests that were performed (open field test, hang test, rotarod, and footprints),
there were no significant differences between the uninfected and C. jejuni infected mice

104

in either the NOD WT or C57BL/6 WT mice. Therefore no clinical manifestation of
neurological disease was observed. Examination of sections of the sciatic nerve for
cellular infiltrates and nerve lesions revealed no significant changes between the
treatment groups of either background. These results suggest that there is no
manifestation of MFS caused by C. jejuni CF93-6 in either C57BL/6 or NOD mice that is
distinguishable from that of the subclinical manifestation of GBS (anti-ganglioside
antibody production) caused by C. jejuni 260.94 in NOD mice in this experiment. They
also suggest that infection with C. jejuni CF93-6 does not result in a neurological
phenotype in NOD WT or C57BL/6 WT mice.

105

INTRODUCTION

Guillain-Barre Syndrome (GBS) is a post-infectious autoimmune-mediated
neuropathy of the peripheral nerves (7, 63, 76). GBS is described as an ascending
symmetrical paralysis with limb weakness, acute flaccid paralysis, and in some
instances sensory loss; muscle weakness frequently affects the respiratory and facial
muscles (7, 147, 161). GBS is defined as many clinical subtypes: 1) acute inflammatory
demyelinating polyneuropathy (AIDP), 2) acute motor axonal neuropathy (AMAN), and
3) Miller Fisher Syndrome (MFS). Studies of pathological changes in AIDP show
macrophages invading the myelin sheath, resulting in myelin damage and
demyelination (134, 147, 154, 160). AIDP is the most common form in the Western
world, affecting approximately 69–90% of GBS patients (134, 147). The clinical subtype
AMAN is more common in China and Central and South America, affecting 38–65% of
GBS patients; only 5–10% of GBS patients in the Western world present with AMAN
(134, 147, 160). Acute motor sensory axonal neuropathy (AMSAN) has also been
recognized as a subtype that affects both the motor and sensory nerves. The targets of
the macrophages are different in AMAN than in AIDP; in AMAN the macrophages target
the nodes of Ranvier and the axolemma of the nerve (147, 161). Worldwide, the mean
annual incidence rate of GBS is 0.6–4 cases per 100,000 with males more often
affected than females (79, 154).
Unlike GBS, MFS is characterized by a descending neuropathy that affects the
central nervous system, especially the brainstem, before affecting the periphery (97).
The classic triad of MFS symptoms includes: ataxia, areflexia, and opthalmoplegia;

106

though not all symptoms are always present (93, 137, 138, 159). MFS is more
commonly found in patients in eastern Asia; 20% of Taiwanese and 25% of Japanese
GBS patients present with MFS (134, 163). Symptoms of MFS peak at 1 week, and
patients tend to begin recovery 1–3 months later, with most patients making a complete
recovery by 6 months (97, 163). The mean annual occurrence of MFS is 0.09 cases per
population (97). MFS is less common than GBS, yet can still produce lifelong
debilitating symptoms.
The onset of GBS is usually preceded by respiratory or gastrointestinal
infections with 2/3 of patients reporting antecedent illness (72). Campylobacter jejuni, a
gram negative bacterium, is the most common preceding infection leading to the onset
of GBS and MFS (67, 76, 105, 147, 162). The role of C. jejuni in the onset of GBS is
thought to be molecular mimicry (7). The lipo-oligosaccharide (LOS) on the surface of
C. jejuni can mimic gangliosides found on host nerves therefore resulting in an
autoimmune response (7). Gangliosides are membrane glycolipids highly enriched in
the nervous tissue (63, 162). Antibodies to gangliosides GD1a and GM1 are more
commonly associated with people displaying the AMAN subtype of GBS (150, 161).
Approximately 75% of AMAN GBS patients and 10–15% AIDP patients have antibodies
reactive with these gangliosides (157). It has been demonstrated that IgG antibodies
reactive to the ganglioside GQ1b are commonly found in patients displaying MFS
phenotype (80, 90, 134, 150). Approximately 83–100% of MFS patients have antibody
reactive to GQ1b gangliosides; only approximately 26% of GBS patients present with
GQ1b antibody (93, 97, 137).

107

Studies have demonstrated that ganglioside GQ1b is highly enriched in the
rd

th

ocular nerves (147). In MFS, the 3 , 4 , and 6

th

cranial nerves are commonly affected

(153, 154). It also has been demonstrated that ganglioside GQ1b is localized in the
nerve fibers enervating extrinsic muscles of the eye. When the immune system attacks
GQ1b on nerves the result is opthalmoplegia (7). Studies have shown that anti-GQ1b
antibodies bind to and affect the function of the neuromuscular junction (NMJ). One
such study demonstrated that antibodies reactive to GQ1b could fix complement
resulting in disruption to the NMJ (54). Plomp et al. (1999) showed that anti-GQ1b IgM
antibody bound to mouse NMJ, which resulted in a release of large amounts of
acetylcholine from the nerve terminals. This phenomenon eventually led to blocked
transmission at the NMJ (117).
We hypothesized that orally infected mice would present with clinical signs and
anti-ganglioside antibodies similar to those seen in humans following infection with a
MFS associated C. jejuni strain. We used non-obese diabetic (NOD) WT and C57BL/6
WT mice inoculated with C. jejuni strain CF93-6 isolated from a patient with MFS. Also,
we inoculated NOD WT mice with C. jejuni 260.94 from a patient with GBS to serve as a
positive control. Mice were analyzed weekly starting before infection for clinical
neurological signs using phenotypic tests; plasma was collected to detect presence of
anti-ganglioside antibodies, and the sciatic nerve was removed to examine cellular
infiltrates and loss of myelin. Our results showed that mice of both genetic backgrounds
can be colonized by the MFS-associated C. jejuni strain CF93-6 with minimal gross
pathological features in the gastrointestinal tract. The NOD WT mice infected with C.
jejuni 260.94 had significant differences in plasma levels of IgG1 antibody reactive to

108

ganglioside mixture GM1/GQ1b. However, there were no significant differences in antiganglioside antibodies or clinical neurological phenotypes between the sham-inoculated
and C. jejuni CF93-6 infected mice of both genetic backgrounds (NOD WT and C57BL/6
WT). No other significant differences were observed between infected and uninfected
mice of either genetic background. Based on our findings we were not able to
distinguish an MFS clinical phenotype; neither the NOD WT nor the C57BL/6 WT mice
displayed any phenotype. More work needs to be done to define a mouse model of
GBS before distinguishing an MFS-like clinical presentation from a GBS-like
presentation will be possible.

109

MATERIALS AND METHODS

Animals
Non-Obese Diabetic wild type mice (stock # 001976) and C57BL/6 wild type mice
(stock # 00064) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA)
at 6 weeks of age and sent directly to the Michigan State University Research
Containment Facility. Mice were housed individually in sterile filter top cages and
allowed to acclimate for 14 days. All mice were given autoclaved reverse osmosis
water; NOD WT mice were fed sterilized mouse diet 7913 (Harlan Teklad) and C57BL/6
mice were fed sterilized diet 7904 (Harlan Teklad). Fecal pellets were taken and DNA
was extracted using a DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) with
modifications to a published protocol (104) to assess the genotype of all mice.
Genotype was confirmed using a modified PCR assay from The Jackson Laboratory
(http://jaxmice.jax.org/pub-cgi/protocols/protocols.sh?objtype=protocol
&protocol_id=346). Fecal samples were also taken from mice to test for presence of
colitogenic bacteria by PCR prior to inoculation; colitogenic bacteria tested included
Campylobacter spp., Enterococcus faecalis, E. faecium, Helicobacter spp., and
Citrobacter rodentium. At 8 weeks of age, mice were inoculated orally with C. jejuni
strains or sham-inoculated with tryptose soya broth (TSB). During the course of the
experiment mice were monitored daily for clinical signs of enteric disease and weekly
for neurological signs. This animal experiment followed all of the guidelines set by NIH
and was approved by the MSU All University Committee on Animal Use and Care under
protocol number 06/09-092-00.

110

Campylobacter jejuni
C. jejuni strains were both purchased from the American Type Culture Collection
(ATCC, Manassas, VA, USA). C. jejuni 260.94 (ATCC BAA-1234) and C. jejuni CF93-6
(ATCC BAA-1456) were isolated from patients that presented with GBS and MFS
respectively. To prepare the inoculum, C. jejuni strains were incubated for 48 hours at
37°C on tryptose soya agar (TSA) plates with 5% sheep’s blood (Cleveland Scientific,
Bath, OH, USA) in anaerobic jars containing CampyGen packs (Oxoid, Basingstoke,
UK) (98). Bacterial growth was collected after two rounds of growth and mixed into a
single master tube of TSB to a final concentration of ~5 x 10

10

CFU/ml (1:10 dilution

approximately 1.0 at OD600). Motility of bacterial cultures was confirmed in a wet mount
and a Gram stain was used to confirm gram negative rods (98).

Experimental Design
The experiment consisted of 5 groups: 1) 15 mice of C57BL/6 WT background
sham-inoculated with TSB, 2) 15 C57BL6/WT mice infected with MFS-associated C.
jejuni CF93-6, 3) 14 sham-inoculated NOD WT mice, 4) 16 MFS-associated C. jejuni
CF93-6 infected NOD WT mice, and 5) 15 NOD WT mice infected with GBS-associated
C. jejuni 260.94. During the course of the experiment mice were tested weekly for
neurological signs with three phenotypic tests and monitored daily for enteric disease.
Mice were euthanized and necropsied at day 53. At necropsy sections of the GI tract,
sciatic nerve, and blood were taken for examination of the presence of C. jejuni

111

colonization and histological evidence of enteritis, lesions and cellular infiltrates in nerve
tissue, and presence of anti-ganglioside antibodies, respectively. Detailed methods are
given below.

Neurological Testing
C. jejuni infected mice and sham-inoculated mice were analyzed for clinical signs
of neurological disease through a series of phenotypic tests given on a weekly basis.
Prior to inoculation with C. jejuni, mice were trained on each test to remove any
influence of a learning curve (22). The tests used in this experiment included footprint
measurements, open field test, and rotarod.

Open field testing
The open field test was used to examine mice for abnormal gait. Mice were set in
the center of a polycarbonate box with four quadrants marked on the bottom and
watched for 30 seconds. Mice were assessed for the number of quadrants crossed,
rearing, and abnormal walking. Mice performed this test 3 times with 15 minute breaks
in between each test interval (14, 24).
During the first interval, mice were recorded using a video camera and the
resulting videos were analyzed with JWatcher™ 1.0 software (Daniel Blumstein
Laboratory, University of California Los Angeles, and The Animal Behaviour Lab,
Macquarie University, Sydney, Australia), by an individual blinded to the group identity
of the mice. JWatcher™ is a computer software program designed to assess and score
the movements of an animal (12). With this software, the videos recorded during the

112

open field test were analyzed by using a keystroke for each movement/clinical sign
detected during the duration of the recorded time. The software then computed the
number of times a keystroke was made and the length of time that keystroke was
maintained (i.e., assessing the length of time a clinical sign was observed) (12).
During the second and third test intervals the number of quadrants crossed was
the only parameter recorded. These results were included in the final analysis with the
number of quadrants recorded with the JWatcher™ program during the first interval.
The number of times each mouse reared was analyzed using JWatcher™ during the
first interval only. Abnormal gait was also recorded if it occurred.

Rotarod testing
A rotarod (IITC Rotarod for Rats and Mice, IITC Inc. Life Science, Woodland
Hills, CA) was used to assess the balance of all mice. Mice were placed on the rotating
wheel for 60 seconds. The speed was set at 3 RPM and was accelerated 0.10 RPM per
second to a final speed of 30 RPM. Thereafter the length of time the mouse remained
on the wheel was recorded. This test was repeated three times with a 15 minute rest
between each test: the longest time spent on the wheel was used for analysis (14, 22).

Footprinting and tracking
Footprint measurements were taken to detect abnormal gait, foot drag, and
splayed limbs (29). To assess walking, we prepared a runway for the mice by cutting a
longitudinal slot out of a cardboard tube to monitor the movement of each mouse. A
filter paper strip was cut to fit the length of the tube and cover the bottom surface. The

113

fore and hind paws of each mouse were painted with different colored non-toxic paint
and the mouse was placed at one end of the tube and allowed to walk to the other end.
The filter paper provided a record of the footprints and distinguished fore and hind feet.
Measurements were then taken by identifying the midpoint of each paw print and
measuring the distances between fore paws and hind paws for strides and stances. The
spread of the toes and length of the paw print were also measured. Stride length and
hind base width were then analyzed and compared between groups (14, 22, 29).

Enzyme-Linked Immunosorbent Assays (ELISA)
Prior to inoculation and 14 days post inoculation, mice were bled from the
saphenous vein. At necropsy mice were bled from the heart post euthanasia. Blood
samples were then centrifuged at 3000 RPM for 5 minutes to collect the plasma. ELISA
assays were performed to detect C. jejuni IgG2b antibody as previously described (98).
ELISA assays were performed to detect the presence of antibodies directed against
gangliosides GQ1b, GM1, and GD1a (Calbiochem/EMD Millipore, Gibbstown, NJ, USA
at 0.2 μg /ml, US Biological, Swampscott, MA, USA at 2.0 μg/ml, and Sigma Aldrich, St.
Louis, MO, USA at 20 μg /ml respectively). Anti-GQ1b was purchased from Associates
of Cape Cod (East Falmouth, MA, USA); anti -GD1a and -GM1 were purchased from
Chemicon/EMD Millipore (Gibbstown, NJ, USA) and were used as positive controls.
ELISA assays were also performed to detect IgG1 antibodies reactive with a 1:1 mixture
of two gangliosides; a mixture of GM1/GQ1b or a mixture of GM1/GD1a. Secondary
antibodies used included anti-mouse IgM (Sigma Aldrich, St. Louis, MO, USA) and antimouse IgG1, IgG2a, IgG2b, IgG2c, and IgG3 (Jackson ImmunoResearch, West Grove,

114

PA, USA). Detection of antibodies was done using extravidin peroxidase (Sigma
Aldrich, St. Louis, MO, USA) and TMB substrate (Rockland Immunochemicals Inc.,
Gilbertsville, PA, USA). All ELISA assays were performed in triplicate.

Assessment of Clinical Signs
During the course of the experiment mice were monitored once a day until first
onset of illness; thereafter mice were monitored twice daily. Enteric disease was
assessed using clinical signs that included rough hair coat, diarrhea, hunched posture,
and lack of movement as described previously (98). Clinical score sheets were
maintained for any mouse exhibiting clinical signs and severe enteric disease. If a
mouse exhibited a high clinical score, it was euthanized with an overdose of CO2 to
prevent any discomfort or distress. Neurological disease was assessed weekly using
the phenotypic tests described above.

Necropsy Procedure and Assessment of Gross Pathological Features
Mice were humanely euthanized with an overdose of CO2 either at humane
endpoints or at end of experiment (53 days) based on approved methods (1).
Immediately following euthanasia, mice were weighed, blood was taken via the heart,
and a puncture to the diaphragm followed to ensure death of the mouse. The entire GI
tract was excised from stomach to rectum and any gross pathological features were
noted. The GI tract was cut into 4 fragments: 1) stomach-duodenum, 2) jejunum, 3)
ileum-cecum-proximal colon, and 4) distal colon-rectum and placed in 10% phosphate
buffered saline to rinse away any contents in the lumen. These fragments were further

115

sectioned into subfragments of the stomach, jejunum, cecum, and colon which were
further cut into three fragments. One fragment was streaked onto TSA + 5% sheep
blood plates supplemented with antibiotics (TSA-CVA: 20 µg/mL cefoperazone, 10
µg/mL vancomycin and 2 µg/mL amphotericin B); these plates were placed in an
anaerobic jar with CampyGen packs (Oxoid, Basingstoke, UK) and incubated for 48
hours are 37°C. The other two fragments were flash frozen for DNA and RNA
extraction. Portions of the cecum, distal ileum, and proximal colon (ileocecocolic
junction) were all placed on a sponge in a histological cassette and placed in 10%
phosphate buffered formalin for fixation. The remaining portions of the GI tract, kidneys,
liver, lungs, reproductive organs, heart, and brain were all placed in 10% phosphate
buffered formalin for 24 hours for fixation and then transferred to 60% ethanol for
storage. The mouse was skinned and the spinal cord was exposed for fixation in 10%
phosphate buffered formalin. Both hind limbs were removed, skinned, the sciatic nerve
was exposed, and the limb placed in 10% phosphate buffered formalin. Upon return to
the laboratory, the sciatic nerve was then removed carefully under a microscope with
some intact muscle and placed in a histological cassette stored in 60% ethanol. Paraffin
embedding, sectioning, and staining were performed at the Michigan State University
Investigative Histopathology Laboratory.

Assessment of Enteric Disease
The TSA-CVA plates that were streaked with fragments of the four sections of
the GI tract were assessed and scored for the presence of C. jejuni growth. Scoring was
done on a scale of 0–4 based on the amount of bacterial growth; 0 for no growth up to 4

116

for confluent growth as previously described (98). Any plates that had bacterial growth
were swabbed, the bacterial growth suspended in TSB containing 15% glycerol, and the
tubes frozen at -80°C. These cultures were then sampled with a toothpick and whole
cell PCR was used to confirm that the culture was C. jejuni, as previously described
(98). Any mouse that was negative for C. jejuni growth on all GI tissue plates and all
control mice had DNA extracted (Qiagen DNeasy tissue kit, Qiagen, Valencia, CA,
USA) from the frozen cecal tissue, and PCR was performed to verify presence/absence
of C. jejuni.

Histopathological Scoring for the GI Tract and Sciatic Nerve
Gastrointestinal tract
At the time of necropsy the ileocecocolic junction was removed, placed in a
histological cassette and fixed in 10% phosphate buffered formalin and placed in 60%
ethanol. These specimens were sent to the Michigan State University Investigative
Histopathology Laboratory to be paraffin embedded and stained with hematoxylin and
eosin. Sections were blinded and scored for the presence of lesions in the lumen,
lamina propria, epithelium, and submucosa as previously described (98). A score from 0
to 44 was given; slides were read by a single veterinary expert in enteric
disease/lesions (LSM).

Sciatic nerve
The sciatic nerve was exposed at time of necropsy, placed in 10% phosphate
buffered formalin for 6 hours, and then placed in 60% ethanol. Nerves were removed

117

from the entire length of the leg and placed on a sponge in a histological cassette for
paraffin embedment at the Michigan State University Investigative Histopathology
Laboratory. After embedment; two 5 μm sections were cut with a Reichert Jung 2030
rotary microtome. One section was stained with hematoxylin and eosin and the second
section with Luxol fast blue. Nerve sections were then examined and scored as
previously described in Chapter 2 for lesions, myelin degradation, and cellular infiltration
by one veterinary pathologist who was blinded and unbiased (BJG).

Statistical Methods
GraphPad Prism version 6.01 (GraphPad Software Inc., La Jolla, CA, USA) was
used for all statistical analyses. For all statistical tests both the NOD WT and C57BL/6
WT were assessed separately from one another. ELISA assays and histological scoring
were analyzed with either the non-parametric Kruskal-Wallis ANOVA on ranks (NOD
WT) followed by Dunn’s if significant differences were noted or Mann-Whitney U test
(C57BL/6 WT). All phenotypic tests were analyzed with a two-way ANOVA with
repeated measures and if significant results were obtained, the Holm-Sidak post hoc
test was used.

118

RESULTS

Screening for Enteric Disease with GBS-associated and MFS-associated C. jejuni
Strains
Examination of the cultures from GI tract tissue and/or C. jejuni specific PCR was
used to determine colonization by the C. jejuni strains. All cultures derived from C. jejuni
infected mice were confirmed by C. jejuni specific PCR. All 29 sham-inoculated mice
(14 NOD WT and 15 C57BL/6 WT) were both culture negative and PCR negative for
detection of C. jejuni. As seen in Figure 3.1, panel A, 11/16 NOD WT mice and 13/15
C57BL/6 WT mice infected with C. jejuni CF93-6 were colonized at end point; 3 mice (1
NOD WT and 2 C57BL/6 WT) were negative by culture but positive when tested by C.
jejuni specific PCR from DNA taken from cecal tissue. All NOD WT infected with C.
jejuni 260.94 were colonized as detected by culture and/or PCR. One NOD WT mouse
infected with C. jejuni CF93-6 had to be euthanized early because of a high clinical
score for GI disease; no neurological disease was observed.
Gross pathological features were assessed at necropsy and noted. Features that
were noted included presence of enlarged ileocecocolic lymph nodes and thickened
wall of the gastrointestinal (GI) tract (Figure 3.1, panel B). No gross pathological
features were observed in the C57BL/6 WT sham-inoculated mice. Seven of fifteen C.
jejuni CF93-6 infected C57BL/6 WT mice exhibited gross pathological changes. Only
1/14 NOD WT negative control mice presented with one feature. Twelve of thirty-one
infected NOD WT mice presented with one of the features; 6/16 C. jejuni CF93-6

119

infected mice and 6/15 C. jejuni 260.94 infected mice. No mouse of any of the treatment
groups presented with both gross pathological changes.
At necropsy, the ileocecocolic junction was removed, fixed, stained, and scored
as described above. Kruskal-Wallis ANOVA on ranks was used to compare the scores
of the NOD WT mice; no significant differences were observed between the treatment
groups. The Mann-Whitney U test was used to examine the C57BL/6 WT mice; no
significant differences were noted between the uninfected and infected mice. The
histological scores for all three infected groups of mice were minimal; no mouse ever
scored above 10 (data not shown).

Detection of Anti-Ganglioside Antibodies with ELISA Assays
Mice were bled at days 0 and 14 post inoculation from the saphenous vein as
described above. The amount of blood taken at these time points was small and
therefore only one ELISA assay was performed. Antibody IgM levels reactive to
ganglioside GQ1b was measured at these time points. No significant differences were
found between any treatment groups in the either NOD WT or the C57BL/6 WT mice at
day 0 and day 14 post inoculation (Table 3.1).
At necropsy mice were bled from the heart, and plasma was used to detect C.
jejuni-specific and auto-antibodies. ELISA assays were performed to detect the
presence of IgG2b antibody levels reactive to C. jejuni antigen to verify an adaptive
immune response to infection with the C. jejuni strains (Figure 3.2, panel A). Significant
differences in the plasma levels of IgG2b reactive to C. jejuni were observed in the NOD
WT mice using Kruskal-Wallis ANOVA on ranks (p<0.0001). Pairwise post hoc Dunn’s

120

tests indicated significant differences between the sham-inoculated mice and both the
C. jejuni CF93-6 infected mice (p≤0.001) and the C. jejuni 260.94 infected mice
(p≤0.01). Significant differences in IgG2b antibody levels were also noted between
infected and uninfected C57BL/6 WT mice (Mann-Whitney U; p<0.0001). ELISA assays
were also performed for the presence of antibodies IgG1, IgG2a/2c, IgG3, and IgM
reactive with gangliosides GD1a, GM1, and GQ1b. NOD WT mice were measured for
levels of IgG2a antibody and C57BL/6 WT mice were measured for levels of the isotype
IgG2c antibody. Presence of IgG1 antibody reactive to a mixture of gangliosides GM1
and GD1a and to a mixture of gangliosides GM1 and GQ1b were also performed with
ELISA assays. Antibody levels of IgG1 reactive to ganglioside mixture GM1/GQ1b were
observed to be significantly different between treatment groups (Kruskal-Wallis;
p=0.0031). Dunn’s post hoc test indicated significant differences in plasma levels of
IgG1 reactive to ganglioside mixture GM1/GQ1b between the sham-inoculated mice
and the C. jejuni 260.94 infected mice (Figure 3.2, panel B; p≤0.01). No significant
differences between the two treatment groups of the C57BL/6 WT mice or among the
three treatment groups of the NOD WT mice were observed in levels of IgG1 antibody
reactive to ganglioside mixture GM1/GD1a (Table 3.1). No significant changes in
antibody levels to any of the three individual gangliosides were observed between
treatment groups in either the NOD WT or the C57BL/6 WT mice (Figure 3.2, panel C;
Table 3.1).

Neurological Testing

121

Mice were assessed on a weekly basis for a neurological phenotype using
neurological tests. These tests included rotarod, open field test, and footprinting. A twoway ANOVA with repeated measures was used to analyze the NOD WT mice and
C57BL/6 WT mice in all phenotype tests. If significant results were obtained then the
Holm-Sidak pairwise post hoc test was performed. Analysis of the footprint
measurements, stride length and hind base width required the removal from the data
set of mice that lacked a complete set of footprint measurements for the entire length of
the experiment due to lack of cooperation.

Rotarod
Mice were placed on the rotating wheel of the rotarod to detect changes in
balance and motor abilities. There were no significant differences in time spent on the
rotarod between treatment groups in both the NOD WT and C57BL/6 WT mice. A
change in the time spent on the rotating wheel over the course of the experimental time
period was observed in all treatment groups of the NOD WT mice (p<0.0001). All
observers noted that qualitatively that the C57BL/6 WT mice generally spent less time
on the rotarod than the NOD WT mice over the course of the experiment (Figure 3.3,
panel A).

Open Field Test
The open field test was used to detect abnormal gait and observe the general
neurological health of the mouse. Mice were videotaped and assessed with the
JWatcher™ program. Analysis of the number of quadrants a mouse crossed during the

122

test showed no significant differences in quadrants crossed between treatment groups
of either the NOD WT mice or the C57BL/6 WT mice. Mice of both genetic backgrounds
did have a significant change in quadrants crossed observed in all treatment groups
over the course of the experimental time (NOD WT: p<0.0001 and C57BL/6 WT:
p=0.0001). Observers noted qualitatively that the C57BL/6 WT mice generally were less
mobile and curious in the open field test during the entire course of the experiment; all
treatment groups of the NOD WT mice moved around much more (Figure 3.3, panels C
and D).
A rear occurs when a mouse stands on its hind legs; this behavior was recorded
during the videotaped interval with the JWatcher™ program. The program recorded the
length of time the mice were rearing; data are shown in Figure 3.3, panel B. No
significant difference in time spent rearing was found between treatment groups of
either the NOD WT mice or the C57BL/6 WT mice. Qualitatively it was observed that the
NOD WT sham-inoculated and both C. jejuni CF93-6 NOD WT infected mice and C.
jejuni 260.94 NOD WT infected mice all spent more time rearing over the course of the
experiment than both uninfected and infected C57BL/6 WT mice.

Footprint Measurements
This test was used to detect abnormal gait and foot drag in experimental mice.
Measurements for stride length, hind base width, print length, stride variability, and toe
spread were all recorded. Only stride length and hind base width (Figure 3.3, panels E–
F) will be discussed here. Some mice were uncooperative during this test (no clean
footprints recorded after 2 trials), and therefore a footprint measurement was not always

123

taken at every week. These mice had to be removed from analysis to enable a two-way
ANOVA with repeated measures to be performed. Two sham-inoculated, three C. jejuni
CF93-6 infected, and two C. jejuni 260.94 infected NOD WT mice had to be removed
from the analysis. Only two sham-inoculated C57BL/6 WT mice had to be removed from
the analysis. Body weight was measured at end of the experiment and no significant
differences in weight were found between treatment groups of either the NOD WT or
C57BL/6 WT mice. Therefore we concluded that body weight of mice did not play a role
in footprint measurements (data not shown).
No significant differences in stride length were found between the shaminoculated mice and either the C. jejuni- 260.94 infected or C. jejuni- CF93-6 infected
NOD WT mice. The C57BL/6 WT mice also showed no significant difference in stride
length between uninfected and infected mice. Measurements of the hind base width
revealed no significant changes between treatment groups of either the NOD WT mice
or the C57BL/6 WT mice. Hind base width changed over the course of the experimental
time in all treatment groups of both mouse strains (NOD WT: p=0.0048; C57BL/6 WT:
p<0.0001), but there were no significant differences between treatment groups within
mice of either genetic background.

Presence of Nerve Lesions
At necropsy, sections of the sciatic nerve were removed, fixed, and stained with
hemotoxylin and eosin and scored as described above. No significant differences in
histopathological scores were observed between treatment groups of NOD WT mice
using a Kruskal-Wallis ANOVA on ranks. The C57BL/6 WT mice did not have a

124

significant difference between the uninfected mice and infected mice as shown by a
Mann-Whitney U test. The presence of cellular infiltrates including mast cells, plasma
cells, neutrophils, lymphocytes, and macrophages was assessed. Fisher’s Exact test
detected no correlation in the presence of cellular infiltrates between treatment groups
of both mouse genetic backgrounds. It was observed that one C57BL/6 WT mouse
infected with C. jejuni CF93-6 had infiltration of mast cells, neutrophils, and
macrophages. One NOD WT C. jejuni CF93-6 infected mouse presented with infiltration
of lymphocytes, neutrophils, and mast cells. It was noted that the lymphocytes and
neutrophils were directly associated with the nerve in this mouse.
Sections of the nerve were also stained with Luxol fast blue to detect any
degradation of the myelin sheath surrounding the axon. Some C57BL/6 mice did exhibit
myelin loss, but Fisher’s Exact test indicated no significant difference between shaminoculated mice and C. jejuni CF93-6 infected mice of the C57BL/6 WT background. No
significant differences were observed between the treatment groups of the NOD WT
mice using Fisher’s Exact test.

125

DISCUSSION

Miller Fisher Syndrome is a clinical subtype of GBS that is distinguished by a
descending paralysis that affects both the central and peripheral nervous systems. In
previous work in our laboratory we have identified a potential murine model to study
GBS using NOD WT mice infected with two different C. jejuni strains, one from an
AMAN patient and one from an AIDP patient (St. Charles, unpublished). We also
observed significant differences in presence of anti-ganglioside antibodies in C57BL/6
IL-10

-/-

mice infected with GBS-associated C. jejuni compared to sham-inoculated mice

(Malik, unpublished). To further our understanding of the autoimmune neuropathies that
follow C. jejuni infection, we sought to develop a mouse model of MFS. To accomplish
this, we infected both NOD WT mice and C57BL/6 WT mice with strains of C. jejuni
associated with MFS and GBS.
Our results showed no significant differences between the sham-inoculated mice
and the C. jejuni CF93-6 infected mice in clinical neurological signs, presence of antiGQ1b antibodies, and cellular infiltration of the sciatic nerve in either mouse genetic
background. NOD WT mice infected with C. jejuni 260.94 had significantly higher levels
of IgG1 antibody reactive to ganglioside mixture GM1/GQ1b; however they showed no
significant differences in any of the phenotypic tests, suggesting a subclinical
phenotype. Furthermore, examination of the sciatic nerve showed no significant
differences between treatment groups of either the NOD WT mice or the C57BL/6 WT
mice for cellular infiltrations and nerve lesions. We were unable to observe significant
differences in anti-ganglioside antibody levels, nerve lesions, and or neurological

126

phenotype in MFS-associated C. jejuni CF93-6 infected animals for either C57BL/6 WT
mice or NOD WT mice. Only the GBS-associated C. jejuni 260.94 NOD WT infected
mice produced anti-ganglioside antibodies. These results were obtained despite the
high level of gastrointestinal colonization of both C. jejuni strains and the strong C. jejuni
specific antibody responses developed against each C. jejuni strain in colonized mice.
Thus, the C. jejuni strains were present in the majority of mice throughout the
experiment, but only the NOD WT mice infected with C. jejuni 260.94 developed
manifestations of autoimmunity in the form of autoantibodies.
Previous experiments performed in this laboratory with NOD WT and the GBSassociated C. jejuni 260.94 strain have shown a neurological phenotype and presence
of macrophages in peripheral nerve sections. Similar results were not seen in this
experiment. The mice used in this experiment were purchased directly from The
Jackson Laboratory; previous experiments used NOD WT mice from our established
specific pathogen free mouse colony. Therefore, differences seen between previous
experiments and this experiment could be a result of different microbiota of purchased
versus breeding colony mice. Our breeding colony mice may also have experienced
some genetic drift during several years of breeding in our colony, resulting in the
differences observed between experiments.
Studies have shown that approximately 80–100% of MFS cases present with
antibodies reactive to ganglioside GQ1b; however, some patients can still develop MFS
without any antibody production to GQ1b (93, 97, 137). Lee et al. (2012), discuss a
case study of a 13 year old boy who presented with the classical clinical profile of MFS
but tested negative for any antibody to ganglioside GQ1b (93). A similar phenomenon

127

could have occurred in our experiment; we performed ELISA assays to detect the
presence of antibodies IgG1, IgG2a/2c, and IgG3 reactive to ganglioside GQ1b, but
found no significant differences between the sham-inoculated mice and those infected
with C. jejuni CF93-6. We could also test for antibodies reactive to gangliosides GT1a
and GD3 which have been shown to cross react with antibodies reactive to GQ1b (117).
Additionally, the primary antibody response may have already waned by the time of
necropsy.
Analysis of the nerve sections examined histologically showed no significant
differences between uninfected mice and infected mice in nerve lesions and cellular
infiltrates. GQ1b is highly enriched in the ocular motor nerves and therefore in future
experiments we should remove and examine these nerves histologically. We should
also identify the NMJ in the mice and determine if there is binding of anti-GQ1b antibody
resulting in disruption of the NMJ.

128

Table 3.1: P values for ELISA assays. Abbreviations: KW, Kruskal-Wallis ANOVA;
MW, Mann-Whitney U.

Genotype
NOD WT
C5BL/6 WT

ELISA
GM1 (IgG1)
GM1 (IgG2a)
GM1 (IgG3)
GM1 (IgM)

KW; p=0.0678
KW; p=0.1666
KW; p=0.9595
KW; p=0.6377

MW; p=0.6491
MW; p=0.6762
MW; p=0.2243
MW; p=0.7042

GD1a (IgG1)
GD1a (IgG2a)
GD1a (IgG3)
GD1a (IgM)

KW; p=0.7730
KW; p=0.1153
KW; p=0.2913
KW; p=0.1446

MW; p=0.4418
MW; p=0.3194
MW; p=0.3253
MW; p=0.6948

GQ1b (IgG1)
GQ1b (IgG2a)
GQ1b (IgG3)
GQ1b (IgM)

KW; p=0.4553
KW; p=0.4867
KW; p=0.4552
KW; p=0.4999

MW; p=0.2000
MW; p=0.9334
MW; p=0.8276
MW; p=0.1329

GM1/GD1a (IgG1)

KW; p=0.3293 MW; p=0.4234

GM1/GQ1b (IgG1)

KW; p=0.0031 MW; p=0.7828

Cj (IgG2b)

KW; p<0.0001 MW; p<0.0001

Day 0 GQ1b (IgM)
Day 14 GQ1b (IgM)

KW; p=0.4751 MW; p=0.1019
KW; p=0.0524 MW; p=0.3611

129

Figure 3.1: Examination of GI disease. Panel A shows colonization of mice based on
culture and/or PCR of the cecum at endpoint. No sham-inoculated mouse was
colonized. Panel B shows the gross pathological features present in mice at time of
necropsy. Features examined were thickened wall of the GI and enlarged lymph
nodes.

A

130

Figure 3.1 (cont’d)

B

131

Figure 3.2: ELISA data. Panel A shows the levels of antibody IgG2b reactive with C.
jejuni antigen. Panel B shows plasma levels of IgG1 reactive with ganglioside mixture
GM1/GQ1b. Panels C-1, C-2, and C-3 show the antibody plasma levels for IgG1,
IgG2a/IgG2c, and IgG3, respectively. NOD WT mice have the IgG2a receptor and were
tested for IgG2a antibody levels; C57/BL6 WT mice have the IgG2c receptor and were
tested for levels of IgG2c antibody. Unless stated otherwise, significance levels in all
graphs are by * (p≤0.05), ** (p ≤ 0.01), *** (p ≤ 0.001) and **** (p ≤ 0.0001).
A

B

132

Figure 3.2 (cont’d)
C-1

C-2

C-3

133

Figure 3.3: Assessment of Clinical Neurological Signs with Phenotype Tests.
Panel A shows time spent on the rotarod; no significant differences between treatment
groups in either the NOD WT mice or the C57BL/6 WT mice were observed. The
C57BL/6 WT mice generally spent less time on the rotarod over the entire course of
the experiment. Panel B shows the time spent rearing in the first interval of the open
field test as calculated with JWatcher™. NOD WT mice spent more time rearing than
C57BL/6 WT mice. Panels C and D show the number of quadrants crossed during the
three intervals of the open field test for NOD WT and C57BL/6 WT mice, respectively.
Stride length and hind base width measurements are shown in panels E and F for
NOD WT mice and C57BL/6 WT mice respectively. No significant differences in either
measurement were observed between the three treatment groups of NOD WT mice or
between uninfected and infected C57BL/6 WT mice.

134

Figure 3.3 (cont’d)

A

B

135

Figure 3.3 (cont’d)
C

D

136

Figure 3.3 (cont’d)
E

137

Figure 3.3 (cont’d)
F

138

CHAPTER 4

Multiple molecular typing schemes applied to define genetic relationships among C.
jejuni isolates that share the same LOS classification

139

ABSTRACT

The Gram negative bacterium Campylobacter jejuni is the leading cause of
bacterial gastroenteritis in the Western world. Symptoms of C. jejuni infection include
fever, abdominal cramping, and watery or bloody diarrhea. Sequelae that can occur
after infection include Guillain Barré Syndrome (GBS), Miller Fisher Syndrome (MFS),
Reiter’s Arthritis, and both Irritable Bowel Syndrome and Inflammatory Bowel Disease.
C. jejuni is the most common antecedent infection associated with development of GBS.
The suggested mechanism of GBS is molecular mimicry of the C. jejuni lipooligosaccharide (LOS) and neurogangliosides resulting in development of antiganglioside antibodies. Gangliosides are sialylated glycolipid structures highly enriched
in nervous tissue, and it has been demonstrated that C. jejuni sialylated LOS can mimic
gangliosides. Serotyping and several molecular typing methods (LOS classification,
multi-locus sequence typing (MLST), flaA SVR sequence typing, and porA allele typing)
have been applied to define possible genetic relationships among C. jejuni isolates from
GBS, MFS, and enteritis patients, but outcomes have been inconsistent. We
hypothesized that there would be associations between MLST, flaA, or porA sequence
types and C. jejuni strains belonging to LOS classes associated with GBS and MFS in a
temporally and geographically limited set of strains. We applied these molecular typing
schemes to 68 C. jejuni isolates from patients that presented with enteritis; a small set
of unrelated strains, including strains from three patients known to have developed
neuropathy were included in the analysis. We expected to see a genetic relationship
between LOS classes and the porA allele types or flaA SVR sequence types. We did

140

expect to find strains of some clonal complexes defined by MLST to contain one
predominant LOS class.
In our study we were able to classify 56/68 C. jejuni isolates into LOS classes A–
F; however no isolate was determined to be in LOS class D. Our MLST studies
indicated that the strain collection included 36 different sequence types that were
clustered into 19 different clonal complexes. DNA sequencing of a portion of the porA
gene and flaA SVR gene resulted in the assignment of 43 different porA alleles and 39
different sequence types respectively. Analysis of the LOS classification, MLST, porA
gene, and flaA SVR gene resulted in 49/68 isolates that were singletons; the remaining
isolates were grouped into 8 different groups of 2 or more isolates based on allele
assignments. These data demonstrate that C. jejuni populations associated with
enteritis are highly diverse based on several genotyping systems. The only clonal
complexes with three or more strains were CC 21 (16), CC 353 (7), CC 45 (5), and CC
206 (3). Furthermore, several clonal complexes had associations with particular LOS
classes.

141

INTRODUCTION

Campylobacter jejuni, a gram negative bacterium, is the most frequent cause of
human bacterial gastroenteritis (19). Common sources of C. jejuni infection include
consumption of contaminated chicken meat, water, or unpasteurized milk (111). Serious
disease sequelae can be triggered by C. jejuni infection, including the acute
neuropathies Guillain Barré Syndrome (GBS) and Miller Fisher Syndrome (MFS) (105).
Incidence of GBS secondary to C. jejuni infection is estimated to be 0.25–0.65 cases
per 1000 cases of C. jejuni infection (163). Neurological symptoms appear 1–3 weeks
after recovery from the initial C. jejuni gastrointestinal infection. Symptoms may start
with weakness of extremities and progress within hours to weeks to paralysis of the
limbs, trunk, or extraocular and facial muscles (7, 80, 116). About 3–10% of GBS
patients die, and up to 10–20% of GBS patients have long term disability (7, 72, 163).
The mechanism thought to play a role in the induction of GBS is a molecular
mimicry between the LOS of C. jejuni and host gangliosides found on peripheral nerves
(7, 145). The LOS structure consists of a lipid A tail that is inserted into the outer
membrane of C. jejuni and a highly variable oligosaccharide moiety that extends into the
extracellular space (112). Gangliosides are glycolipids that are enriched on the cell
surface of neural tissues. They consist of variable oligosaccharide moieties containing
one or more sialic acid residues that protrude into the extracellular space and a
ceramide tail that is inserted into the lipid bilayer of the membrane (7, 80, 156).
Gangliosides play a role in maintenance of cell membrane structure and are important

142

in cell-cell recognition; they have been shown to be present in the nerve myelin sheaths
and axons of peripheral nerves (7).
In 2000, completion of the genome sequence of the C. jejuni strain 11168
defined the genes involved in LOS biosynthesis and their arrangement in the LOS
biosynthesis locus (112). Since then the genes of LOS biosynthesis loci have been
identified in both sequenced and unsequenced strains of C. jejuni; the results have
shown that the gene content of the LOS loci is highly variable (47, 112), as are the LOS
structures produced by the pathway (45, 48, 50, 113). The LOS biosynthesis locus
extends from Cj1131c to Cj1151c in the genome of C. jejuni 11168. Genes that have
been identified in the locus include glycosyltransferases and genes involved in sialic
acid biosynthesis. To date, 23 different classes (A–W) of C. jejuni strains have been
defined based on both gene content as detected by PCR and organization of the genes
within the LOS biosynthesis loci (111, 112, 125). Previous work by Gilbert et al. (2002)
identified a variable region of the LOS that contains the cgtA (orf5abI) and cgtB (orf6ab)
genes that results in 2 alternative allele types: 1) orf5abI1/orf6ab1 (allele 1) or 2)
orf5abI2/orf6ab2 (allele 2). Both LOS class A and LOS class B can be further defined by
the presence of these alleles; either class can contain either allele type 1 or allele type 2
based on the presence of the cgtA and cgtB genes (47, 112). A study by Godschalk et
al. (2004) demonstrated an association between GBS and C. jejuni LOS classes that
contain the genes necessary to synthesize a sialylated LOS that mimics gangliosides
(49). The genes needed to synthesize sialic acid and transfer it to the LOS sugar
backbone are neuBCA (biosynthesis of sialic acid) and cstII and III (sialic acid
transferases) (47, 49, 111, 112, 125); cstIII and one of the two alleles of cstII have

143

different transferase activities and produce different ganglioside mimics. C. jejuni strains
in LOS classes A, B, C, M, R, and V possess these genes.
Other outer membrane surface molecules have been examined to define
possible epidemiological genetic relationships between these molecules and C. jejuni
strains associated with GBS and MFS. One such molecule that has been studied is the
flaA gene of the polar flagellum of C. jejuni. Meinersmann et al. (1997) showed that
sequencing of a short variable region (SVR) in the flaA gene produced results similar to
those found with sequences of the entire flaA gene (101). Thus a quicker method of
typing C. jejuni isolates using flaA SVR was proposed (101). Many studies have been
performed to determine possible associations between flaA SVR sequence typing and
GBS-associated strains of C. jejuni. Tsang et al. (2001) reported that they had identified
a flaA SVR sequence type that was associated with C. jejuni strains from GBS patients
(144), while Dingle et al. (2001) found no correlation between the flaA SVR sequence
type identified by Tsang et al. and GBS or MFS in a study of 25 C. jejuni isolates (35).
Another outer membrane surface molecule that has been studied is the porA
gene encoding the major outer membrane protein. Clark et al. (2007) characterized the
porA sequence of 105 C. jejuni isolates and defined 3 sequence types (27). They found
no association in their study of any particular sequence type with GBS-associated or
MFS-associated C. jejuni isolates; all 3 sequence types were associated with GBSassociated and MFS-associated C. jejuni strains (27).
Another typing scheme used to identify genetic relationships among C. jejuni
isolates is multi-locus sequence typing (MLST). MLST was first described for C. jejuni in
2001 by Dingle et al. using 194 C. jejuni isolates (34). MSLT is based on use of

144

sequence variation in 7 conserved housekeeping genes to compare genetic
relationships among different bacterial isolates. The partial gene sequences used vary
in length and are assigned allele numbers based on sequence variation. In the study by
Dingle et al. (2001), the authors were able to assign 155 sequence types (STs) to 62
different clonal complexes (34). Multiple studies using MLST have been performed to
compare the genetic relationships among C. jejuni strains isolated from GBS and MFS
patients. Revez et al. (2012) combined the molecular typing schemes of LOS
classification and MLST to compare the genetic relationships among 403 C. jejuni
isolates revealed by the two methods and were able to show an association between
LOS classes and sequence types of MLST (124).
The PCR-based classification scheme for the LOS locus and MLST have been
very useful tools in comparing and contrasting relatedness and LOS genetic content of
enteric and neuropathic strains of C. jejuni; however there are conflicting data for the
molecular typing of C. jejuni with flaA and porA. Therefore, we conducted our own
analysis of 68 C. jejuni isolates. In our study, 5/68 isolates were previously sequenced;
these strains included C. jejuni 11168; 2 strains isolated from GBS patients and 1 from
an MFS patient; and C. jejuni CG8421, which has been demonstrated to have an LOS
that does not mimic gangliosides (143). We determined the LOS classification of the 68
C. jejuni isolates; 86% of the isolates typed as LOS classes A–C. To determine genetic
relationships between the LOS classification and other variable outer membrane
structures (porA and flaA) of the bacterium, we performed molecular typing of the porA
and flaA SVR; we also included MLST to examine the association of highly conserved
genes. We hypothesized that known isolates that were associated with GBS and MFS

145

would be of similar sequence types as those isolates that only caused enteritis; there
would be a genetic relationship between other outer membrane surface molecules and
LOS class. We analyzed each molecular typing method individually and then combined
them for a comparative study of the techniques. Based on our results, the 68 isolates of
C. jejuni were highly diverse. Analysis of combined data from the three sequence-based
typing schemes resulted in 49 singletons with 8 different groups of isolates of 2 or more
isolates; LOS classes A–C were distributed among all groups.

146

MATERIALS AND METHODS

Campylobacter jejuni Strains and Growth Conditions
Sixty-eight isolates of Campylobacter jejuni were collected and maintained for
molecular typing (Table 4.1); of the 68 isolates 5 have previously been sequenced. Of
the previously sequenced strains, two are from GBS cases (C. jejuni 260.94 and C.
jejuni HB93-13), one is from an MFS patient (C. jejuni CF93-6), one is the genome
strain C. jejuni 11168, and one is C. jejuni CG8421, which has been demonstrated not
to mimic gangliosides (143). C. jejuni 260.94 was isolated from a patient in the Red
Cross Children’s Hospital in Cape Town, South Africa who presented with the acute
inflammatory demyelinating polyneuropathy (AIDP) subtype of GBS (2). C. jejuni HB9313 was isolated from an 8 year boy who presented with the acute motor axonal
neuropathy (AMAN) subtype of GBS (2). The MFS-associated C. jejuni strain CF93-6
was isolated from an MFS patient in Japan (2). These strains were streaked for growth
on Bolton agar plates (Bolton Broth, Thermo Fisher Scientific, Pittsburgh, PA, containing
15% Bacteriological Agar, Neogen, Lansing, MI) for DNA extraction (10).
Forty-nine of sixty-eight isolates were recovered from patients by the Michigan
Department of Community Health. These isolates were de-identified and graciously
supplied by Dr. Shannon Manning (Michigan State University Microbial Evolution
Laboratory, East Lansing, MI). Upon arrival in the laboratory, samples were streaked for
isolation onto Bolton agar plates and grown for 48 hours at 37°C in anaerobic jars with
CampyGen packs (Oxoid, Basingstoke, UK). Isolated colonies were collected and
streaked for growth under the same conditions. After 48 hours, bacterial growth was

147

harvested and suspended in tryptose soya broth (TSB) containing 15% glycerol to
produce a stock culture and stored at -80°C.
Four of sixty-eight isolates were obtained from soil samples of calf pens from
dairy farms in Pennsylvania and Virginia. These isolates were graciously supplied by Dr.
Terence Marsh, Michigan State University Department of Microbiology and Molecular
Genetics; his laboratory isolated the C. jejuni strains and supplied us with stock
cultures. We streaked the stock cultures on Bolton agar plates (Bolton Broth, Thermo
Fisher Scientific, Pittsburgh, PA, containing 15% Bacteriological Agar, Neogen, Lansing,
MI) to check viability as described above. Bacterial growth was harvested and stored at
-80°C in TSB + 15% glycerol.
The 10/68 remaining isolates were supplied by the Centers for Disease Control
(CDC; Atlanta, Georgia) from different C. jejuni infections in both the U.S. and Canada
(10, 11). One isolate included in this group, C. jejuni D8942, was obtained from a GBS
patient in New Mexico. All other isolates have no association with GBS to the best of
our knowledge. All strains were previously acquired by our laboratory; stock cultures
were maintained in TSB + 15% glycerol at -80°C (10, 11).

Preparation of Genomic DNA
Frozen cultures were streaked on Bolton agar plates (Bolton Broth, Thermo
Fisher Scientific, Pittsburgh, PA, containing 15% Bacteriological Agar, Neogen, Lansing,
MI) for growth at 37°C for 48 hours in anaerobic jars containing CampyGen packs
(Oxoid, Basingstoke, UK). Growth was collected and DNA was isolated using Wizard®

148

Genomic DNA Purification Kits (Promega, Madison, Wisconsin) according to the
manufacturer’s instructions. All genomic DNA was diluted to a final concentration of 25
ng/μL.

Determination of Campylobacter spp.
Verification of isolates as C. jejuni or C. coli was performed using a previously
described multiplex PCR (96). Hot Start Syzygy Mean Green (Integrated Scientific
Solutions, San Diego, CA) was used as the DNA polymerase with addition of buffer,
primers, and water. Enzyme activation was for 15 minutes at 95°C for followed by 25
cycles of denaturation at 95°C for 30 seconds, annealing for 1.5 minutes at 58°C, and
extension for 1 minute at 72°C, with a final extension of 7 minutes at 72°C.

Classification of LOS Classes by PCR
Determination of LOS classes was performed as previously described (112). To
date, 23 classes (A–W) have been described; in this study we only classified isolates
into 6 classes (A–F) using the 12 primers previously described by Parker et al. (2005)
(112). Some isolates were unable to be typed into one of the 6 classes, however more
primer sets could be used to further classify these isolates (111). As described above,
we also further distinguished LOS class A and class B by the presence of allele type 1
or type 2. LOS class A and class B were assigned alleles based on the
presence/absence of orf5abI1/orf6ab1 and orf5abI2/orf6ab2.

149

DNA Sequencing and Allele Assignment

Multi-locus sequence typing (MLST)
MLST was performed on the CDC and environmental isolates as previously
described only using the inner primers (http://pubmlst.org/campylobacter/) as described
by Bell et al. (2009) (10, 11, 83). A protocol using KAPA2G Fast HotStart DNA
(KAPABiosystems, Woburn, MA) polymerase was optimized for MLST of the 49 isolates
from Michigan. KAPA2G Fast HotStart DNA polymerase (KAPABiosystems, Woburn,
MA) was used with 10 μM of both the forward and reverse primer and water. The
reaction consisted of 35 cycles of denaturation at 95°C for 15 sec, annealing for 15 sec
at 60°C, followed by extension for 5 seconds at 72°C.

flaA SVR
flaA SVR sequences were obtained with PCR. The reaction consisted of 35
cycles with a denaturation step at 94°C for 1 minute, annealing at 45°C for 1 minute,
and extension for 3 minutes at 72°C. Primers used were previously described (101).
Platinum® Taq DNA Polymerase High Fidelity (Life Technologies, Grand Island, NY)
was used in reactions with 25 ng/μl of template DNA and 40 pmol of primer.

porA
Three different primers were used to sequence the entire length of the porA
gene; these primers may be found on the PubMLST website

150

(http://pubmlst.org/campylobacter/info/porA_method.shtml) (83). We encountered
problems with these primers: they did not amplify the entire portion of the porA gene we
wished to sequence. Therefore we developed four additional primers to amplify the
entire porA gene: 1) porA A (5’-TTTCCTAARRAAGCACCTTC-3’), 2) porA B (5’GATGCWASYCTTGGTGGTTT-3’), 3) porA C (5’-GGWCAATTMAACCCACAA-3’), and
4) porA D (5’-ACAWAWCRAAGATATTACC-3’). Primers were added at the
concentration of 40 pmol; enzyme used was Platinum® Taq DNA Polymerase High
Fidelity (Life Technologies, Grand Island, NY) with 25 ng/μl of template DNA. The PCR
reactions were 40 cycles beginning with a denaturation step at 94°C for 30 seconds
followed by annealing for 30 seconds at 45°C and ending with extension for 90 seconds
at 72°C.

Sequence Analysis
All PCR products were submitted to the Michigan State University Genomics
Technology Support Facility for sequencing. PCR products were fluorescently labeled
and then separated on an ABI 3730 Genetic Analyzer (Life Technologies, Grand Island,
NY) by capillary electrophoresis. Purified PCR product was sequenced in both the
forward and reverse directions. DNA sequences were compiled and analyzed with
SeqMan 5.06 (DNASTAR, Madison, WI); alleles of each gene, sequence type, and
clonal complex were assigned using the PubMLST http://pubmlst.org/campylobacter/
website (83).

151

C. jejuni strains 260.94 (GenBank® #AANK01000001.1), HB93-13 (GenBank®
#AANQ01000001.1), CF93-6 (GenBank® #AANJ01000005.1), CG8421 (GenBank®
#AANT02000001.1), and 11168 (GenBank® #NC 018521.1) have all previously been
sequenced; therefore, DNA sequences for the portions of the flaA SVR gene and the
porA gene of those strains were derived from the NCBI database using Basic Local
Alignment Search Tool (BLAST) (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi).

Statistical Analysis
DNA sequences for MLST, porA, and flaA SVR were aligned; phylogenetic trees
and tree files were created using phylogeny.fr using a maximum likelihood calculation
(23, 33, 39).
The allelic profiles of MLST, porA, and flaA SVR were analyzed with PAST
version 1.89 (Øyvind Hammer, Oslo, Norway); cluster analysis of allelic profiles was
performed using the Hamming similarity coefficient and unweighted pair group with
arithmetic averaging (UPGMA). All bootstrapping was performed at 1000 repetitions.
Dendrograms of the MLST, porA, and flaA SVR DNA sequences were compared using
Compare2Trees: pairwise comparison of phylogenies
(http://www.mas.ncl.ac.uk/~ntmwn/compare2trees/index.html). This test results in an
overall topological score that describes the percent similarity between the two trees
being analyzed (106). eBurst was used to assess the similarity of the results obtained
using the different molecular typing methods based on allele assignments
(http://eburst.mlst.net/default.asp).

152

RESULTS

LOS Classification
Of the 68 (63 unknown and 5 previously sequenced) C. jejuni isolates analyzed
we were able to classify 56 (82%) into LOS classes A–F (Table 4.1). Forty-eight of fiftysix (86%) isolates were classified into LOS classes A–C, the three classes shown to
have the ability to sialylate the LOS structure to mimic gangliosides (47). Seven of the
fifty-six (13%) isolates were determined to belong to LOS class A1 and 1/56 (2%)
belonged to class A2. Four of fifty-six (7%) were classified as class B1 and 11/56 (20%)
were class B2. The remainder of the 56 classified isolates were categorized as LOS
class C (25/56; 45%), class E (5/56; 9%), and class F (3/56; 5%). Twelve of sixty-three
unknown isolates could not be assigned to a single LOS class (A–F) because the gene
content of their LOS loci as determined by PCR was a composite of two classes; LOS
class E was one of the two classes in all 12 isolates.
The two GBS-associated strains, C. jejuni 260.94 and C. jejuni HB93-13, and the
MFS-associated strain, C. jejuni CF93-6, all belong to LOS class A1. The genome
strain, C. jejuni 11168, belongs to LOS class C. C. jejuni CG8421, a strain shown to not
mimic gangliosides (143), was classified as LOS class E.
Cluster analysis of the gene content of the LOS loci used in classification (0 for
absent; 1 for present) of the 68 isolates is shown in Figure 4.1. The cophenetic
coefficient for this tree was 0.97, indicating that the tree reflects the data accurately.
Bootstrapping analysis indicated three well supported groups (bootstrap values ≥50%).

153

MLST Analysis
The 68 C. jejuni isolates included 37 different sequence types (ST) in19 clonal
complexes (CC). Three of sixty-eight (4%) isolates did not match any sequence type in
the PubMLST database and were classified in this study as “unknown.” Four of sixtyeight (6%) isolates could not be assigned to described clonal complexes. Clonal
complexes CC 21 and CC 45 have been demonstrated to be the most common among
C. jejuni (124); CC 21 was the most common complex in this study. Twenty-one of sixtyeight (31%) C. jejuni isolates belonged to CC 21 and 8/68 (12%) belonged to CC 45. C.
jejuni 11168 and C. jejuni CF93-6 both belonged to CC 21. C. jejuni HB93-13, C. jejuni
CG8421, and C. jejuni 260.94 belonged to CC 22, CC 52, and CC 362 respectively.
Comparison of the MLST results with LOS classification showed that some clonal
complexes contained isolates of only one LOS class. Four of four CC48 isolates
belonged to LOS class B2; two of two CC 508 isolates belonged to LOS class A1, and
two of two CC 607 isolates belonged to LOS class B1. The remaining clonal complexes
with more than one isolate had a mixture of LOS classes (Figure 4.2). The majority of C.
jejuni CC 21 isolates belonged to LOS class C (18/20; 90%); one CC 21 isolate
belonged to LOS class A1 (1/20; 5%); and 2/20 (10%) isolates could not be classified.
Strains of LOS class A1 were found in 5 clonal complexes: CC 21(1/20; 5%), CC 22
(2/4; 50%), CC 206 (1/3; 33%), CC 362 (1/1; 100%), and CC 508 (2/2; 100%). The LOS
class A2 was only present in one isolate that did not belong to a known clonal complex.
Two of two CC 607 isolates belonged to LOS class B1 (2/2:100%); LOS class B1 was
also present in clonal complexes CC 42 (1/1; 100%) and CC 353 (1/9; 11%). Four of
four CC 48 isolates belonged to LOS class B2, but LOS class B2 was also found in CC

154

257 (1/1; 100%), CC 353 (3/9; 33%), and CC 403 (1/1; 100%). Two other isolates that
were classified as LOS class B2 could not be assigned to known clonal complexes.
LOS class C was represented in four clonal complexes: CC 21 (18/20; 90%), CC 206
(2/3; 67%), CC 353 (4/9; 44%), and CC 658 (1/1; 100%). The 5 isolates that belonged
to LOS class E were distributed among 4 clonal complexes: CC 45 (2/9; 22%), CC 52
(1/2; 50%), CC 61 (1/1; 100%), and one isolate could not be assigned to a known clonal
complex. LOS class F was present in 3 clonal complexes: CC 49 (1/1; 100%), CC 52
(1/2; 50%), and CC 446 (1/1; 100%).

porA Typing
Analysis of the porA gene sequences in the 68 different C. jejuni isolates
revealed 43 different alleles. Figure 4.3 represents a phylogenetic tree of the C. jejuni
isolates based on porA sequences; two groups were well supported (bootstrap values
≥50%). Allele 119 was the most common among the 68 isolates with seven (10%)
isolates. The 5 previously sequenced strains had different porA allele types. C. jejuni
11168 and C. jejuni CG8421 had alleles 27 and 606 respectively. The GBS-associated
strains C. jejuni 260.94 and C. jejuni HB93-13 had alleles 866 and 70 respectively. The
MFS-associated strain C. jejuni CF93-6 had allele 131.
Comparison of the porA allele types with the LOS classification showed that
strains of five allele types contained isolates of the same LOS class. Allele 14 (2/2;
100%) and allele 808 (2/2; 100%) were associated only with LOS class B2. LOS class C
was associated with isolates that had 3 different porA alleles; allele 27 (2/2; 100%),
allele 119 (7/7; 100%) and allele 738 (3/3; 100%). Three of three isolates having allele

155

44 could not be classified in LOS typing; all 3 isolates had the same gene content based
on the LOS PCR. All other allele types were associated with more than one LOS class.

flaA SVR Sequence Types
The short variable region (SVR) of the flaA gene was sequenced and analyzed
(data not shown) for the 68 C. jejuni isolates; 39 different sequence types were present
in the strain collection. ST-47 was the most common sequence type (9/68; 13%). Each
of the 5 sequenced strains was assigned to a different flaA SVR sequence type. Based
on the DNA sequences of the flaA SVR two clusters were identified (bootstrap values
≥50%).
Comparison of the flaA SVR sequence types with the LOS classes showed that
two sequence types were only associated with LOS class C isolates, ST-103 (3/3;
100%) and ST-784 (2/2; 100%). Sequence types ST-22 and ST-106 were both (2/2;
100%) associated only with isolates that could not be classified using the LOS
classification scheme used in this paper; however, the LOS gene content of each pair of
isolates having each ST was identical based on PCR. All other sequence types of flaA
SVR were associated with a mixture of LOS classes.

Comparison of Isolates with All Molecular Typing Methods
Analysis of the alleles assigned to each DNA sequence in MLST, porA, and flaA
SVR was performed with eBurstV3. This analysis identified 6 different groups of 2 or
more isolates and 41 (59%) singletons. Addition of the LOS classification to the MLST,
flaA SVR sequence type, and porA allele type resulted in 49/68 (72%) isolates that were

156

singletons. Eight different groups containing more than one isolate were identified in the
combined allelic profile and LOS class data; 3/8 groups had 3 isolates per group and
5/8 groups were pairs of isolates that contained identical alleles and sequence types.
Pairwise analyses of the dendrograms generated from the MLST, porA, and flaA
SVR DNA sequences were performed; all pairs of dendrograms were found to have low
similarity. The overall topological score for the pairwise comparison of the MLST data
and the porA data was 37.9%. The overall topological scores for the comparison of the
MLST data and the flaA data was 33.8%; analysis of the flaA SVR and porA data
resulted in an overall topological score of 33.6%.

157

DISCUSSION

Multiple studies have been performed to determine the gene content of the LOS
loci of C. jejuni isolates from both enteric and GBS/MFS patients. The results have
indicated that LOS classes strongly associated with GBS (A and B) are also present in
patients that clinically presented with enteritis only and that some C. jejuni isolates from
GBS patients had LOS classes that lacked genes needed to synthesize a sialylated
LOS (112).
We analyzed 68 C. jejuni isolates, 5 of which had published genome sequences.
Of the 68 isolates 56 were categorized into LOS classes A–F; with a plurality classifying
as LOS class C (45%). LOS class A represented 14% of the isolates; class B, 27% of
the isolates; class E, 9% of the isolates, and class F, 5% of the isolates. No isolate was
classified as LOS class D. Eighty-six percent of the isolates analyzed were found to
belong to LOS classes A–C. Previous studies have been done with human C. jejuni
isolates to associate the prevalence of LOS classes A–C among the population, but
because these studies varied in location, duration of the study period, and duration of
the study period, direct comparisons are difficult. One study in Finland found 32% of
human C. jejuni isolates were assigned to LOS classes A, B, and C (124). Two other
studies performed in Belgium and Bangladesh found 53% and 61% respectively, of LOS
classes A–C in the collections of human isolates they sampled (57, 78). In South Africa,
50% of human isolates positive for C. jejuni were of the LOS classes A–C (123).
Furthermore, another study done in New Zealand identified 77% of C. jejuni human
isolates as belonging to LOS classes A–C (69). Additionally, in this study 34/49 (69%)

158

C. jejuni strains associated with enteritis cases in Michigan had LOS class assignments
consistent with strains documented from GBS patients, suggesting that the risk of
acquiring a strain capable of eliciting neurological disease is relatively high in Michigan.
However, these isolates from Michigan were collected during a short period of time and
a limited geographical location; therefore a more extensive study including a bigger
geographical area and more C. jejuni isolates would need to be conducted to conclude
that Michigan does have a high prevalence of GBS-associated C. jejuni strains.
Twelve isolates were not categorized into LOS classes A–F because based on
gene content of their LOS loci as determined by PCR, these isolates could be classified
into two or more LOS classes. Additional primers would be needed to further classify
these isolates beyond LOS class A–F (111). All twelve of these isolates were a mixture
of LOS class E with either LOS class A, class B, or class C; 7/12 have the necessary
cstII or cstIII gene needed to transfer sialic acid to the LOS. This result suggests that
further typing of these isolates would place them in LOS class M, class R, or class V.
The remaining 5/12 isolates do not contain either cstII or cstIII and therefore are
hypothesized to belong to LOS classes not associated with sialylation of the LOS.
We were able to determine that CC 21 was the most common clonal complex
assigned in this study with 29% of our isolates belonging to it; this result coincides with
previous reports of CC 21 and CC 45 as the two most prevalent clonal complexes
among the C. jejuni human population (124). By comparison of results from the variable
LOS locus classification to those from the more conserved loci of MLST, we determined
that 90% of the isolates in CC 21 in our own study belonged to LOS class C. Other
studies that have compared MLST data to LOS classification have also identified LOS

159

class C to as the most common class associated with CC 21 (57, 69, 78, 123, 124).
Previous studies have also commonly found C. jejuni LOS classes that lack the ability to
sialylate an LOS to be prevalent in CC 45 (69, 124). These studies have associated
LOS classes E and H specifically with CC 45. In our study we found that 22% of our
isolates assigned to LOS class E to belong to CC 45; we did not classify beyond LOS
class F and therefore have no data about LOS class H. The majority of the LOS classes
belonging to CC 45 in our study were unknown LOS class types. Further studies to
determine these LOS classes may reveal that they belong to LOS class H, which would
correspond to previous studies.
Analysis of the LOS biosynthesis loci resulted in 86% of the isolates categorized
into LOS classes A–C; therefore we wanted to explore the genetic relationship of other
outer membrane surface molecules. MLST and porA allele typing, and flaA SVR
sequence typing did not reveal strong relationships of LOS classes A–C with any
sequence type by any method used. MLST analysis resulted in 37 different sequence
types that could be grouped into 18 different clonal complexes. The flaA SVR sequence
typing resulted in 39 different sequence types and the porA assigned 43 different
alleles, therefore porA molecular typing method produced the greatest diversity among
the C. jejuni isolates.
Pairwise comparisons of the separate dendrograms and cluster analyses of the
four molecular typing methods indicated that the 68 isolates examined in this paper
were genetically diverse. eBurst analysis of combined MLST allele profiles, porA alleles,
and flaA SVR sequences indicated 41 unique strains (60%) and 6 groups of 2 or more
isolates that were identical to one another within each group. Addition of the LOS

160

classification to this analysis resulted in 49 unique strains (72%) with 8 groups of 2 or
more isolates that were genetically identical based on allele types; 3/8 were groups of 3
isolates and 5/8 were groups of 2 isolates that contained identical alleles and sequence
types. Addition of the LOS classification thus revealed even more diversity among the
isolates. Combination of these typing schemes should be useful in the analysis of the
genetic relationships among C. jejuni isolates in a population. We are able to conclude
from the individual analyses of these molecular typing schemes that there is no
relationship between these schemes and the LOS classes of the isolates except for the
association of some LOS classes with some MLST clonal complexes. However, there
were also instances of members of a clonal complex having LOS loci of multiple
classes. We also were able to conclude that the Michigan strains form a genetically
diverse group of isolates with both the individual analyses and the combined results of
the four molecular schemes.
Recently, Clark et al. (2012) presented a new molecular method for typing C.
jejuni isolates: comparative genomic fingerprinting (CGF); this method employs a
multiplex PCR to detect highly variable genes in Campylobacter spp. (28). These
authors were able to demonstrate that molecular typing of C. jejuni isolates with CGF
was potentially useful in detecting genetic clusters of Campylobacter isolates in
population surveys of Campylobacter spp. (28). There are 4 LOS biosynthesis locus
genes that are in common with the LOS classification and the CGF molecular typing:
Cj1134, Cj1136, Cj1141 (needed for sialic acid biosynthesis) and Cj1151. These genes
alone with both methods would not be sufficient to identify genetic relationship to an
LOS class. With the addition of this new typing tool and the molecular typing methods

161

employed above, genetic relationships among C. jejuni populations could be well
defined.

162

Table 4.1: Summary of C. jejuni Isolates and their assigned allele types. The
table below corresponds the isolate name with the ID’s given throughout the chapter in
the first two columns. The ID’s were given based on source of the isolate: CDC# for
strains received from the Center for Disease Control, ENV# for the C. jejuni isolates
taken from cow pen soil samples, MI# for the isolates collected in Michigan, and REF#
are the previously sequenced C. jejuni strains. The allele types for the MLST genes
(aspA, glnA, gltA, glyA, pgm, tkt, and uncA) are given along with the assigned
sequence type (ST) and clonal complex (CC). The next column is the sequence type
of the flaA SVR sequence. The porA column indicates the allele assigned based on the
DNA sequences. The final column is the LOS classification based on PCR for
presence/absence of gene content. Some isolates were unable to be classified and/or
sequenced; therefore are given “unk” for unknown.

163

Table 4.1 (cont’d)

Isolate
D2586
D0835
D0121
D6844
D6845
D6846
D6847
D6848
D6849
D8942
LM28
LM31
LM36
LM41

I.D.
CDC1
CDC2
CDC3
CDC4
CDC5
CDC6
CDC7
CDC8
CDC9
CDC10
ENV11
ENV12
ENV13
ENV14

aspA glnA gltA glyA pgm
2
1
5
3
4
7
4
5
2
11
4
7 10 4
1
1
3
6
4
3
7
17 5
2 10
9
2
2
2
11
7
17 5
2 10
2
1 12 3
2
2
4
1
2
7
1
6 22 24 12
4
7 10 4
1
4
7 10 4
1
2
1
1
3
1
10
1 59 19 10

164

tkt uncA ST CC flaA porA
1
5
43 21 335 27
1
5
429 48 970 1728
7
1
45 45 270 44
3
3
22 22 1434 48
3
6
353 353 663
2
5
6
824 257 270 98
3
6
353 353 147 55
1
5
50 21 174 92
1
5
48 48 147 14
28
1
132 508 289 790
7
1
45 45 270 44
7
1
45 45 270 44
1
3
186 21 37
8
1
7
933 403 122 1539

LOS
C
B2
unk
A1
B2
B2
unk
C
B2
A1
unk
unk
C
B2

Table 4.1 (cont’d)

Isolate
I.D.
TW16361 1 MI15
TW16362 1 MI16
TW16370 1 MI17
TW16373 2 MI18
TW16374 2 MI19
TW16397 3 MI20
TW16398 1 MI21
TW16399 1 MI22
TW16400 2 MI23
TW16401 1 MI24
TW16402 3 MI25
TW16409 2 1 MI26
TW16420 1 MI27
TW16422 1 1 MI28
TW16429 1 MI29
TW16430 1 1 MI30
TW16431 1 MI31
TW16435 1 MI32
TW16438 1 MI33
TW16441 1 MI34
TW16442 1 MI35
TW16443 1 MI36

aspA
1
1
2
7
2
2
2
7
2
10
4
6
2
47
7
8
2
4
7
4
4
7

glnA
3
3
1
17
4
21
1
17
21
2
7
4
1
55
4
2
1
7
84
7
7
17

gltA
6
6
2
5
1
5
2
5
5
50
10
5
12
5
5
5
12
10
5
10
40
5

glyA
4
4
3
2
2
2
3
2
2
62
4
2
3
10
2
53
3
4
10
4
4
2

165

pgm
3
3
2
13
7
59
2
10
59
91
1
2
2
11
11
11
2
1
119
42
42
10

tkt
3
3
1
3
1
1
1
3
1
1
7
1
1
3
3
3
1
7
178
7
51
3

uncA
3
3
5
6
5
5
5
6
5
45
1
5
5
8
165
1
23
1
26
1
1
6

ST
22
22
982
3510
48
222
982
353
222
unk
45
122
50
862
1838
607
3574
45
1911
137
267
353

CC
22
22
21
353
48
206
21
353
206
unk
45
206
21
446
353
607
21
45
unk
45
283
353

flaA
106
106
49
142
147
103
46
147
103
632
249
49
147
222
121
1011
78
249
350
239
239
604

porA
48
48
1696
808
14
131
752
119
131
766
49
7
6
116
1479
116
529
49
1101
53
73
164

LOS
unk
unk
unk
B2
B2
C
C
C
C
E
E
A1
C
F
B1
B1
C
unk
B2
E
unk
B2

Table 4.1 (cont’d)

TW16444 1
TW16445 1
TW16446 1
TW16451 1
TW16452 1
TW16453 1
TW16455 1
TW16463 2
TW16464 1
TW16467 1
TW16469 1
TW164711
TW16475 2
TW16478 2
TW16491 1
TW16493 1
TW16494 1
TW16495 1
TW16498 1
TW16499 2
TW16506 3
TW16510 3
TW16511 2

MI37
MI38
MI39
MI40
MI41
MI42
MI43
MI44
MI45
MI46
MI47
MI48
MI49
MI50
MI51
MI52
MI53
MI54
MI55
MI56
MI57
MI58
MI59

2
7
1
2
2
3
166
4
4
1
8
2
2
1
1
2
2
2
2
2
9
2
7

1
17
2
1
1
1
2
7
7
6
1
1
1
1
1
1
1
1
1
17
1
1
17

1
5
3
12
1
5
5
10
10
60
6
12
2
2
2
2
2
12
2
1
2
1
2

3
2
1
2 10 3
3
5
9
3
2
1
3
2
1
84 11 11
72 151 3
4
1
7
4
1
7
24 12 28
3
2
1
3
2
1
83 2
3
2 225 3
83 2
3
3
2
1
3
2
1
3
2
1
3
2
1
2
7
1
10 22 3
3
2
1
2 22 3

166

6
6
3
5
6
6
1
1
1
1
1
5
6
17
6
5
5
5
5
5
6
6
6

8
353
459
50
8
467
2310
45
45
508
44
50
3007
1244
922
982
982
50
982
453
4506
8
unk

21
353
42
21
21
49
607
45
45
508
21
21
658
61
unk
21
21
21
21
48
52
21
unk

122
965
274
147
122
247
825
22
22
289
142
147
211
249
14
46
46
147
46
147
337
1623
142

1231
119
199
119
745
55
1087
53
53
17
839
119
1653
37
744
738
738
119
738
808
1
745
745

C
C
B1
C
unk
F
B1
unk
unk
A1
C
C
C
E
A2
C
C
C
C
B2
F
C
B2

Table 4.1 (cont’d)

TW16512 1
TW16513 1
TW16514 1
TW16515 3
11168
CF93-6
HB93-13
260.94
CG8421

MI60
MI61
MI62
MI63
Ref64
Ref65
Ref66
Ref67
Ref68

2
2
7
7
2
2
1
1
9

1
1
17
17
1
17
3
2
2

1
1
5
5
5
2
6
49
2

3 140 3
3 140 3
2 10 3
2 10 3
3
4
1
3
2
1
4
1
3
4
11 66
10 10 3

167

5
806
5
806
6
353
6
353
5
43
5
883
3
unk
51 362
5 1919

21
21
353
353
21
21
22
362
52

1069
1600
784
784
103
285
1256
1058
46

843
48
119
119
27
131
70
866
606

C
C
C
C
C
A1
A1
A1
E

Figure 4.1: Dendogram of the LOS biosynthetic loci. Cluster analysis of the gene
content at the LOS biosynthesis loci based on our PCR classification.

40

MI28-F
MI42-F
MI57-F
2
CDC9-B2
2
CDC2-B2
2
CDC5-B2
MI18-B2
MI19-B2
MI33-B2
MI56-B2
MI59-B2
MI50-A2
MI60-C
MI61-C
B2
ENV14-B2
unk
unk
MI36-B2
1
MI15-unk
1
MI16-unk
1
CDC4-A1
Ref66-A1
Ref67-A1
Ref65-A1
A1
MI26-A1
MI46-A1
CDC10-A1
MI29-B1
2
MI30-B1
MI39-B1
MI43-B1
CDC6-B2
Ref68-E
unk
MI24-E
unk
MI25-E
unk
MI34-E unk
unk
MI51-E
ENV11-unk
CDC3-unk
MI35-unk
ENV12-unk
MI32-unk

16
71
14
168

Figure 4.1 (cont’d)

71
46
169

CDC8-C
CDC1-C
Ref64-C
MI20-C
MI21-C
MI22-C
MI23-C
MI27-C
MI31-C
MI37-C
MI38-C
MI40-C
MI47-C
MI48-C
MI49-C
MI52-C
MI53-C
MI54-C
MI55-C
MI58-C
MI62-C
MI63-C
ENV13-C
C
unk
CDC7-unk
unk
MI17-unk
unk
MI45-unk
unk
MI44-unk
unk
MI41-unk

Figure 4.2: Comparison of LOS class to MLST Clonal Complexes. The figure
below shows the distribution of the 68 isolates based on their LOS class into the 19
MLST clonal complexes.

170

Figure 4.3: Phylogenic tree of the porA based on DNA sequences. Data
indicates 2 well supported clusters (bootstrap values ≥50%). The LOS class
assigned to each isolate follows the isolate ID.

MI36-B2
Ref68-E

CDC7-unk
MI42-F

100

100

MI38-C
CDC8-C
MI29-B1
MI54-C
MI22-C
MI63-C
MI62-C
MI48-C
MI40-C
MI45-unk
MI44-unk
MI34-E
MI35-unk
ENV14-B2
Ref66-A1
Ref67-A1
MI15-unk
CDC4-A1
MI61-C
MI16-unk
MI60-C
MI24-E
MI32-unk
MI25-E
CDC10-A1
MI46-A1

0.1

171

Figure 4.3 (cont’d)

CDC5-B2
ENV12-unk
CDC3-unk
ENV11-unk
MI27-C
MI30-B1
MI3I56-B2
MI18-B2
MI43-B1
MI28-F
MI21-C
CDC9-B2
MI19-B2
MI31-C
MI26-A1
MI47-C
CDC1-C
Ref64-C
CDC6-B2
ENV13-C
MI57-F
MI23-C
MI33-B2
CDC2-B2
Ref65-A1

0.1

MI53-C
MI37-C
MI50-E
MI49-C
MI39-B1
MI41-unk
MI59-B2
MI58-C
MI17-unk
MI52-C
MI55-C

172

CHAPTER 5

Campylobacter jejuni isolates from calves have A, B, and C lipo-oligosaccharide (LOS)
biosynthetic locus classes similar to human Guillain Barré Syndrome associated strains

173

ABSTRACT

Campylobacter jejuni, a Gram negative zoonotic bacterium, is a frequent cause
of human gastroenteritis. The acute neuropathy Guillain Barré Syndrome (GBS) is a
post-infectious polyradiculoneuropathy triggered by C. jejuni through molecular mimicry.
The lipo-oligosaccharide (LOS) of C. jejuni is variable; some forms can mimic
gangliosides enriched on peripheral nerves leading to autoimmunity. Classification of
the LOS can be determined by PCR based on gene content of the complex LOS
biosynthesis locus; 23 classes have been identified. Infection with C. jejuni strains of
LOS classes A–C has been strongly associated with GBS. Members of a family
managing a large southwest Michigan dairy operation reported a history of C. jejuni
infections that lasted several days and recurred over the course of two years. Because
up to 37.7% of dairy cattle have been found to shed C. jejuni, we sought to determine
whether calves were the source of the family infections. Fecal samples were obtained
from 25 randomly selected calves, 1 dog, and all family members and cultured for C.
jejuni. Human and calf isolates were characterized by LOS biosynthesis locus class
typing, multi-locus sequence typing (MLST), porA allele typing, and flaA SVR sequence
typing. C. jejuni was cultured from 15 (60%) of calves and one asymptomatic family
member; the dog was negative. Campylobacter coli was cultured from 2 (1%) of calves.
Some calves had diarrhea, but most were clinically normal. Typing of LOS biosynthetic
loci showed that 10 of 15 calf C. jejuni isolates fell into LOS classifications A, B, and C.
The human isolate belonged to LOS class E, which is associated with enteric disease.
Two calves had C. jejuni with LOS class E. Thus, the typing results suggest that some

174

calf and human C. jejuni isolates have similar characteristics. Molecular typing with
MLST, porA allele typing, and flaA SVR typing indicated that one calf strain and the
human isolate of the same LOS class share the same phylogenetic lineage, thereby
providing direct support for zoonotic transmission. Furthermore, finding multiple MLST
sequence types, flaA SVR sequence types, porA alleles, and LOS classes in isolates
from these calves demonstrates that there are diverse C. jejuni populations present on
this farm.

175

INTRODUCTION

C. jejuni is the leading cause of human bacterial gastrointestinal infection in the
Western world (25, 127, 146). It is estimated that 2.1–2.5 million cases of
campylobacterosis occur annually in the U.S. (41). C. jejuni infection symptoms include
fever, abdominal cramping, and watery or bloody diarrhea (6, 127). The majority of C.
jejuni infections are the result of ingestion of undercooked chicken, but it has been
shown that contact with household pets and the ingestion of raw milk, contaminated
water, or undercooked beef or pork can lead to infection (6, 127). Farm animals and
milk are natural reservoirs for C. jejuni. C. jejuni has been isolated from the intestines of
chickens and found in the feces of domesticated household pets, beef cattle, and dairy
cattle (41, 146). Dairy products are highly associated with C. jejuni outbreaks, while
poultry products are mostly associated with sporadic infections (81). A Swiss study by
Busato et al. (1999) examined 395 calves from cow-calf farms for enteropathogenic
bacteria and showed that C. jejuni was present in 32% of the calves (18). Results of a
molecular typing study of C. jejuni isolates from farm animals by Fitzgerald et al. (2001)
demonstrated a link between Campylobacter isolates from farm animals with isolates
from human cases within the community (41). Other studies have shown a correlation
between C. jejuni infections and people exposed to farm animals (44, 88).
Guillain Barré Syndrome (GBS) is an acute neuromuscular autoimmune disease
of the peripheral nervous system and is the most common cause of acute flaccid
paralysis since the near eradication of polio (25, 119). GBS commonly develops
following gastrointestinal infections with C. jejuni; recently a systematic review of

176

published studies between 1998 and 2010 was performed to estimate the number of
GBS cases as a result of C. jejuni infection. It was determined that C. jejuni infection
preceded 13–72% of GBS cases (119). Molecular mimicry of the LOS structures found
on the outer membrane of C. jejuni that resemble gangliosides enriched in nervous
tissue is thought to be the mechanism behind the pathogenesis of GBS (7). Symptoms
may start with limb weakness and loss of tendon reflexes, but may develop into
paralysis of the limbs, trunk, and facial muscles; 25% of patients may require
mechanical ventilation (73, 80, 148). Ten to twenty percent of patients do not completely
recover and have life-long disabilities; 3–10% die (7, 148). It has been estimated that
the annual incidence of GBS is 1–2 individuals per 100,000 people in North America
and Europe (136).
In 2007, Price et al. showed that there is an increase in development of
peripheral neuropathy in people who work with poultry (122). Additionally, results of one
study in Iowa and North Carolina by Vegosen et al. (2012) demonstrated that
neurological symptoms associated with exposure to C. jejuni are more commonly
reported among people who handle beef cattle (146).
In this study, residents of a family-owned southwest Michigan dairy farm
complained of recurring C. jejuni infections over a course of two years. To explore this
problem, stool samples were taken from 25 calves and 1 dog on the dairy farm and
tested for presence of C. jejuni. Fifteen of twenty-five calves tested positive for C. jejuni.
We hypothesized that transmission of C. jejuni was occurring between the family
members and the calves. To test this hypothesis, we used multiple molecular typing
schemes to determine the genetic relationships among the C. jejuni isolates. Isolates

177

were examined using MLST, LOS locus class determination, flaA SVR typing, and porA
typing. Based on our results, the majority of the calf isolates (67%) belonged to LOS
classes that are associated with the development of GBS; this result is consistent with
previous studies of increased rates of neurological symptoms in farm workers. One calf
isolate and the isolate from the only family member colonized by C. jejuni had identical
alleles for the seven MLST loci, flaA SVR, and porA; both isolates belonged to LOS
class E. These data strongly suggest that zoonotic transmission occurred between the
dairy calves and the family members.

178

MATERIALS AND METHODS

Case Study
Residents of a family-owned dairy farm in the southwest Michigan reported C.
jejuni infection among the family that lasted several days in summer, 2012. The family
had experienced previous recurring C. jejuni infections for the previous two years. A
veterinarian obtained 2 stool samples from dairy calves to eliminate them as possible
carriers of C. jejuni and sent the samples to the Diagnostic Center for Population and
Animal Health (DCPAH) at Michigan State University (East Lansing, MI). Using
bacteriological tests, DCPAH staff confirmed that the stool samples were positive for C.
jejuni. DCPAH staff then contacted our laboratory, the Comparative Enteric Disease
Laboratory at Michigan State University (East Lansing, MI). We contacted the
veterinarian, who obtained 25 randomly chosen calf samples along with one stool
sample from the farm dog and submitted them for further testing. The family members
enrolled in concurrent study conducted by Dr. Shannon Manning and the Microbial
Evolution Laboratory (Michigan State University, East Lansing, MI). The single positive
family sample was recovered by the Michigan Department of Community Health and deidentified. It was graciously given to us by Dr. Shannon Manning. Our laboratory had no
interaction with the family members or farm workers.

Campylobacter jejuni Isolation
Stool samples from the calves and dog were streaked on tryptose soya agar
plates with 5% sheep’s blood (Cleveland Scientific, Bath, OH, USA) and supplemented

179

with antibiotics (TSA-CVA: 20 µg/mL cefoperazone, 10 µg/mL vancomycin and 2 µg/mL
amphotericin B) upon arrival. The plates were then placed in anaerobic jars containing
CampyGen packs (Oxoid, Basingstoke, UK) and incubated for 48 hours at 37°C.
Isolates were selected and streaked again for growth and purification; bacterial growth
was collected and stored at -80°C in TSB + 15% glycerol.
The isolate from the family member who tested positive for C. jejuni was streaked
onto Bolton agar plates (Bolton Broth, Thermo Fisher Scientific, Pittsburgh, PA,
containing 15% Bacteriological Agar, Neogen, Lansing, MI) upon arrival in the
laboratory. The plates were incubated at 37°C for 48 hours in anaerobic jars containing
CampyGen packs (Oxoid, Basingstoke, UK). Bacterial growth was harvested and stored
as described above.

Preparation of Genomic DNA and Confirmation of Campylobacter spp.
Frozen cultures were streaked for growth onto Bolton agar plates (Bolton Broth,
Thermo Fisher Scientific, Pittsburgh, PA, containing Bacteriological Agar, Neogen,
Lansing, MI) and incubated for 48 hours at 37°C in anaerobic jars with CampyGen
packs (Oxoid, Basingstoke, UK). Bacterial growth was collected and DNA was extracted
using Wizard® Genomic DNA Purification Kits (Promega, Madison, Wisconsin)
according to the manufacturer’s instructions. All genomic DNA was diluted to a final
concentration of 25 ng/μL for PCR.
Identification of Campylobacter isolates as C. jejuni or C. coli was performed with
a multiplex PCR that has previously been described (96). Hot Start Syzygy Mean Green

180

DNA polymerase was used (Integrated Scientific Solutions, San Diego, CA). The
reaction had an enzyme activation step of 15 minutes at 95°C followed by 25 cycles of
denaturation for 30 seconds at 95°C, annealing for 1.5 minutes at 58°C, and extension
for 1 minute at 72°C, with a final extension of 7 minutes at 72°C. In addition, C. jejuni
isolates were verified with a previously described C. jejuni gyrA-specific PCR used in
our lab (96).

Classification of LOS Biosynthesis Loci
LOS classification was performed as previously described by Parker et al. (2005)
(112). Parker et al. (2005 and 2008) described 19 classes (A–S) of LOS based on gene
content. This was supplemented by the procedures of Stanhope et al. (2013) who
described an additional 4 LOS classes (T–W) (111, 112, 125).

Multi-locus Sequence Typing (MLST)
MLST based on 7 conserved housekeeping genes was performed on DNA from
the calf isolates as previously described (10, 11); only the inner primers were used to
amplify the partial gene sequences. Platinum® Taq DNA Polymerase High Fidelity (Life
Technologies, Grand Island, NY) was used; primer concentrations used were 37.5
pmol. The PCR reaction contained 35 cycles of denaturation of 2 minutes at 95°C,
annealing for 1 minute at 50°C, followed by an extension for 1 minute at 72°C. The
family member isolate was included in a concurrent study with the MSU Microbial

181

Evolution Laboratory; and the protocol for MLST differed. MLST PCR reactions were
performed on DNA from the family member isolate using KAPA2G HotStart DNA
polymerase (KAPABiosystems, Woburn, MA). Briefly, 10 μM of both the forward and
reverse primer, 25 ng/μl of genomic DNA template, water, and the DNA polymerase
were used for the reaction. The reactions began with an enzyme activation step of 3
minutes at 95°C and consisted of 35 cycles of denaturation for 15 seconds at 95°C,
annealing for 15 seconds at 60°C, and an extension for 5 seconds at 72°C.

porA Allele Typing and flaA SVR Sequence Typing
porA allele typing was done using 7 different primers previously defined in
Chapter 4 of this dissertation or on the PubMLST website
(http://pubmlst.org/campylobacter/info/porA_method.shtml) (83). The PCR reaction
consisted of 40 cycles of denaturation for 30 seconds at 94°C followed by an annealing
step for 30 seconds at 45°C and extension for 90 seconds at 72°C. Platinum® Taq
DNA Polymerase High Fidelity (Life Technologies, Grand Island, NY) was used with 40
pmol of primer.
The flaA SVR sequence typing was done with Platinum® Taq DNA Polymerase
High Fidelity (Life Technologies, Grand Island, NY) and 40 pmol of primer. The PCR
reaction consisted of a denaturation step at 94°C for 1 minute, annealing at 45°C for 1
minute, and extension for 3 minutes at 72°C for 35 cycles. The primers used were
previously described (101).

182

Sequence Analysis
PCR products were purified with QIAquick PCR Purification Kit (Qiagen,
Germantown, MD) according to the manufacturer’s instructions and submitted for
sequencing to the Michigan State University Genomics Technology Support Facility
(East Lansing, MI) on an ABI 3730 Genetic Analyzer (Life Technologies, Grand Island,
NY). Each PCR product was sequenced in both the forward and reverse directions.
SeqMan 5.06 (DNASTAR, Madison, WI) software was used to align the sequences; the
PubMLST website was used to determine allele, sequence type, and clonal complex
assignments (http://pubmlst.org/campylobacter/) (83).

Statistical Analysis
Alignments, cluster analysis, and dendrograms for porA and flaA SVR DNA
sequences were performed using phylogeny.fr, which uses a maximum likelihood
clustering method (23, 33, 39).
The allelic profiles of MLST, porA, and flaA SVR were analyzed with LIAN 3.5
available at the PubMLST website (http://pubmlst.org/campylobacter/) (65, 83). The
standardized Index of Association (sIA) was calculated and a Monte-Carlo test of the
hypothesis that the population was in linkage equilibrium (that is, that alleles at the
different loci were randomly distributed among the isolates) was run with 1000 bootstrap
repetitions.
Cluster analysis was performed using the Hamming similarity coefficient and
unweighted pair group with arithmetic averaging (UPGMA). All bootstrapping was

183

performed at 1000 repetitions. Tree files of the porA and flaA SVR DNA sequences
were compared using Compare2Trees: pairwise comparison of phylogenies (106). This
algorithm produces an overall topological score that describes the percent of similarity
between the trees (106).

184

RESULTS

Verification of C. jejuni
Based on culture, 15/25 calf stool samples were positive for C. jejuni and 2/25
calf stool samples were positive for C. coli. The dog sample was negative. One family
member tested positive for C. jejuni without reporting any symptoms; therefore, this
family member was considered an asymptomatic carrier (Table 5.1).

LOS Classification
Based on the PCR results of 12 different genes in LOS biosynthesis loci, 10/15
(67%) of the calf isolates were classified into one of the LOS classes A–C. One of
fifteen calf isolates belonged in class A2, 5/15 were in class B2, and 4/15 belonged to
LOS class C. Two of fifteen calf isolates were determined to belong to LOS class E; the
family member isolate also belonged to class E. Three of fifteen isolates were in the
LOS class F. No isolates were determined to belong in the LOS class D.

MLST Analysis
MLST allelic profiles are shown in Table 5.1. Six sequence types (ST; 806, 922,
928, 982, 1712, and 6227) were observed based on the DNA sequences compiled for
15 calf isolates and the single family member. Five of these sequence types could be
further classified into 2 clonal complexes (CC; 21 and 257); ST-922 is currently not
assigned to a defined clonal complex according to the C. jejuni/C. coli PubMLST
website (http://pubmlst.org/campylobacter/) (83). The family member isolate also

185

belonged to ST-922; the clonal complex of the four ST-922 isolates is designated
“unknown’ in Table 5.1. One of the 15 calf isolates (C2) and the single family member
isolate (FM1) had identical alleles at all loci. Calf isolate C7 was found to have the same
LOS class (E), MLST sequence type (ST-922), and clonal complex (unknown) as FM1,
and had identical allele sequences for 6 of the 7 MLST loci, but varied in the allele of the
pgm gene, which encodes a phosphoglucomutase.
Linkage analysis of the MLST alleles using LIAN 3.5 indicated that the population
is in linkage disequilibrium (Monte Carlo p<0.0001), that is, particular alleles at different
loci are more commonly associated with each other than expected by chance. The
Index of Association sIA was 0.47, consistent with a population in linkage disequilibrium
with low levels of recombination between strains. Figure 5.1 shows a split
decomposition analysis of the MLST allelic profiles; the presence of the network
suggests that recombination may be occurring among the population of C. jejuni,
although at low levels.
One of 9 isolates of CC 21 belonged to LOS class A2 (1/1); 5 of 9 isolates, to
LOS class B2 (5/5); and 3/9 isolates, to LOS class C. The 3 CC 257 isolates included
one isolate of LOS class C and two isolates of LOS class F. The 4 ST-922 strains that
could not be assigned to a clonal complex included 3 LOS class E and 1 LOS class F
isolates.

porA Allele Typing and flaA SVR Sequence Typing
DNA sequences for the porA gene were analyzed and assigned allele types; 4
different alleles were detected. Two alleles, 749 and 1696, were each present in 5/15 of

186

the calf isolates. Allele 744 was found in 4/15 of the calf isolates and the family member
isolate. Allele 351 was found in a single calf isolate. Cluster analysis of the porA DNA
sequences revealed 2 clusters of strains that were well supported by bootstrap values
≥50% (Figure 5.2, panel A).
porA allele 749 was present in all five isolates of LOS class B2. Allele type 744
was present in both LOS class E calf isolates, the family member isolate, the single
LOS class A2 calf isolate, and 1 of the 3 calf isolates in LOS class F. The other LOS
class F calf isolates had porA allele 1696. LOS class C isolates contained 2 porA alleles
351 and allele 1696.
Cluster analysis of flaA SVR sequences produced 3 strongly supported clusters
of strains (Figure 5.2, panel B). DNA sequence analysis resulted in the detection of 6
different flaA SVR sequence types among the isolates. Four of 15 calf isolates had flaA
SVR sequence type 16. Two of fifteen calf isolates had SVR sequence type 14; 2/15
calf isolates had flaA SVR sequence type 227; and 5/15 calf isolates had flaA SVR
sequence type 41.The remaining sequence types, 327 and 111, were each found in a
single calf isolate. The family member isolate carried flaA SVR sequence type 14.
Comparison of the flaA SVR sequence types with the LOS classification showed
that all of the LOS class E isolates carried flaA SVR sequence type, ST-14; as did the
family member isolate. The single isolate that classified as LOS class A2 carried flaA
SVR sequence type 41; LOS class B2 isolates carried flaA SVR sequence types 41
(4/5) and 222 (1/5). The four calf isolates with LOS class C carried flaA SVR sequence
types 16 (2/4), 227 (1/4) and 111 (1/4). The three calf isolates with LOS class F were
isolates carried flaA SVR sequence types 16 (2/3) and 327 (1/3). Comparison of the

187

porA and flaA SVR DNA sequence dendrograms resulted in an overall topological score
of 67%; reinforcing the similarity of the results from the two molecular typing schemes.

Composite Profiles
The allele profiles derived from MLST, flaA SVR sequence typing, and porA
allele typing were assessed together; results are shown in Figure 5.3. LIAN 3.5 was
used to calculate linkage equilibrium parameters; the results indicated that the
population is in linkage disequilibrium (Monte Carlo p<0.0001) with an index of
association sIA of 0.43. The average genetic diversity, H, for the composite allele
profiles from MLST (7 loci), flaA SVR, and porA was calculated using LIAN 3.5; H was
0.67 ± 0.05. It was also observed that within the 16 different isolates (15 calf and 1
family member) analyzed in this study, there were 13 distinct composite profiles based
on MLST, porA alleles, and flaA SVR sequence type; this demonstrated a very diverse
population of isolates.

188

DISCUSSION

Campylobacter jejuni is the leading cause of bacterial gastrointestinal infection
and is commonly found in the intestines of farm animals, including chickens and cattle
(146). C. jejuni is the most common antecedent infection to the development of the
autoimmune muscular neuropathy, Guillain Barré Syndrome (GBS). Studies have been
performed that show an increase in neurological symptoms in people who work with or
handle farm animals compared to people who do not (122, 146). Previous studies have
demonstrated that C. jejuni is prevalent in 0–51.2% of dairy farms in the U.S. (64, 131).
Therefore when a small outbreak of C. jejuni occurred in humans living on a on a dairy
farm in southwest Michigan, we not only analyzed the isolates to determine whether
zoonotic transmission had occurred between family members on the farm and the dairy
cattle, but also to determine the genetic relationships among the isolates and to
evaluate whether dairy calves could be a significant reservoir of C. jejuni strains having
LOS locus types associated with GBS.

Zoonotic Transmission Between the Calves and Family Members
Analysis of the LOS classification of the 15 C. jejuni calf isolates and 1 family
member isolate showed that 2 calf isolates (C2 and C7) shared the same LOS class as
the family member isolate (FM1); all carried LOS locus class E, which lacks the genes
needed to synthesize sialic acid and transfer it to the LOS outer core sugar residues.
Calf isolate C2 and family member isolate FM1 were identical in LOS classification
(class E), MSLT sequence type (ST-922; unknown clonal complex), flaA sequence

189

(sequence type 14), and porA sequence (allele 744). Calf isolate C7 differed from
isolates C2 and FM1 only in the sequence of the pgm gene in the MLST analysis, but
still shared the same sequence type (ST-922). It has been demonstrated that a great
deal of recombination occurs in the pgm and uncA loci of C. jejuni (43). We suspect that
the C7 calf isolate experienced a mutation at the pgm locus resulting in a different pgm
allele assignment than that of isolates C2 and FM1; BLAST analysis of the two
sequences resulted in a difference of 2 nucleotides. Based on these data, we concluded
that C. jejuni isolate C2 could have been transmitted from the calf to the family member.
However, it is impossible to exclude the possibility that the family member infected the
calf.

Examination of Genetic Relationships Among Calf Isolates with LOS Locus
Classes A–C
We determined that 67% of the calf isolates belonged to LOS classes A–C,
which are commonly associated with GBS; these locus classes contain the genes
necessary to synthesize a sialylated LOS structure on the outer membrane of C. jejuni.
To dissect further the population structure of these isolates, we typed them by a
standard MLST protocol and by sequencing portions of the variable porA gene and flaA
SVR, both of which are outer membrane structures of C. jejuni.
Our results are similar to those obtained in other studies of C. jejuni in dairy
cattle. A study done by Sanad et al. (2013) on a dairy farm in Ohio showed that 36.6%
of the cattle were positive for C. jejuni (131). In this study they examined the genetic
relationships of the C. jejuni isolates found in both dairy cattle and starling birds with

190

MLST and found the most common clonal complexes in the cattle to be CC 21, CC 42,
CC 45, and CC 62 (131). Clonal complex 21 was also the most common complex
identified in our study. In a Finnish study, that authors performed MLST on 102 bovine
C. jejuni isolates and found that 51% of the isolates belonged to CC 21 (32). Also,
Dingle et al. (2001) determined that clonal complexes CC 21 and CC 45 were the most
common clonal complexes among human C. jejuni infections (34).
Jay-Russell et al. (2013) utilized porA sequencing and MLST sequences of
human C. jejuni isolates to isolates taken from dairy cattle to study a milk-borne
outbreak of campylobacterosis (81). These authors concluded that the level of strain
discrimination provided by porA typing was very useful in identifying an association of
dairy farm C. jejuni isolates with the milk-borne outbreak strain. In addition, MLST CC
21 was the most prevalent clonal complex among the cow isolates analyzed in that
study (81). Our results from porA sequencing also provided important support for our
hypothesis of zoonotic transmission. We identified 4 porA alleles in our sample and
determined that the family member isolate and the calf isolate, C2, both shared the
same porA allele.
The C. jejuni strains in our study were isolated from a small set of calves from a
single breeder and probably reflect a larger C. jejuni population in the breeding herd.
The high incidence (67%) of strains carrying LOS locus classes associated with GBS
even in this small sample justify larger studies in US dairy herds, especially given the
growing popularity of raw milk consumption. Based on our study and studies by others
there is compelling evidence to look for a link between C. jejuni from cattle and
development of GBS.

191

Table 5.1: Summary of Molecular Typing. The first column shows the strain ID’s for
the 15 calf C. jejuni isolates (C1-C15) and the single family member C. jejuni isolate
(FM1). The next 7 columns are the allele assignments for the genes analyzed in MLST
followed by the sequence type (ST) and clonal complexes (CC). The column labeled
flaA shows the alleles assigned to the flaA SVR DNA sequences. The porA column
shows the alleles assigned to the porA sequences. The last column is the LOS
biosynthesis locus classification as determined by PCR. Three calf isolates and the
single family member isolate were not assigned to a clonal complex and therefore
labeled “unk” for unknown.

ID aspA glnA gltA glyA pgm tkt uncA ST cc flaA porA
C1
9
2
4
62
4
5
1
1712 257 16 1696
C2
1
1
2
83
2
3
6
922 unk 14 744
6227 21 227 351
C3
2
1
1
362
2
1
6
C4
2
1
1
3
140 3
5
806 21 41 749
C5
2
1
1
3
140 3
5
806 21 227 749
C6
2
1
1
3
140 3
5
806 21 41 744
C7
1
1
2
83 447 3
6
922 unk 14 744
C8
2
1
1
3
140 3
5
806 21 41 749
6227 21 16 1696
C9
2
1
1
362
2
1
6
C10
9
2
4
62
4
5
38
928 257 16 1696
C11
1
1
16
83
1
3
6
922 unk 327 744
C12
9
2
4
62
4
5
38
928 257 16 1696
C13
2
1
2
3
2
1
5
982 21 111 1696
C14
2
1
1
2
140 3
5
806 21 41 749
C15
2
1
1
3
140 3
5
806 21 41 749
FM1
1
1
2
83
2
3
6
922 unk 14 744

192

LOS
C
E
C
B2
B2
A2
E
B2
C
F
F
F
C
B2
B2
E

Figure 5.1: SplitsTree diagram based on MLST alleles. As seen in the tree below,
C2 and FM1, which are 1 calf isolate and the single family member positive for C.
jejuni, are identical in allele types based on MLST. The presence of the network
suggests the occurrence of low levels of recombination.

ST 922
CC unknown

ST 806
CC 21

193

Figure 5.2: Dendograms of porA and flaA SVR DNA sequences. Panel A is a
dendogram of the porA sequences; two well supported clusters were observed.
Panel B is a dendogram of the flaA SVR DNA sequences; 3 well supported clusters
were observed. These clusters are defined based on bootstrap values ≥50% and
indicated with brackets. Isolates C2 and FM1 both shared the same porA allele type
and flaA SVR sequence type; they are indicated by a

A
100

C6-A2

C7-E
C2-E
FM1-E
C11-F

0.02

100
84
100
88

0.09

194

C3-C

C1-C
C10-F
C9-C
C12-F
C13-C
C14-B2
C4-B2
C5-B2
C15-B2
C8-B2

100

B

.

C1-C
C9-C
C10-F
C12-F
FM1-E
C7-E
C2-E
C11-F
C15-B2
C13-C
C4-B2
C3-C
C5-B2
C6-A2
C8-B2
C14-B2

Figure 5.3: SplitsTree analysis of combined allelic profiles of 7 MLST loci, porA
alleles, and flaA SVR sequence types.

ST 922
CC unknown

ST 806
CC 21

195

CHAPTER 6

Conclusions

196

SUMMARY

All disease results from one or more of the following factors: pathogens (and their
genetics), host genetics/immune responses, the microbiota, and the environment.
These factors interact in many diseases such as the role of the hygiene hypothesis in
the development of allergies and the role of parasites (155). Some autoimmune
diseases are associated with prior infection; Guillain Barré Syndrome (GBS) are
prominent examples (7). GBS is an autoimmune disease of the peripheral nerves in
which both the myelin sheath and axons may be involved; it is the most frequent cause
of acute flaccid paralysis (116, 160). GBS frequently follows gastrointestinal infections
caused by Campylobacter jejuni (105). C. jejuni may also result in the development of
other autoimmune diseases such as Reiter’s Syndrome, Inflammatory Bowel Disease,
and Irritable Bowel Syndrome (44, 68, 75, 87, 125). The overall aim of my work was on
GBS induced by C. jejuni with focus on two aspects of the disease—host
genetics/immune responses and pathogen genetics. Pathogen genetics is thought to
provide the primary mechanism of pathogenesis of the disease. The pathogenesis of
GBS following C. jejuni infection is believed to involve immune responses to
ganglioside-like structures displayed by the lipo-oligosaccharides (LOS) on the outer
surface of C. jejuni cells that mimic neurogangliosides (7).
The LOS loci of C. jejuni can be assigned to classes A–W based on the presence
or absence and organization of the LOS biosynthesis genes (47, 111, 112, 125). LOS
classes A, B, C, M, R, and V all have the necessary genes to synthesize a sialylated
LOS to mimic gangliosides enriched in the nervous tissues. C. jejuni strains with LOS

197

classes A and B have been isolated from GBS patients and therefore are highly
associated with development of GBS. However, it has been documented that C. jejuni
strains that express an LOS with class A or B have also been isolated from patients who
only developed enteritis (47, 111). This result suggests that the host’s genetics and
immune response is important to development of disease, but the basis for this
interaction is currently unknown. C. jejuni isolates from GBS patients have also been
shown to have LOS loci (F, H, K and P) that do not have the necessary genes to
synthesize an LOS structure capable of mimicking a ganglioside. These data suggest
that there may be alternative mechanisms in C. jejuni that result in the sialylation of the
LOS and subsequent production of anti-ganglioside antibodies; there could also be
involvement of specific pathogen or host genes that have not been identified. There
may also be alternative targets on nerves that are shared on C. jejuni strains. A better
understanding of the genetic relationships among LOS classes A–C is needed.
Molecular typing schemes have been applied to assess the genetic relationship
of LOS classes A–C with variation in other genes within the genomes of C. jejuni
strains. These schemes include multi-locus sequence typing (MLST), porA allele typing,
and flaA SVR sequence typing (27, 28, 34, 101). MLST is a widely used typing tool
based on the partial sequences of 7 highly conserved housekeeping genes in C. jejuni.
Sequence typing of the major outer membrane protein, porA, and the short variable
region (SVR) of the flaA gene of the flagellum, both of which are outer surface
molecules of C. jejuni, has provided conflicting data about the genetic relationships
among C. jejuni isolates in LOS classes A–C versus those that have LOS classes not
associated with ganglioside mimicry. These two outer surface molecules are highly

198

variable, and therefore it has been suggested that combined molecular typing using all
of these techniques (MLST, porA, flaA SVR, and LOS) might provide less conflicting
data than when these typing schemes are used separately.
GBS is characterized as a group of disorders affecting the peripheral nerves that
include acute inflammatory demyelinating polyneuropathy (AIDP), acute motor axonal
neuropathy (AMAN), and Miller Fisher Syndrome (MFS). AIDP is the most common
subtype of GBS found in humans, is characterized by myelin damage, and affects most
myelinated motor nerves of the limbs as well as lower cranial motor and sensory nerves
(145). AMAN is the most common subtype identified in Asia and is characterized by
damage to motor neurons, specifically the nodes of Ranvier. Tissue damage in AMAN
occurs mainly in the nerve roots or distal nerve terminals (145). MFS is an anatomically
localized variant of GBS characterized by acute onset of ophthalmoplegia, ataxia, and
areflexia. In some instances, patients with MFS progress to GBS (109, 138, 159). Antiganglioside antibodies have been detected in approximately 60% of GBS patients (84).
The current animal models used to study the pathogenesis of GBS include an
AMAN model in rabbits, an AIDP model in mice, and an AMAN model in chickens. The
rabbit model is based on immunization of rabbits with purified C. jejuni LOS that is
known to mimic gangliosides. The rabbits develop neuropathy and anti-ganglioside
antibodies, but this is not a model of GBS secondary to C. jejuni infection (natural
model) to study development of GBS (164, 165). In the mouse model of AIDP, mice are
immunized with bovine myelin and develop nerve lesions; demyelination occurs that is
similar to that in GBS patients that present with AIDP (8, 126). Like the rabbit model,
this mouse model does not develop GBS secondary to C. jejuni infection and therefore

199

is not a natural model of GBS. The chicken model of AMAN is the only natural animal
model used to study GBS. Chickens inoculated with C. jejuni isolated from a GBS
patient developed paralysis and nerve lesions similar to those seen in AMAN patients
(94). However, chickens are anatomically and physiologically different from humans:
therefore a mouse model is needed to study the natural development of GBS secondary
to C. jejuni infection.

Chapter 2 Summary: Non-Obese Diabetic (NOD) mice as working models of
Guillain Barré Syndrome following Campylobacter jejuni infection
We attempted to develop a mouse model of GBS using wild type Non-Obese
Diabetic (NOD) mice and their congenic knockouts NOD IL-10

-/-

and NOD B7-2

-/-

mice

infected with C. jejuni isolates from GBS patients. These GBS-associated strains were
C. jejuni HB93-13, which was isolated from a AMAN patient in China, and C. jejuni
260.94, which was isolated from an AIDP patient in South Africa (2). Our experiments
included confirmation of C. jejuni colonization, detection of anti-ganglioside antibodies,
assessment of clinical neurological phenotypes, and detection of nerve lesions (myelin
damage and cellular infiltrates).
Based on our results, we determined that NOD WT mice have the greatest
potential as a suitable model. When inoculated with the two C. jejuni isolates taken from
GBS patients, these mice were colonized with C. jejuni but had minimal enteric disease
as assessed by evaluation of both gross and histopathological changes in the GI tract.
Thus enteric disease should not have influenced clinical evaluation of neurological
signs. During infection with C. jejuni, NOD WT mice developed anti-ganglioside

200

antibodies to both individual gangliosides and ganglioside mixtures and displayed a
neurological phenotype. The infected NOD WT mice also had significantly greater
infiltration of macrophages into the sciatic nerve. However, nerve lesions were not
detected in the sciatic nerve. This likely resulted because we only removed a small
portion of the sciatic nerve when following standard technique for removing nerves from
the mouse. The lesions could also occur in locations other than the sciatic nerve, such
as the dorsal root ganglia. Therefore other sections of the peripheral nervous system
should be removed in the future for analysis. Another reason for not detecting lesions
could be a problem with selecting the appropriate time to sacrifice the animals. We may
not have waited long enough for lesions to form since infiltration of macrophages could
be an early step in lesion development. More work is needed to confirm that the NOD
WT mice are a true murine model of GBS secondary to C. jejuni infection and to select
the optimal time for sampling after lesion development.
NOD IL-10

-/-

mice developed enteritis from the C. jejuni infection in Experiment 1

and therefore were given chloramphenicol (CMP) in Experiment 3 to attempt to prevent
enteritis from occurring and potentially interfering with the assessment of clinical
neurological phenotypes. The NOD IL-10

-/-

mice were colonized with C. jejuni,

developed gross pathological changes (enlarged lymph nodes and thickened wall of the
gastrointestinal tract), displayed high scores for histological assessment of the GI tract
lesions, developed anti-ganglioside antibodies, and displayed clinical neurological
phenotypes as determined by the DigiGait™. In Experiment 3 we also improved our
techniques for removal of the sciatic nerve: we removed the entire length of the nerve,

201

the nerve roots (L3–L5), and also removed the brachial plexus for analysis of nerve
lesions and cellular infiltrates. Here, we identified a significant increase in the number of
macrophages in the nerve roots of infected mice compared to uninfected mice. We also
identified structures we termed “fat vacuoles” in the sciatic nerves of all treatment
groups of the NOD IL-10

-/-

mice. These fat vacuoles could actually be macrophages

that were ingesting damaged myelin. Myelin has lipid components, and therefore these
fat vacuoles could indicate myelin damage. To confirm this hypothesis, we could use a
stain for lipids or myelin proteins in order to assess the significance of this finding.
In Experiment 3, treatment with CMP alone resulted in development of enteritis.
This was a somewhat surprising finding because C. jejuni are susceptible to CMP.
However, these mice were IL-10 deficient and therefore would not have been able to
down-regulate an inflammatory response, whether it was a result of C. jejuni infection or
a disruption in the intestinal microbiota. To continue studying these mice as a potential
GBS model, a ganglioside mimicking C. jejuni strain that colonizes but doesn’t cause
enteric disease should be identified.
In both the NOD WT and the NOD IL-10

-/-

mice, we saw an increase in infiltration

of macrophages in the sciatic nerve sections and nerve roots, respectively. In both
locations there are resident macrophages and neutrophils present; however we found a
significant increase in the number of macrophages in both locations in mice infected
with C. jejuni 260.94 when compared to their sham-inoculated counterparts. This
elevation in macrophage numbers suggests that we were observing additional
macrophages in the mice due to the infection with C. jejuni.

202

NOD B7-2

-/-

mice were colonized by the C. jejuni strains we used for our

infection studies with minimal enteric disease. These mice developed anti-ganglioside
antibodies after infection and displayed clinical neurological phenotypes. However, no
lesions or cellular infiltrates of the nerves were identified in these mice. Perhaps, for the
same reasons no lesions were detected in NOD WT mice, we missed the right location
and time for taking nerve samples. Nevertheless, we did not pursue these animals as a
model because these mice develop spontaneous neuropathy and nerve lesions at or
after 20 weeks of age, which would be a disadvantage in long term studies.
As stated above, GBS is characterized as a group of disorders that all result in
damage to the motor nerves. Each of the NOD background mice examined in our
experiments displayed neurological phenotypes; however not all mice developed such a
phenotype. During the course of the experiments we used or developed neurological
tests for detection of motor function failure in the peripheral nervous system. The tests
we used included the open field test (OFT), hang test, rotarod, and footprint and
DigiGait™ analysis that allowed detection of ataxia, loss of motor strength, loss of
balance, and abnormal gait, respectively. Each of these tests focused only on the loss
of motor function. However, the first onset of symptoms in humans usually includes
tingling sensations, suggesting that sensory nerves are involved. Sensory nerves have
no Schwann cells, are poorly myelinated, and have a slow conduction velocity while
motor nerves are highly myelinated, have Schwann cells, and have a fast conduction
velocity. Given these differences between the two types of nerves, it would be beneficial
to develop sensory tests that could be applied to mice since our current neurological
tests focus solely on the motor nerves. Therefore, we suggest that future work with the

203

NOD WT murine model include sensory tests to determine if there is loss of sensory
function.
Sensory tests that could be performed include tail flick and hotplate tests. The tail
flick test consists of applying a thermal stimulus to a mouse’s tail that results in a simple
spinal reflex of the tail flicking (30). The hotplate test consists of placing a mouse on the
surface of a hotplate that is maintained at 50–55°C. The length of time required for the
mouse to raise and lick its forepaw or jump up is then recorded. If no response is
observed after 30 seconds then the mouse is removed from the plate (30). Mice
infected with C. jejuni strains associated with MFS could also be tested for the palpebral
reflex by touching the eyes noting whether the eyelid closes (42). These tests could all
be added to future experiments. As described above, many of the subtypes of GBS are
mainly associated with motor nerves, and therefore our previously defined tests would
continue to be used.

Chapter 3 Summary: Examination of Miller Fisher Syndrome in NOD WT and
C57BL/6 WT mice subsequent to Campylobacter jejuni CF93-6 infection
This experiment assessed the development of MFS and GBS in mice infected
with a MFS-associated C. jejuni strain compared to mice infected with a GBSassociated C. jejuni strain. There were two groups of infected NOD WT mice: one group
received the GBS-associated C. jejuni strain and the other received the MFS-associated
C. jejuni strain. The C57BL/6 WT mice only received the MFS-associated C. jejuni
strain. Results from this experiment were all insignificant except for presence of antiganglioside antibodies in the NOD WT mice infected with the GBS-associated C. jejuni

204

strain compared to sham-inoculated NOD WT mice. All other treatment groups of both
the NOD WT and C57BL/6 WT mice were not significantly different from shaminoculated mice in any of the parameters examined. The origin of the mice could
provide an explanation for the lack of significant results in the NOD WT mice in this
experiment. The NOD WT mice used in Experiment 1 were all bred in our laboratory’s
specific pathogen free breeding colony; the NOD WT mice used in this experiment were
purchased directly from The Jackson Laboratory and sent straight to the MSU
containment facility for inoculation. These mice may have a different GI tract microbiota
from the mice from our breeding colony, which could result in the different results seen
between experiments. Furthermore, the NOD WT mice in the breeding colony could
have diverged genetically from mice at The Jackson Laboratory since the breeding
colony was established five years prior to Experiment 1 from only three pairs of
breeding mice. Regardless, the results suggest that further work to develop a mouse
model of MFS should include reconsideration of other inbred mouse strains and other
strains of C. jejuni from patients with MFS.

Chapter 4 Summary: Multiple molecular typing schemes applied to define genetic
relationships among C. jejuni isolates that share the same LOS classification
The LOS biosynthesis loci have been analyzed in numerous sequenced and
unsequenced strains of C. jejuni and a classification scheme has been developed
based on the gene content of the loci. In this study we characterized the LOS loci of C.
jejuni isolates collected in our laboratory (68 total: 63 unsequenced and 5 previously
sequenced); the majority were collected in Michigan (49/68), but a subset came from

205

other locations in the US. One strain was from Canada. To date there have been 23
LOS classes (A–W) identified, but in our analysis we only classified the LOS into seven
groups: classes A–F or unknown (unclassified). We identified 12/68 isolates that could
be classified as members of more than 1 of the previously described LOS classes and
therefore were termed “unknown.” Of the remaining 56 isolates, 86% belonged to LOS
classes A–C and the remaining 14% were classified as LOS class E or class F; no
isolate was classified as LOS class D. An examination of only the 49 Michigan isolates
showed that 69% of them belong to LOS classes A–C; this percentage is near the upper
end of the range shown by previous studies (32–77%) (57, 69, 78, 123, 124),
suggesting that there is a higher than average likelihood of becoming infected with a C.
jejuni strain associated with GBS in Michigan.
In these studies, the identification of the LOS classes showed that a high
proportion of C. jejuni isolates typed with LOS classes A–C. Therefore we wanted to
examine possible genetic relationships of other outer membrane surface molecules with
the LOS classes in these isolates using previously described molecular typing
schemes—porA allele type and flaA SVR sequence type—both of which are highly
variable. We also included MLST analysis since it assesses the gene content of
conserved housekeeping genes and provides a phylogenetic framework for the strain
collection.
Analysis of MLST data showed that the collection contained 37 different
sequence types (ST) that were further grouped into 19 clonal complexes. Comparison
of the MLST data with the LOS classification showed that only 3 clonal complexes of the
19 contained only one kind of LOS class. The alleles assigned to the porA sequences

206

resulted in 43 different allele types with only 2 porA alleles associated with a single LOS
class. flaA SVR sequence typing resulted in 39 different sequence types; only 2 of
which were associated with a single LOS class. Overall analysis of these molecular
typing schemes indicated the possibility that some recombination occurred; the group of
isolates was genetically diverse.
To further understand the role of the LOS biosynthesis locus genes, the flaA
SVR, and the porA gene of the C. jejuni isolates we examined, future experiments could
be performed to determine whether these genes are being expressed during infection.
We believe that determining whether a gene is present or absent or the sequence of
that gene does not provide enough not information to discern the characteristics of
GBS-associated C. jejuni strains. We need to determine whether that gene is being
expressed in the host during infection.

Chapter 5 Summary: Campylobacter jejuni isolates from calves have A, B, and C
lipo-oligosaccharide (LOS) biosynthetic locus classes similar to human Guillain
Barré Syndrome associated strains
Chapter 5 focused on a small epidemiological study of a C. jejuni outbreak on a
Southwest Michigan dairy farm. Family members on the farm had complained of
reoccurrence of C. jejuni infections in previous years, so 25 stool samples were
obtained from calves on the farm, along with 1 stool sample from a pet dog. All family
members were tested for C. jejuni as well. Our results indicated that 15/25 (60%) of the
calves were positive for C. jejuni and 2/25 (8%) were positive for C. coli; no calf was
positive for both C. jejuni and C. coli. The dog and all family members but one tested

207

negative. The 15 calf and 1 human C. jejuni isolates were analyzed with PCR-based
LOS classification and MLST to determine whether zoonotic transmission occurred
between the family member and the calves. The LOS classification of the calf strains
showed that 10 of the calf isolates typed as LOS class A, B, or C. Two calf isolates fell
into LOS class E and 3 calf isolates class F. The family isolate was classified as LOS
class E. MLST results showed that there were 5 different sequence types among the
strains; the sequence types were further grouped into 2 defined clonal complexes and 1
unknown clonal complex. All isolates typed as LOS class A2 or class B2 fell into the
same clonal complex. The family member isolate grouped in the same sequence type
(ST-922; unknown clonal complex) as the LOS class E isolates from the calves. DNA
sequence analysis of the porA gene grouped the calf isolates into 4 different allele
types; the family member isolate carried the same allele of the other LOS class E
isolates. flaA SVR sequence typing defined 6 different sequence types; the LOS class E
isolates had the same flaA SVR sequence type as the LOS class E isolate from the
family member. Analysis of the combined allele profiles from the three typing schemes
with LIAN 3.5 indicated that minimal recombination occurred.
Based on MLST data, one calf isolate (C2) and the family member isolate had
identical MLST allele types and were both LOS class E. The family member isolate also
shared the same porA allele type and flaA SVR sequence type as calf isolate C2. These
results therefore suggest zoonotic transmission between the calf and the family
member. Since all but one calf isolate had different allele and sequence types, it
suggests that the family member contracted the C. jejuni from the single calf that carried
a C. jejuni strain that shared sequence type, LOS class and porA allele and flaA SVR

208

sequence type with the human isolate. Alternatively, the calf may have been infected
from the family member. The direction of transfer cannot be discerned from this
analysis.
One other C. jejuni calf isolate, C7, belonged to the LOS class E and shared the
same MLST sequence type and clonal complex, porA allele type, and flaA SVR
sequence types as calf isolate C2 and the family member isolate. The C7 calf isolate
differed from the calf isolate C2 and family isolate in the allele assignment to the MLST
housekeeping gene, pgm, which has been shown to be subject to high rates of
recombination. These data suggest recombination occurred in the C7 calf isolate.

209

FUTURE DIRECTIONS

Defining the NOD WT mouse as a natural model of GBS
As stated previously, more work needs to be done to determine if NOD WT mice
are suitable models of GBS disease that occurs secondary to C. jejuni infection. First, a
time course experiment should be performed. An important goal in this experiment
would be to determine the time at which anti-ganglioside antibodies peak. We propose
to employ the DigiGait™ apparatus to detect clinical neurological phenotypes since it
has been shown to be sensitive enough to detect the subtle changes seen in GBS. The
DigiGait™ was not available during experiments performed with NOD WT mice;
therefore, assessment of NOD WT mice using this apparatus could result in a more
detailed assessment of the clinical neurological phenotype than in our previous
experiments. In future experiments an assessment of sensory nervous system function
will also be included as described above.
Examination of the nerve root for infiltration of macrophages should also be done
with NOD WT mice exhibiting GBS. In our previous experiments with NOD WT mice, we
only examined a small section of the sciatic nerve; future experiments should include
the entire length of the sciatic nerve, including the nerve roots because we have shown
this to be effective in finding peripheral nerve lesions wherever they occur. Previous
experiments also suggest that light microscopy does not provide enough detail to detect
the exact location of cellular infiltrates along the nerve. We propose to identify the
presence of these cell types (i.e. macrophages, neutrophils, lymphocytes) with light
microscopy, and to then employ transmission electron microscopy to localize these cell

210

types to various locations on the nerve—i.e. nodes of Ranvier, the paranodal region,
mid-axon, or Schwann cells—where damage is occurring. In addition, the presence of
anti-ganglioside antibodies at the neuromuscular junction (NMJ) should be assessed
using IHC, since it has been demonstrated that anti-GM1 and anti-GQ1b antibodies
both bind at the NMJ in a complement deficient murine model (62, 160). Gangliosides
are highly concentrated at the NMJ, and this site lacks the nerve-blood barrier (157),
therefore allowing easy access for antibodies reactive to the gangliosides. To determine
which side of the NMJ the anti-gangliosides are binding to, we could stain with alpha
bungarotoxin to identify the post-synaptic side. Furthermore, to assess the binding of
anti-ganglioside antibodies to the pre-synaptic side of the NMJ, we could stain for the
soluble NSF attachment protein receptor (SNARE) proteins associated with this side of
the NMJ. By performing these stains, we can isolate which part of the NMJ is involved in
GBS in the mice. We could also perform additional stains for Schwann cell integrity,
components of the complement cascade, and for antibody binding directly to the nerve.
In addition to these approaches, we could switch to plastic embedding of the
nervous tissue in place of paraffin embedment. In our previous experiments with
paraffin embedment we were only able to get 2–3 slides per section of nerve. With
plastic embedding, thinner sections (1–2 μm) can be made, resulting in more slides per
animal. This would allow us to do the additional histology and immunohistochemistry
analyses described above. In addition to providing thinner sections, plastic embedding
also allows more accurate orientation of the nerve sections.

Potential for the NOD IL-10

-/-

mouse as a natural model of GBS

211

The NOD IL-10

-/-

mice developed enteritis and spontaneous colitis that

complicated the assessment of clinical neurological phenotypes. If work with this strain
of mouse is to continue, then a C. jejuni strain associated with GBS that colonizes
without causing enteritis in these mice would have to be identified or developed in the
laboratory. We could also try to identify an antibiotic that would kill the C. jejuni
selectively but not disrupt the gut microbiota as the CMP did in Experiment 3. Another
approach to understand what occurs in these IL10-deficient mice would be to move
away from the natural model of GBS and immunize them with purified LOS from a C.
jejuni isolate of a GBS patient. This immunization model would allow us to understand
the immune system products and effects resulting from the LOS in the absence of other
outer membrane components. However, this immunization would remove the interaction
of the bacteria with the gut microbiota and immune system that results in the
development of enteritis, and, as stated earlier, this would be then no longer a natural
model of GBS.
If a natural model is one that develops GBS secondary to infection with C. jejuni,
another option would be to investigate C57BL/6 IL-10

-/-

mice as a model for GBS.

Previous studies in our lab have shown that these mice are colonized by the GBSassociated C. jejuni strain 260.94 without developing enteric disease (99). Antiganglioside antibodies have been detected in plasma of C57BL/6 IL-10

-/-

mice infected

with GBS-associated C. jejuni (Malik and Mansfield, unpublished). However, these mice
have never been assessed for a clinical neurological phenotype or nerve lesions.

212

Preliminary experiments with C57BL/6 IL-10

-/-

mice would have to be performed to

determine the potential of these mice as a mouse model of GBS.

Examination of other factors contributing to the pathogenesis of GBS
The LOS biosynthesis locus of C. jejuni has been highly scrutinized with respect
to its role in molecular mimicry leading to the development of GBS. Strains classified as
having LOS classes A and B have been found in both GBS patients and enteric
patients; furthermore, strains having LOS loci that do not contain the necessary genes
to synthesize an LOS capable of mimicking gangliosides have been identified in GBS
patients (111). These findings suggest there are other factors contributing to the
development of GBS that need to be explored.
One such mechanism that should be explored more thoroughly to determine its
possible role in the development of GBS is the protein C-Dps. C-Dps is a DNA binding
protein produced by C. jejuni in response to nutritional or oxidative stress; it is thought
to protect the bacterial DNA from damage. Sulfatide, a glycosphingolipid, is needed for
paranodal junction formation and maintenance of ion channels along axons; C-Dps has
been shown to bind to sulfatide (114, 115). In one study by Piao et al. (2010), the
authors injected C-Dps directly into sciatic nerves of rats and found that it localized to
the outermost parts of the myelin sheath and nodes of Ranvier, resulting in myelin
detachment and axonal degeneration. These findings were confirmed by a reduction in
action potentials in nerve conduction studies of dissected tissue (115). In the same
study, the authors infused C-Dps intrathecally into rats and showed that it bound to the
nodal regions and myelin sheath (115). In another study, Piao et al. (2011) examined

213

the binding properties of C-Dps immunohistochemically and found that not only does
the protein bind to the nodes of Ranvier, the outer part of the myelin sheath, and the
basement membrane of the peripheral nerves, but it also binds to the anterior horn cells
in the spinal cord grey matter and to myelin in the white matter (114). These findings
suggest that C-Dps could play a role in the pathogenesis of GBS.
Kawamura et al. (2011) examined sera from GBS patients for the C-Dps protein
and antibodies to the protein and found that 5/27 patients had C-Dps protein in their
sera and that 15/24 patients had antibodies to the C-Dps protein in their sera (89). This
study offers further evidence that this protein, C-Dps, may play a role in pathogenesis of
GBS. Therefore, our future work with the mouse model could include detection of antiC-Dps antibodies and the protein itself in the plasma of GBS-associated C. jejuni
infected mice. To do this, we could perform ELISA on stored frozen plasma from mice in
our previous experiments as well as plasma obtained in future experiments.
Development of a C-Dps knockout strain of C. jejuni could be used to assess the role of
C-Dps in vitro and in vivo. If the C-Dps gene is found to play a role in the pathogenesis
of GBS, further experiments could be performed to determine if both C-Dps and
sialylated LOS are needed in the pathogenesis of GBS. This could be done by taking a
GBS-associated strain of C. jejuni that causes GBS in mice and making 3 knockouts: 1)
knock out only the C-Dps gene to assess its role in pathogenesis of GBS, 2) knock out
only the sialic acid related genes in the LOS (cst and neuBCA) to examine their role in
the pathogenesis of GBS, and 3) knock out both the C-Dps gene and sialic acid related
genes to determine if GBS still develops.

214

Host factors should also be assessed; they may play an important role in the
development of GBS as well, since C. jejuni with LOS classes A or B can be isolated
from patients who developed enteritis but not neurological disease. Some host factors
that have been studied include temperature, CD1 antigen presenting molecules, and
KM immunoglobulin genes. One study showed that C. jejuni grown at 37°C is more
likely to produce an LOS capable of mimicking gangliosides than C. jejuni grown at
42°C (133); the internal body temperature of a mice and humans is approximately
37°C. In another study, polymorphisms of the CD1 antigen-presenting molecules in 65
GBS patients were examined; CD1 has recently been shown to present self-glycolipids
to antigen specific T cells (21, 118). Caporale et al. (2006) examined 3 genes of the
CD1 molecule (CD1A, CD1D, and CD1E) at exon 2 where known polymorphisms exist.
They differentiated the alleles by PCR and found that subjects with a certain CD1E
allele are 2.5 times more likely to develop GBS (21). In another study of genetic
markers in the constant region of immunoglobulin kappa chains, Pandey et al. (2003)
genotyped 83 GBS patients for the KM1 and KM3 alleles using PCR-RFLP. They found
that KM3 homozygosity was significantly associated with GBS while KM1/KM3
heterozygosity was significantly associated with control patients; they did not examine
any patients with KM1 homozygosity (110). These data suggest that host genotype for
specific KM and CD1 alleles could play a role in the development of GBS.

215

APPENDICES

216

APPENDIX A

Chapter 2 Experiment 1

217

Figure A-1: Examination of enteric disease from C. jejuni infection in Experiment
1. Panel A shows the gross pathological features present at the endpoint of the
experiment. Features examined were thickened wall of the GI tract and enlarged lymph
nodes. Panel B shows the histopathological scores at the ileocecocolic junction.
A

B

218

Figure A-2: Experiment 1 ELISA assay data. Panel A shows plasma levels of IgG1
reactive to ganglioside GD1a on day 0 (A-1) and day 14 P.I. (A-2). Panel B shows the
plasma levels for IgM reactive to ganglioside GM1 on day 0 (B-1) and day 14 P.I. (B2). Panel C shows ELISA data for plasma levels of antibody reactive with ganglioside
GM1 in samples taken at the end of the experiment (56 days); antibody isotypes
IgG1, IgG2a, and IgG3 are shown in panels C-1, C-2, and C-3, respectively. Panel D
shows endpoint ELISA assay data for ganglioside GD1a with plasma levels of
antibodies IgG1, IgG2a, and IgG3 shown in panels D-1, D-2, and D-3, respectively. In
panel E, plasma levels of antibody isotype IgG1 reactive with the ganglioside mixture
GM1/GD1a at the endpoint of the experiment are shown. No significant differences in
levels of antibody were detected in treatment groups of any of the genotypes in any
of these assays.

219

Figure A-2 (cont’d)
A-1

A-2

220

Figure A-2 (cont’d)
B

B-2

221

Figure A-2 (cont’d)
C-1

C-2

C-3

222

Figure A-2 (cont’d)
D-1

D-2

D-3

223

Figure A-2 (cont’d)
E

224

Figure A-3: Phenotypic data for Experiment 1. Panel A shows the number of
quadrants crossed during the open field test for NOD IL-10-/- mice (A-1) and NOD
B7-2-/- mice (A-2). Panels B-D show stride length and hind base width
measurements. NOD WT are in panels B-1 (stride length) and B-2 (hind base width),
NOD IL-10-/- mice are in panels C-1 (stride length) and C-2 (hind base width), and
panel D are the NOD B7-2-/- mice (panel D-1, stride length; panel D-2, hind base
width). Unless stated otherwise, significance levels in all graphs are by * (p≤0.05), **
(p≤0.01), *** (p≤0.001) and **** (p≤0.0001).

225

Figure A-3 (cont’d)
A-1

A-2

226

Figure A-3 (cont’d)
B-1

B-2

227

Figure A-3 (cont’d)
C-1

C-2

228

Figure A-3 (cont’d)
D-1

D-2

229

APPENDIX B

Chapter 2 Experiment 2

230

Figure B-1: Examination of enteric disease caused by C. jejuni infection in
Experiment 2. Histopathological scores of the ileocecocolic junction taken from mice
at necropsy (80 days). No significant differences were found between uninfected
mice and infected mice in either genotype (Mann-Whitney U).

231

Figure B-2: ELISA assay data for days 0 and 14 P.I. Panel A shows data for
plasma levels of IgM antibody reactive to ganglioside GM1 at day 0 (A-1) and day 14
P.I. (A-2). Panel B indicates the IgG1 antibody plasma levels reactive to ganglioside
GD1a at day 0 (B-1) and day 14 P.I. (B-2). Levels of IgG2b reactive with C. jejuni
crude antigen are shown in panel C. Panel D shows levels of antibody isotypes IgM,
IgG1, IgG2a, and IgG3 to ganglioside GM1 (D-1, D-2, D-3, and D-4 respectively).
Panel E shows the plasma levels of antibody isotypes IgG1 and IgG2a reactive with
ganglioside GD1a (E-1 and E-2). Panel F shows the plasma levels of IgG1 antibody
to ganglioside mixture GM1/GD1a. Unless stated otherwise, significance levels in all
graphs are by * (p≤0.05), ** (p≤0.01), *** (p≤0.001) and **** (p≤0.0001).

232

Figure B-2 (cont’d)
A-1

A-2

233

Figure B-2 (cont’d)
B-1

B-2

234

Figure B-2 (cont’d)
C

D-1

235

Figure B-2 (cont’d)
D-2

D-3

D-4

236

Figure B-2 (cont’d)
E-1

E-2

F

237

Figure B-3: Experiment 2 Phenotype Data. Panel A shows the length of time mice
spent on the hang test; NOD WT (A-1) and NOD B7-2-/- (A-2). No significant
differences were found between treatment groups in either mouse genotype. Panels
B and C show footprint measurements taken for stride length and hind base width;
panels B-1 and B-2 show stride length and hind base width measurements,
respectively, for NOD WT mice. Panels C-1 and C-2 show stride length and hind base
width measurements, respectively, for NOD B7-2-/- mice.

238

Figure B-3 (cont’d)
A-1

A-2

239

Figure B-3 (cont’d)
B-1

B-2

240

Figure B-3 (cont’d)
C-1

C-2

241

Figure B-4: Presence of Nerve Lesions in Experiment 2. Histological scores of
sciatic nerve sections examined with H & E staining. No significant differences were
observed between treatment groups in either of the genotypes.

242

APPENDIX C

Chapter 2 Experiment 3

243

Figure C-1: Evidence of enteric disease in Experiment 3. Panel A shows the
colonization of C. jejuni 260.94 in the cecum at end point. Note that one shaminoculated mouse was PCR positive for C. jejuni 260.94, but produced no other
evidence of colonization with C. jejuni based on ELISA assay data, gross pathological
features, and presence of clinical neurological or enteric signs. Panel B shows the
histological scores for the ileocecocolic junction at endpoint. No significant differences
were found.

A

244

Figure C-1 (cont’d)

B

245

Figure C-2: ELISA assay data for Experiment 3. Panel A displays the plasma levels
of IgG2 antibody reactive with C. jejuni. Panels B1–B4 show the antibody levels of IgM,
IgG1, IgG2a, and IgG3, respectively, to gangliosides GM1. Panels C-1–C-3 show the
plasma levels of reactive antibody IgG1, IgG2a, and IgG3, respectively, to ganglioside
GD1a. Panel D shows the IgG2a antibody levels reactive to ganglioisde GQ1b. Unless
stated otherwise, significance levels in all graphs are by * (p≤0.05), ** (p≤0.01), ***
(p≤0.001) and **** (p≤0.0001).
A

B-1

246

Figure C-2 (cont’d)
B-2

B-3

B-4

247

Figure C-2 (cont’d)
C-1

C-2

C-3

248

Figure C-2 (cont’d).
D

249

APPENDIX D

Chapter 2 Rotarod Controls

250

Figure D-1: Rotarod Control Experiment with C57/BL6 mice. This graph shows
rotarod data for uninfected C57/BL6 WT mice and C57/BL6 IL-10-/- mice over time to
determine whether if these mice experience a drop in time spent on the rotarod. The
experiment was maintained for 84 days; these mice did display a decrease in time
spent on the rotarod as the experiment progressed.

251

REFERENCES

252

REFERENCES

1.

AVMA Guidelines for the Euthanasia of Animals, 2013 ed. American Veterinary
Medical Association, Schaumburg, IL.

2.

J. Craig Venter Institute 2012, posting date. Campylobacter Genome Project.
http://gsc.jcvi.org/projects/msc/campylobacter/index.shtml. [Online.]

3.

The Jackson Laboratory, posting date. Jax ® Mice Database.
http://jaxmice.jax.org/query/ 6.1. [Online.]

4.

Allos, B. M. 1997. Association between Campylobacter infection and GuillainBarre syndrome. J Infect Dis 176 Suppl 2:S125-8.

5.

Allos, B. M. 2001. Campylobacter jejuni Infections: update on emerging issues
and trends. Clin Infect Dis 32:1201-6.

6.

Altekruse, S. F., N. J. Stern, P. I. Fields, and D. L. Swerdlow. 1999.
Campylobacter jejuni--an emerging foodborne pathogen. Emerg Infect Dis 5:2835.

7.

Ang, C. W., B. C. Jacobs, and J. D. Laman. 2004. The Guillain-Barre
syndrome: a true case of molecular mimicry. Trends Immunol 25:61-6.

8.

Archelos, J. J., T. Fortwangler, and H. P. Hartung. 1997. Attenuation of
experimental autoimmune neuritis in the Lewis rat by treatment with an antibody
to L-selectin. Neurosci Lett 235:9-12.

9.

Asbury, A. K., and D. R. Cornblath. 1990. Assessment of current diagnostic
criteria for Guillain-Barre syndrome. Ann Neurol 27 Suppl:S21-4.

10.

Bell, J. A., J. P. Jerome, A. E. Plovanich-Jones, E. J. Smith, J. R. Gettings,
H. Y. Kim, J. R. Landgraf, T. Lefebure, J. J. Kopper, V. A. Rathinam, J. L. St
Charles, B. A. Buffa, A. P. Brooks, S. A. Poe, K. A. Eaton, M. J. Stanhope,
and L. S. Mansfield. 2013. Outcome of infection of C57BL/6 IL-10(-/-) mice with

253

Campylobacter jejuni strains is correlated with genome content of open reading
frames up- and down-regulated in vivo. Microb Pathog 54:1-19.
11.

Bell, J. A., J. L. St Charles, A. J. Murphy, V. A. Rathinam, A. E. PlovanichJones, E. L. Stanley, J. E. Wolf, J. R. Gettings, T. S. Whittam, and L. S.
Mansfield. 2009. Multiple factors interact to produce responses resembling
spectrum of human disease in Campylobacter jejuni infected C57BL/6 IL-10-/mice. BMC Microbiol 9:57.

12.

Blumstein, D. T., and J. C. Daniel. 2007. Quantifying Behavior the JWatcher
Way. Sinauer Associates, Inc., Sunderland, MA.

13.

Bour-Jordan, H., H. L. Thompson, and J. A. Bluestone. 2005. Distinct effector
mechanisms in the development of autoimmune neuropathy versus diabetes in
nonobese diabetic mice. J Immunol 175:5649-55.

14.

Brooks, S. P., and S. B. Dunnett. 2009. Tests to assess motor phenotype in
mice: a user's guide. Nat Rev Neurosci 10:519-29.

15.

Bullens, R. W., G. M. O'Hanlon, E. Wagner, P. C. Molenaar, K. Furukawa, K.
Furukawa, J. J. Plomp, and H. J. Willison. 2002. Complex gangliosides at the
neuromuscular junction are membrane receptors for autoantibodies and
botulinum neurotoxin but redundant for normal synaptic function. J Neurosci
22:6876-84.

16.

Bullens, R. W., G. M. O'Hanlon, E. Wagner, P. C. Molenaar, K. Furukawa, K.
Furukawa, J. J. Plomp, and H. J. Willison. 2003. Roles of complex
gangliosides at the neuromuscular junction. Ann N Y Acad Sci 998:401-3.

17.

Burns, T. M. 2008. Guillain-Barre syndrome. Semin Neurol 28:152-67.

18.

Busato, A., D. Hofer, T. Lentze, C. Gaillard, and A. Burnens. 1999.
Prevalence and infection risks of zoonotic enteropathogenic bacteria in Swiss
cow-calf farms. Vet Microbiol 69:251-63.

19.

Butzler, J. P. 2004. Campylobacter, from obscurity to celebrity. Clin Microbiol
Infect 10:868-76.

254

20.

Calida, D. M., S. G. Kremlev, T. Fujioka, B. Hilliard, E. Ventura, C. S.
Constantinescu, E. Lavi, and A. Rostami. 2000. Experimental allergic neuritis
in the SJL/J mouse: induction of severe and reproducible disease with bovine
peripheral nerve myelin and pertussis toxin with or without interleukin-12. J
Neuroimmunol 107:1-7.

21.

Caporale, C. M., F. Papola, M. A. Fioroni, A. Aureli, A. Giovannini, F.
Notturno, D. Adorno, V. Caporale, and A. Uncini. 2006. Susceptibility to
Guillain-Barre syndrome is associated to polymorphisms of CD1 genes. J
Neuroimmunol 177:112-8.

22.

Carter, R. J., J. Morton, and S. B. Dunnett. 2001. Motor coordination and
balance in rodents. Curr Protoc Neurosci Chapter 8:Unit 8 12.

23.

Castresana, J. 2000. Selection of conserved blocks from multiple alignments for
their use in phylogenetic analysis. Mol Biol Evol 17:540-52.

24.

Choleris, E., A. W. Thomas, M. Kavaliers, and F. S. Prato. 2001. A detailed
ethological analysis of the mouse open field test: effects of diazepam,
chlordiazepoxide and an extremely low frequency pulsed magnetic field.
Neurosci Biobehav Rev 25:235-60.

25.

Church Potter, R., and J. B. Kaneene. 2003. A descriptive study of GuillainBarre syndrome in high and low Campylobacter jejuni incidence regions of
Michigan: 1992-1999. Neuroepidemiology 22:245-8.

26.

Cingoz, F., M. Tavlasoglu, M. Kurkluoglu, and M. A. Sahin. 2012. GuillainBarre syndrome after coronary artery bypass surgery. Interact Cardiovasc
Thorac Surg 15:918-9.

27.

Clark, C. G., A. Beeston, L. Bryden, G. Wang, C. Barton, W. Cuff, M. W.
Gilmour, and L. K. Ng. 2007. Phylogenetic relationships of Campylobacter jejuni
based on porA sequences. Can J Microbiol 53:27-38.

28.

Clark, C. G., E. Taboada, C. C. Grant, C. Blakeston, F. Pollari, B. Marshall, K.
Rahn, J. Mackinnon, D. Daignault, D. Pillai, and L. K. Ng. 2012. Comparison
of molecular typing methods useful for detecting clusters of Campylobacter jejuni
and C. coli isolates through routine surveillance. J Clin Microbiol 50:798-809.

255

29.

Crawely, J. N. 2007. What's wrong with my mouse? Behavioral phenotyping of
transgenic and knockout mice, 2 ed. John Wiley & Sons.

30.

Crawley, J. N. 1999. Behavioral phenotyping of transgenic and knockout mice:
experimental design and evaluation of general health, sensory functions, motor
abilities, and specific behavioral tests. Brain Res 835:18-26.

31.

Cullen, T. W., J. P. O'Brien, D. R. Hendrixson, D. K. Giles, R. I. Hobb, S. A.
Thompson, J. S. Brodbelt, and M. S. Trent. 2013. EptC of Campylobacter
jejuni mediates phenotypes involved in host interactions and virulence. Infect
Immun 81:430-40.

32.

de Haan, C. P., R. I. Kivisto, M. Hakkinen, J. Corander, and M. L. Hanninen.
2010. Multilocus sequence types of Finnish bovine Campylobacter jejuni isolates
and their attribution to human infections. BMC Microbiol 10:200.

33.

Dereeper, A., V. Guignon, G. Blanc, S. Audic, S. Buffet, F. Chevenet, J. F.
Dufayard, S. Guindon, V. Lefort, M. Lescot, J. M. Claverie, and O. Gascuel.
2008. Phylogeny.fr: robust phylogenetic analysis for the non-specialist. Nucleic
Acids Res 36:W465-9.

34.

Dingle, K. E., F. M. Colles, D. R. Wareing, R. Ure, A. J. Fox, F. E. Bolton, H.
J. Bootsma, R. J. Willems, R. Urwin, and M. C. Maiden. 2001. Multilocus
sequence typing system for Campylobacter jejuni. J Clin Microbiol 39:14-23.

35.

Dingle, K. E., N. Van Den Braak, F. M. Colles, L. J. Price, D. L. Woodward, F.
G. Rodgers, H. P. Endtz, A. Van Belkum, and M. C. Maiden. 2001. Sequence
typing confirms that Campylobacter jejuni strains associated with Guillain-Barre
and Miller-Fisher syndromes are of diverse genetic lineage, serotype, and
flagella type. J Clin Microbiol 39:3346-9.

36.

Duim, B., P. C. Godschalk, N. van den Braak, K. E. Dingle, J. R. Dijkstra, E.
Leyde, J. van der Plas, F. M. Colles, H. P. Endtz, J. A. Wagenaar, M. C.
Maiden, and A. van Belkum. 2003. Molecular evidence for dissemination of
unique Campylobacter jejuni clones in Curacao, Netherlands Antilles. J Clin
Microbiol 41:5593-7.

37.

Duncan, R., and P. G. Kennedy. 1987. Guillain-Barre syndrome following acute
head trauma. Postgrad Med J 63:479-80.

256

38.

Dzieciatkowska, M., X. Liu, A. P. Heikema, R. S. Houliston, A. van Belkum,
E. K. Schweda, M. Gilbert, J. C. Richards, and J. Li. 2008. Rapid method for
sensitive screening of oligosaccharide epitopes in the lipooligosaccharide from
Campylobacter jejuni strains isolated from Guillain-Barre syndrome and Miller
Fisher syndrome patients. J Clin Microbiol 46:3429-36.

39.

Edgar, R. C. 2004. MUSCLE: multiple sequence alignment with high accuracy
and high throughput. Nucleic Acids Res 32:1792-7.

40.

Fearnhead, P., N. G. Smith, M. Barrigas, A. Fox, and N. French. 2005.
Analysis of recombination in Campylobacter jejuni from MLST population data. J
Mol Evol 61:333-40.

41.

Fitzgerald, C., K. Stanley, S. Andrew, and K. Jones. 2001. Use of pulsed-field
gel electrophoresis and flagellin gene typing in identifying clonal groups of
Campylobacter jejuni and Campylobacter coli in farm and clinical environments.
Appl Environ Microbiol 67:1429-36.

42.

Fox, W. M. 1965. Reflex-ontogeny and behavioural development of the mouse.
Anim Behav 13:234-41.

43.

French, N., M. Barrigas, P. Brown, P. Ribiero, N. Williams, H. Leatherbarrow,
R. Birtles, E. Bolton, P. Fearnhead, and A. Fox. 2005. Spatial epidemiology
and natural population structure of Campylobacter jejuni colonizing a farmland
ecosystem. Environ Microbiol 7:1116-26.

44.

Friedman, C. R., R. M. Hoekstra, M. Samuel, R. Marcus, J. Bender, B.
Shiferaw, S. Reddy, S. D. Ahuja, D. L. Helfrick, F. Hardnett, M. Carter, B.
Anderson, and R. V. Tauxe. 2004. Risk factors for sporadic Campylobacter
infection in the United States: A case-control study in FoodNet sites. Clin Infect
Dis 38 Suppl 3:S285-96.

45.

Fry, B. N., V. Korolik, J. A. ten Brinke, M. T. Pennings, R. Zalm, B. J. Teunis,
P. J. Coloe, and B. A. van der Zeijst. 1998. The lipopolysaccharide
biosynthesis locus of Campylobacter jejuni 81116. Microbiology 144 (Pt 8):204961.

46.

Gilbert, M., P. C. Godschalk, M. F. Karwaski, C. W. Ang, A. van Belkum, J.
Li, W. W. Wakarchuk, and H. P. Endtz. 2004. Evidence for acquisition of the
lipooligosaccharide biosynthesis locus in Campylobacter jejuni GB11, a strain

257

isolated from a patient with Guillain-Barre syndrome, by horizontal exchange.
Infect Immun 72:1162-5.
47.

Gilbert, M., M. F. Karwaski, S. Bernatchez, N. M. Young, E. Taboada, J.
Michniewicz, A. M. Cunningham, and W. W. Wakarchuk. 2002. The genetic
bases for the variation in the lipo-oligosaccharide of the mucosal pathogen,
Campylobacter jejuni. Biosynthesis of sialylated ganglioside mimics in the core
oligosaccharide. J Biol Chem 277:327-37.

48.

Gilbert, M., C. T. Parker, and A. P. Moran. 2008. Campylobacter jejuni
lipooligosaccharides: structures and biosynthesis, Campylobacter, 3 ed. ASM
Press, Washington D.C.

49.

Godschalk, P. C., A. P. Heikema, M. Gilbert, T. Komagamine, C. W. Ang, J.
Glerum, D. Brochu, J. Li, N. Yuki, B. C. Jacobs, A. van Belkum, and H. P.
Endtz. 2004. The crucial role of Campylobacter jejuni genes in anti-ganglioside
antibody induction in Guillain-Barre syndrome. J Clin Invest 114:1659-65.

50.

Godschalk, P. C., M. L. Kuijf, J. Li, F. St Michael, C. W. Ang, B. C. Jacobs, M.
F. Karwaski, D. Brochu, A. Moterassed, H. P. Endtz, A. van Belkum, and M.
Gilbert. 2007. Structural characterization of Campylobacter jejuni
lipooligosaccharide outer cores associated with Guillain-Barre and Miller Fisher
syndromes. Infect Immun 75:1245-54.

51.

Godschalk, P. C., A. van Belkum, N. van den Braak, D. van Netten, C. W.
Ang, B. C. Jacobs, M. Gilbert, and H. P. Endtz. 2007. PCR-restriction fragment
length polymorphism analysis of Campylobacter jejuni genes involved in
lipooligosaccharide biosynthesis identifies putative molecular markers for
Guillain-Barre syndrome. J Clin Microbiol 45:2316-20.

52.

Gold, R., H. P. Hartung, and K. V. Toyka. 2000. Animal models for autoimmune
demyelinating disorders of the nervous system. Mol Med Today 6:88-91.

53.

Gong, Y., Y. Tagawa, M. P. Lunn, W. Laroy, M. Heffer-Lauc, C. Y. Li, J. W.
Griffin, R. L. Schnaar, and K. A. Sheikh. 2002. Localization of major
gangliosides in the PNS: implications for immune neuropathies. Brain 125:2491506.

54.

Goodyear, C. S., G. M. O'Hanlon, J. J. Plomp, E. R. Wagner, I. Morrison, J.
Veitch, L. Cochrane, R. W. Bullens, P. C. Molenaar, J. Conner, and H. J.
Willison. 1999. Monoclonal antibodies raised against Guillain-Barre syndrome-

258

associated Campylobacter jejuni lipopolysaccharides react with neuronal
gangliosides and paralyze muscle-nerve preparations. J Clin Invest 104:697-708.
55.

Govoni, V., E. Granieri, I. Casetta, M. R. Tola, E. Paolino, E. Fainardi, and V.
C. Monetti. 1996. The incidence of Guillain-Barre syndrome in Ferrara, Italy: is
the disease really increasing? J Neurol Sci 137:62-8.

56.

Guerry, P., C. M. Szymanski, M. M. Prendergast, T. E. Hickey, C. P. Ewing,
D. L. Pattarini, and A. P. Moran. 2002. Phase variation of Campylobacter jejuni
81-176 lipooligosaccharide affects ganglioside mimicry and invasiveness in vitro.
Infect Immun 70:787-93.

57.

Habib, I., R. Louwen, M. Uyttendaele, K. Houf, O. Vandenberg, E. E.
Nieuwenhuis, W. G. Miller, A. van Belkum, and L. De Zutter. 2009.
Correlation between genotypic diversity, lipooligosaccharide gene locus class
variation, and caco-2 cell invasion potential of Campylobacter jejuni isolates from
chicken meat and humans: contribution to virulotyping. Appl Environ Microbiol
75:4277-88.

58.

Hadden, R. D., H. Karch, H. P. Hartung, J. Zielasek, B. Weissbrich, J.
Schubert, A. Weishaupt, D. R. Cornblath, A. V. Swan, R. A. Hughes, and K.
V. Toyka. 2001. Preceding infections, immune factors, and outcome in GuillainBarre syndrome. Neurology 56:758-65.

59.

Hafer-Macko, C., S. T. Hsieh, C. Y. Li, T. W. Ho, K. Sheikh, D. R. Cornblath,
G. M. McKhann, A. K. Asbury, and J. W. Griffin. 1996. Acute motor axonal
neuropathy: an antibody-mediated attack on axolemma. Ann Neurol 40:635-44.

60.

Hafer-Macko, C. E., K. A. Sheikh, C. Y. Li, T. W. Ho, D. R. Cornblath, G. M.
McKhann, A. K. Asbury, and J. W. Griffin. 1996. Immune attack on the
Schwann cell surface in acute inflammatory demyelinating polyneuropathy. Ann
Neurol 39:625-35.

61.

Hahn, A. F. 1996. Experimental allergic neuritis (EAN) as a model for the
immune-mediated demyelinating neuropathies. Rev Neurol (Paris) 152:328-32.

62.

Halstead, S. K., G. M. O'Hanlon, P. D. Humphreys, D. B. Morrison, B. P.
Morgan, A. J. Todd, J. J. Plomp, and H. J. Willison. 2004. Anti-disialoside
antibodies kill perisynaptic Schwann cells and damage motor nerve terminals via
membrane attack complex in a murine model of neuropathy. Brain 127:2109-23.

259

63.

Hardy, C. G., L. G. Lackey, J. Cannon, L. B. Price, and E. K. Silbergeld.
2011. Prevalence of potentially neuropathic Campylobacter jejuni strains on
commercial broiler chicken products. Int J Food Microbiol 145:395-9.

64.

Harvey, R. B., R. E. Droleskey, C. L. Sheffield, T. S. Edrington, T. R.
Callaway, R. C. Anderson, D. L. Drinnon, R. L. Ziprin, H. M. Scott, and D. J.
Nisbet. 2004. Campylobacter prevalence in lactating dairy cows in the United
States. J Food Prot 67:1476-9.

65.

Haubold, B., and R. R. Hudson. 2000. LIAN 3.0: detecting linkage
disequilibrium in multilocus data. Linkage Analysis. Bioinformatics 16:847-8.

66.

Hayashi, T., and D. Faustman. 1999. NOD mice are defective in proteasome
production and activation of NF-kappaB. Mol Cell Biol 19:8646-59.

67.

Ho, T. W., B. Mishu, C. Y. Li, C. Y. Gao, D. R. Cornblath, J. W. Griffin, A. K.
Asbury, M. J. Blaser, and G. M. McKhann. 1995. Guillain-Barre syndrome in
northern China. Relationship to Campylobacter jejuni infection and anti-glycolipid
antibodies. Brain 118 (Pt 3):597-605.

68.

Horrocks, S. M., R. C. Anderson, D. J. Nisbet, and S. C. Ricke. 2009.
Incidence and ecology of Campylobacter jejuni and coli in animals. Anaerobe
15:18-25.

69.

Hotter, G. S., I. H. Li, and N. P. French. 2010. Binary genomotyping using
lipooligosaccharide biosynthesis genes distinguishes between Campylobacter
jejuni isolates within poultry-associated multilocus sequence types. Epidemiol
Infect 138:992-1003.

70.

Hughes, R. A., and D. R. Cornblath. 2005. Guillain-Barre syndrome. Lancet
366:1653-66.

71.

Hughes, R. A., R. D. Hadden, N. A. Gregson, and K. J. Smith. 1999.
Pathogenesis of Guillain-Barre syndrome. J Neuroimmunol 100:74-97.

72.

Hughes, R. A., and J. H. Rees. 1997. Clinical and epidemiologic features of
Guillain-Barre syndrome. J Infect Dis 176 Suppl 2:S92-8.

260

73.

Hughes, R. A., and J. H. Rees. 1994. Guillain-Barre syndrome. Curr Opin
Neurol 7:386-92.

74.

Ilyas, A. A., H. J. Willison, R. H. Quarles, F. B. Jungalwala, D. R. Cornblath,
B. D. Trapp, D. E. Griffin, J. W. Griffin, and G. M. McKhann. 1988. Serum
antibodies to gangliosides in Guillain-Barre syndrome. Ann Neurol 23:440-7.

75.

Iovine, N. M. 2013. Resistance mechanisms in Campylobacter jejuni. Virulence
4.

76.

Islam, Z., M. Gilbert, Q. D. Mohammad, K. Klaij, J. Li, W. van Rijs, A. P. TioGillen, K. A. Talukder, H. J. Willison, A. van Belkum, H. P. Endtz, and B. C.
Jacobs. 2012. Guillain-Barre Syndrome-Related Campylobacter jejuni in
Bangladesh: Ganglioside Mimicry and Cross-Reactive Antibodies. PLoS ONE
7:e43976.

77.

Islam, Z., A. van Belkum, A. J. Cody, H. Tabor, B. C. Jacobs, K. A. Talukder,
and H. P. Endtz. 2009. Campylobacter jejuni HS:23 and Guillain-Barre
syndrome, Bangladesh. Emerg Infect Dis 15:1315-7.

78.

Islam, Z., A. van Belkum, J. A. Wagenaar, A. J. Cody, A. G. de Boer, H.
Tabor, B. C. Jacobs, K. A. Talukder, and H. P. Endtz. 2009. Comparative
genotyping of Campylobacter jejuni strains from patients with Guillain-Barre
syndrome in Bangladesh. PLoS One 4:e7257.

79.

Israeli, E., Agmon-Levin, Nancy. 2012. Guillain-Barre Syndrome-A Classical
Autoimmune Disease Triggered by Infection or Vaccination. Clinical Reviews In
Allergy and Immunology 42:121-130.

80.

Jacobs, B., A. van Belkum, and H. Endtz. 2008. Guillain-Barre syndrome and
Campylobacter Infection, p. 245-261, Campylobacter, 3 ed.

81.

Jay-Russell, M. T., R. E. Mandrell, J. Yuan, A. Bates, R. Manalac, J. MohleBoetani, A. Kimura, J. Lidgard, and W. G. Miller. Using major outer membrane
protein typing as an epidemiological tool to investigate outbreaks caused by milkborne Campylobacter jejuni isolates in California. J Clin Microbiol 51:195-201.

82.

Jin, T., L. S. Hu, M. Chang, J. Wu, B. Winblad, and J. Zhu. 2007. Proteomic
identification of potential protein markers in cerebrospinal fluid of GBS patients.
Eur J Neurol 14:563-8.

261

83.

Jolley, K. A., and M. C. Maiden. BIGSdb: Scalable analysis of bacterial genome
variation at the population level. BMC Bioinformatics 11:595.

84.

Kaida, K., T. Ariga, and R. K. Yu. 2009. Antiganglioside antibodies and their
pathophysiological effects on Guillain-Barre syndrome and related disorders--a
review. Glycobiology 19:676-92.

85.

Kaida, K., D. Morita, M. Kanzaki, K. Kamakura, K. Motoyoshi, M. Hirakawa,
and S. Kusunoki. 2007. Anti-ganglioside complex antibodies associated with
severe disability in GBS. J Neuroimmunol 182:212-8.

86.

Kaida, K., D. Morita, M. Kanzaki, K. Kamakura, K. Motoyoshi, M. Hirakawa,
and S. Kusunoki. 2004. Ganglioside complexes as new target antigens in
Guillain-Barre syndrome. Ann Neurol 56:567-71.

87.

Kalischuk, L. D., and A. G. Buret. 2010. A role for Campylobacter jejuniinduced enteritis in inflammatory bowel disease? Am J Physiol Gastrointest Liver
Physiol 298:G1-9.

88.

Kapperud, G., E. Skjerve, N. H. Bean, S. M. Ostroff, and J. Lassen. 1992.
Risk factors for sporadic Campylobacter infections: results of a case-control
study in southeastern Norway. J Clin Microbiol 30:3117-21.

89.

Kawamura, N., H. Piao, M. Minohara, T. Matsushita, S. Kusunoki, H.
Matsumoto, K. Ikenaka, Y. Mizunoe, and J. Kira. 2011. Campylobacter jejuni
DNA-binding protein from starved cells in Guillain-Barre syndrome patients. J
Neuroimmunol 240-241:74-8.

90.

Koga, M., M. Gilbert, J. Li, S. Koike, M. Takahashi, K. Furukawa, K. Hirata,
and N. Yuki. 2005. Antecedent infections in Fisher syndrome: a common
pathogenesis of molecular mimicry. Neurology 64:1605-11.

91.

Konkel, M. E., M. R. Monteville, V. Rivera-Amill, and L. A. Joens. 2001. The
pathogenesis of Campylobacter jejuni-mediated enteritis. Curr Issues Intest
Microbiol 2:55-71.

92.

Kuroki, S., T. Saida, M. Nukina, T. Haruta, M. Yoshioka, Y. Kobayashi, and
H. Nakanishi. 1993. Campylobacter jejuni strains from patients with GuillainBarre syndrome belong mostly to Penner serogroup 19 and contain beta-Nacetylglucosamine residues. Ann Neurol 33:243-7.

262

93.

Lee, K. Y. 2012. Anti-GQ1b-negative Miller Fisher syndrome after
Campylobacter jejuni enteritis. Pediatr Neurol 47:213-5.

94.

Li, C. Y., P. Xue, W. Q. Tian, R. C. Liu, and C. Yang. 1996. Experimental
Campylobacter jejuni infection in the chicken: an animal model of axonal GuillainBarre syndrome. J Neurol Neurosurg Psychiatry 61:279-84.

95.

Lin, T. M., S. S. Lee, R. T. Lin, C. S. Lai, and S. D. Lin. 2006. Guillain-Barre
syndrome following facial bone fracture. J Plast Reconstr Aesthet Surg 59:543-6.

96.

Linton, D., R. J. Owen, and J. Stanley. 1996. Rapid identification by PCR of the
genus Campylobacter and of five Campylobacter species enteropathogenic for
man and animals. Res Microbiol 147:707-18.

97.

Lo, Y. L. 2007. Clinical and immunological spectrum of the Miller Fisher
syndrome. Muscle Nerve 36:615-27.

98.

Mansfield, L. S., J. A. Bell, D. L. Wilson, A. J. Murphy, H. M. Elsheikha, V. A.
Rathinam, B. R. Fierro, J. E. Linz, and V. B. Young. 2007. C57BL/6 and
congenic interleukin-10-deficient mice can serve as models of Campylobacter
jejuni colonization and enteritis. Infect Immun 75:1099-115.

99.

Mansfield, L. S., J. S. Patterson, B. R. Fierro, A. J. Murphy, V. A. Rathinam,
J. J. Kopper, N. I. Barbu, T. J. Onifade, and J. A. Bell. 2008. Genetic
background of IL-10(-/-) mice alters host-pathogen interactions with
Campylobacter jejuni and influences disease phenotype. Microb Pathog 45:24157.

100.

Markoula, S., S. Giannopoulos, I. Sarmas, S. Tzavidi, A. P. Kyritsis, and G.
Lagos. 2007. Guillain-Barre syndrome in northwest Greece. Acta Neurol Scand
115:167-73.

101.

Meinersmann, R. J., L. O. Helsel, P. I. Fields, and K. L. Hiett. 1997.
Discrimination of Campylobacter jejuni isolates by fla gene sequencing. J Clin
Microbiol 35:2810-4.

102.

Meinersmann, R. J., R. W. Phillips, K. L. Hiett, and P. Fedorka-Cray. 2005.
Differentiation of campylobacter populations as demonstrated by flagellin short
variable region sequences. Appl Environ Microbiol 71:6368-74.

263

103.

Moran, A. P., H. Annuk, and M. M. Prendergast. 2005. Antibodies induced by
ganglioside-mimicking Campylobacter jejuni lipooligosaccharides recognise
epitopes at the nodes of Ranvier. J Neuroimmunol 165:179-85.

104.

Murgatroyd, C., D. Bilko, and D. Spengler. 2006. Isolation of high-quality DNA
for genotyping from feces of rodents. Anal Biochem 348:160-2.

105.

Nachamkin, I., B. M. Allos, and T. Ho. 1998. Campylobacter species and
Guillain-Barre syndrome. Clin Microbiol Rev 11:555-67.

106.

Nye, T. M., P. Lio, and W. R. Gilks. 2006. A novel algorithm and web-based tool
for comparing two alternative phylogenetic trees. Bioinformatics 22:117-9.

107.

O'Hanlon, G. M., G. J. Paterson, J. Veitch, G. Wilson, and H. J. Willison.
1998. Mapping immunoreactive epitopes in the human peripheral nervous
system using human monoclonal anti-GM1 ganglioside antibodies. Acta
Neuropathol 95:605-16.

108.

Oh, S. J., C. LaGanke, and G. C. Claussen. 2001. Sensory Guillain-Barre
syndrome. Neurology 56:82-6.

109.

Overell, J. R., and H. J. Willison. 2005. Recent developments in Miller Fisher
syndrome and related disorders. Curr Opin Neurol 18:562-6.

110.

Pandey, J. P., and C. A. Vedeler. 2003. Immunoglobulin KM genes in GuillainBarre syndrome. Neurogenetics 4:147-9.

111.

Parker, C. T., M. Gilbert, N. Yuki, H. P. Endtz, and R. E. Mandrell. 2008.
Characterization of lipooligosaccharide-biosynthetic loci of Campylobacter jejuni
reveals new lipooligosaccharide classes: evidence of mosaic organizations. J
Bacteriol 190:5681-9.

112.

Parker, C. T., S. T. Horn, M. Gilbert, W. G. Miller, D. L. Woodward, and R. E.
Mandrell. 2005. Comparison of Campylobacter jejuni lipooligosaccharide
biosynthesis loci from a variety of sources. J Clin Microbiol 43:2771-81.

113.

Phongsisay, V., V. N. Perera, and B. N. Fry. 2006. Exchange of
lipooligosaccharide synthesis genes creates potential Guillain-Barre syndromeinducible strains of Campylobacter jejuni. Infect Immun 74:1368-72.

264

114.

Piao, H., M. Minohara, N. Kawamura, W. Li, T. Matsushita, R. Yamasaki, Y.
Mizunoe, and J. Kira. 2011. Tissue binding patterns and in vitro effects of
Campylobacter jejuni DNA-binding protein from starved cells. Neurochem Res
36:58-66.

115.

Piao, H., M. Minohara, N. Kawamura, W. Li, Y. Mizunoe, F. Umehara, Y.
Goto, S. Kusunoki, T. Matsushita, K. Ikenaka, T. Maejima, J. Nabekura, R.
Yamasaki, and J. Kira. 2010. Induction of paranodal myelin detachment and
sodium channel loss in vivo by Campylobacter jejuni DNA-binding protein from
starved cells (C-Dps) in myelinated nerve fibers. J Neurol Sci 288:54-62.

116.

Pithadia, A. B., and N. Kakadia. 2010. Guillain-Barre syndrome (GBS).
Pharmacol Rep 62:220-32.

117.

Plomp, J. J., P. C. Molenaar, G. M. O'Hanlon, B. C. Jacobs, J. Veitch, M. R.
Daha, P. A. van Doorn, F. G. van der Meche, A. Vincent, B. P. Morgan, and
H. J. Willison. 1999. Miller Fisher anti-GQ1b antibodies: alpha-latrotoxin-like
effects on motor end plates. Ann Neurol 45:189-99.

118.

Porcelli, S. A., and R. L. Modlin. 1999. The CD1 system: antigen-presenting
molecules for T cell recognition of lipids and glycolipids. Annu Rev Immunol
17:297-329.

119.

Poropatich, K. O., C. L. Walker, and R. E. Black. 2010. Quantifying the
association between Campylobacter infection and Guillain-Barre syndrome: a
systematic review. J Health Popul Nutr 28:545-52.

120.

Prendergast, M. M., A. J. Lastovica, and A. P. Moran. 1998.
Lipopolysaccharides from Campylobacter jejuni O:41 strains associated with
Guillain-Barre syndrome exhibit mimicry of GM1 ganglioside. Infect Immun
66:3649-55.

121.

Prendergast, M. M., D. R. Tribble, S. Baqar, D. A. Scott, J. A. Ferris, R. I.
Walker, and A. P. Moran. 2004. In vivo phase variation and serologic response
to lipooligosaccharide of Campylobacter jejuni in experimental human infection.
Infect Immun 72:916-22.

122.

Price, L. B., A. Roess, J. P. Graham, S. Baqar, R. Vailes, K. A. Sheikh, and E.
Silbergeld. 2007. Neurologic symptoms and neuropathologic antibodies in
poultry workers exposed to Campylobacter jejuni. J Occup Environ Med 49:74855.

265

123.

Quinones, B., M. R. Guilhabert, W. G. Miller, R. E. Mandrell, A. J. Lastovica,
and C. T. Parker. 2008. Comparative genomic analysis of clinical strains of
Campylobacter jejuni from South Africa. PLoS One 3:e2015.

124.

Revez, J., and M. L. Hanninen. 2012. Lipooligosaccharide locus classes are
associated with certain Campylobacter jejuni multilocus sequence types. Eur J
Clin Microbiol Infect Dis 31:2203-9.

125.

Richards, V. P., T. Lefebure, P. D. Pavinski Bitar, and M. J. Stanhope. 2013.
Comparative characterization of the virulence gene clusters (lipooligosaccharide
[LOS] and capsular polysaccharide [CPS]) for Campylobacter coli,
Campylobacter jejuni subsp. jejuni and related Campylobacter species. Infect
Genet Evol 14:200-13.

126.

Rostami, A. M. 1997. P2-reactive T cells in inflammatory demyelination of the
peripheral nerve. J Infect Dis 176 Suppl 2:S160-3.

127.

Sahin, O., T. Y. Morishita, and Q. Zhang. 2002. Campylobacter colonization in
poultry: sources of infection and modes of transmission. Anim Health Res Rev
3:95-105.

128.

Salloway, S., L. A. Mermel, M. Seamans, G. O. Aspinall, J. E. Nam Shin, L. A.
Kurjanczyk, and J. L. Penner. 1996. Miller-Fisher syndrome associated with
Campylobacter jejuni bearing lipopolysaccharide molecules that mimic human
ganglioside GD3. Infect Immun 64:2945-9.

129.

Salmon, D. A., M. Proschan, R. Forshee, P. Gargiullo, W. Bleser, D. R.
Burwen, F. Cunningham, P. Garman, S. K. Greene, G. M. Lee, C. Vellozzi, W.
K. Yih, B. Gellin, and N. Lurie. 2013. Association between Guillain-Barre
syndrome and influenza A (H1N1) 2009 monovalent inactivated vaccines in the
USA: a meta-analysis. Lancet.

130.

Salomon, B., L. Rhee, H. Bour-Jordan, H. Hsin, A. Montag, B. Soliven, J.
Arcella, A. M. Girvin, J. Padilla, S. D. Miller, and J. A. Bluestone. 2001.
Development of spontaneous autoimmune peripheral polyneuropathy in B7-2deficient NOD mice. J Exp Med 194:677-84.

131.

Sanad, Y. M., G. Closs, Jr., A. Kumar, J. T. Lejeune, and G. Rajashekara.
2013. Molecular epidemiology and public health relevance of campylobacter
isolated from dairy cattle and European starlings in ohio, USA. Foodborne
Pathog Dis 10:229-36.

266

132.

Sellon, R. K., S. Tonkonogy, M. Schultz, L. A. Dieleman, W. Grenther, E.
Balish, D. M. Rennick, and R. B. Sartor. 1998. Resident enteric bacteria are
necessary for development of spontaneous colitis and immune system activation
in interleukin-10-deficient mice. Infect Immun 66:5224-31.

133.

Semchenko, E. A., C. J. Day, J. C. Wilson, I. D. Grice, A. P. Moran, and V.
Korolik. 2010. Temperature-dependent phenotypic variation of Campylobacter
jejuni lipooligosaccharides. BMC Microbiol 10:305.

134.

Shahrizaila, N., and N. Yuki. 2011. Antiganglioside antibodies in Guillain-Barre
syndrome and its related conditions. Expert Rev Neurother 11:1305-13.

135.

Shahrizaila, N., and N. Yuki. 2010. Guillain-barre syndrome animal model: the
first proof of molecular mimicry in human autoimmune disorder. J Biomed
Biotechnol 2011:829129.

136.

Shui, I. M., M. D. Rett, E. Weintraub, M. Marcy, A. A. Amato, S. I. Sheikh, D.
Ho, G. M. Lee, and W. K. Yih. 2012. Guillain-Barre syndrome incidence in a
large United States cohort (2000-2009). Neuroepidemiology 39:109-15.

137.

Snyder, L. A., V. Rismondo, and N. R. Miller. 2009. The Fisher variant of
Guillain-Barre syndrome (Fisher syndrome). J Neuroophthalmol 29:312-24.

138.

Suerbaum, S., M. Lohrengel, A. Sonnevend, F. Ruberg, and M. Kist. 2001.
Allelic diversity and recombination in Campylobacter jejuni. J Bacteriol 183:25539.

139.

Takahashi, M., M. Koga, K. Yokoyama, and N. Yuki. 2005. Epidemiology of
Campylobacter jejuni isolated from patients with Guillain-Barre and Fisher
syndromes in Japan. J Clin Microbiol 43:335-9.

140.

Tam, C. C., L. C. Rodrigues, and S. J. O'Brien. 2003. Guillain-Barre syndrome
associated with Campylobacter jejuni infection in England, 2000-2001. Clin Infect
Dis 37:307-10.

141.

Tam, C. C., L. C. Rodrigues, I. Petersen, A. Islam, A. Hayward, and S. J.
O'Brien. 2006. Incidence of Guillain-Barre syndrome among patients with
Campylobacter infection: a general practice research database study. J Infect Dis
194:95-7.

267

142.

Tan, I. L., T. Ng, and S. Vucic. 2010. Severe Guillain-Barre syndrome following
head trauma. J Clin Neurosci 17:1452-4.

143.

Tribble, D. R., S. Baqar, M. P. Carmolli, C. Porter, K. K. Pierce, K. Sadigh, P.
Guerry, C. J. Larsson, D. Rockabrand, C. H. Ventone, F. Poly, C. E. Lyon, S.
Dakdouk, A. Fingar, T. Gilliland, P. Daunais, E. Jones, S. Rymarchyk, C.
Huston, M. Darsley, and B. D. Kirkpatrick. 2009. Campylobacter jejuni strain
CG8421: a refined model for the study of Campylobacteriosis and evaluation of
Campylobacter vaccines in human subjects. Clin Infect Dis 49:1512-9.

144.

Tsang, R. S., G. Figueroa, L. Bryden, and L. Ng. 2001. Flagella as a potential
marker for Campylobacter jejuni strains associated with Guillain-Barre syndrome.
J Clin Microbiol 39:762-4.

145.

Van Koningsveld, R., P. I. Schmitz, C. W. Ang, J. Groen, A. D. Osterhaus, F.
G. Van der Meche, and P. A. Van Doorn. 2002. Infections and course of
disease in mild forms of Guillain-Barre syndrome. Neurology 58:610-4.

146.

Vegosen, L., M. F. Davis, E. Silbergeld, P. N. Breysse, J. Agnew, G. Gray, L.
B. Freeman, and F. Kamel. 2012. Neurologic symptoms associated with cattle
farming in the agricultural health study. J Occup Environ Med 54:1253-8.

147.

Vucic, S., M. C. Kiernan, and D. R. Cornblath. 2009. Guillain-Barre syndrome:
an update. J Clin Neurosci 16:733-41.

148.

Walling, A. D., and G. Dickson. 2013. Guillain-barre syndrome. Am Fam
Physician 87:191-7.

149.

Walsh, F. S., M. Cronin, S. Koblar, P. Doherty, J. Winer, A. Leon, and R. A.
Hughes. 1991. Association between glycoconjugate antibodies and
Campylobacter infection in patients with Guillain-Barre syndrome. J
Neuroimmunol 34:43-51.

150.

Willison, H. J. 2005. The immunobiology of Guillain-Barre syndromes. J
Peripher Nerv Syst 10:94-112.

151.

Willison, H. J., and G. M. O'Hanlon. 1999. The immunopathogenesis of Miller
Fisher syndrome. J Neuroimmunol 100:3-12.

268

152.

Willison, H. J., and J. J. Plomp. 2008. Anti-ganglioside antibodies and the
presynaptic motor nerve terminal. Ann N Y Acad Sci 1132:114-23.

153.

Winer, J. B. 2011. Guillain-Barre syndrome: clinical variants and their
pathogenesis. J Neuroimmunol 231:70-2.

154.

Winer, J. B. 2001. Guillain Barre syndrome. Mol Pathol 54:381-5.

155.

Yazdanbakhsh, M., P. G. Kremsner, and R. van Ree. 2002. Allergy, parasites,
and the hygiene hypothesis. Science 296:490-4.

156.

Young, K. T., L. M. Davis, and V. J. Dirita. 2007. Campylobacter jejuni:
molecular biology and pathogenesis. Nat Rev Microbiol 5:665-79.

157.

Yu, R. K., T. Ariga, S. Usuki, and K. Kaida. 2011. Pathological roles of
ganglioside mimicry in Guillain-Barre syndrome and related neuropathies. Adv
Exp Med Biol 705:349-65.

158.

Yuki, N. 2005. Carbohydrate mimicry: a new paradigm of autoimmune diseases.
Curr Opin Immunol 17:577-82.

159.

Yuki, N. 2007. Ganglioside mimicry and peripheral nerve disease. Muscle Nerve
35:691-711.

160.

Yuki, N. 2010. Human gangliosides and bacterial lipo-oligosaccharides in the
development of autoimmune neuropathies. Methods Mol Biol 600:51-65.

161.

Yuki, N. 2001. Infectious origins of, and molecular mimicry in, Guillain-Barre and
Fisher syndromes. Lancet Infect Dis 1:29-37.

162.

Yuki, N. 1997. Molecular mimicry between gangliosides and lipopolysaccharides
of Campylobacter jejuni isolated from patients with Guillain-Barre syndrome and
Miller Fisher syndrome. J Infect Dis 176 Suppl 2:S150-3.

163.

Yuki, N., and H. P. Hartung. 2012. Guillain-Barre syndrome. N Engl J Med
366:2294-304.

269

164.

Yuki, N., and S. Kuwabara. 2007. Axonal Guillain-Barre syndrome:
carbohydrate mimicry and pathophysiology. J Peripher Nerv Syst 12:238-49.

165.

Yuki, N., K. Susuki, M. Koga, Y. Nishimoto, M. Odaka, K. Hirata, K. Taguchi,
T. Miyatake, K. Furukawa, T. Kobata, and M. Yamada. 2004. Carbohydrate
mimicry between human ganglioside GM1 and Campylobacter jejuni
lipooligosaccharide causes Guillain-Barre syndrome. Proc Natl Acad Sci U S A
101:11404-9.

166.

Yuki, N., T. Taki, F. Inagaki, T. Kasama, M. Takahashi, K. Saito, S. Handa,
and T. Miyatake. 1993. A bacterium lipopolysaccharide that elicits Guillain-Barre
syndrome has a GM1 ganglioside-like structure. J Exp Med 178:1771-5.

167.

Yuki, N., M. Yamada, M. Koga, M. Odaka, K. Susuki, Y. Tagawa, S. Ueda, T.
Kasama, A. Ohnishi, S. Hayashi, H. Takahashi, M. Kamijo, and K. Hirata.
2001. Animal model of axonal Guillain-Barre syndrome induced by sensitization
with GM1 ganglioside. Ann Neurol 49:712-20.

168.

Yuki, N., H. Yoshino, S. Sato, and T. Miyatake. 1990. Acute axonal
polyneuropathy associated with anti-GM1 antibodies following Campylobacter
enteritis. Neurology 40:1900-2.

169.

Zhang, M., L. He, Q. Li, H. Sun, Y. Gu, Y. You, F. Meng, and J. Zhang. 2010.
Genomic characterization of the Guillain-Barre syndrome-associated
Campylobacter jejuni ICDCCJ07001 Isolate. PLoS One 5:e15060.

270