UNIVERSITY OF CINCINNATI
Date: February 22, 2007
I, _ Samuel Lee Hayes______, hereby submit this work as part of the requirements for the degree of: Doctor of Philosophy in: Biological Sciences It is entitled: Response of Mammalian Models to Exposure of Bacteria from the Genus Aeromonas Evaluated using Transcriptional Analysis and Conjectures on Disease Mechanisms
This work and its defense approved by:
Chair: _Brian K. Kinkle
_Dennis W. Grogan
_Richard D. Karp
_Mario Medvedovic
_Stephen J. Vesper
Response of Mammalian Models to Exposure of Bacteria from the Genus Aeromonas Evaluated using Transcriptional Analysis and Conjectures on Disease Mechanisms
A dissertation submitted to the
Division of Graduate Studies and Research of the University of Cincinnati
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
in the Department of Biological Sciences of the College of Arts and Sciences
2007
by
Samuel Lee Hayes
B.S. Ohio University, 1978 M.S. University of Cincinnati, 1986
Committee Chair: Dr. Brian K. Kinkle Abstract
The genus Aeromonas contains virulent bacteria implicated in waterborne disease, as well
as avirulent strains. One of my research objectives was to identify and characterize host-
pathogen relationships specific to Aeromonas spp. Aeromonas virulence was assessed using
changes in host mRNA expression after infecting cell cultures and live animals. Messenger
RNA extracts were hybridized to murine genomic microarrays. Initially, these model systems
were infected with two virulent A. hydrophila strains, causing up-regulation of over 200 and 50
genes in animal and cell culture tissues, respectively. Twenty-six genes were common between
the two model systems.
The live animal model was used to define virulence for many Aeromonas spp. Strains
that demonstrated mortality and produced an average up-regulation of > 3-fold, at challenged
doses of 107-108 CFU, were considered virulent. Mortality results correlated well with dose and
transcript up-regulation.
Cell cultures were infected with representative virulent and avirulent Aeromonas strains.
Transcriptional response from live animal and cell culture models were compared to find
common transcripts unique to virulent infections. Two genes with potential for predicting
virulence (Jun and Fos) were identified. Confirmation testing (qRT-PCR) indicated that the Jun
oncogene is potentially predictive of Aeromonas virulence using cell culture.
Aeromonas caviae is associated with gastrointestinal disease but lacks obvious virulence
factors (VFs). Microarray profiling of mouse intestinal extracts after A. caviae infection
demonstrated a Th1 immune response characterized by gamma-interferon induced genes. This
suggests A. caviae causes a dysregulatory cytokine response leading to an irritable bowel-like syndrome.
To evaluate loss of single VFs, isogenic mutants were produced using a transposable element. Mutations in VFs associated with lateral flagella, O-antigen and secretion systems were created. Swarming motility, associated with intestinal colonization, was eliminated in a lateral flagella mutant, but subsequent colonization testing was inconclusive. Murine monolayers demonstrated no difference in gene expression after infection with lateral flagella mutant and wild type organisms.
As a result of this research, I suggest that Aeromonas has at least two pathways to virulence. The first functions by activation of multiple VFs to pathway(s) thru Jun. The second, as demonstrated by some A. caviae strains, is activated through gamma interferon, producing a chronic infection.
ACKNOWLEDGMENTS
Many individuals were instrumental in helping me complete this dissertation and associated research. I owe the following people an enormous amount of gratitude.
First, many thanks to my advisor, Dr. Brian Kinkle for his advice and guidance, especially for keeping me focused on the science. My EPA contact, Dr. Steve Vesper, a very passionate and driven scientist, infused his enthusiasm for this project into me and was very inspirational. I want to also thank the other members of my committee, Dr. Dennis Grogan, Dr.
Richard Karp and Dr. Mario Medvedovic for sticking with me over the past four years and for their input and expertise.
I have also had the support and advice of a group of scientists (our Aeromonas research group), with whom I have been meeting with almost weekly over this same time period. Dr.
Dennis Lye, Dr. Maura Donohue, Dr. Mark Rodgers and Dr. Gerard Stelma have been incredibly supportive and have offered invaluable advice to me regarding research direction. I am very grateful to this group and look forward to a continuing collaboration on similar research efforts.
The following people have provided technical assistance to me during various phases of the research, Ms. Bethany Lye, Ms. Shannon Murphy, Mr. Brent Bertke, Ms. Karen White and
Ms. Cathy Kelty. Many of these marvelous professionals have moved on to bigger and better endeavors, though I have the pleasure of still having some as colleagues. Thanks for their assistance over the years.
As she always has, the most important support I received was from my wife, Kim. The biggest lesson I have learned from her is that with the right support, anything is possible. She has been, and always will be, my rock. Table of Contents
Table of Contents ...... i List of Tables ...... iii List of Figures...... iv LITERATURE REVIEW ...... 1 1.0 Aeromonas, Taxonomy and General Information ...... 1 1.1 Aeromonas Occurrence/Epidemiology/Significance ...... 4 1.2 Known Aeromonas Virulence Factors (VFs)...... 12 1.2.1 Enterotoxins ...... 13 1.2.1.1 Aerolysin...... 13 1.2.1.2 Act ...... 14 1.2.1.3 Ast and Alt ...... 15 1.2.2 Hemolysins ...... 16 1.2.3 Lipase ...... 17 1.2.4 Adhesion Factors...... 17 1.2.5 Metallo- and serine proteases ...... 20 1.2.6 Type III Secretion Systems ...... 21 1.2.7 S-Layer ...... 21 1.2.8 Plasmid associated virulence factor...... 22 1.3 Additional Research of VFs and Aeromonas Virulence ...... 22 1.4 DNA chip technology (Affymetrix) ...... 25 1.4.1 Gene Chip Preparation and Technology Description...... 26 1.4.2 Gene Chip Data Analysis ...... 28 1.5 Bacterial-host interaction research using microarray technology...... 31 1.5.1 Research of host bacterial interactions, epithelial cell models ...... 33 1.6 Animal models for assessing Aeromonas virulence...... 36 1.7 References ...... 41 Chapter 1, Identification by microarray of a common pattern of gene expression in intact intestine and cultured intestinal cells exposed to virulent Aeromonas hydrophila isolates...60 Abstract ...... 61 INTRODUCTION...... 62 MATERIALS and METODS...... 63 RESULTS and DISCUSSION ...... 67 CONCLUSIONS ...... 70 REFERENCES...... 72 Chapter 2, Evaluating virulence of waterborne and clinical Aeromonas isolates using gene expression and mortality in neonatal mice followed by assessing cell culture's ability to predict virulence based on transcriptional response ...... 81 Abstract ...... 82 INTRODUCTION...... 83 MATERIALS and METHODS...... 84 RESULTS ...... 89 DISCUSSION ...... 92 ACKNOWLEDGEMENT ...... 96 REFERENCES...... 97 i Chapter 3, Proposal for disease presentation of Aeromonas caviae: irritable bowel-like syndrome ...... 110 Abstract ...... 111 INTRODUCTION...... 112 MATERIALS and METHODS...... 113 RESULTS ...... 115 DISCUSSION ...... 115 REFERENCES...... 119 Chapter 4, Creation of isogenic strains of a virulent Aeromonas hydrophila strain by transposon insertion with subsequent evaluation of effects on pathogenic characteristics124 Abstract ...... 125 INTRODUCTION...... 126 MATERIALS and METHODS...... 126 RESULTS ...... 130 DISCUSSION ...... 134 REFERENCES...... 139 SUMMARY...... 149
ii List of Tables
Page
Literature Review
Table 1. Phenospecies and associated genospecies 3
Table 2. Human pathogens and disease states 4
Table 3. Mouse GeneChip sequence summary 28
Chapter 1
Table 1. Numbers of genes up- and downregulated from A. hydrophila infection 78
Table 2. Common genes up-regulated and downregulated in cell culture and neonates 79
Chapter 2
Table 1. Aeromonas isolates used in this study; their source and identification number 103
Table 2. Number of significantly up-regulated (>2 fold) genes in the intestinal tissue
of neonatal mice after exposure to Aeromonas isolates 104
Table 3. Summary of probe set up-regulation of listed genes in neonatal mouse model
from Aeromonas infection 105
Table 4. Summary of Aeromonas infections and intestinal murine cell line 109
Chapter 3
Table 1. Up-regulated transcripts after 24 hours, neonatal mouse exposure 122
Table 2. Results of confirmation testing by qRT-PCR 123
Chapter 4
Table 1. Listing of mutants created to date 142
iii List of Figures
Page
Literature Review
Figure 1. Affymetrix GeneChip analysis summary 25
Figure 2. GeneChip photolithography technique 26
Chapter 4
Figure 1. Gel images of PCRs indicating transposon insertions and PCR products to
be sequenced 147
Figure 2. Swarming assay, lateral flagella mutant versus wild-type 148
iv LITERATURE REVIEW
1.0 Aeromonas, Taxonomy and General Information
Listed below is the current classification of the genus Aeromonas as presented in
Bergey's Manual of Systemic Bacteriology (Martin-Carnahan and Joseph 2005).
Phylum: Proteobacteria
Class: Gammaproteobacteria
Order: Aeromonadales
Family: Aeromonadaceae
Genus: Aeromonas
The genus Aeromonas is comprised of gram-negative, facultatively anaerobic rods that measure
0.3 to 1.0 µm x 1.0 to 3.5 µm. and are of the γ-3 subgroup of Proteobacteria. Members of the genus are generally motile but some species are non-motile. Members of the genus are chemoorganotrophic and live in aquatic environments. Some species are pathogenic to humans and other warm-blooded mammals as well as in fish, eels, amphibians and leeches. Aeromonads are found in source waters used to produce drinking water and are also found in sewage and biofilms (e.g., on drinking water distribution system pipes).
In the 1984 edition of Bergey's Manual of Systemic Bacteriology, the genus was included
in the family Vibrionaceae. Colwell et al. (1986) provided evidence from genetic analyses (i.e.,
16S rRNA cataloguing, 5S rRNA gene sequence comparisons and rRNA-DNA hybridizations) that aeromonads needed to be separated into their own family. Analysis showed approximate equal divergence from Vibrionaceae and Enterobacteriaceae. Additional research using genetic 1 analyses of type strains confirmed Colwell's determination. (Borrell et al. 1997, Figueras et al.
2000).
The taxonomy of genus Aeromonas is confusing. The species are classified in terms of genospecies and phenospecies. In 1988, five species were recognized (three were phenospecies based upon biochemical testing). The genus is also characterized by mesophilic and psychrophilic species. The mesophilic species include those bacteria that infect mammals while the psychrophilic species are primarily fish pathogens. In the early 1980s, the mesophilic species were categorized within three phenotypic groups, designated as A. hydrophila, A. caviae, and A. sobria. The fish pathogens were named A. salmonicida. Abbot et al. (2003) published a comprehensive classification of the genus based on biochemical testing that identified 14 species. They evaluated 60 biochemical tests and 400 strains. Only four of those tests exhibited uniform reactions and there were many atypical reactions.
Aeromonas spp. were originally classified based on phenotypic expressions of carbohydrate metabolism or specific protein production. To date, 14 well-defined and 3 newly described genomic species (representing 17 DNA-DNA hybridization groups [HGs]) are recognized within the genus (Table 1) (Huys et al. 1997, Janda and Abbott 1998). Currently, the genus is grouped into 17 genospecies based on HGs. Using the type strain from each HG, new isolates can be categorized. Two additional genomic species, A. ichthiosmia and A. enteropelogenes, have been proposed. The three phenotypic groups are no longer used in the identification of Aeromonas spp., but these designations exist in the older literature.
2 Table 1. Phenospecies and associated genospecies
Phenotypic Group (Phenospecies) Hybridization Group Species A. hydrophila 1 A. hydrophila 2 A. bestiarum 3 A. salmonicida A. caviae 4 A. caviae 5 A. media 6 A. eucrenophila A. sobria 7 A. sobria 8 A. veronii biotype sobria 9 A. jandaei 10 A. veronii biotype veronii 11 A. encheleia 12 A. schubertii 13 Aeromonas Group 501 14 A. trota 15 A. allosaccharophila 16 A. enchelia 17 A. popoffi
Aeromonads are not recognized as part of the normal human gut flora, and researchers estimate that in developed countries, less than 1 percent of people harbor Aeromonas in their gastrointestinal tract (Janda 1999; Smith and Cheasty, 1998). Five species have been established as human pathogens thus far and are listed in Table 2. Three of these species, A. hydrophila, A. caviae, and A. veronii (biotype sobria) are considered major pathogens and likely to be clinically significant. Two identified species, A. jandaei and A. schubertii, have been established as potential pathogens through isolation from extra-intestinal (wound) infections, but very little information on occurrence and pathogenicity is available for them (Janda and Kokka 1991).
Janda and Kokka also found one A. salmonicida (HG3) and one A. veronii biotype veronii
(HG10) to exhibit virulence using an intra-peritoneal (i.p.) LD50 assay with 8-10 week old Swiss
Webster mice.
3 Table 2. Human pathogens and disease states
Species Hybridization Group Human Disease Association A. hydrophila 1 Gastroenteritis, skin/soft tissue infections, bacteremia, meningitis, peritonitis, respiratory tract infections, ocular infections, hemolytic, uremic syndrome
A. caviae 4 Gastroenteritis, skin/soft tissue infections, bacteremia, peritonitis
A. veronii 8 Gastroenteritis, skin/soft tissue infections, biotype sobria bacteremia, peritonitis
A. media 5 Gastroenteritis
A. sobria 7 Gastroenteritis, skin/soft tissue infections, bacteremia, peritonitis
1.1 Aeromonas Occurrence/Epidemiology/Significance
The prevalence of Aeromonas infection varies as infection occurs in both children and adults, in symptomatic and asymptomatic people, and in developed and developing countries.
The predominant site of isolation from healthy humans is the gastrointestinal tract, where aeromonads are present in low frequency. In humans with Aeromonas-associated gastrointestinal disease, the frequency of isolation was highest among children (under 5 years of age) and older adults (over 60 years of age) (USEPA, 2000).
Aeromonas spp. are found in surface water, groundwater, marine and estuarine environments, and even in chlorinated water supplies. They have been shown to contribute to biofilms (Gavriel et al. 1998), and they experience regrowth in drinking water distribution systems (Gavriel et al. 1998; Smith and Cheasty, 1998). Aeromonas hydrophila has been detected in hospital water supplies (Picard and Goullet, 1987). Environmental occurrence varies by season, with increases seen in warmer summer months (Burke et al. 1984; Gavriel et al. 1998;
4 Smith and Cheasty, 1998). Infection rates are correlated with peak rates of occurrence in the
environment. Two studies recovered more aeromonads in surface water than in groundwater
(Burke et al. 1984; Legnani et al. 1998).
The only documented human feeding study for Aeromonas was performed by Morgan et
al. (1985). Five strains of Aeromonas that were each positive for cytotoxicity, hemolysin,
DNase, lysine decarboxylase and acetylmethylcarbinol were used. Doses of 104-1010 organisms were fed to healthy human test subjects. Only two strains caused illness, 6Y and 3647. For strain 6Y, one person fed a dose of 109 organisms developed mild diarrhea, although 11 of 20
volunteers had Aeromonas recovered from stool samples. Three of sixteen individuals fed strain
3647 (107 organisms) developed mild diarrhea. The other three strains were never recovered
from stool samples nor caused any symptoms.
Transmission is generally limited to sporadic, individual infection from ingestion of food
or water, or from exposure of skin and mucus to water. Person-to-person transmission of
Aeromonas is rare, but has been reported (Cahill 1990; Janda and Abbott 1998). No single
point-source outbreaks have been confirmed (Janda and Abbott 1998). In addition, the hospital
environment is a reservoir of Aeromonas species and can represent a source of infection
(Altwegg and Geiss 1989). Most infections of healthy individuals involve an injury exposed to
an environmental source, usually recreational or occupational water exposure.
The one human study suggests that a very large dose (>108-1010 CFUs) of particular
strains is required for infection (Morgan et al. 1985). Patients with a serious underlying
condition may be more susceptible to infection via smaller doses. Alternatively, compromised
patients may also be infected at the same rate but suffer more frequent and serious illness.
Aeromonas virulence and the host immune response are not well characterized. However,
5 potential virulence factors (VFs), such as adhesions, enterotoxins, hemolysins, and protease are produced in some form by all three major pathogenic species.
In one study of 80 infected adult patients, symptoms of Aeromonas-related gastroenteritis ranged from acute, self-limited diarrhea to chronic, indolent diarrhea, with 16 percent confirmed cases of colitis (George et al. 1985). A study of 36 patients of all ages showed that adults tended to suffer chronic symptoms, while children’s illnesses were more severe and acute (Holmberg et al. 1986). The clinical course of infection in children seems to occur most frequently in the youngest children, a few years old or less. Several investigators have described cases of children whose illness progressed to chronic diarrhea, dysentery-like illness, and dehydration (Gluskin et al. 1992; Gracey et al. 1982; San Joaquin and Pickett 1988). Vomiting and/or fever sometimes accompanied the diarrhea. Bloody stools or fecal leukocytes are occasionally detected
(Ashdown and Koehler 1993; Challapalli et al. 1988; Gluskin et al. 1992). Severe complications, such as bacteremia are rarely reported, but do occur (San Joaquin and Pickett
1988). Cases of hemolytic uremic syndrome, most recently been associated with E. coli
O157:H7, have also been associated with Aeromonas (Bogdanovic et al. 1991; Robson et al.
1992; San Joaquin and Pickett 1988). In children, Aeromonas gastrointestinal disease rarely progresses to hemolytic uremic syndrome (Robson et al. 1992; Janda and Abbott 1998).
Aeromonas caviae has been isolated most frequently in children with gastroenteritis.
However, A. caviae isolates associated with illness do not always exhibit typical virulence properties like cytotoxin production (Janda 1999; Havelaar et al. 1992; San Joaquin and Pickett
1988; Smith and Cheasty 1998) and at times their virulence properties are not substantially different from the virulence properties of other isolates (Challapalli et al. 1988; Figura et al.
1986). Challapalli et al. (1988) did not correlate different isolates with levels of disease severity,
6 but A. caviae has been associated with more serious pathology in children, such as chronic diarrhea (San Joaquin and Pickett 1988).
There is controversy regarding the origin of Aeromonas gastrointestinal infections.
Borchardt et al. (2003) found no relationship between water isolates and stool samples using
DNA based assays (pulsed field gel electrophoresis). Borchardt's team analyzed 2,565 stool samples from diarrheal patients. Water samples were taken from private wells throughout
Wisconsin and only 0.7% were positive for Aeromonas. Moyer et al. (1992) proved aeromonads: 1) are often found in well waters, 2) colonize treatment systems prior to chlorination, 3) do not easily colonize distribution systems with residual chlorine levels, and 4) isolates from treatment systems are not the same strains found in epidemiological studies of populations served.
Moyer (1987) also performed a two year epidemiological in Iowa where 3,334 fecal samples were analyzed. Of these, 238 were positive for Aeromonas spp., representing 214 patients. One conclusions reached from this study was that A. caviae was common in children aged 0 - 5 years with mixed infections common in this group. Aeromonas hydrophila and A. veronii biotype sobria were associated with acute infections and A. caviae more associated with chronic infections. Most infections could be linked to patients who were on antibiotic therapy, drank or were exposed to untreated drinking water, or had eaten shellfish.
In another U.S. study, King et al. (1992) compiled data on Aeromonas occurrence when
California made Aeromonas a reportable condition. Over a one year period, 280 cases of
Aeromonas infection were reported with 218 cases investigated. This equated to 10.6 cases per 1 million individuals. Eighty-one percent were cases of gastroenteritis, 9% were wound infections and 2% died of complications due to the Aeromonas infection (in these five cases, there were
7 other underlying conditions). The conclusions of this study were that Aeromonas was not a large health care concern and was largely non-preventable. Also, they conjectured that Aeromonas infections were likely under reported. Young and old patients were most affected, but those in the middle-aged classification might have been affected as well, without reporting any disease due to mild symptoms. The authors did conclude that Aeromonas is a human pathogen.
Similar to the above California report, Kuijper et al. (1989) saw a pattern of age
distribution from a five year study of Aeromonas infections in the Netherlands. Aeromonas was
found in 208 of 34, 311 fecal samples analyzed (0.61%). Nineteen percent of the Aeromonas
isolations were associated with mixed infections, 5% were associated with other underlying
diseases and 15% were found in patients pre-disposed to intestinal disease (e.g., on antibiotic
therapy). Aeromonas caviae strains isolated did not produce cytotoxins, whereas strains from
HG groups 1 and 8 did. The non-cytotoxic A. caviae strains were predominant in children and in
adults over the age of 50.
Legnani et al. (1998) performed a three year Aeromonas occurrence study in an Italian
mountain area and analyzed springs, wells, rivers and reservoirs. They found no correlation with
temperature but conceded that water temperatures in this area were consistently cold. They also
found no correlation to fecal indicator organisms. Twenty two percent of all samples collected
were positive for Aeromonas with concentrations ranging from 1 - 240 CFU/100 mL. Seventy-
two percent of the isolates were A. hydrophila, 15 % were A. caviae and 13% were A. veronii
biotype sobria. The percent recovery of Aeromonas from reservoirs was approximately equal to
the percentage found in distribution systems waters, indicating there was no regrowth in the
distribution system.
8 Contrary to the above, presence of Aeromonas in drinking water is thought by some to be primarily a distribution system issue. Water distribution systems are conjectured to become colonized with Aeromonas over long time periods, as water leaving treatment is typically negative for the presence of Aeromonas spp. Chauret et al. (2001) assessed an Indiana (USA) drinking water distribution system for the presence of Aeromonas. Aeromonas counts were made: 1) of source water, 2) within the treatment plant and 3) within the distribution system
(both total bulk water and biofilm samples). Annular reactors set up in a laboratory were also tested. The source water contained 103-104 CFU/100 mL of Aeromonas, with the majority of these isolates (75%) identified as A. hydrophila. Within the treatment plant, concentrations of
Aeromonas ranged from below the detection limit to 490 CFU/100 mL, with apparent colonization of the granular activated carbon (GAC) unit (there is no residual disinfectant within the GAC). The treatment plant effluent samples were all negative (i.e., below the detection limit) for Aeromonas. Distribution system biofilms were assessed using water meter replacement operations and 7.7% of biofilm samples were positive for Aeromonas. Chloroamine was used to provide the residual disinfection in the distribution system. These authors saw seasonal variation in source water samples. Seeding studies using the annular reactors failed to produce Aeromonas populations in these test systems. The conclusion was that insufficient testing time and high residual chlorine concentrations were the cause of the lack of Aeromonas colonization in the annular reactors.
In Spain, Borrell et al. (1998) assessed a variety of matrices for the presence of
Aeromonas. The types and numbers of samples analyzed were: surface waters (218), drinking water (148), food (439) and human feces (212). Aeromonads were speciated to the genotype level by phenotypic assays. The drinking water and food samples had low levels of detection
9 (<10% for drinking water), with the exception of shellfish where 31.1% of the samples were
positive for Aeromonas. Reservoirs and rivers samples were positive 95 and 88% of time,
respectively, with concentrations ranging from 103-104 CFU/100 mL. Aeromonas veronii biotype sobria was the most frequent isolate from clinical (32%) and environmental (22%) samples. Aeromonas caviae was the second most frequent isolate in food and stools, with A. hydrophila the second most common isolate in lakes and reservoirs. Atypical isolations of A. bestiarum were not uncommon in fecal samples. In another assessment of Spanish waters,
Figueras et al. (2005) also found similar levels of aeromonads as described above as well as atypical Aeromonas isolates. This group identified three A. culicicola isolates (an extremely rare species not yet listed in Bergey's and previously only found in mosquitoes in India) each harboring a gene (act) for a known enterotoxin (act).
Burke et al. (1984) found a seasonal correlation in Aeromonas incidence in Perth,
Australia. In drinking water samples, this group found a large number of Aeromonas spp. even though these same samples meet international standards for E. coli. Aeromonas spp. were also found in distribution system samples when E. coli was either very low or not detected. The largest number of detections were found in the summer months. Patients with Aeromonas- associated gastroenteritis appeared to follow the distribution of organism detection in drinking water. Aeromonas detections did not correlate well with indicator bacteria populations.
Albert et al. (2000) evaluated 1,735 children with diarrhea in Bangladesh. Aeromonads were found in 7.2% of the cases with 28 having aeromonads as the only isolate. Mixed infections were therefore common. Aeromonas hydrophila, A. veronii biotype sobria and A.
caviae accounted for 85% of the clinical isolates (A caviae > A. veronii biotype sobria > A.
10 hydrophila). Aeromonas eucrenophila was an atypical species isolated from one case of diarrhea.
In a Sydney, Australia survey, Ashbolt et al. (1995) examined wastewater treatment plant
(WTP) primary and tertiary effluents, storm water, freshwater and ocean water for the presence of aeromonads. They found levels of Aeromonas spp. to be as numerous as total coliforms in each matrix. Aeromonads were found in higher concentrations in samples taken downstream of the WTP as compared to upstream samples. Aeromonas hydrophila was the most frequent species identified.
DeMarta et al. (2000) looked at the epidemiological relationships between Aeromonas isolates from children in Switzerland and from environmental samples. They assessed approximately 30 cases and attempted to define the route of transmission as well as trying to determine why children are predisposed to infection. The study evaluated 133 total Aeromonas isolates. Once a case of Aeromonas related diarrhea was found, extensive sampling was performed of tap water, wet surfaces and family members' stools. Ribotyping was performed on the isolates. Conclusions of this study were that: 1) there was no correlation between ribotypes found in the stool samples and environmental isolates, 2) person to person transmission was not a significant factor, and 3) A. caviae was the most common species isolated, followed next by A. hydrophila.
Sinha et al. (2004) performed a two year study (2000-2001) in Kolkata, India. A total of
1,648 and 1,853 stool samples were analyzed in years 2000 and 2001, respectively. In 2000
6.5% of the stools were positive for Aeromonas spp. and 3.1% were positive in 2001. A seasonal trend was noted in 2000, but not in 2001. Of the total samples positive for Aeromonas, 58% had only Aeromonas isolated and 42% were mixed infections. Aeromonas caviae was the most
11 common isolate (45%), followed by A. hydrophila (25%) and A. veronii biotype sobria (15%).
This group analyzed the Aeromonas isolates for Act, Alt and Ast enterotoxins. Aeromonas
caviae isolates were rarely positive for the presence of Act or Ast enterotoxins, or for
hemolysins. Most A. caviae were positive for the Alt enterotoxin (see later discussion regarding
this toxin). Aeromonas veronii biotype sobria and A. hydrophila were characterized by common
production of both hemolysins and Act (pore forming toxin).
1.2 Known Aeromonas Virulence Factors (VFs)
Known Aeromonas VFs include a variety of toxins, hemolysins, adhesion factors, proteases and lipases. The cytotoxic pore-forming enterotoxins, Act and aerolysin, are perhaps the most well-known and characterized of the Aeromonas toxins. Other known enterotoxins are
Ast (heat stabile toxin) and Alt (heat labile toxin). These two toxins are classified as being
cytotonic rather than cytotoxic. A number of cytotoxic hemolysins are also common among
Aeromonas spp.
The virulence factors of Aeromonas are largely associated with chromosomal DNA.
Brown et al. (1997) found no evidence that plasmids play a role in Aeromonas virulence. This
group found that only 16% of aeromonads contain plasmids. Plasmids appeared to be more
frequent in environmental than in clinical isolated (at least for A. veronii biotype sobria). This
group verified the two putative VFs, aerolysin and Type IV pili, as being chromosomal. This
same publication did note a report of a shiga toxin plasmid being found in Aeromonas.
12
1.2.1 Enterotoxins
1.2.1.1 Aerolysin
As mentioned above, aerolysin is a major VF of Aeromonas spp. Aerolysin (AerA, pore
forming toxin) is a cytotoxin similar to the cholera toxin HlyA (Wong et al. 1998). Aerolysin is
a hydrophilic non-enzymatic channel forming protein. It is secreted as a dimer that polymerizes
into heptamers when inserted into host cell membranes (Parker et al. 1996). The formed pores
allow ions to pass but not proteins. Chakraborty et al. (1986) identified and sequenced aerA as
the gene producing the 54 kDa aerolysin protein. This gene has both upstream and downstream
modulators, aerB and aerC, respectively. This group incorporated a 5.8 kilobase EcoRI fragment
into E. coli K-12 to produce hemolytic and cytotoxic effects. The actual size of the aerA, aerB,
aerC complex was later determined to be 1.4 kilobases in length. There are three forms of the
native protein. The first form's folding and assembly occurs in the periplasma. This first form is
modified before being excreted into the extracellular matrix by a type II secretion system. This
excreted protoxin is modified by serine proteases to cleave forty amino acids from the C-terminal
end to produce the active toxin. Trypsin (digestive tract enzyme) and furin can activate the
protoxin (Fivaz et al. 2001). AerA binds to glycophosphatidylinositol (GPI) anchored proteins
of eukaryotic cells which, in turn, activate G proteins, induce cytokine production, and cause
vacuolization of the endoplasmic reticulum (ER) (Fivaz et al. 2001). Ion fluxes result, causing
either Ca2+ entry or the depletion of cytoplasmic components. Takahashi et al. (2005) isolated a
toxin they called aerolysin from A. sobria (likely A. veronii biotype sobria) and compared the
protein to aerolysin from A. hydrophila. They reported a 68.5% homology at the amino acid level. Their aerolysin reportedly increased chloride ion secretion because of increased
13 intracellular calcium ion levels.
1.2.1.2 Act
An important aerolysin-related toxin harbored by Aeromonas spp is Act. There is some controversy regarding whether aerolysin and Act are actually different proteins (see Letter to
Editor, Infection and Immunity, 67:466-467). The toxin was fully characterized by Chopra et al.
(1993a). Chopra provided the following for argument for differentiating Act from aerolysin; 1) difference in neutralization with specific antibodies; 2) difference in amino acid sequence in one portion of peptide chain; and 3) the inability of Act to bind to aerolysin receptors (see Ferguson et al. 1997, below). This research group discovered that Act induces cytokine production and activates arachidonic acid (AA) metabolism in murine macrophages (Chopra et al. 2000). The
AA pathway plays a role in fluid secretion in animals. They demonstrated that Act induces tumor necrosis factor alpha (TNFα) as well as interleukin 1β (IL-1β) and IL-6 in murine macrophages. Galindo et al. (2003, 2004, 2005, also part of Chopra's laboratory) published a series of articles regarding host gene response to Act exposure. Cell lines used were murine macrophages and human HT-29 colon epithelial cells. Microarray technology was used to evaluate host gene response. Transcripts up-regulated included genes associated with cell growth, adhesion, cell signaling, immune response and apoptosis. This group also reported the phosphorylation of JNK and MAPK kinases, indicating activation of immune response and apoptotic pathways. In murine macrophages, c-Jun mRNA was strongly up-regulated within two hours. This protein is a major subunit of transcription factors for many immune and apoptosis pathways. Also, transcription factor inhibitors (Ikappaβ) to NF-кB were up-regulated in the murine macrophage cell line demonstrating the activation of feedback loops associated with
14 regulation of cytokine production. Using the SSU strain of Aeromonas, gene knockouts of act, ast and alt were prepared by Sha et al. (2002, also part of Chopra's laboratory group) using suicide vectors. Of these three toxins, Act produced the most fluid secretion in a ligated mouse ileal loop model as compared to Ast and Alt. Xu et al. (1998, also from Chopra's laboratory) demonstrated an increase of the LD50 of act knockout mutants (SSU) when injected intraperitoneally into mice. The lethality was restored when a plasmid containing the act gene
was replaced in the mutant.
Ferguson et al. (1997) studied the pore forming mechanisms of Act by a number of techniques including electron microscope imaging and saccharide size exclusion. They also found that the Act toxin was neutralized by cholesterol. The 3'-OH group of cholesterol interacts
5 8 with Act. LD50 levels were raised from 10 organisms to 10 organisms when testing was
performed using isogenic act mutants of Aeromonas strains. This group found Act to be
processed at the N-terminal end to produce the protoxin and at the C-terminal end to get the
mature toxin and also demonstrated that different receptors interact with Act and aerolysin at the
host cell surface.
1.2.1.3 Ast and Alt
Ast and Alt are cytotonic enterotoxins. Chopra et al. (1993b) isolated both a heat stabile
and a heat labile enterotoxin both causing elevation of cyclic adenosine monophosphate (cAMP)
in Chinese hamster ovary (CHO) cells. Sha et al. (2002) demonstrated that Ast caused
elongation of CHO cells as well as causing fluid secretion in ligated mouse ileal loops.
A membrane bound toxin, Alt, was sequenced by Chopra et al. (1996) and was identified
as a 38kDa protein with 45-51% homology to phospholipase C. This toxin produced fluid
15 response in rat ileal loop assays. Alt also increased intracellular levels of PGE2. Merino et al.
(1999) reported on phospholipase A1 and identified the product of this pla gene as Alt. Alt was shown to be cytotonic rather than cytotoxic by Sha et al. (2002). In this research, Alt caused elongation of CHO cells as well as causing fluid secretion in ligated mouse ileal loops.
Aeromonas strains with both Ast and Alt appear to cause watery diarrhea, whereas bloody diarrhea is associated with the added presence of the Act toxin (Chopra et al. 1996 and Albert et al. 2000). Albert et al. (2000) speculated that Alt and Ast act synergistically to cause diarrhea.
1.2.2 Hemolysins
A known β-hemolysin in Aeromonas spp. is the product of the hlyA gene. HlyA is similar to cholera hemolysin. Epple et al. (2004) did an extensive investigation of Aeromonas- related beta hemolysins. After performing a search of GeneBank, they discovered 10 different aerolysin-related β-hemolysins with 30-90% homology to each other. Using electrophysiological techniques, they determined that an A. hydrophila β-hemolysin induced chloride ion secretion from colon epithelial cells (HT-29 cell line). They measured a decrease in transepithelial resistance and an increase of short circuit current of the HT-29 monolayers when infected with a β-hemolysin positive A. hydrophila strain and no changes when infected with a β- hemolysin negative strain of A. veronii. The β-hemolysin gene was cloned and inserted into a β- hemolysin negative strain and obtained similar results as with the wild-type testing above.
A GeneBank entry for an A. salmonicida β-hemolysin was found in an A. veronii biotype sobria isolate. The A. salmonicida β-hemolysin was identified earlier by Hirono et al. (1992).
This group identified four different associated hemolysins from various Aeromonas spp. They also found some Aeromonas isolates that were β-hemolytic on BAP, but negative for the PCR
16 primers that identified their four specific hemolysins. Therefore, there were additional β- hemolysins to the four for which they had PCR primers.
1.2.3 Lipase
Lipase activity is another known VF. Two studies report on either phospholipases or extracellular lipase and how they might affect virulence. Eliminating phospholipase C
(lecithinase, reported as being nearly identical to a 65 kDa hemolysin [ASH1] from A. salmonicida, 99.1% amino acid overlap) activity from A. hydrophila increased the LD50
(approximately 10X increase) in rainbow trout and mice. Knocking out phospholipase A1 (Alt, lipase activity on tributyrin medium) had no apparent affect on virulence (Merino et al. 1999).
The lecithinase protein described by Merino et al. was non-hemolytic, but was cytotoxic, leading them to speculate that the A. salmonicida protein was misidentified as a hemolysin.
A 67 kDa purified lipase was identified as an extracellullar product from an of A. hydrophila strain (Anguita et al. 1993). Extracellular lipase is reported to interact with human leukocytes and also affect several immune system functions. Free fatty acids are generated by the lipase activity and interfere with chemotactic peptides as well as neutrophil function. The amino acid sequence of an extracellular lipase was characterized by Anquita et al. (1993). No virulence testing was performed with regards to this lipase.
1.2.4 Adhesion Factors
The ability to adhere is a putative virulence factor. The presence of pili, especially bundle forming pili (Bfp) have been implicated. However, Bartkova and Cižnár (1994) reported on the adherence of a non-piliated strain of A. hydrophila to tissue culture. They speculated that
17 flagella, outer membrane proteins, LPS or cell surface hydrophobicity were possible explanations for adherence. Grey and Kirov (1993) investigated the adherence of Aeromonas spp. to a Hep-2 cell line. They found A veronii biotype sobria were the most adherent (58%), followed by clinical isolates of A. caviae (33%), with A. hydrophila (11%) being the least adherent. Environmental isolates of A. caviae were less adherent than clinical isolates. There was no difference found between clinical and environmental isolates of A. veronii biotype sobria and A. hydrophila in terms of adherence. The percent adherence was based on enumerating bacteria still attached to the Hep-2 cells after incubation, fixation and staining. In contrast,
Merino et al. (1996) reported that an O:34 serotype A. hydrophila adhered well to Hep-2 cell culture and adherence was enhanced at 20oC over 37oC.
Kirov et al. (2004) assessed adherence of aeromonads to Henle 407 and Caco-2 cell lines
(both cancerous human intestinal epithelial cells). Aeromonads are known to possess both polar and lateral flagella. Polar flagella are associated with swimming in liquid matrices while lateral flagella are associated with swarming on surfaces. This group found that 60% of mesophilic aeromonads exhibit lateral flagella. Polar and lateral flagella A. caviae mutants were tested for adherence to the above cell lines. Those mutants with no polar flagella exhibited no adherence and the lateral flagella mutants showed 60% less adherence to the cell lines. This same group had investigated the importance of type IV pili (Tap) and Bfp on Aeromonas virulence (Kirov et al. 2000). Tap mutants of Aeromonas strains A. hydrophila and A. veronii biotype sobria were prepared by inactivating the tapA gene which encodes the type IV pilus subunit protein, TapA.
Cytotoxic activities were unaffected by the mutation in tapA, and bacterial adherence was also unaffected for these two mutant strains with Hep-2 cells. For the A. veronii biotype sobria isolate, adhesion to Henle 407 intestinal cells was also unaffected. There was no significant
18 effect on the duration of colonization by the A. veronii biotype sobria mutant strain when tested in the removable intestinal tie adult rabbit diarrhea (RITARD) model. This study suggested that
Tap pili may not be as significant as Bfp pili for Aeromonas intestinal colonization.
Nishikawa et al. (1994) tested four A. hydrophila and two A. sobria (likely veronii biotype sobria) strains for their ability to adhere to Caco-2 cells. Their conclusion was that the mechanism of adherence was not similar to the attachment and effacement (A/E) mechanism of particular E. coli strains. The Aeromonas strains did not react with probes for the eae (A/E) or ipaB (invasin) genes of E. coli. The strains were also negative for a fluorescent stain procedure for actin reorganization associated with A/E by E. coli.
Merino et al. (2003) reported the presence of an insertion sequence (IS3) into the lafA gene in A. salmonicida. Therefore, no lateral flagella were produced in this species. They concluded that this may be the reason A. salmonicida was a specialized pathogen with a narrow host range. A. salmonicida is known to possess a number of other putative VFs including hemolysins and aerolysin-like toxins.
Non-fimbrial adhesion associated with A. caviae was demonstrated by Rocha-de-Souza et. al. (2001). A 43 kDa outer membrane protein (OMP) was isolated by this group from an A. caviae culture. The isolated protein was bonded to glass beads and these beads exhibited adhesion to Hep-2 cells. An antibody to the OMP blocked this adhesion. This antibody also blocked adhesion of A. caviae to Hep-2 cells. This group also reported on the presence of a 43 kDa OMP hemagglutinin in A. hydrophila that binds to mammalian cell H-antigens.
19 1.2.5 Metallo- and serine proteases
Metallo- and serine proteases (e.g. elastase) are known virulence factors for Aeromonas
spp. and are associated with cell-to-cell spread of pathogens by processing and degrading
phospholipids, peptides, immunoglobulins and hormones. Elastase is a metalloprotease that
degrades elastin, a component of the extracellular matrix. Proteases also break down cellular
matrices to provide nutrients for proliferating organisms. Song et al. (2004) identified and
characterized a number of Aeromonas proteases from various species. From A. hydrophila, this
group identified a 38 kDa metalloprotease with elastase activity (produced by the gene ahyB), a
19kDA zinc protease and a 68 kDa temperature sensitive serine protease. From A. caviae, they identified a 34 kDa metalloprotease (different from the A. hydrophila 38 kDa protease because of no elastase activity) and also a 19 kDa protease. A 34 kDa protease from an A. veronii biotype sobria strain activated aerolysin from proaerolysin. Various mutants prepared using suicide vectors were less hemolytic and cytotoxic by up to a factor of 8. Cascón et al. (2000) demonstrated a higher LD50 in rainbow trout when elastase was eliminated from an A.
hydrophila strain (isogenic mutant for ahyB gene). Hasan et al. (1992) tested numerous
Aeromonas spp. for elastase activity using bi-layer plates. Virtually all A. hydrophila strains
were positive, with A. veronii biotype sobria, caviae, jandaei, trota, Group 501 and schubertii
being largely negative. Esteve and Birkbeck (2004) determined that a serine protease (caseinase)
and a metalloprotease (elastase) purified from A. hydrophila were cytotoxic.
20
1.2.6 Type III Secretion Systems
Type III secretion systems (TTSSs) are common virulence factors among gram negative
organisms. Chacón et al. (2004) identified TTSS system genes in Aeromonas as part of a
pathogenicity island. Also identified by this group was the ADP ribosylating protein (AexT
toxin) excreted by the Aeromonas TTSS. Fifty percent of tested Aeromonas spp. were positive
for TTSS genes ascF-ascG, ascV and aexT, with the majority of these being either A. hydrophila
or A. veronii biotype sobria. All A. caviae strains tested were negative for the TTSS. Vilches et al. (2004) identified the complete Type III secretion system genes of A. hydrophila and made
probes to perform dot blots on other clinical and environmental A. hydrophila strains and
Aeromonas species. Thirty five genes were identified in this chromosomal cluster. Complete
TTSSs were found in: 28/35 clinical and 4/25 environmental A. hydrophila isolates, 32/40
clinical and environmental 9/20 A. veronii biotype sobria isolates, 5/40 clinical and 4/20
environmental A. caviae isolates, and 1/2 clinical A. jandaei isolates. Mutation of an essential
assembly gene demonstrated that TTSS is required for virulence. TTSS genes were located on a
plasmid in A. salmonicida.
1.2.7 S-Layer
Another potential VF associated with Aeromonas spp. is the presence of an S-layer.
Dooley et al. (1988) characterized the Aeromonas S-layer as a paracrystaline, single species
protein array, with each subunit an approximately 52 kDa protein. The Aeromonas S-layer had
no homology to S-layers from other bacterial species. This group speculated that the S-layer
21 functions as a physical barrier to lytic agents (e.g., serum proteins and bacteriophages). S-layers
from A. salmonicida were found to have binding sites for immunoglobulins. Janda et al. (1994)
investigated the susceptibility of Aeromonas with S-layers to complement lysis. This research
found all Aeromonas of serotype O:11 were S-layer positive. Serotype O:11 strains that were
susceptible to complement lysis had a one log higher LD50 than those that were not susceptible.
They concluded that S-layer presence is not always predictive of virulence, but complement lysis
was predictive.
1.2.8 Plasmid associated virulence factor
Haque et al. (1996) found some aeromonads (3 strains of A. hydrophila and 1 strain of A.
caviae) with plasmids encoding a shiga like toxin (SLT). The SLT produced by these strains
was neutralized with an E. coli SLT antitoxin. PCR primers for E. coli SLT1 amplified a product
from the Aeromonas plasmid DNA preparations. The A. hydrophila strains were the sole
potential pathogens isolated from patient stool samples. The A. caviae isolate was from a
wastewater sample.
1.3 Additional Research of VFs and Aeromonas Virulence
Heuzenroeder et al (1999) assessed a number of Aeromonas spp. for the presence of aerA
and hlyA and the association of their presence with virulence. For A. caviae isolates, 41% were
aerA+ and 35% were hlyA+. However, some A. caviae that were hlyA+ were not hemolytic on
blood agar plates. Aeromonas hydrophila strains that were aerA+ and hlyA+ were all virulent.
Some A. veronii biotype sobria strains that were aerA+ and hlyA- were also virulent.
22 Chacón et al. (2003) assessed the distribution of virulence factors in clinical versus
environmental isolates. This effort genotyped 234 Aeromonas isolates (including 14 ATCC type
strains) using the 16rDNA-RFLP method of Borrell (1997). These strains were assayed for the
presence of aerolysin, serine protease, GCAT, lipase and DNase. Aerolysin was found in 72.6%
of the total strains tested and was more frequent in clinical versus environmental isolates. DNase
and beta-hemolytic activity were the only other two VFs found more commonly in clinical over
environmental isolates. The serine protease was positively correlated with the presence of
aerolysin and the authors concluded that this may be due to the need of the bacteria to convert
pro-aerolysin to the active form.
Yu et al. (2005) used suppressive subtractive hybridization to identify virulent genes in
A. hydrophila. This group used blue gourami fish as the test system and measured the LD50 using insertion and deletion mutants. An approximate 1-log reduction in the LD50 was found
when knocking out the Type III secretion system (TTSS), hlyA or aerA genes. A triple knockout of serine and metalloproteases also caused a 1-log10 reduction in the LD50. A single knockout of
genes associated with the S-layer had no affect on the LD50.
Kingombe (1999) used one primer set to identify both act and aerA in Aeromonas strains.
This was possible due to the nucleotide homology between the two genes. Not only did the
primer set detect the presence of these two genes, but also identified eight other virulence genes
contained in GenBank (one being a β-hemolysin). Testing of human isolates from Switzerland
showed all A. caviae to be negative for PCR products using this primer set. However, 25% of A. caviae from Bangladesh isolates exhibited VF genes with this same set of primers. Verification was made that these Bangladesh isolates did belong to HG4.
Aeromonas caviae are generally considered to be non-hemolytic. However, some
23 research efforts did find exceptions. Imzlin et al. (1998) found 27 of 100 A. caviae isolates tested to be hemolytic. These hemolytic A. caviae also exhibited cytotoxic and cytotonic abilities. This same group proposed a "potential virulence index" (PVI) based on hemolytic ability, cytotoxic or cytotonic effects and adhesion. This PVI would be to used score Aeromonas spp. and was an attempt to quantify virulence. In a large study in Tasmania, Australia, Kirov et al. (1986) did not find any A. caviae isolates that were positive for hemolysin or enterotoxin production. Barer et al. (1986) also found no cytotoxic strains of A. caviae when assessing 95 strains for this activity as well as for haemolysins and protease activity.
While aerolysin and β-hemolysins are major VFs for Aeromonas, Gonzales-Serrano et al.
(2002) demonstrated suckling mouse enterotoxicity with an A. veronii biotype sobria strain that was AerA- and HlyA-. Knocking out the two genes for AerA and HlyA production reduced A.
hydrophila virulence, but did not eliminate it in a neonatal mouse assay (Wong et al. 1998).
Apparently, a sufficient number of other virulence factors were present in ample quantity to
cause host damage.
Alavandi and Ananthan (2003) found Aeromonas serogroups appeared to segregate
depending on whether isolates were of clinical or environmental origin. However, they found no
difference in virulence when the same serogroups of clinical and environmental isolates were
tested in relation to suckling mouse toxicity or Vero cell cytotoxicity.
24 1.4 DNA chip technology (Affymetrix)
Thousands of genes and their products (i.e., RNA and proteins) in a given living organism function in an orchestrated manner to determine the outcome of a stimulus. However, traditional methods in molecular biology generally work on a "one gene in one experiment" basis, which means that the throughput is very limited and the "whole picture" of gene function is hard to obtain. DNA microarray techniques monitor the whole genome on a single chip so that researchers can have a better picture of the interactions among thousands of genes simultaneously. The process begins with the extraction of total RNA from tissues or cells.
The procedure to hybridize extracted RNA to the microarray is summarized in Figure 1.
The hybridized probe array is stained with streptavidin phycoerythrin conjugate and scanned by the GeneArray Scanner at the excitation wavelength of 488 nm. The amount of light emitted at
570 nm is proportional to the bound target at each location on the probe array.
Figure 1. Reproduced from Affymetrix website www.affymetrix.com
25
1.4.1 Gene Chip Preparation and Technology Description
Affymetrix GeneChip probe arrays are manufactured using technology that combines photolithography and combinational chemistry. Synthetic linkers that are modified with photochemically removable protecting groups are attached to a glass substrate. Light is directed through a photolithographic mask to specific areas on the surface to produce localized photodeprotection. The first of a series of chemical building blocks, hydroxyl-protected deoxynucleosides, is incubated with the surface, and chemical coupling occurs at those sites that have been illuminated in the preceding step. Next, light is directed to different regions of the substrate by a new mask, and the chemical cycle is repeated until oligonucleotides of around 20- mer length are created (see Figure 2).
Figure 2. Reproduced from Lipshutz et al. 1999.
The number of probe sets created on an array is limited only by the physical size of the array and the achievable lithographic resolution. Current large scale commercial manufacturing
26 methods allow for approximately 300,000 probe sets to be located on a 1.28 X 1.28 cm array.
Each 20-25-mer oligo is located in a specific area on the array called a probe cell. Each probe cell contains millions of copies of a given oligonucleotide sequence or probe.
Oligonucleotide arrays for murine expression monitoring are designed and synthesized based on sequence information from various databases (UniGene, dbEST, GenBank, RefSeq) and the draft assembly of the mouse genome (Whitehead Institute Center for Genome Research).
Sets of probes are designed to specifically monitor the expression levels of as many genes as possible. The cDNA or extended sequence tag (EST) sequences are used to develop independent
20-25-mer oligonucleotides to serve as sequence-specific detectors. The use of multiple oligonucleotides of different sequence is designed to hybridize to different regions of the same
RNA. Eleven probe sets are typically used to represent a cDNA or EST sequence. According to
Affymetrix, "the use of multiple independent detectors for the same molecule greatly improves signal-to-noise ratios (due to averaging over the intensities of multiple array features), improves the accuracy of RNA quantitation (averaging and outlier rejection), increases the dynamic range, mitigates effects due to cross-hybridization, and drastically reduces the rate of false positives and miscalls." (Lipshutz et al. 1999). Also built into the array is an additional level of redundancy using mismatch (MM) control probes that are identical to their perfect match (PM) partners except for a single base difference in a central position. The MM probes act as specificity controls that allow the direct subtraction of both background and cross-hybridization noise for a given gene or EST. (Lipshutz et al. 1999).
Table 3 summarizes the sequences used on the mouse GeneChips (Series 430) used for this project. Each sequence is represented by 11 probe sets and there are also a number of
27 control probe sets on each array. As mentioned above, the approximately 300,000 probe sets per
array provides full coverage of the mouse genome.
Table 3. Mouse GeneChip sequence summary
Classification Mouse Set 430
Full lengths 14,484
Non-ESTs (excluding full lengths) 9,450
ESTs 21,103
Total number of sequences 45,037
EST=extended sequence tags
1.4.2 Gene Chip Data Analysis
Comparing exposed and unexposed animals or cell cultures using Gene Chip technology
creates approximately 39,000 comparisons simultaneously. If a conventional t test on each comparison is used at a significance level of p = 0.01, approximately 390 genes would be identified as having significantly different expression levels when in fact there is no difference.
To decrease the number of incorrect "calls," data produced must be pre-processed and analyzed using robust statistical methodologies developed specifically for multiple testing procedures.
Data from microarrays need to be background corrected, normalized, and then subjected to an appropriate multiple testing regime that accounts for the false discovery rate (FDR).
The Affymetrix software performs background correction when converting the raw data file (*.DAT) to a cellular intensity file (*.CEL). Basically, the scanned raw image is converted to a digitized file where each probe cell is assigned a mean intensity value. To account for
28 differences attributed to non-biological factors (e.g., sample processing that might lead to different starting concentrations of mRNA), data are normalized using a log scale robust multi- array analysis (RMA) method (Irizarry et al. 2003; reviewed by Saviozzi and Calogero 2003).
Normalization is a process of scaling values in a data set to improve the accuracy of the subsequent numeric computations. This is needed to ensure that expression levels seen in the experiment are comparable to the expression levels in the controls. In the RMA procedure, intensity data from analysis are log-transformed to base 2, adjusted for log-scale affinity effects, and also adjusted with an error estimate based on the variance of probe pair data. Normalization is performed using the RMA module of an R-based program language, specific to Affymetrix applications (downloaded from the Bioconductor web site www.bioconductor.org).
Once data were normalized, mRNA intensities are assessed any number of significant analyses. One such analysis significance of microarrays (SAM, Tusher et al. 2001). In SAM, a
"relative difference" test statistic is calculated for each gene. This test statistic compares change in gene expression (control vs. treatment) to the standard deviation in the data for a particular gene. In addition to calculating the above test statistic, SAM also calculates an "expected" relative difference for each gene based on results of data set permutations. The ratio of the observed test statistic to the expected test statistic should be approximately 1. (That is, plotting the observed test statistic versus the expected test statistic should result in a plot with a slope of
1.) In reality, points fall at some distance away from this line. By setting a "threshold", or limits around this line, one can locate genes deviating from this line by a fixed distance. The higher this threshold, the more likely genes outside the limits are significantly different between the control and the treatment.
29 SAM also computes the number of falsely significant genes at the selected threshold.
This is done by performing permutations of the control and treatment data sets in a random manner (i.e., creating nonsense sets) and then obtaining an average number of genes that are called significant in the nonsense sets. The ratio of falsely discovered significant genes versus the number of significant genes in the original testing creates the FDR. SAM allows the user to select a threshold that maximizes the number of genes called significant versus an acceptable
FDR. The version of SAM (Version 1.21) used for parts of this project was obtained from
Stanford University's web site (http://www-stat.stanford.edu/~tibs/SAM/). It is an add-in to
Microsoft EXCEL.
Another statistical procedure that assesses differential expression is an empirical Bayes analysis (Efron et al. 2001). This procedure is based on nonparametric statistics and produces a posteriori probabilities of effects for individual genes. The term "a posteriori" refers to obtaining a proposition based on past experience. The procedure connects well with the FDR theory mentioned above and applies itself well to microarray results, i.e., estimating probabilities of events from the simultaneous analysis of tens of thousand data points. The Bayes approach calculates a single expression value (Z) for each gene based on the number of probes sets representing each gene and the number of biological replicates. A simple Bayesian model assumes there two classes of genes, those that are differentially expressed and those that are not.
A balanced permutation of arrays in the test and control data sets provides a density plot that should mirror the Student's t-distribution with the appropriate degrees of freedom. This density curve creates a smooth curve that represents the case when no genes are differentially expressed.
A density plot of the actual Z values will be wider assuming some genes are differentially expressed (Efron 2003). Taken together, the curves fit to the density plots can provide functions
30 that, in turn, can be used to provide probability estimates of genes that are differentially expressed. As with SAM, a module is available at the Bioconductor web site for performing this analysis.
1.5 Bacterial-host interaction research using microarray technology
Cummings and Relman (2000) proposed using microarrays for transcriptional profiling to investigate host-pathogen interactions. They noted a recent publication that reported perfect correlation of microarray results with confirmatory Northern blots (72/72 mRNA transcripts detected via microarrays) in a research effort regarding host response to cytomegalovirus (CMV) infection. This article describes experimental design variables that need to be considered when applying microarray technology to host-pathogen research. Time series analyses are beneficial to follow transcriptional induction and any subsequent repression. Time series experiments would also determine the order of events following an encounter, with apoptosis or necrosis often being determinant endpoints. One weakness is that microarrays cannot detect post- transcriptional events, so they recommend assessment of protein expression as an amendment to microarray analysis. Test model considerations (tissues versus monolayer cell culture) are presented, noting the fact that infected tissues would lead to a very complicated host response whereas cell culture would reduce this complexity. The article states that interrogation of the transcriptome should lead to the identification of interesting cellular events, thereby leading to further exploratory investigations. The authors conjecture that a comparison of host response to related strains of the same pathogen could explain the differences in pathogenicity. Use of isogenic mutants or exposing cells to related toxins might also lead to hypotheses regarding mode of action of single virulence factors. Relman published another article with Manger (2000)
31 that expanded on the above. Manger and Relman investigated the degree to which the host response to any microbial pathogen is "stereotyped" or limited to certain patterned transcripts.
The responses could possibly be categorized based on the predominant type of locally recruited leukocytes or T-cells or by the pattern of cytokines elicited.
A review article by Schoolnik (2002) expanded on the above introduction to microarray analysis. Schoolnik noted that, in principle, assessing infections of tissues and/or cell cultures via microarrays could simultaneously identify host genes regulated during the infection process.
Cell culture variables to consider include cell type, multiplicity of infection (MOI) and the use of primary versus transformed cell lines (or cancerous lines). He also notes that the manner in which the bacteria are propagated prior to the infection experiments would potentially affect the host response. While cell culture would reduce response complexity, Schoolnik notes that host cell communication information would be lost. The review noted that publications to date had relied on transformed cell lines, used high MOIs and short time courses. Therefore, current studies reflected only early events in the infection process.
In another review, Yowe et al. (2001) discussed using microarray technologies and host- pathogen relationships with a focus on using the information to develop drugs to modulate host response. This review covered research to date regarding both viral and bacterial interactions with host cells. The review reported on a summary of host genes that were regulated and included emphasis on ligand/receptor responses, cell signal transduction and transcriptional factor activation. Yowe noted that some studies identified up-regulated transcripts that had never before been implicated in disease pathways, and that the physiological relevance of some of these newly discovered up-regulations was not always clear. Similar to Cummings and Relman,
Yowe stated that protein chip technology needed to be developed to support transcriptional
32 findings. In addition, Yowe cautioned that just because a transcript is up-regulated, it does not
necessarily mean it is important to the model under investigation.
Bryant et al. (2004) noted that microarray based approaches allow researchers to interrogate host genomes without prior bias regarding which gene/pathways are involved. This could lead to discovery of previously unknown functions and possibly different drug treatments and vaccine development. Another avenue of research proposed by these authors is to assess differences in host characteristics such as age, sex, time of day and even antigenic differences of host cell receptors between individuals. A final suggestion by this article was to compare host responses elicited by intra- versus extra-cellular organisms.
1.5.1 Research of host bacterial interactions, epithelial cell models
Using a cancerous gastric epithelial cell line, Maeda et al. (2001) and Bach et al. (2002)
used cDNA microarrays to monitor host gene regulation from Helicobacter pylori infection. In
the Maeda research, an isogenic mutant was used to look for differences in expression. H. pylori
were incubated with the gastric cells for three hours. The cagE gene function (a Type IV
secretion system component in the H. pylori pathogenicity island [PAI]) was eliminated in the
mutant. The microarray in this effort included 2,304 gene probes. The wild type up-regulated
eight genes and the mutant did not cause any gene up-regulation. The eight genes included
Interleukin 8 (IL-8), IкBα (an inhibitor to NF-кB), ERF-1 and A20 (Tnfaip3) which are all
associated with innate immune response. They proposed that A20 is a negative regulator of NF-
кB activity. The Bach group also used an insertional mutation, specifically targeting cagA in the
same PAI of H. pylori. This group also used a cancerous gastric epithelial cell line, infected the
tissue culture at 80% confluency, using a MOI of 100 and incubated the infected flasks for 4.5
33 hours. A 588 probe microarray was used to monitor gene regulation of cell signaling, apoptosis,
cell adhesion and transcription factors. Using a fold change threshold of 2.5, the most notable
probe up-regulated was MIP-2 (a macrophage signaling chemokine). Northern blots were used
to confirm microarray results and were found to be inconsistently correlated.
Coombs and Mahoney (2001) used a 268 spot array, consisting mainly of chemokine,
growth factors and cell receptors probe sets, to investigate Chlamydia pneumoniae infections of
epithelial (Hep-2) and endothelial (HMEC-1) cells. C. pneumoniae is an obligate intracellular
pathogen causing respiratory illness. After infection, the monolayers were centrifuged at 1000xg
for 60 minutes and incubated for one hour. This incubation time was felt sufficient because they
reported that NF-кB translocates to the nucleus within 15 minutes of infection in endothelial cells. Twenty five genes were found to be up-regulated from C. pneumoniae infection.
Interleukin cytokines, interferon receptors and regulatory proteins and tumor necrosis factor-
related transcripts were prominent among the differentially expressed genes. Confirmation of
gene expression was performed using RT-PCR and agreement was seen in those transcripts
displaying large up-regulations.
Interaction of Pseudomonas aeruginosa with a lung cancer epithelial cell line (ATCC
CCL165) was investigated using a 1,506 probe set array (Ichikawa et al. 2000). This group used a MOI of 50:1 and incubated the cultures three hours post infection. A mutant lacking Type IV pili (∆pilA) was also tested. Their focus was on the host gene IRF-1, known to be part of multiple pathways associated with the immune response. In addition, purified LPS was tested on this cell line. A 2-fold change in expression was used as the cut-off for differentially expressed genes. IRF-1 did exhibit up-regulation as did c-Jun, a variety of transcription factors and G- protein binding targets, as well as Tnfaip3 (A20). There were nine differentially expressed genes
34 in the wild type organism infections that were not detected in the mutant infections. There was
no up-regulation seen in those cultures when purified LPS was added.
Eckmann et al. (2000) used high density cDNA microarrays to investigate altered gene
expression in human intestinal epithelial cells after infection with the intracellular pathogen
Salmonella dublin. Two different human colon epithelial cell lines were used, ATCC HTB-38
and T84 (from an earlier effort by Eckmann). Monolayers were infected with 108 organisms and incubated for one hour, followed by antibiotic treatment. The monolayers were then assayed at various time periods up to 20 hours. Two different microarrays were used, one that included approximately 4,000 probes (GF211, Research Genetics, Inc.) and the other 277 probes (Atlas,
CLONTECH). A small number of differentially expressed genes were confirmed using RT-
PCR. The atlas array focused on chemokine and cell receptor genes. The two arrays had little overlap of probes. Relatively few genes (approx. 5%) exhibited fold changes of greater than 2 with the 4,000 gene array. The second array which had a more focused set of probes related to innate immune response showed 65% genes up-regulated. Most up-regulated genes could be associated with the immune response, but with some there was no known association. RT-PCR confirmation was variable and differences were attributed to its inability to detect small variances in mRNA concentrations.
With a major focus on developing a model to assess microarray infections experiments while minimizing replication, Stekel et al. (2005) assessed E. coli infection of Caco-2 monolayers. A primary reason for the research was to find a way to minimize the number of replicates because of the high cost associated with microarray testing. The desire was to obtain a listing of differentially expressed genes from a variety of experimental conditions where only one array was analyzed per variable. The statistical model was built upon array to array
35 variability, array feature to feature variability and dependence of error on signal intensity.
Enteropathogenic and entero-hemorrhagic strains of E. coli (EPEC and EHEC, respectively) were used for infections at a MOI of 50, with an incubation time of 2.5 hours. Heat killed organisms were also tested. A microarray of 850 probes (60-mers) was used, with the array consisting of probes representing apoptosis, cell cycle, chemo-cytokines, cell receptors, adhesion and signaling molecules. Both organisms produce the attachment and effacement (A/E) lesions, but possess different virulence factors otherwise. Using their proposed model, differences in gene regulation were noted between the EHEC and EPEC strains, as well as similarities. One noted difference was the induction of gamma interferon receptors in host cells infected with the
EHEC strain.
1.6 Animal models for assessing Aeromonas virulence
To my knowledge there have been five live animal models used to investigate Aeromonas virulence. One model uses rabbits and was a diarrheal model. Another model uses intra- peritoneal injections of live Aeromonas organisms and monitors the LD50. A third model uses fluid accumulation in a ligated ileal loop of mice. The fourth model uses suckling mice which are fed live Aeromonas via gastric lavage. Again, this model uses LD50 as an endpoint. Last, streptomycin treated adult mice were fed Aeromonas and tested for colonization. A brief overview of research using each model follows.
Pazzaglia et al. 1990 tested Aeromonas spp. in a removable intestinal tie adult rabbit diarrheal (RITARD) model. In this model, a loop of the terminal ileum is isolated using slipknots of umbilical tape. The tied off segment is then injected with the organisms and after a specified time period, the ileum loop is restored to original function. The endpoints can be fluid
36 accumulation or diarrhea production (Davis and Banks 1991). Species tested included A.
hydrophila, A. veronii biotype sobria and A. caviae. All A. hydrophila and A. veronii biotype
sobria strains demonstrated virulence by either producing death or diarrhea. Only one of three
A. caviae strains produced diarrhea. One A. hydrophila strain (6Y) used in this study was the
same as used in the Morgan et al. human feeding study. This strain was highly virulent in the
RITARD model as compared to the mild virulence seem in the human study.
Janda and Kokka (1991) reported on virulence of 80 Aeromonas strains using intra-
peritoneal injection of mice and monitoring the LD50. They found a 4-log10 difference in the
LD50 among the species/strains tested. HGs 9, 1, 12, 10 and 8 had the lowest LD50s (in
increasing order). HG4 (A. caviae) exhibited a high LD50 indicating it is not a virulent organism.
Sha et al. (2002) used an ileal loop assay in mice. It is similar to the RITARD method
except performed in mice. The endpoint is a measurement of fluid accumulation. As described
earlier, Sha was interested in assessing the effects of enterotoxin gene knockouts.
In the suckling mouse assay described by Wong et al (1996), live organisms are
administered to neonatal mice (4-6 days old) via gastric lavage. The LD50 was monitored after
43 hours. Any death was indicative of virulence. The range of LD50 for strains deemed
"virulent" was from log10 7.53 to log10 8.88. The found no correlation between the LD50 and the source of the isolate or hemolytic activity.
Sanderson et al. (1996) used streptomycin treated mice and administered Aeromonas spp. via oral gavage. Adult C57BL mice were pre-treated with streptomycin to reduce competing normal flora and colonization was assessed by culturing the fecal material of the mice for 10 days post inoculation. Comparisons were made between colonization and adherence assays on cell culture. Colonization did occur in the streptomycin treated mice, but not in untreated mice.
37 This group hoped to produce a diarrheal model, but none of the treated mice exhibited
symptoms.
Based on the literature search, there is much known regarding Aeromonas virulence
factors. There is also much that is unknown, specifically regarding host response to these VFs
(i.e., disease pathways). It is obvious that some Aeromonas strains harbor a multitude of VFs.
There is also evidence that virulence of Aeromonas can be attenuated by knocking out a
particular VF. Specific and quantifiable response data would help interpret cause and effect
relationships between pathogens and host systems.
Besides the three species most often associated with human gastrointestinal disease (A.
hydrophila, A. caviae, and A. veronii biotype sobria), other species are occasionally implicated in human disease (A. jandaei, A. schubertii, A. salmonicida (HG3), A. veronii biotype veronii
(HG10), A. media and other atypical isolates). There is also variation of virulence within species
(i.e., each species appears to harbor both virulent and avirulent strains). The literature reports that species share common VFs, indicating lateral exchange of VFs. Enzymes such as proteases and lipases are not considered as major VFs, yet isogenic mutants relative to these enzymes also show higher LD50s.
In terms of adhesion factors, A. hydrophila strains are reported to attach weakly to cell culture, but this species is considered the most virulent. The role of lateral flagella in adhesion and associated virulence is uncertain. There is little correlation between the presence of a
"major" VF and a strain's virulence (for example, A. salmonicida and other species are aerolysin positive, yet do not [or rarely] cause disease in humans). Host response data, obtained using appropriate models, may assist in determining which VFs are essential for disease establishment.
38 Infection assays would benefit from more standardization. A number of animal models have been used, each with a different endpoint. Mice have been inoculated intraperitoneally and orally, with the LD50 typically used as an endpoint. Ligated ileal loops have also been used in mice and rabbits with fluid secretion being the endpoint measured. Only one human dose study has been performed and it was lacking in comprehensive data. Because human dose response data are unlikely to be obtained for this (or any other) organism, other host-response models need to be developed. Disease mechanisms discovered in animal (or better yet in non-animal) models require discovery and correlation to human systems.
Aeromonas are globally distributed and demonstrate population differences regarding the percentages of species found at a particular location. It is still being debated in the U.S. whether
Aeromonas is an organism that merits attention regarding regulation in food and water.
However, organisms from this genus are definitely an issue in developing countries where exposure to high levels of Aeromonas is common and diarrhea remains a major cause of death.
Knowledge of disease mechanisms may assist in the development of drug strategies for countries where treatment of water and food are lacking.
One question noted in the literature review is, how much of the host response is stereotyped? Microarray data can be useful in answering this question. Stereotyped responses to a variety of organisms could indicate common treatment.
There are a multitude of methods for analyzing microarray data. This includes pre- processing of raw data as well as statistical analysis of final intensity values. Again, the community of researchers utilizing microarray technology would benefit from a user-friendly and standard approach.
39 The goal of this research effort will be to provide insight into some of the uncertainties associated with Aeromonas associated disease, specific to gastrointestinal infection. Use of the innovative microarray molecular technique and associated data will play a major role in this effort. Cell culture and animal models will be assessed for their ability to define host pathogen relationships and assessed for commonalities. A model's value for defining virulence or important disease mechanisms will also be evaluated.
40 1.7 References
Abbott, S.L., Cheung, W.K.W. and Janda, J.M. 2003. The genus Aeromonas: biochemical characteristics, atypical reactions, and phenotypic identification schemes. Journal of Clinical
Microbiology. 41:2348-2357.
Alavandi, S.V. and Ananthan, S. 2003. Biochemical characteristics, serogroups, and virulence factors of Aeromonas species isolated from cases of diarrhea and domestic water samples in
Chennai. Indian Journal of Medical Microbiology. 21:233-238.
Albert, M.J., Ansaruzzaman, M., Talukder, K., Chopra, A.K., Kuhn, I., Rahman, M., Faruque,
A.S.G., Islam, M.S., Sack, R.B. and Mollby, R. 2000. Prevalence of enterotoxin genes in
Aeromonas spp. isolated from children with diarrhea, healthy controls, and the environment.
Journal of Clinical Microbiology. 38:3785-3790.
Altwegg, M. and Geiss, H.K. 1989. Aeromonas as a human pathogen. CRC Critical Reviews in
Microbiology. 16: 253-286.
Anguita, J., Aparicio, A.B.R. and Naharro, G. 1993. Purification, gene cloning, amino acid sequence analysis, and expression of an extracellular lipase from an Aeromonas hydrophila human isolate. Applied and Environmental Microbiology. 59:2411-2417.
41 Ashbolt, N.J., Ball, A., Dorsch, M., Turner, C., Cox, P., Chapman, A. and Kirov, S.M. 1995.
The identification and human health significance of environmental aeromonads. Water Science
Technology. 31:263-269.
Ashdown, L.R. and Koehler, J.M. 1993. The spectrum of Aeromonas-associated diarrhea in tropical Queensland, Australia. Southeast Asian Journal of Tropical Medicine and Public
Health. 24: 347-353.
Bach, S., Makristathis, A., Rotter, M. and Hirschl, A.H. 2002. Gene expression profiling in AGS cells stimulated with Helicobacter pylori isogenic strains (cagA positive or cagA negative).
Infection and Immunity. 70:988-992.
Barer, M.R., Millership, S.E. and Tabaqchali, S. 1986. Relationship of toxin production to species in the genus Aeromonas. Journal of Medical Microbiology. 22:303-309.
Bartkovǎ, G. and Čižnár, I. 1994. Adherence pattern of non-pilated Aeromonas hydrophila strains to tissue cultures. Microbios. 77:47-55.
Bogdanovic, R., Cobeljic, M., Markovic, M., Nikolic, V., Ognjanovic, M., Sarjanovic, L. and
Makic, D. 1991. Haemolytic-uraemic syndrome associated with Aeromonas hydrophila enterocolitis. Pediatric Nephrology. 5: 293-295.
42 Borchardt, M.A., Stemper, M.E. and Standridge, J.H. 2003. Aeromonas isolates from human diarrheic stool and groundwater compared by pulsed-field gel electrophoresis. Emerging
Infectious Diseases. 9:224-228.
Borrell, N., Acinas, S.G., Figueras, M.-J. and Martinez-Murcia, A.J. 1997. Identification of
Aeromonas clinical isolates by restriction fragment length polymorphism of PCR-amplified 16S rRNA genes. Journal of Clinical Microbiology 35:1671-1674.
Borrell, N., Figueras, M.J. and Guarro, J. 1998. Phenotypic identification of Aeromonas genomospecies from clinical and environmental sources. Canadian Journal of Microbiology.
44:103-108.
Brown, R.L., Sanderson, K. and Kirov, S.M. 1997. Plasmids and Aeromonas virulence. FEMS
Immunology and Medical Microbiology. 17:217-223.
Bryant, P.A., Venter, D., Robins-Browne, R. and Curtis, N. 2004. Chips with everything: DNA microarrays in infectious diseases. The Lancet Infectious Diseases. 4:100-111.
Burke, V., Robinson, J., Gracey, M., Peterson, D. and Partridge, K. 1984. Isolation of
Aeromonas hydrophila from a metropolitan water supply: seasonal correlation with clinical isolates. Applied and Environmental Microbiology. 48:361-366.
43 Cahill, M.M. 1990. Virulence factors in motile Aeromonas species. Journal of Applied
Bacteriology. 69: 1-16.
Cascón, A., Yugueros, J., Temprano, A., Sánchez, M., Hernanz, C., Luengo, J.M. and Naharro,
G. 2000. A major secreted elastase if essential for pathogenicity of Aeromonas hydrophila.
Infection and Immunity. 68:3233-3241.
Chacón, M.R., Figueras, M.J., Castro-Escarpulli, G., Soler, L. and Guarro, J. 2003. Distribution of virulence genes in clinical and environmental isolates of Aeromonas spp. Antonie van
Leeuwenhoek. 84:269-278.
Chacón, M.R., Soler, L., Groisman, E.A., Guarro, J. and Figueras, M.J. 2004. Type III secretion system genes in clinical Aeromonas isolates. Journal of Clinical Microbiology. 42:1285-1287.
Chakraborty, T., Huhle, G., Gergbauer, J. and Goeber, W. 1986. Cloning, expression, and mapping of the Aeromonas hydrophila aerolysin gene determinant in Escherichia coli K-12.
Journal of Bacteriology. 167:368-374.
Challapalli, M., B.R. Tess, D.G. Cunningham, A.K. Chopra, and C.W. Houston. 1988.
Aeromonas-associated diarrhea in children. Pediatric Infectious Disease Journal. 7: 693-698.
44 Chauret, C., Volk, C., Cresson, R., Jars, J., Robinson, J. and Warns, C. 2001. Detection of
Aeromonas hydrophila in a drinking-water distribution system: a field and pilot study. Canadian
Journal of Microbiology. 47:782-786.
Chopra, A.K., Houston, C.W., Peterson, J.W. and Jin, G-F. 1993a. Cloning, expression, and
sequence analysis of a cytolytic enterotoxin gene from Aeromonas hydrophila. Canadian Journal
of Microbiology. 39:513-523.
Chopra, A.K., Pham, R. and Houston, C.W. 1993b. Cloning and expression of putative cytotonic
enterotoxin-encoding genes from A. hydrophila. Gene. 139:87-91.
Chopra, A.K., Peterson, J.W., Xu, X.J., Coppenhaver, D.H. and Houston, C.W. 1996. Molecular
and biochemical characterization of a heat-labile cytotonic enterotoxin from Aeromonas
hydrophila. Microbial Pathogenesis. 21:357-377.
Chopra, A.K., Xu, X.J., Ribardo, D., Gonzalez, M., Kuhl, K., Peterson, J.W. and Houston, C.W.
2000. The cytotoxic enterotoxin of Aeromonas hydrophila induces proinflammatory cytokine production and activates arachidonic acid metabolism in macrophages. Infection and Immunity.
68:2808-2818.
Colwell, R.R., MacDonell, M.T. and DeLey, J. 1986. Proposal to recognize the family
Aeromonadaceae fam. nov. International Journal of Systematic Bacteriology. 36: 473-477.
45 Coombs, B. K. and Mahony, J.B. 2001. cDNA array analysis of altered gene expression in human epithelial cells in response to Chlamydia pneumoniae infection. Infection and Immunity.
69:1420-1427.
Cummings, C.A. and Relman, D.A. 2000. Using DNA microarrays to study host-microbe interactions. Genomics. 6:513-525.
Davis, J.A. and Banks, R.E. 1991. Modification to the RITARD surgical model. Journal of
Investigative Surgery. 4:499-504.
Demarta, A., Tonolla, M., Caminada, A., Beretta, M. and Peduzzi, R. 2000. Epidemiological relationships between Aeromonas strains isolated from symptomatic children and household environments as determined by ribotyping. European Journal of Epidemiology. 16:447-453.
Dooley, J.S.G., McCubbin, W.D., Kay, C.M. and Trust, T.J. 1988. Isolation and biochemical characterization of the S-layer protein from a pathogenic Aeromonas hydrophila strain. Journal of Bacteriology. 170:2631-2638.
Eckmann, L., Smith, J.R., Housley, M.P., Dwinell, M.B. and Kangoff, M.F. 2000. Analysis by high density cDNA arrays of altered gene expression in human intestinal epithelial cells in response to infection with the invasive enteric bacteria Salmonella. Journal of Biological
Chemistry. 275:14084-14094.
46 Efron, B., Tibshirani, R., Storey, J.D. and Tusher, V. 2001. Empirical Bayes analysis of a
microarray experiment. Journal of the American Statistical Association. 96:1151-1160.
Efron, B. 2003. Robbins, empirical Bayes and microarrays. The Annals of Statistics. 31:366-
378.
Epple, H.J., Mankertz, J., Ignatius, R., Liesenfeld, O., Fromm, M., Zeitz, M., Chakraborty, T.
and Schulzke, J.D. 2004. Aeromonas hydrophila beta-hemolysin induces active chloride
secretion in colon epithelial cells (HT-29/B6). Infection and Immunity. 72:4848-4858.
Esteve, C. and Birkbeck, T.H. 2004. Secretion of haemolysins and proteases by Aeromonas
hydrophila EO63: separation and characterization of the serine protease (caseinase) and the
metalloprotease (elastase). Journal of Applied Microbiology. 96:994-1001.
Ferguson, M.R., Xu, X.J., Houston, C.W., Peterson, J.W., Coppenhaver, D.H., Popov, V.L. and
Chopra, A.K. 1997. Hyperproduction, purification, and mechanism of action of the cytotoxic
enterotoxin produced by Aeromonas hydrophila. Infection and Immunity. 65:4299-4308.
Figueras, M.J., Soler, L., Chacón, M.R., Guarro, J. and Martĭnez-Murcia, A.J. 2000. Extended method for discrimination of Aeromonas spp. by 16S rDNA analysis. International Journal of
Systematic and Evolutionary Microbiology. 50:2069-2073.
47 Figueras, M.J., Suarez-Franquet, A., Chacón, M.R., Soler, L., Navarro, M., Alejandre, C., Grasa,
B., Martĭnez-Murcia, A.J. and Guarro, J. 2005. First record of the rare species Aeromonas culicicola from a drinking water supply. Applied and Environmental Microbiology. 71:538-
541.
Figura, N., Marri, L., Verdiani, S., Ceccherini, C. and Barber, A. 1986. Prevalence, species differentiation, and toxigenicity of Aeromonas strains in cases of childhood gastroenteritis and in controls. Journal of Clinical Microbiology. 23: 595-599.
Fivaz, M., Abrami, L., Tstrin, Y. and van der Goot, F.G. 2001. Not as simple as just punching a hole. Toxicon. 39:1637-1645.
Galindo, C.L., Sha, J., Ribardo, D.A., Fadl, A.A., Pillai, L. and Chopra, A.K. 2003.
Identification of Aeromonas hydrophila cytotoxic enterotoxin-induced genes in macrophages using microarrays. The Journal of Biological Chemistry. 278;40198-40212.
Galindo, C.L., Fadl, A.A., Sha, J., Gutierrez, C., Popov, V.L., Boldogh, I., Aggarwal, B.B. and
Chopra, A.K. 2004. Aeromonas hydrophila cytotoxic enterotoxin activates mitogen-activated protein kinases and induces apoptosis in murine macrophages and human intestinal epithelial cells. The Journal of Biological Chemistry. 279:37597-37612.
48 Galindo, C.L., Fadl, A.A., Sha, J., Pillai, L., Gutierrez, C. and Chopra, A.K. 2005. Microarray
and proteomics analyses of human intestinal epithelial cells treated with the Aeromonas
hydrophila cytotoxic enterotoxin. Infection and Immunity. 73:2628-2643.
Gavriel, A.A., Landre, J.P.B. and Lamb, A.J. 1998. Incidence of mesophilic Aeromonas within a
public drinking water supply in north-east Scotland. Journal of Applied Bacteriology. 84: 383-
392.
George, W.L., Nakata, M.M., Thompson, J.J, and White, M.L.. 1985. Aeromonas-related diarrhea in adults. Archives of Internal Medicine. 145: 2207-2211.
Gluskin, I., Batash, D., Shoseyov, D., Mor, A., Kazak, R., Azizi, E. and Boldur, I. 1992. A 15- year study of the role of Aeromonas spp. in gastroenteritis in hospitalized children. Journal of
Medical Microbiology. 37: 315-318.
González-Serrano, C.J., Santos, J.A., García-López, M.L. and Otero, A. 2002. Virulence markers in Aeromonas hydrophila and Aeromonas veronii biovar sobria isolates from freshwater fish and from diarrhoea case. Journal of Applied Microbiology. 93:414-419.
Gracey, M., Burke, V. and Robinson, J. 1982. Aeromonas-associated gastroenteritis. Lancet. 2:
1304-1306.
49 Grey, P.A. and Kirov, S.M. 1993. Adherence to Hep-2 cells and enteropathogenic potential of
Aeromonas spp. Epidemiology and Infections. 110:279-287.
Haque, Q.M., Sugiyama, A., Iwade, Y., Midorikawa, Y. and Yamauchi, T. 1996. Diarrheal and environmental isolates of Aeromonas spp. produce a toxin similar to Shiga-like toxin 1. Current
Microbiology. 32:239-245.
Hasan, J.A.K., Macaluso, P., Carnahan, A.M. and Joseph, S.W. 1992. Elastolytic activity among
Aeromonas spp. using a modified bilayer plate assay. Diagnostic Microbiology and Infectious
Diseases. 15:201-206.
Havelaar, A.H., Schets, F.M., van Silfhout, A., Jansen, W.H., Wieten, G. and van der Kooij, D.
1992. Typing of Aeromonas strains from patients with diarrhoea and from drinking water.
Journal of Applied Bacteriology. 72:435-444.
Heuzenroeder, M.W., Wong, C.Y.F., and Flower, R.L.P. 1999. Distribution of two hemolytic toxin genes in clinical and environmental isolates of Aeromonas spp.: correlation with virulence in a suckling mouse model. FEMS Microbiology Letters. 174:131-136.
Hirono, I., Aoki, T., Asao, T. and Kozaki, S. 1992. Nucleotide sequences and characterization of haemolysin genes from Aeromonas hydrophila and Aeromonas sobria. Microbial Pathogenesis.
13:433-446.
50 Holmberg, S.D., Schell, W.L., Fanning, G.R., Wachsmuth, I.K., Hickman-Brenner, F.W., Blake,
P.A., Brenner, D.J. and Farmer III, J.J. 1986. Aeromonas intestinal infections in the United
States. Annals of Internal Medicine. 105: 683-689.
Huys, G., Kampfer, P., Altwegg, M., Kersters, I., Lamb, A., Coopman, R., Luthy-Hottenstein, J.,
Vancannneyt, M., Jannsen, P. and Kersters, K. 1997. Aeromonas popoffii sp. Nov.: a mesophilic bacterium isolated from drinking water production plants and reservoirs. International Journal of
Systemic Bacteriology. 47: 1165-1171.
Ichikawa, J.K., Norris, A., Bangera, M.G., Geiss, G.K., vant't Wout, A.B., Bumgarner, R.E. and
Lory, S. 2000. Interaction of Pseudomonas aeruginosa with epithelial cells: identification of differentially regulated genes by expression microarray analysis of human cDNAs. Proceedings of the National Academy of Science. 97:9656-9664.
Imziln, B., Krovacek, K., Baloda, S.B., Kühn, I., Gonzalez-Rey, C. and Svenson, S.B. 1998.
Characterization of potential virulence markers in Aeromonas caviae isolated from polluted and
unpolluted aquatic environments in Morocco. FEMS Microbiology Ecology. 27:153-161.
Irizarry, R. A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B. and Speed, T.P. 2003.
Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Research. 31: e15
Janda, J.M. 1991. Recent advances in the study of taxonomy, pathogenicity, and infectious
syndromes associated with the genus Aeromonas. Clinical Microbiological Reviews. 4: 397-410.
51
Janda, J.M., Kokka, R.P. and Guthertz, L.S. 1994. The susceptibility of S-layer-positive and S- layer-negative Aeromonas strains to complement-mediated lysis. Microbiology. 140:2899-
2905.
Janda, J. M. and S. L. Abbott. 1998. Evolving concepts regarding the genus Aeromonas: an expanding panorama of species, disease presentations, and unanswered questions. Clinical
Infectious Diseases. 27:333-344.
Janda, J.M. 1999. Vibrio, Aeromonas and Plesiomonas. W.J. Hausler, M. Sussman (Eds.),
Topley & Wilson’s Microbiology and Microbial Infections. (pp.1065-1084). New York: Oxford
University Press.
King, G.E., Werner, B. and Kizer, K.W. 1992. Epidemiology of Aeromonas infections in
California. Clinical Infectious Diseases. 15:449-452.
Kingombe, C.I.B., Huys, G., Tonolla, M., Albert, M.J., Swings, J., Peduzzi, R. and Jemmi, T.
1999. PCR detection, characterization, and distribution of virulence genes in Aeromonas spp.
Applied and Environmental Microbiology. 65:5293-5302.
Kirov, S.M., Rees, B., Wellock, R.C., Goldsmid, J.M., and van Galen, A.D. 1986. Virulence characteristics of Aeromonas spp. in relation to source and biotype. Journal of Clinical
Microbiology. 24:827-834.
52
Kirov, S.M., Barnett, T.C., Pepe, C.M., Strom, M.S. and Albert, M.J. 2000. Investigation of the role of type IV Aeromonas pilus (tap) in the pathogenesis of Aeromonas gastrointestinal infection. Infection and Immunity. 68:4040-4048.
Kirov, S.M., Castrisios, M., and Shaw, J.G. 2004. Aeromonas flagella (polar and lateral) are enterocyte adhesions that contribute to biofilm formation on surfaces. Infection and Immunity.
72:1939-1945.
Kuijper, E.J., Bol, P., Peeters, M.F., Steigerwalt, A.G., Zanen, H.C. and Brenner, D.J. 1989.
Clinical and epidemiologic aspects of members of Aeromonas DNA hybridization groups isolated from human feces. Journal of Clinical Microbiology. 27:1531-1537.
Legnani, P., Leoni, E., Soppeisa, F. and Burigo, R. 1998. The occurrence of Aeromonas species in drinking water supplies of an area of the Dolomite Mountains, Italy. Journal of Applied
Microbiology. 85:271-276.
Lipshutz, R.J., Fodor, S.P.A., Gingeras, T.R. and Lockhart, D.J. 1999. High density synthetic oligonucleotide arrays. Nature Genetics Supplement. 21:20-24.
Maeda, S., Otsuka, M., Hirata, Y., Mitsuno, Y., Yoshida, H., Shiratori, Y., Masuho, Y.,
Muramatsu. M., Seki, N. and Omata, M. 2001. cDNA microarray analysis of Helicobacter
53 pylori-mediated alteration of gene expression in gastric cancer cells. Biochemical and
Biophysical Research Communications. 284:443-449.
Manger, I.D. and Relman, D.A. 2000. How the host 'sees' pathogens: global gene expression responses to infection. Current Opinion in Immunology. 12:215-218.
Martin-Carnahan, A. and Joseph, S.W. 2005. Order XII. Aeromonadales ord. nov. In George
M. Garrity (Editor in Chief), Bergey's Manual of Systematic Bacteriology. 2nd Edition, Vol.2
Part B. (pp. 556-580). New York: Springer.
Merino, S., Rubires, X., Aguilar, A. and Tomás, J.M. 1996. The O:34-antigen lipopolysaccharide as an adhesin in Aeromonas hydrophila. FEMS Microbiology Letters.
139:97-101.
Merino, S., Aguilar, A., Nogueras, M.M., Regue, M., Swift, S. and Tomás, J.M. 1999. Cloning, sequencing, and role in virulence of two phospholipases (A1 and C) from mesophilic Aeromonas sp. Serogroup O:34. Infection and Immunity. 67:4008-4013.
Merino, S., Gavĭn, R., Vilches, S., Shaw, J.G. and Tomás. J.M. 2003. A colonization factor
(production of lateral flagella) of mesophilic Aeromonas spp. is inactive in Aeromonas salmonicida strains. Applied and Environmental Microbiology. 69:663-667.
54 Morgan, D.R., Johnson, P.C., DuPont, H.L., Satterwhite, T.K. and Wood, L.V. 1985. Lack of correlation between known virulence properties of Aeromonas hydrophila and enteropathogenicity for humans. Infection and Immunity. 50:62-65.
Moyer, N.P. 1987. Clinical significance of Aeromonas species isolated from patients with diarrhea. Journal of Clinical Microbiology. 25:2044-2048.
Moyer, N.P., Luccini, G.M., Holcomb, L.A., Hall, N.H. and Altwegg, M. 1992. Application of ribotyping for differentiating aeromonads isolated from clinical and environmental sources.
Applied and Environmental Microbiology. 58:1940-1944.
Nishikawa, Y., Hase, A., Ogawasare, J., Scotland, S.M., Smith, H.R. and Kimura, T. 1994.
Adhesion to and invasion of human colon carcinoma Caco-2 cells by Aeromonas strains. Journal of Medical Microbiology. 40:55-61.
Parker, M.W., van der Goot, F.G. and Buckley, J.T. 1996. Aerolysin – the ins and outs of a model channel-forming toxin. Molecular Microbiology. 19:205-212.
Pazzaglia, G., Sack, R.B., Bourgeois, A.L., Froehlich, J. and Ekstein, J. 1990. Diarrhea and intestinal invasiveness of Aeromonas strains in the removable intestinal tie rabbit model.
Infection and Immunity. 58:1924-1931.
55 Picard, B., and Goullet, P. 1987. Seasonal prevalence of nosocomial Aeromonas hydrophila infection related to Aeromonas in hospital water. Journal of Hospital Infections. 10: 152-155.
Rocha-de-Souza, C.M., Colombo, A.V., Hirata, R., Mattos-Guarasdi, A.L., Monteiro-Leal, L.H.,
Previato, J.O., Freitas, A.C. and Andrade, A.F. 2001. Identification of a 43-kDa outer- membrane protein as an adhesin in Aeromonas caviae. Journal of Medical Microbiology.
50:313-319.
Robson, W.L.M., Leung, A.D.C. and Trevenen, C.L. 1992. Haemolytic-uremic syndrome associated with Aeromonas hydrophila enterocolitis. Pediatric Nephrology. 6:221-222.
Sanderson, K., Ghazali, F.M. and Kirov, S.M. 1996. Colonization of streptomycin treated mice by Aeromonas spp. Journal of Diarrhoeal Disease Research. 14:27-32.
San Joaquin, V.H. and Pickett, D.A. 1988. Aeromonas-associated gastroenteritis in children.
Pediatric Infectious Disease Journal. 7: 53-57.
Saviozzi, S. and Calogero, R.A. 2003. Microarray probe expression measures, data normalization and statistical validation. Comparative and Functional Genomics. 4:442-446.
Schoolnik, G.K. 2002. Microarray analysis of bacterial pathogenicity. Advances in Microbial
Physiology. 46:1-45.
56 Sha, J., Kozlova, E.V. and Chopra, A.K. 2002. Role of various enterotoxins in Aeromonas
hydrophila-induced gastroenteritis: generation of enterotoxin gene-deficient mutants and
evaluation of their enterotoxic activity. Infection and Immunity. 70:1924-1935.
Sinha, S., Shimada, T., Ramamurthy, T., Bhattacharya, S.K., Yamasaki, S., Takeda, Y. and
Balakrish Nair, G. 2004. Prevalence, serotype distribution, antibiotic susceptibility and genetic
profiles of mesophilic Aeromonas species isolated from hospitalized diarrhoeal cases in Kolkata,
India. Journal of Medical Microbiology. 53:527-534.
Smith, H.R. and Cheasty, T. 1998. Diarrhoeal disease due to Escherichia coli and Aeromonas.
In W.J. Hausler & M. Sussman (Eds.), Microbiology and Microbial Infections. (pp. 513-529).
New York: Oxford University Press.
Song, T., Toma, C., Nakasone, N. and Iwanaga, M. 2004. Aerolysin is activated by metalloprotease in Aeromonas veronii biovar sobria. Journal of Medical Microbiology. 53:477-
482.
Stekel, D.J., Sarti, D., Trevino, V., Zhang, L., Salmon, M., Buckley, C.D., Stevens, M., Pallen,
M.J., Penn, C. and Falciani, F. 2005. Analysis of host response to bacterial infection using error model based gene expression microarray experiments. Nucleic Acids Research. 33:e53.
57 Takahashi, A., Tanoue, N., Nakano, M., Hamamoto, A., Okamoto, K., Fujii, Y., Harada, N. and
Nakaya, Y. 2005. A pore-forming toxin produced by Aeromonas sobria activates Ca2+ dependent Cl- secretion. Microbial Pathogenesis. 38:173-180.
Tusher, V.G., Tibshirani, R. and Chu, G. 2001. Significance analysis of microarrays applied to
the ionizing radiation response. Proceedings of the National Academy of Sciences. 98:5116-
5121.
USEPA. 2000. Sept. 25, 2000. Aeromonas Criteria Document. Office of Water.
Vilches, S., Urgell, C., Merino, S., Chacón, M.R., Soler, L., Castro-Escarpulli, G., Figueras, M.J.
and Thomás, J.M. 2004. Complete type III secretion system of a mesophilic Aeromonas
hydrophila strain. Applied and Environmental Microbiology. 70:6914-6919.
Wong, C.Y.F., Mayrhofer, G., Heuzenroeder, M.W., Atkinson, H.M., Quinn, D.M. and Flower,
R.L. 1996. Measurement of virulence of aeromonads using a suckling mouse model of infection.
FEMS Immunology and Medical Microbiology. 15:233-241.
Wong, C.Y.F., Heuzenroeder, M.W. and Flower, R.L.P. 1998. Inactivation of two haemolytic
toxin genes in Aeromonas hydrophila attenuates virulence in a suckling mouse model.
Microbiology. 144:291-298.
58 Xu, X.J., Ferguson, M.R., Popov, V.L., Houston, C.W., Peterson, J.W. and Chopra, A.K. 1998.
Role of a cytotoxic entertoxin in Aeromonas-mediated infections: development of transposon and isogenic mutants. Infection and Immunity. 66:3501-3509.
Yowe, D., Cook, W.J. and Gutierrez-Ramos, J-C. 2001. Microarrays for studying the host transcriptional response to microbial infection and for the identification of host drug targets.
Microbes and Infection. 3:813-821.
Yu, H.B., Zhang, Y.L., Lau, Y.L., Yao, F., Vilches, S., Merino, S., Tomas, J.M., Howard, S.P. and Leung, K.Y. 2005. Identification and characterization of putative virulence genes and gene clusters in Aeromonas hydrophila PPD134/91. Applied and Environmental Microbiology.
71:4469-4477.
59
Published in Journal of Water and Health 4:381-388.
Chapter 1
Identification by microarray of a common pattern of gene expression in intact intestine and
cultured intestinal cells exposed to virulent Aeromonas hydrophila isolates
60
Abstract
The genus Aeromonas comprises known virulent and avirulent isolates and has been implicated in waterborne disease. A common infection model of human gastroenteritis associated with A. hydrophila uses neonatal mice. The goal of this research was to evaluate whether a murine small intestinal cell line could provide comparable results to the gene expression changes in the neonatal mouse model. Changes in mRNA expression in host cell cultures and intestinal tissues were measured after exposure to virulent Aeromonas hydrophila strains. A. hydrophila caused the up-regulation of more than 200 genes in neonates and over 50 genes in cell culture. Twenty- six genes were found to be in common between the two models, of which the majority of these genes are associated with the innate immune response.
Keywords: Aeromonas hydrophila, animal model, cell culture, microarray, virulence factors
61 INTRODUCTION
Aeromonas hydrophila is a gram negative bacterium commonly found in surface water, groundwater, marine and estuarine environments, and even in chlorinated water supplies. This organism is currently listed on the U.S. Environmental Protection Agency’s Candidate
Contaminant List (CCL) (U.S. Federal Register, March 2, 1998) primarily because it has been implicated in waterborne disease and it is commonly found in source water. A. hydrophila can cause wound infections and septicemia in immuno-compromised people and some evidence suggests that it causes gastrointestinal disease in healthy individuals. Aeromonas spp. contribute to biofilms (Gavriel et al. 1998), and experience regrowth in drinking water distribution systems
(Gavriel et al. 1998; Smith and Cheasty, 1998). A. hydrophila has been detected in hospital water supplies (Picard and Goullet, 1987). Environmental occurrence varies by season, with increases seen in warmer summer months (Burke et al. 1984; Gavriel et al. 1998; Smith and
Cheasty, 1998). Infection rates are correlated with peak rates of occurrence in the environment.
Two studies recovered more aeromonads in surface water than in groundwater (Burke et al.
1984; Legnani et al. 1998).
The consequences resulting from exposure of host cells to Aeromonas and their associated virulence factors (VFs) are not fully defined. The published literature presents confounding information on the relative importance of a particular Aeromonas VF or which VFs are absolutely essential for causing disease. Therefore, it is not currently possible to assess health risks associated with the large population of different Aeromonas isolates found in drinking water sources. A neonatal mouse model for Aeromonas, as described by Wong et al.
1996, has been used in past research to quantitate virulence of A. hydrophila. Gene Chip
62 technology provides an innovative technique to monitor host cell transcriptome changes due to infection. Analyzing excised intestinal tissue after artificial infection will produce a holistic, yet very complicated, picture of host response. Epithelial cell cultures model the site of infection for many enteric pathogens. Results obtained from infection of an epithelial cell monolayer should reduce the complexity and be predictive of disease initiation. In a review of microarray research
(Cummings and Relman 2000), examples are provided to support this approach with various known pathogens. Advantages noted include rapid detection of pathogen exposure and using a single water sample to diagnose exposure to multiple disease agents.
In this work, a similar approach has been used to evaluate host cell exposure to A. hydrophila isolates. The intent was to demonstrate that the expression of specific host cell genes
(previously reported to be up- or downregulated in bacteria challenged host cells) can be similarly regulated in response to virulent A. hydrophila. An additional goal was to determine if the mRNA response in a mouse model of intestinal epithelial cells is comparable to the response seen in the whole animal small intestinal tissue.
MATERIALS and METODS
Aeromonas Strains. Aeromonas hydrophila isolate, EPA1 (40707D1), was obtained from a 2000-2001 water distribution system study conducted by the United States Environmental
Protection Agency (USEPA). Aeromonas hydrophila isolate, EPA2 (MC12723W), was a clinical isolate obtained from Dr. Amy Carnahan of the University of Maryland, Baltimore,
USA. Pure Aeromonas cultures were stored long-term in 15 % glycerol at -70ºC. High virulence of both strains was demonstrated in immunocompromised, adult Swiss Webster mice.
63 Phenotypic assays showed each strain was positive for elastase and lipase activity, hemolytic to
both sheep and rabbit erythrocytes, and cytotoxic to Vero cells (data not shown).
Suspensions of EPA1 and EPA2 were prepared by growing the cells in tryptic soy broth
(TSB) (Becton Dickinison, Franklin Lakes, NJ, USA) for 16 hours at 35ºC. This broth culture
was swabbed across the entire surface of a sheep blood agar (SBA) plate (Becton Dickinson,
Franklin Lakes, NJ, USA) and incubated for 5 hours at 37ºC. The resultant lawn of bacteria was
suspended in 20 mL of phosphate buffered saline (PBS) at pH 7.4 to give a concentration of 109
CFU/mL. In the tissue culture tests, Aeromonas suspensions were diluted to about 106 CFU/mL in PBS (1:1000 dilution). Viable cell counts were performed to confirm actual dosage levels.
The initial concentrations for bacterial cell suspensions harvested from the SBA plates were 1.9 x 109 and 1.8 x 109 for EPA1 and EPA2, respectively, for the tissue culture experiments and 6.5
x 109 and 6.3 x 109 for EPA1 and EPA2, respectively, for the neonatal experiments.
Controls consisted of either sterile PBS or combined UV and heat killed suspensions of
EPA1 (heating at 50oC for 20 minutes while exposed to UV light) (Model 11SC-1, Spectronics
Corp., Westbury, NY, USA). The loss of viability was confirmed by a culture assay.
Neonatal Mice. All animal experiments were performed under a protocol approved by the US EPA Animal Facility Oversight Committee. Timed pregnant Swiss Webster dams were received at 15 days gestation (Charles River Laboratories, Wilmington, MA, USA). Five neonates, aged 4-6 days, were intubated with 20µL of 109 CFU/mL suspension of either EPA1,
EPA2, sterile 1X PBS (as control), or UV/Heat killed suspension (as control) through a 100 µL
Hamilton syringe fit with a 24 gauge neonatal mouse intubation canula.
64 Neonates were euthanized by asphyxiation in CO2 chamber (w/ 5% O2) five hrs post-exposure
and necropsied, or necropsied within 0.5 hrs of death. Small intestines were excised and rinsed
in pre-chilled 1X PBS and then minced into sections no longer than 0.5 cm. The processed
intestines were stored at 4ºC in RNALater solution (Ambion, Austin, TX, USA).
Cell Cultures. In vitro experiments were performed using an intestinal murine cell line
(mICcl2) which have been derived from isolated crypts of small intestinal villi of a L-PK/Tag1
transgenic mouse (Bens et al. 1996). Previous studies have demonstrated that these cells
expressed many features of crypt epithelial cells from which they were derived (Peng et al. 1999,
Hornef et al. 2002, Luangsay et al. 2003). Cells were cultured in 25 cm2 canted, vented cell
culture flasks (Corning Inc., Corning, NY, USA) with 5 mL of Dulbecco’s Modified Eagle’s
Medium/Ham's Nutrient Mixture F12 (D-MEM/F12, GIBCO, Grand Island, NY, USA)
supplemented with 2% fetal bovine serum (FBS). Growth factors additions included (all from
Sigma, St. Louis, MO, USA) insulin (5µg/mL), dexamethasone (5X10-8M), selenium (60nM),
transferrin (5µg/mL), triiodothyronine (10-9M), EGF (10ng/mL), sodium bicarbonate (1.2 g/L),
D-glucose (22.4 mL/L of 10% solution) and antibiotics (100U penicillin/mL; 100µg
streptomycin/mL). Cells were maintained at 37ºC in a humidified incubator containing 7.0 %
CO2.
Prior to use, cells became 90-100% confluent. The cells were rinsed twice with 5 mL of
antibiotic-free and serum-free D-MEM/F-12 medium. The mono-layers were then refreshed
with 5 mL of antibiotic-free and serum-free D-MEM/F-12 medium. Control flasks (n=5) were
treated with 0.1 mL of sterile 1X PBS or killed cell suspension. Each experimental flask (n=5)
was inoculated with suspension of EPA1 or EPA2 at a ratio of approximately1:100 (bacteria:
host cells). Exposed and control flasks were centrifuged for 10 min. at 1000 rpm (Sorvall
Instruments, Model RC3B, Newtown, CT, USA). All flasks were incubated at 37ºC in 65 humidified air containing 7% CO2 for five hours. Following incubation, the medium in each flask
was decanted and samples extracted using TRIZOL™.
RNA Extraction/Isolation. All samples were extracted for RNA content within 7 days
with TRIZOL™ (Invitrogen, Carlsbad, CA, USA), as per manufacturer's instructions. Intestinal
samples were disrupted in TRIZOL™ using a ten second burst with a homogenizer (Fisher
Scientific, Model PowerGen700, Pittsburg, PA) but this was not necessary for the mICcl2 cells.
The RNA concentrations were determined spectrophotometrically (Perkin Elmer UV/VIS, Model
Lambda 20,Wellesley, MA, USA) at 260 nm (260/280 ratios also determined). Total RNA
samples were stored at -20ºC until purified using a glass-fiber procedure (RNAqueous, Ambion,
Austin, TX, USA). RNA quality was assessed for each sample using an Agilent 2100
bioanalyzer and associated RNA LabChip kits (Agilent Corp., Palo Alto, CA, USA).
Following RNAqueous purification, samples were stored at -20ºC.
Gene Chips. High quality double-stranded cDNA was created from total RNA using a
SuperScript™ Double-Stranded cDNA Synthesis kit (Invitrogen, Carlsbad, CA, USA) using
manufacturer’s protocol. Biotinylated cRNA targets were produced in vitro from cDNA using a
Enzo BioArray™ HighYield™ RNA Transcript labeling kit (Affymetrix, Santa Clara, CA, USA)
using manufacturer’s protocol. The cRNA targets were then hybridized to prefabricated Mouse
430A and B GeneChip probe arrays (Affymetrix, Santa Clara, CA, USA) and scanned according to the manufacturer’s protocol.
Microarray Data Processing and Analysis. To account for differences attributable to non-biological factors (e.g., sample processing that might lead to different starting concentrations of mRNA) data were normalized using a log scale, robust multi-array analysis (RMA) method
(Irizarry et al. 2003; reviewed by Saviozzi and Calogero, 2003). Normalization was performed
66 using the RMA module of an R-based program language, specific to Affymetrix applications,
downloaded from the Bioconductor web site (www.bioconductor.org).
Once data were normalized, mRNA intensities were assessed using significant analysis of
microarrays (SAM) (Version 1.21) obtained from Stanford University's web site (http://www-
stat.stanford.edu/~tibs/SAM/) (Tusher et al. 2001). The SAM program was used to select an
acceptable false discovery rate of 5% (FDR) for this study.
RESULTS and DISCUSSION
Up- and Downregulated Genes. The number of genes with at least 2-fold change in expression when compared to a phosphate buffered saline (PBS) control is shown in Table 1.
For this study, a 2-fold change in gene expression is considered significant. None of the UV, heat-killed controls showed any significant changes in gene expression. The larger number of significantly changed genes in neonates compared to the cell cultures was likely due to the use of the entire intestine for the analysis. However, there were 26 genes that were upregulated and two genes downregulated in both the neonates and the cell culture, after exposure to these virulent strains of A. hydrophila (Table 2). These 26 upregulated genes are considered as possible indicators of Aeromonas virulence.
The observation that a greater number of genes are upregulated compared to downregulated in response to exposure to a bacterial pathogen was also reported by Cohen et al.
(2000) using infection of promyelocyctic cells with Listeria moncytogenes. Their explanation
was that mRNA levels decrease from events such as repression of basal transcriptional
67 machinery and mRNA turnover, but these events are less likely to cause a large change in mRNA when compared to a positive regulatory event such as induction of transcription.
In the present study, the absolute number of up- and downregulated genes was similar when fold change was not a criterion (data not shown). However, when a fold change of greater than 2 was applied, most downregulated genes were discarded. If a criterion of greater than 3 fold change was applied, the downregulated genes almost completely disappear. As a modified t-test, SAM is sensitive to variability. Small fold changes between the control and experimental replicate average responses, combined with the variation inherent in true replicates (as were used in this research), causes downregulated genes with small fold changes to be considered as not significant by SAM. When Cohen assessed reproducibility using independent infections (i.e., true replicates) the percent genes commonly found between the replicates were 57 and 21% for up- and downregulated genes, respectively. If we consider EPA1 and EPA2 as replicate experiments (both being A. hydrophila, but acknowledging that they are not clonal), similar percentages are observed. These similar percentages are in light of the different microarray systems used (Affymetrix oligo arrays versus cDNA spotted arrays [200-500 bp]).
Reproducibility of this magnitude is encouraging considering the complexity of the assay involved.
Gene Expression Upregulated in Neonates and Intestinal Cell Culture. Table 2 lists the 26 genes whose expression was significantly changed in both the neonates and cell cultures.
Seven of these genes (Ccl2, Csf3, Socs3, Tnfaip3, Nfkbia, Cepbd, and Icam1) were also reported by Galindo et al. (2003) to be up-regulated in murine macrophages exposed to the purified
Aeromonas cytotoxic enterotoxin Act. A majority of these 26 upregulated genes have been associated with the innate immune response, e.g., cytokines and transcription factors that act to
68 regulate cytokine gene expression. In addition, four surface receptor molecule genes (Icam1,
Vcam1, CD14, Tlr2) were also up-regulated.
NF-кB is a transcription factor known to be a central regulator of the innate immune response to entero-invasive bacteria (Elewaut et al. 1999). The Nfkbia gene encodes one of three inhibitors which maintain NF-кB in an inactive state. If cells are stimulated by exposure to bacterial lipopolysaccharide (LPS), the complex bound to NF-кB is phosphorylated and subsequently degraded by proteases, allowing NF-кB to translocate to the nucleus. There the
NF-кB binds to a number of promoters inducing the transcription of a variety of genes including those coding for several cytokines and chemokines. Nfkbia is also upregulated as a mechanism to shut down the cytokine production before extensive cellular damage occurs.
CD14 and toll-like receptor 2 (Tlr2) are both known mediators of bacterial induced cellular signaling. Soluble CD14 (sCD14) has been shown to be up-regulated in LPS-stimulated human intestinal cells (Funda et al. 2001) and is thought to activate membrane CD14 (mCD14) positive cells (e.g., monocytes and macrophages). Toll-like receptor 4 (Tlr4) is the toll receptor known to bind LPS and activate the NF-кB pathway. Recent research has demonstrated that the
Tlr4-LPS complex is localized in the Golgi apparatus (Hornef et al. 2002, Hornef et al. 2003).
However, Tlr2 also mediates trans-membrane LPS signaling via the NF-кB pathway and is enhanced by CD14 (Yang et al. 1998). Tlr2 detects structural variants of LPS. In addition, LPS- binding protein (LBP), a required participant in the LPS/CD14 complex leading to NF-кB nuclear translocation, is produced by epithelial cells (Blas and Hiemstra 2004). This provides an explanation of Tlr2 being upregulated rather than Tlr4. Tlr4, the Toll-like receptor involved in the activation of LPS and localized in the Golgi apparatus, is not activated by A. hydrophila. In the membrane, Vcam1 causes recruitment of leucocytes to sites of infection and was induced in
69 immortalized murine small intestinal cells by LPS (Li et al. 1997). Also, Vcam-1 is a product of
NF-кB induced genes and upregulated in a number of inflammatory conditions of the gut (Jobin
and Sartor 2000).
Socs3 and Tnfaip3 are genes involved with cell signaling. For example, suppressor of
cytokine signaling 3 (Socs3) takes part in inhibiting the production of pro-inflammatory signals
and favors the expression of anti-inflammatory molecules (Berlato et al. 2002). Tumor necrosis factor alpha-induced protein 3 (Tnfaip3) is associated with negative feedback in cellular systems.
Tnfaip3 inhibits NF-кB activation and suggests a role for this gene in limiting inflammation by terminating NF-кB responses (Lee et al. 2000). In addition, Tnfpia3 has an anti-apoptotic effect by potently inhibiting NF-кB activation induced by tumor necrosis factor receptor 1 (TNFR1)
(He and Ting 2002).
The gene Icam1, part of the immunoglobulin super-family, is expressed on epithelial cell
surfaces and is upregulated in response to bacterial infections. For example, Icam-1 was up-
regulated in human colonic epithelial cells within 4-9 hours after bacterial infection and
produced an increase in adherence of neutrophils to epithelial cells (Huang et al. 1996).
Studies are underway comparing isogenic avirulent mutants of virulent A. hydrophila
strains, as well as virulent and avirulent A. caviae and A. veronii (biotype sobria) isolates.
Future studies focused on additional Aeromonas species and other pathogenic genera will
determine if the measurement of specific gene expression patterns might be useful in defining
potential virulence of environmental bacteria.
CONCLUSIONS
This study has identified a group of genes whose expression is up- or downregulated in
70 response to exposure to viable and virulent A. hydrophila bacteria. The results show that a subset of upregulated genes in a mouse intestinal cell line responds in a similar fashion to intestinal cells in a whole animal model. This is an important finding given that the use of whole animals in determining bacterial virulence is laborious and not suited to the screening of large numbers of isolates. Few similarities were seen in the downregulated genes in the two models.
Downregulated genes did not exhibit large fold changes in either model and therefore consistent differences are more difficult to distinguish statistically. Gene chip technology provides the basis for a more rapid process of identifying virulent A. hydrophila isolated from drinking water.
71 REFERENCES
Anguita, J., Aparicio, L.B.R. and Naharro, G. 1993 Purification, gene cloning, amino acid
sequence analysis, and expression of an extracellular lipase from an Aeromonas hydrophila human isolate. Appl. Environ. Micro. 59, 2411-2417.
Bals, R and Hiemstra, P.S. 2004 Innate immunity in the lung: how epithelial cells fight against respiratory pathogens. Eur. Respir. J. 23, 327-333.
Bens, M., Bogdanova, A., Cluzeaud, F., Miquerol, L., Kerneis, S., Kraehenbuhl, J.P., Kahn, A.,
Pringault, E. and Vandewalle, A. 1996 Transimmortalized mouse intestinal cells (m-ICc12) that maintain a crypt phenotype. Amer. J. Phys. 270, C1666-1674.
Berlato, C., Cassatella, M.A., Kinjyo, I., Gatto, L., Yoshimura, A. and Bazzoni, F. 2002
Involvement of suppressor of cytokine signaling-3 as a mediator of the inhibitory effects of IL-
10 on liposaccharide-induced macrophage activation. J. Immun. 168, 6404-6411.
Chopra, A.K., Peterson, J.W., Xu, X.-J., Coppenhaver, D.H. and Houston, C.W. 1996 Molecular
and biochemical characterization of a heat-labile cytotonic enterotoxin from Aeromonas
hydrophila. Microb. Path. 21, 357-377.
Cohen, P., Bouaboula, M., Bellis, M., Baron, V., Jblio, O., Poinot-Chazel, C., Galiègue, S.,
Hadibi, E-H. and Casellas, P. 2000 Monitoring cellular response to Listeria monocytogenes with
72 oligonucleotide arrays. J. Biol. Chem. 275, 11181-11190.
Elewaut, D., DiDonato, J.A., Kim, J.M., Truong, F., Eckmann, L. and Kagnoff, M.F. 1999 NF-
кB is a central regulator of the intestinal epithelial cell innate immune response induced by
infection with enteroinvasive bacteria. J. Immun. 163, 1457-1466.
Funda, D.P., Tučková, L., Farré, M.A., Iwase, T., Moro, I. and Tlaskalová-Hogenová, H. 2001
CD14 is expressed and released as soluble CD14 by human intestinal epithelial cells in vitro: lipopolysaccharide activation of epithelial cells revisited. Infect. Immun. 69, 3772-3781.
Galindo, C.L., Sha, J., Ribardo, D.A., Fadl, A.A., Pillai, L. and Chopra, A.K. 2003 Identification of Aeromonas hydrophila cytotoxic enterotoxin-induced genes in macrophages using microarrays. J. Biol. Chem. 278, 40198-40212.
He, K.-L. and Ting, A.T. 2002 A20 inhibits tumor necrosis factor (TNF) alpha-induced apoptosis by disrupting recruitment of TRADD and RIP to the TNF receptor 1 complex in Jurkat T cells.
Mol. Cell. Biol. 22, 6034-6045.
Hoffman, A., Levchenko, A., Scott, M.L. Baltimore, D. 2002 The Iк-NF-кB signaling module: temporal control and selective gene activation. Science. 298, 1241-1245.
Hornef, M.W., Frisan, T., Vandewalle, A., Normark, S. and Richter-Dahlfors, A. 2002 Toll-like
receptor 4 resides in the Golgi apparatus and colocalizes with internalized lipopolysaccharide in
73 intestinal epithelial cells. J. Exp. Med. 195, 559-570.
Hornef, M.W., Normark, B.H., Vandewalle, A., and Normark, S. 2003 Intracellular recognition of lipopolysaccharide by Toll-like receptor 4 in intestinal epithelial cells. J. Exp. Med. 198,
1225-1235.
Huang, G.T.-J., Eckmann, L., Savidge, T.C. and Kangnoff, M.F. 1996 Infection of human intestinal epithelial cells with invasive bacteria upregulates apical intercellular adhesion molecile-1 (ICAM-1) expression and neutrophil adhesion. J. Clin. Invest. 98, 572-583.
Irizarry, R. A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B. and Speed, T.P. 2003
Summaries of Affymetrix GeneChip probe level data. Nuc. Acids Res. 31, e15
Janda, J.M. 1991 Recent advances in the study of taxonomy, pathogenicity, and infectious syndromes associated with the genus Aeromonas. Clin. Micro. Rev. 4, 397-410.
Janda, J.M. and Abbott, S.L. 1998 Evolving concepts regarding the genus Aeromonas: an
expanding panorama of species, disease presentations, and unanswered questions. Clin. Infect.
Dis. 27, 332-344.
Jobin, C. and Sartor, R.B. 2000 The IкB/NF-кB system: a key determinant of mucosal
inflammation and protection. Am. J. Physiol. Cell Physiol. 278, C451-C462.
74 Kampik, D., Schulte, R. and Autenrieth, I.B. 2000 Yersinia enterocolitica invasin protein triggers
differential production of interleukin-1, interleukin-8, monocyte chemoattractant protein 1,
granulocyte-macrophage colony-stimulating factor, and tumor necrosis factor alpha in epithelial
cells: implications for understanding the early cytokine network in Yersinia infections. Infec.
Immun. 68, 2484-2492.
Kitamura, H., Kanehira, K., Okita, K., Morimatsu, M. and Saito, M. 2000 MAIL, a novel nuclear
IкB protein that potentiates LPS-induced IL-6 production. FEBS Ltrs. 485, 53-56.
Luangsay, S., Kasper, L.H., Rachinel, N., Minns, L.A., Mennechet, F.J.D., Vandewalle, A. and
Buzoni-Gatel, D. 2003 CCR5 mediates specific migration of Toxoplasma gondii-primed CD8+ lymphocytes to inflammatory intestinal epithelial cells. Gastroenterology. 125, 491-500.
Lee, E.G., Boone, D.L., Chai, S., Libby, S.L., Chien, M., Lodolce, J.P. and Ma, A. 2000 Failure to regulate TNF-induced NF-кB and cell death responses in A20-deficient mice. Science. 289,
2350-2354.
Li, X.C., Jevnikar, A.M. and Grant, D.R. 1997 Expression of functional ICAM-1 and VCAM-1 adhesion molecules by an immortalized epithelial cell clone derived from the small intestine.
Cell. Immun. 175, 58-66.
Merino, S., Aguilar, A., Nogueras, M.M., Regue, M., Swift, S. and Tomás, J.M. 1999 Cloning, sequencing, and role in virulence of two phospholipases (A1 and C) from mesophilic Aeromonas
75 spp. serogroup O:34. Infec. Immun. 67, 4008-4013.
Matussek, A., Lauber, J., Bergau, A., Hansen, W., Rohde, M., Dittmar, K.E.J., Gunzer, M.,
Mengel, M., Gatzlaff, P., Hartmann, M., Buer, J. and Gunzer, F. 2003 Molecular and functional analysis of Shiga toxin-induced response patterns in human vascular endothelial cells. Blood.
102, 1323-1332.
Peng, K-C., Cluzeaud, F., Bens, M., Van Huyen, J-P.D., Wioland, M.A., Lacave, R. and
Vandewalle, A. 1999 Tissue and cell disruption of the multidrug resistance-associated protein
(MRP) in mouse intestine and kidney. J. Histochem. Cytochem. 47, 747-767.
Saviozzi, S. and Calogero, R.A. 2003 Microarray probe expression measures, data normalization and statistical validation. Comp. Func. Genom. 4, 442-446.
Schivano, G. and van der Goot, F.G. 2001 The bacterial toxin toolkit. Nat. Rev. 2, 530-537.
Sha, J., Kozlova, E.V. and Chopra, A.K. 2002 Role of various enterotoxins in Aeromonas hydrophila-induced gastroenteritis: generation of enterotoxin gene-deficient mutants and evaluation of their enterotoxic activity. Infec. Immun. 70, 1924-1935.
Thorpe, C.M., Smith, W.E., Hurley, B.P. and Acheson, D.W. 2001 Shiga toxins induce, superinduce, and stabilize a variety of C-X-C chemokine mRNAs in intestinal epithelial cells, resulting in increased chemokine expression. Infec. Immun. 69, 6140-6147.
76
Tusher, V.G., Tibshirani, R. and Chu, G. 2001 Significance analysis of microarrays applied to
the ionizing radiation response. Proc. Nat. Acad. Sci. 98, 5116-5121.
Wong, C.Y.F., Heuzenroeder, M.W. and Flower, R.L.P. 1998 Inactivation of two haemolytic
toxin genes in Aeromonas hydrophila attenuates virulence in a suckling mouse model.
Microbiol. 144, 291-298.
Yang, R-B., Mark, M.R., Gray, A., Huang, A., Xie, M.H., Zhang, M., Goddard, A., Wood, W.I.,
Gurney, A.L. and Godowski, P.J. 1998 Toll-like receptor-2 mediates lipopolysaccharide-induced cellular signaling. Nature. 395, 284-288.
77 Table 1. Numbers of genes up- and downregulated from A. hydrophila infection
Control, Common Common to both UV, EPA1 EPA2 genes, EPA1 animal and cell heat- and EPA2 culture killed Number of upregulated 427 316 239 genes in animal model 0 26 Number of upregulated 58 58 37 genes in cell culture 0
Number of downregulated genes in 0 97 221 46 animal model 2 Number of downregulated genes in 0 245 125 72 cell culture
78
Table 2. Common genes up-regulated and down-regulated in cell culture and neonates. Fold changes for EPA1 are listed first. EPA2 are presented in parentheses. F.C. cell F.C. Probe Set ID Gene Title Gene Symbol Public ID culture neonates Signaling molecules chemokine (C-X-C motif) 1418930_at Cxcl10a NM_021274 4.5 (4.8) 8.3 (9.6) ligand 10
1419209_at NM_008176 10 (7.3) 67 (55) chemokine (C-X-C motif) 1441855_x_at Cxcl1b BB554288 5.0 (3.5) 3.3 (3.4) ligand 1 1457644_s_at BB554288 9.6 (6.7) 16 (18)
chemokine (C-X-C motif) 1419728_at Cxcl5c NM_009141 10 (20) 13 (4.0) ligand 5
chemokine (C-C motif) ligand 1420380_at Ccl2d AF065933 9.2 (4.0) 6.1 (5.4) 2
chemokine (C-C motif) ligand 1422029_at Ccl20e AF099052 3.9 28 (12) 20
chemokine (C-X-C motif) 1449984_at Cxc12f NM_009140 34 (13) 49 (50) ligand 2
colony stimulating factor 3 1419427_at Csf3 NM_009971 4.6 (2.3) 8.0 (7.0) (granulocyte)
colony stimulating factor 1 1460220_a_at Csf1 BM233698 3.7 (2.1) (2.4) (macrophage)
1455899_x_at suppressor of cytokine BB241535 2.3 (2.2) 14 (13) Socs3 1456212_x_at signaling 3 BB831725 2.0 13 (10)
1450829_at tumor necrosis factor, alpha- NM_009397 3.1 6.3 (4.2) Tnfaip3 1433699_at induced protein 3 BM241351 8.0 (3.6) 18 (14) Transcription factors/enhancers/regulators 1420088_at Al462015 3.4 (2.8) 4.0 (5.2) 1420089_at nuclear factor of kappa light Al462015 4.0 (3.0) 2.8 (4.1) 1448306_at chain gene enhancer in B-cells Nfkbia NM_010907 8.9 (4.8) 13 (10) 1449731_s_at inhibitor, alpha Al462015 8.1 (5.0) 8.8 (10) 1438157_s_at BB096843 6.5 (4.6) 8.0 (8.7)
79 nuclear factor of kappa light 1458299_s_at polypeptide gene enhancer in Nfkbie BB820441 9.3 (6.2) 3.5 (4.4) B-cells inhibitor, epsilon
1417483_at expressed sequence AB026551 4.0 (3.1) 10 (8.9) AA408868 1448728_a_at AA408868 AB026551 4.2 (2.3) 7.2 (4.6)
CCAAT/enhancer binding 1423233_at Cebpd BB831146 6.2 (5.6) 12 (10) protein (C/EBP), delta
cold inducible RNA binding 1416332_at Cirbp NM_007705 2.1 (2.2) 2.5 protein
1418133_at B-cell leukemia/lymphoma 3 Bcl3 NM_033601 2.8 (2.6) 5.4 (4.5)
1416916_at E74-like factor 3 Elf3 NM_007921 2.7 3.2 (2.5) Surface receptor molecules 1417268_at CD14 antigen Cd14 NM_009841 2.1 (2.1) 3.9 (3.6)
1419132_at toll-like receptor 2 Tlr2 NM_011905 2.1 3.0 (2.6)
intercellular adhesion 1424067_at Icam1 BC008626 2.2 10 (9.1) molecule
vascular cell adhesion 1448162_at Vcam1 BB250384 3.7 (2.3) 2.1 molecule 1 Miscellaneous upregulated genes 1427348_at cDNA sequence BC036563 BC036563 BC006817 3.0 (2.2) 8.1 (2.5)
1427747_a_at lipocalin 2 Lcn2 X14607 2.4 (2.6) 10 (3.1)
guanine nucleotide binding 1430295_at Gna13 BG094302 (2.0) 2.2 protein, alpha 13
1451924_a_at endothelin 1 Edn1 D43775 2.1 (2.3)
1455197_at Rho family GTPase 1 Rnd1 BE852181 3.6 3.9 (5.6) a-interferon activating gene-10 b-GRO1 oncogene (neutrophil specific) c-GCP-2, granulocyte chemotactic protein-2 d-MCP-1 (monocyte specific) e-MIP-3a, macrophage inflammatory protein-3, chemoattractant for T- and B-cells f-MIP-2, Gro2 80
Accepted for publication in Journal of Applied Microbiology
Chapter 2
Evaluating virulence of waterborne and clinical Aeromonas isolates using gene expression
and mortality in neonatal mice followed by assessing cell culture's ability to predict
virulence based on transcriptional response
81 Abstract
Aims: To assess the virulence of Aeromonas spp. using two models, a neonatal mouse assay and
a mouse intestinal cell culture.
Methods and Results: After artificial infection with a variety of Aeromonas spp., mRNA
extracts from the two models were processed and hybridized to murine microarrays to determine
host gene response. Definition of virulence was determined based on host mRNA production in
murine neonatal intestinal tissue and mortality of infected animals. Infections of mouse
intestinal cell cultures were then performed to determine whether this simpler model system's
mRNA responses correlated to neonatal results and therefore be predictive of virulence of
Aeromonas spp. Virulent aeromonads up-regulated transcripts in both models including multiple
host defense gene products (chemokines, regulation of transcription and apoptosis, cell
signaling). Avirulent species exhibited little or no host response in neonates. Mortality results
correlated well with both bacterial dose and average fold change of up-regulated transcripts in
the neonatal mice.
Conclusions: Cell culture results were less discriminating but showed promise as potentially being able to be predictive of virulence. Jun oncogene up-regulation in murine cell culture is potentially predictive of Aeromonas virulence.
Significance and Impact of the Study: Having the ability to determine virulence of waterborne pathogens quickly would potentially assist public health officials to rapidly assess exposure risks.
Keywords. Aeromonas; Virulence; Gene expression; Host response
82
INTRODUCTION
Currently there are 17 recognized Aeromonas species (Martin-Carnahan and Joseph
2005). Isolates of Aeromonas can cause infections of mammals, birds and fish (Smith and
Cheasty 1998). In humans, the most common clinical infection is gastroenteritis but aeromonads
have been implicated in septicemia, meningitis, peritonitis, wound infections, respiratory disease
and ocular infections (Szewzyk et al. 2000). The source of the bacteria causing these infections
is not clear but in some cases drinking water is suspected. A method is needed to quickly
evaluate the virulence potential of an Aeromonas isolate.
Intra-peritoneal (i.p.) injection of mice with bacteria has been found to be useful in
establishing the virulence of some bacteria, especially when combined with immune function
modulation and evaluated using the LD50 (Stelma et al. 1992, Stelma et al.1987). However,
current animal experimental guidelines make this an undesirable endpoint to describe virulence.
An animal model used to evaluate the virulence of Aeromonas isolates is gastric lavage of neonatal mice (Wong et al. 1996). However, the endpoint in this study was again the LD50.
While bioassays using live animals are still the most reliable means for identifying the spectrum
of virulent strains of known opportunistic pathogens and for detecting previously unknown
opportunistic pathogens, development of a model that does not rely on the use of live animals is
desirable. Analyzing excised intestinal tissue after artificial infection produces a holistic, yet
very complicated, picture of host response. Cell culture offers an alternative and has been used
in previous research to study bacterial virulence factor effects (Galindo et al. 2004, Xia et al.
2003, Nagasako et al. 2003, Belcher et al. 2000, Eckmann et al. 2000). Cell cultures are simpler test systems and potentially reduce the amount of variation seen in animal models.
83 The primary goal of the current study was to demonstrate that the intestinal tissue of
infected neonatal mouse and the murine intestinal cell line mICcl2 share in common a set of
up-regulated transcripts after inoculation with virulent strains of Aeromonas spp., which are
currently recovered from environmental or clinical samples. In one model, neonatal mice were
infected with Aeromonas strains followed by transcriptional analysis of excised intestinal tissue
via microarrays. In addition, neonatal mortality was considered when assessing virulence. The
other model was a murine intestinal epithelial cell monolayer (mICcl2) and its response
(transcriptional changes) to Aeromonas inoculation. Because epithelial cell layers are the first
defense against an invading pathogen, transcriptional profiles from infected epithelial cell
monolayers were predicted to be able to define virulence.
Thus, the ultimate goal would be to define a subset of host cell transcripts from cell
culture infections rather than using whole genome microarrays to define virulence. This would
eventually lead to a rapid and reproducible method for determining virulence of Aeromonas spp. detected in water systems.
MATERIALS and METHODS
Bacterial Isolates and Culture Conditions. Aeromonas spp. used in this study are described in Table 1. These isolates are maintained in EPA-Cincinnati laboratories. For exposure studies, all isolates were grown overnight at 35oC in tryptic soy broth (TSB) (Becton,
Dickinson and Co., Sparks, MD, USA) before being spread onto the surface of 5 % sheep’s
blood agar (SBA) (Becton, Dickinson and Co., Sparks, MD, USA) and then incubated for 5 hr at
35oC. The bacteria were harvested from the SBA surface and resuspended in 20 mL of sterile
phosphate buffered saline (PBS, pH 7.0). Dilutions were prepared in PBS to give a final
84 concentration of approximately 1 x 109 colony forming units (CFU) per 1 mL of suspension.
Viable cell counts were performed to confirm actual dosage levels. The bacterial suspensions
were used within 2 hr of preparation. The Aeromonas spp. tested represented those found
frequently from an USEPA occurrence survey (Sen and Rodgers 2004), other clinical isolates,
and reference ATCC strains.
Virulence Assay. All animal testing was performed in an AAALAC accredited facility
under an approved protocol by an Institutional Animal Care and Use Committee (IACUC).
Timed-pregnant Swiss Webster dams (Charles River Laboratories, Wilmington, MA, USA) were
received at 15 days gestation. Five neonates, aged 4-6 days, were intubated through a 100 µL
Hamilton syringe fit with a 24 gauge neonatal mouse intubation canula with 20 µL of 109
CFU/mL suspension (dose of approximately 107 total organisms/animals) of either: selected
Aeromonas strains; sterile 1X PBS (as control); or an ultraviolet (UV)-heat killed suspension of a
virulent strain (50oC for 20 min while exposed to UV light source, Model 11SC-1, Spectronics
Corp., Westbury, NY, USA). Loss of viability of ultraviolet (UV)-heat killed organisms was confirmed by lack of growth on tryptic soy agar plates. A decision was made to assess virulence using one time point at 5 hrs because intestinal cellular response to pathogens is known to be rapid (Eckmann et al. 1993, Bohn et al. 2003). A five hour time period allowed the performance
of infections and subsequent harvesting of samples during one work day as well as allowing
sufficient time for measurable host response. Additional experiments using higher doses of
avirulent Aeromonas spp. were performed based on preliminary results.
Neonates were monitored every hour for mortality. Neonates were euthanized by
asphyxiation in CO2 chamber (with 5% O2) 5 hrs post-exposure and necropsied, or necropsied
within 0.5 hrs of death. Small intestines were excised and rinsed in pre-chilled 1X PBS and then
85 minced into sections no longer than 0.5 cm and stored at 4ºC in RNALater solution (Ambion,
Austin, TX, USA). Five biological replicates were used in data analysis for unexposed controls
and infection experiments.
Cell Cultures. In vitro experiments were performed using an intestinal murine cell line
(mICcl2) derived from isolated crypts of small intestinal villi of a L-PK/Tag1 transgenic mouse
(Bens et al. 1996). Previous studies have demonstrated that these cells expressed many features
of crypt epithelial cells from which they were derived (Hornef et al. 2002, Luangsay et al. 2003).
Cells were cultured in 25 cm2 canted, vented cell culture flasks (Corning Inc., Corning, NY,
USA) with 5 mL of Dulbecco’s Modified Eagle’s Medium/Ham's Nutrient Mixture F12 (D-
MEM/F12, GIBCO, Grand Island, NY, USA) supplemented with 2% fetal bovine serum (FBS).
Growth factors additions included (all from Sigma, St. Louis, MO, USA) insulin (5µg/mL), dexamethasone (5X10-8M), selenium (60nM), transferrin (5µg/mL), triiodothyronine (10-9M),
EGF (10ng/mL), sodium bicarbonate (1.2 g/L), D-glucose (22.4 mL/L of 10% solution) and
antibiotics (100U penicillin/mL; 100µg streptomycin/mL). Cells were maintained at 37ºC in a
humidified incubator containing 7 % CO2.
Prior to use, cells became 90-100% confluent. The cells were rinsed twice with 5 mL of
antibiotic-free and serum-free D-MEM/F-12 medium. The mono-layers were then refreshed
with 5 mL of antibiotic-free and serum-free D-MEM/F-12 medium. Control flasks (n=5) were
treated with 0.1 mL of sterile 1X PBS or killed cell suspension. Each experimental flask (n=5)
was inoculated with a bacterial suspension at a ratio of approximately 1:100 (bacteria: host
cells). Exposed and control flasks were centrifuged for 10 min. at 1000 rpm (Sorvall
Instruments, Model RC3B, Newtown, CT, USA). All flasks were incubated at 37ºC in
humidified air containing 7% CO2 for five hours.
86 Seven Aeromonas strains were selected for repeat testing in mICcl2 infections, EPA designations 1, 9, 61, 63, 71, 73 and 100. This subset reflected three virulent strains and four avirulent strains as determined by neonatal testing. This group represented two strains from each of the three species known to cause human gastroenteritis as well as the Aer. allosaccharophila
strain that was weakly virulent.
RNA Extraction and Isolation. All samples were extracted for total RNA with
TRIZOL™ (Invitrogen, Carlsbad, CA, USA), as per manufacturer's instructions. Intestinal
samples stored in RNALater were disrupted in TRIZOL™ using a ten second burst with a
homogenizer (Fisher Scientific, Model PowerGen700, Pittsburg, PA) and processed within seven
days of excision. Following incubation of cell cultures, the medium in each flask was decanted
and samples extracted immediately using TRIZOL™. RNA concentrations were determined
spectrophotometrically (Perkin Elmer UV/VIS, Model Lambda 20, Wellesley, MA, USA) at 260
nm and 260/280 ratios were also determined. Total RNA samples were stored at -20ºC until
further purified using a glass-fiber procedure (RNAqueous, Ambion, Austin, TX, USA). Further
purification is required to obtain pristine RNA for use in microarray testing. RNA quality was
assessed for each sample using an Agilent 2100 bioanalyzer and associated RNA LabChip
kits (Agilent Corp., Palo Alto, CA, USA). Following RNAqueous purification, samples
were stored at -20ºC.
Microarray Hybridization and Analysis. High quality double-stranded cDNA was
created from total RNA using a SuperScript™ Double-Stranded cDNA Synthesis kit (Invitrogen,
Carlsbad, CA, USA) using manufacturer’s protocol. Biotinylated cRNA targets were produced
in vitro from cDNA using an Enzo BioArray™ HighYield™ RNA Transcript labeling kit
(Affymetrix, Santa Clara, CA, USA) using manufacturer’s protocol. The cRNA targets were
87 then hybridized to prefabricated Mouse 430_2, 430A and B GeneChip probe arrays (Affymetrix,
Santa Clara, CA, USA) and scanned according to the manufacturer’s protocol. GeneChip arrays
were scanned using an Agilent GeneArray 2500® or GeneChip3000® scanners (both supplied
by Affymetrix, Santa Clara, CA, USA) and raw data images analyzed using Microarray Suite
(MAS) 5.0 software or GCOS software. Default values for signal processing were as follows:
target signal scaling 500; normalization 1; Alpha1 0.05; Alpha2 0.065; Tau 0.015; Gamma1L
and 1H 0.0045; Gamma 2L and 2H 0.06; and Pertubation 1.1. Samples were deposited in the
Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/projects/geo/) under series number
GSE6065.
qRT-PCR single tube gene expression assay. Single tube gene expression assays were
pre-designed TaqMan® Gene Expression Assays (Applied Biosystems, Foster City, CA, USA).
The assays are a single tube format using TaqMan® Universal PCR Master Mix without
AmpErase® UNG. The cDNA sample was prepared as described above. Universal cycling
conditions were used (1 cycle of 10 minutes at 95°C and 45 cycles of 15 seconds at 95°C, 15
seconds at 60°C, and 45 seconds at 72°C). The samples were analyzed on the ABI PRISM®
7000 Sequence Detection Systems (Applied Biosystems, Foster City, CA, USA).
Microarray Data Processing and Analysis. To account for differences attributable to non-biological factors (e.g., sample processing that might lead to different starting concentrations of mRNA) data were normalized on the log scale using robust multi-array analysis (RMA) methods (Irizarry et al. 2003 and reviewed in Saviozzi and Calogero, 2003). Normalization was performed using the RMA module of Bioconductor specific to Affymetrix applications
(www.bioconductor.org).
88 The LIMMA (Linear Models for Microarray Data) module of Bioconductor was used for the analysis of the normalized gene expression data. LIMMA treats the expression intensity of each spot on a microarray as the dependent variable in a linear model, and the treatment condition of each microarray as the independent factor variables for a designed experiment.
Inference of differential gene expression is made using the fitted model coefficients using an empirical Bayes approach which moderates the standard errors of the estimated fold changes
(FCs) on the log scale. A q-statistic of a probe set estimates the conditional probability of any false positive if the said probe set was flagged as differentially expressed. For this study, probe sets with a q-value less than or equal to 0.05 were retained.
RESULTS
Table 2 presents a summary of doses, average FCs of up-regulated transcripts and mortality data from Aeromonas infections tested in the neonatal mouse model. Virulence of each isolate at each particular dose was assessed on the basis of its ability to kill challenged mice and also by response of mouse intestinal tissue to infection. Thus, we defined as “virulent” those strains that were able to kill some or all challenged mice and also produced an average FC of up- regulated transcripts in mouse intestinal cells greater than or equal to 3, at challenged doses of
107-108 CFU/mice. This target dose of 107-108 CFU/mice was chosen according other authors
(Wong et al. 1996) as well as on the basis of preliminary testing (dose response curves on two
Aer. hydrophila strains, data not shown). EPA strains 2, 5, 72, 73 and 100 were tested at higher doses to determine if transcriptional responses and/or mortality responses would be affected and thereby cause a virulent designation. EPA 2 and 5 were tested using doses of 1.3E+8 and
2.5E+8, respectively, and the latter three strains were tested at concentrations that exceeded
89 1.0E+9 organisms. At the higher doses, EPA 2 and 5 both demonstrated mortality and average
FCs of 4.5 and 3.5, respectively. Based on this, EPA 2 and 5 were considered as virulent at these
higher doses. For EPA 72 results at the higher dose showed no mortality and average FC
remained less than three. For EPA73, some mortality was seen and the average FC approached
three. Therefore, at this high dose EPA73 approached being designated as virulent. At the high
dose, EPA100 would be considered as virulent, demonstrating both mortality and also an average
FC of 3.1. Host cell response with higher doses of EPA72, 73 and 100 showed little association
with host immune response in terms of NF-кB activation and chemokine production. These Aer. caviae and Aer. allosaccharophila strains did appear to up-regulate cytoskeletal associated transcripts as well a defensin gene product.
The repeated testing of EPA61 (Aer. veronii biotype sobria strain) at a slightly higher dose (approx. 0.6 log CFU greater than initial testing) did produce mortality in more than one neonate but no difference in average FC. Therefore, this isolate is considered as avirulent at both doses, even though some mortality was observed. Killed Aeromonas did not cause a host response either in cell culture or in neonates.
Table 3 summarizes the individual FC of transcripts up-regulated at least 2 times or greater in the neonatal mouse model. Transcripts with low FCs (<3 FC) and found in infection experiments with only one species were not included. Whenever strains were tested at higher doses (see above), the FC values in Table 3 are reflective of these experiments. The up-regulated transcripts are grouped by known or suspected biological function as defined by gene ontology
(GO) classifications. Classifications that are associated with tissue damage, cell signaling events adhesion, cell death, actin re-arrangement or defense response are included. For example, angiogenesis is a normal process in growth and development, as well as in wound healing. The
90 "regulation of transcription" classification is relevant because infection is viewed as a positive regulatory event and many transcripts listed are known to be transcription factors for immune response functions. Actin/cytoskeleton re-arrangement is known to occur in response to bacterial adhesion. The strains designated as virulent are represented in most GO categories, while avirulent strains are not. The avirulent strains do appear to stimulate defensins (response to biotic stimulus/stress), transcripts integral to the membrane, as well as actin/cytoskeleton transcripts.
The ultimate goal was to define a subset of host cell transcripts from cell culture infections rather than using whole genome microarrays to define virulence. Therefore, a stepwise procedure identified up-regulated transcripts that were unique and in common to
virulent infections of both models and thus predictive of virulence. For mICcl2 infection experiments, EPA 1, 9 and 63 were designated as "virulent" strains with others designated as
"avirulent". Table 2 indicates EPA100 as virulent, but this was at the higher dose. The first step was to identify up-regulated transcripts in each model that were unique to the virulent infections.
This produced two sets of data, specifically, those up-regulated transcripts in each the neonate and cell culture model that were unique to infections using virulent strains.
Table 4 presents a summary of the total up-regulated transcripts in the neonatal infections and in cell line mICc12 after inoculation with Aeromonas and the number of up-regulated transcripts that are unique to infections using virulent strains. The total up-regulated column reflects a summation of all transcripts noted as having a FC of greater than 2 from infections of all strains tested using that particular model. The next column represents the number of transcripts, of the total, that were unique to virulent strain infections. The last column indicates the percentage of transcripts that are specific to virulent strain infections. The table indicates that
91 cell culture was less discriminating than the neonatal model for identifying transcripts useful for
defining virulence. An intersection of the two unique, virulent data sets from each model
produced a listing of 15 transcripts (identified also within Table 3). Only two of these 15
transcripts, Jun and Fos, were found as being common to most virulent strain infections. The
remaining 13 transcripts could only be attributed to one of the virulent infections in either model.
These two transcripts, along with two housekeeping genes (Pgk1 and Trfp) were tested
using qRT-PCR. The selected housekeeping genes (Trfp and Pgk1) demonstrated consistent
response in the microarray results when comparing controls and treatment exposures and were
taken from a candidate list provided by the qRT-PCR manufacturer. Adjusting gene expression
profiles to housekeeping gene expression provided no advantage over using the unadjusted data.
For Jun, the average ∆CT values for virulent and avirulent strains were 2.1 (+1.1) and 0.6 (+0.4),
respectively. For Fos, the average ∆CT values for virulent and avirulent strains were 2.4 (+1.6) and 1.7 (+1.3), respectively. The calculated p-values for Welch’s 2-sample T-tests (equal
variance assumption relaxed) on differences in CT cycle between gene exposure to virulent vs.
avirulent exposure were 0.009 and 0.310 for Jun and Fos, respectively. Therefore, the single
tube gene expression assays performed using qRT-PCR indicate virulent strain infections cause
up-regulation of Jun in mICcl2 cultures more so than do avirulent strains after a five hour
incubation time period.
DISCUSSION
Animal models are still the gold standard for bacterial virulence testing. We are
attempting to correlate transcriptional responses seen from infecting an animal model with a less
complex and more ethically acceptable cell culture infection model. Intestinal epithelial cells
92 were chosen for in vitro testing because they represent the initial site of host defense and their
response to bacterial intestinal infection has been characterized in past publications (Eckmann et al. 1993, Eckmann et al. 2000, Strober 1998). Besides Aer. hydrophila, the other two
Aeromonas species implicated in gastrointestinal disease are Aer. veronii biotype sobria and
Aer. caviae. While Aer. caviae has been implicated as a cause of waterborne disease (Sinha et
al, 2004, Demarta et al. 2000, Kühn et al. 1997, Havelaar et al. 1992, Moyer et al. 1992) it has
been largely accepted as an opportunistic pathogen (Moyer 1987). The rationale for testing Aer.
salmonicida strains was because we found atypical isolates that grew well at 37oC. Aer.
encheleia was included because it was a common drinking water isolate from the USEPA
survey, was positive for the presence of the Act cytotoxin gene (Sen and Rodgers 2004) and is
beta-hemolytic on sheep blood agar. These latter two characteristics were also common to the
Aer. salmonicida isolates. Aer. allosaccharophila was added because little is known regarding
its ability to cause disease in warm-blooded animals.
Neonate results (Table 2) demonstrate a distinction between virulent and avirulent
species in most instances. At doses of approximately107 - 108 organisms, the average FC and
mortality results were in general agreement. Inoculation of neonates with Aer. hydrophila and
Aer. veronii biotype sobria strains listed in Table 1 commonly caused substantial mortality. Our testing supports the assertion that these two species contain strains that are virulent in mammalian systems.
It was interesting to observe a relatively weak immune host response from the intestinal tissues of the animals dosed with high levels of Aer. caviae and Aer. allosaccharophila. The increased mortality noted may have been from a more generalized and systemic reaction to endotoxins, rather than a localized response from the intestines. The increased mortality seen at
93 the higher dose for EPA61 is likely not the same as for Aer. caviae and Aer. allosaccharophila.
Possible explanations are variability in neonate robustness due to maternal care for this particular
cage or replicate variability within GeneChip arrays that precluded detection of statistical
significance. This research appears to indicate that Aer. caviae is likely an opportunistic
pathogen because it does not induce significant host response in neonatal mice at the same dose
as other known human pathogens. Aer. allosaccharophila has potential to be a pathogen (or
opportune pathogen) at high doses.
Based on results of the heat/UV-killed organisms, Aeromonas need to be metabolically
active to induce a host response. Other gram negative organisms such as Salmonella,
Helicobacter and Chlamydia also demonstrate the need to be metabolically active (Kampik et al.
2000) to invoke a immune response. There is no apparent passive attachment from receptor
ligand interactions with Aeromonas spp.
Previous work by this group introduced the notion that cell culture might be useful for predicting virulence of Aer. hydrophila. (Hayes et al. 2006). This would be desirable because of the simplicity of the test system and the need to develop models other than live animals for predicting virulence. In our initial report, host cell cultures and intestinal tissues were assessed after exposure to two virulent Aer. hydrophila strains. Aer. hydrophila caused the up-regulation of more than 200 genes in neonates and over 50 genes in cell culture. Twenty-six genes were found to be in common between the two models. The majority of these genes were associated with the innate immune response. The variety of species/strains used in the current effort, as well as the attempt to classify them as being virulent or avirulent, increased the complexity of the analysis. Because each Aeromonas strain potentially causes a different host response, the list of up-regulated transcripts grew larger when compared to our first effort. For this same reason, the
94 list of commonly up-regulated genes between our virulent and avirulent designations was small.
Ultimately, analysis of a subset of mRNA targets would be desirable rather that needing to run
an entire microarray assay. The qRT-PCR results did support the use Jun as a predictor of
virulence. It is not unreasonable to assume that many host responses to infection would cause
the up-regulation of transcription factors common to pathways associated with apoptosis and/or
cytokine production. Jun is a central transcription factor in the cellular apoptosis pathway as
well as being integral to AP-1 transcription factor complexes for cytokine activation. The qRT-
PCR results did support the use Jun as a predictor of virulence as its up-regulation was specific
to virulent infections in both models and in both assays (microarray and qRT-PCR). The
significant difference of Jun up-regulation is a potential indicator for differentiating Aeromonas spp. in terms of virulence.
A more robust bioinformatics approach might identify other useful transcripts and data analysis of the current data set is on-going. Smaller FCs may require more replicates to ensure validity of significance. Some smaller FCs are more significant than larger ones because level of transcriptional change is not always correlated with biological activity. Temporal changes also need to be investigated more thoroughly because of the likelihood that virulent species cause faster host response due to toxins or other virulence factors not found in less virulent species.
Kinetic curves for key host response transcripts might better characterize virulence.
The original hypothesis that cell culture might be able to replace animal testing as a test model for predicting Aeromonas virulence was not fully realized. The limitation is potentially on our ability to interpret the massive database. Drawing correct inferences is contingent on bioinformational techniques and on variation of test systems. Other cell types may be better
95 predictors rather than the intestinal cell line used in this study. Current and continuing research efforts will go forward to fine tune both aspects.
ACKNOWLEDGEMENT
We would like to acknowledge Alain Vandewalle for the gift of the m-ICcl2 cell line, present address INSERM, U773, Centre de Recherche Biomédicale Bichat-Beaujon, UFR de Médecine,
Site Bichat, Paris, Fr.
96 REFERENCES
Belcher, C.E., Drenkow, J., Kehoe, B., Gingeras, T.R., McNamara, N., Lemjabbar, H., et al.
(2000) From the cover: the transcriptional responses of respiratory epithelial cells to Bordetella pertusis reveal host defensive and pathogen counter-defensive strategies. Proceedings National
Academy of Sciences USA 97, 13847-13852.
Bens, M., Bogdanova, A., Cluzeaud, F., Miquerol, L., Kerneis, S., Kraehenbuhl, J.P., Kahn, A.,
Pringault, E. and Vandewalle, A. (1996) Transimmortalized mouse intestinal cells (m-ICc12) that maintain a crypt phenotype. American Journal of Physiology 270, C1666-1674.
Bohn, E., Müller, S., Lauber, J., Geffers, R., Speer, N., Spieth, C., Krejci, J., Manncke, B., Buer,
J., Zell, A. and Autenrieth, I.B. (2003) Gene expression patterns of epithelial cells modulated by pathogenicity factors of Yersinia enterocolitica. Cellular Microbiology 6, 129-141.
Demarta, A., Tonolla, M., Caminada, A., Beretta, M. and Peduzzi, R. (2000) Epidemiological relationships between Aeromonas strains isolated from symptomatic children and household environments as determined by ribotyping. European Journal of Epidemiology 16, 447-453.
Eckmann, L., Kangnoff, M.F. and Fierer, J. (1993) Epithelial cells secrete the chemokine interleukin-8 in response to bacterial entry. Infection and Immunity 61, 4569-4574.
97 Eckmann, L., Smith, J.R., Housley, M.P., Dwinell, M.B. and Kangnoff, M.F. (2000) Analysis by
high density cDNA arrays of altered gene expression in human intestinal epithelial cells in
response to infection with the invasive enteric bacteria Salmonella. Journal of Biological
Chemistry 275, 14084-14094.
Galindo, C.L., Fadl, A.A., Sha, J., Gutierrez, C., Popov, V.L., Boldogh, I.B., Aggarwal, B.B. and
Chopra, A.K. (2004) Aeromonas hydrophila cytotoxic enterotoxin activates mitogen-activated
protein kinases and induces apoptosis in murine macrophages and human intestinal epithelial
cells. Journal of Biological Chemistry 279, 37597-37612.
Havelaar, A.H., Schets, F.M., van Silfout, A., Jansen, W.H., Wieten, G. and van der Kooij, D.
(1992) Typing of Aeromonas strains from patients with diarrhoea and from drinking water.
Journal of Applied Microbiology 72, 435-444.
Hayes, S.L., Lye, B.R., Lye, D.J., Rodgers, M.R., Stelma, G., Malard, J.M., Vandewalle, A. and
Vesper, S.J. (2006) Identification by microarray of a common pattern of gene expression in intact intestine and cultured intestinal cells exposed to virulent Aeromonas hydrophila isolates.
Journal of Water and Health 4, 381-388.
Hornef, M.W., Frisan, T., Vandewalle, A., Normark, S. and Richter-Dahlfors, A. (2002) Toll- like receptor 4 resides in the Golgi apparatus and colocalizes with internalized lipopolysaccharide in intestinal epithelial cells. Journal of Experimental Medicine 195, 559-570.
98 Irizarry, R. A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B. and Speed, T.P. (2003)
Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Research 31, e15.
Kampik, D., Schulte, R. and Autenrieth, I.B. (2000) Yersinia enterocolitica invasin protein triggers differential production of interleukin-1, interleukin-8, monocyte chemoattractant protein-
1, granulocyte-macrophage colony-stimulating factor, and tumor necrosis factor alpha in epithelial cells: implications for understanding the early cytokine network in Yersinia infections.
Infection and Immunity 68, 2484-2492.
Kühn, I., Albert, M.J., Ansaruzzaman, M., Bhuiyan, N.A., Alabi, S.A., Islam, M.S., Neogi,
P.K.B., Huys, G., Janssen, P., Kersters, K. and Mollby, R. (1997) Characterization of Aeromonas spp. isolated from humans with diarrhea, from healthy controls, and from surface water in
Bangladesh. Journal of Clinical Microbiology 35, 369-373.
Luangsay, S., Kasper, L.H., Rachinel, N., Minns, L.A., Mennechet, F.J.D., Vandewalle, A. and
Buzoni-Gatel, D. (2003) CCR5 mediates specific migration of Toxoplasma gondii-primed CD8+ lymphocytes to inflammatory intestinal epithelial cells. Gastroenterology 125, 491-500.
Martin-Carnahan, A. and Joseph, S.W. 2005. Order XII. Aeromonadales ord. nov. In George
M. Garrity (Editor in Chief), Bergey's Manual of Systematic Bacteriology. 2nd Edition, Vol.2
Part B. (pp. 556-580). New York: Springer.
99 Moyer, N.P., Martinetti, G., Holcomb, L.A., Hall, N.H. and Altwegg, M. (1992) Application of ribotyping for differentiating aeromonads isolated from clinical and environmental samples.
Applied and Environmental Microbiology 58, 1940-1944.
Moyer, N. P. (1987) Clinical Significance of Aeromonas species isolated from patients with diarrhea. Journal of Clinical Microbiology 25, 2044-2048.
Nagasako, T., Sugiyama, T., Mizushima, T., Miura, Y., Kato, M. and Asaka, M. (2003) Up- regulated Smad5 mediates apoptosis of gastric epithelial cells induced by Helicobacter pylori infection. Journal of Biological Chemistry 278, 4821-4825.
Saviozzi, S. and Calogero, R.A. (2003) Microarray probe expression measures, data normalization and statistical validation. Computational Functional Genomics 4, 442-446.
Sen, K. and Rodgers, M. (2004) Distribution of six virulence factors in Aeromonas species isolated from US drinking water utilities: a PCR identification. Journal of Applied Microbiology
97, 1077-1086.
Sinha, S., Shimada, T., Ramamurthy, T., Bhattacharya, S.K., Yamasaki, S., Takeda, Y. and
Balakrish Nair, B. (2004) Prevalence, serotype distribution, antibiotic susceptibility and genetic profiles of mesophilic Aeromonas species isolated from hospitalized diarrhoeal cases in Kolkata,
Indian Journal of Medical Microbiology 53, 527-534.
100 Smith, H.R. and Cheasty, T. (1998) Diarrhoeal disease due to Escherichia coli and Aeromonas
In: Microbiology and Microbial Infections ed. W.J. Hausler & M. Sussman, pp. 513-529. New
York: Oxford University Press.
Smyth, G. K. (2004) Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Statistical Applications in Genetics and Molecular
Biology 3, No. 1, Article 3.
Stelma, G. N., Reyes, A.L., Peeler, J.T., Johnson, C.H., Spaulding, P.L. (1992) Virulence characteristics of clinical and environmental isolates of Vibrio vulnificus. Applied and
Environmental Microbiology 58, 2776-2782.
Stelma, G. N., Reyes, A.L., Peeler, J.T., Francis, D.W., Hunt, J.M., Spaulding, P.L., Johnson,
C.H, and Lovett, J. (1987) Pathogenicity test for Listeria monocytogenes using
immunocompromised mice. Journal of Clinical Microbiology 25, 2085-2089.
Strober, W. (1998) Interactions between epithelial cells and immune cells in the intestine.
Annals of the New York Academy of Sciences 859, 37-45.
Szewzyk, R., Manz, W. and Schleifer, K-H. (2000) Microbiological safety of drinking water.
Annual Review of Microbiology 54, 81-127.
101 Wong, C. Y. F., Mayrhofer, G., Heuzenroeder, M.W., Atkinson, H.M., Quinn, D.M., and
Flower, R.L.P. (1996) Measurement of virulence of aeromonads using a suckling mouse model of infection. FEMS Immunology and Medical Microbiology 15, 233-241.
Xia, M., Bumgarner, R.E., Lampe, M.F. and Stamm, W.E. (2003) Chlamydia trachomatis infection alters host cell transcription in diverse cellular pathways. Journal of Infectious Disease
187, 424-434.
102 Table 1. Aeromonas isolates used in this study; their source and identification number.
Bacterium Source Number Aer. hydrophila EPA1 distribution water* 407-07-D1 EPA2 non-environmental 12723W** EPA5 non-environmental ATCC7966 *** EPA9 non-environmental 15(B) EPA10 distribution water* 407-07-D4 Aer. veronii biotype sobria EPA61 distribution water* 492-26-D1 EPA63 distribution water* 492-32-D1 Aer. caviae EPA 71 distribution water* 649-09-E1 EPA72 environmental isolate ATCC15468*** EPA 73 non-environmental 1685 Aer. salmonicida EPA51 distribution water* 492-07-B3 EPA52 distribution water* 492-06-D8 Aer. encheleia EPA80 distribution water* 492-35-B6 EPA82 distribution water* 407-07-C10 Aer. allosaccharophila EPA100 environmental isolate ATCC 51208***
* Distribution system water (Sen and Rodgers, 2004) **AMC – Dr. Amy Horneman collection ***ATCC – American Type Culture Collection
103 Table 2. Number of significantly up-regulated (>2 fold) genes in the intestinal tissue of neonatal mice after exposure to Aeromonas isolates, as determined by Affymetrix® microarray analysis. Strain ID Dose Average FC of up-regulated Mortality* Virulent genes* Aer. hydrophila EPA1 1.1E+7 4.0 18% Y EPA2 1.3E+8 4.5 12% Y EPA2 1.2E+7 2.7 0 % N EPA5 2.5E+8 3.5 50% Y EPA5 5.2E+7 2.5 11% N EPA9 2.4E+7 3.5 22% Y EPA10 2.2E+7 3.4 33% Y Aer. caviae EPA71 4.6E+7 2.7 0 % N EPA72 >1.0E9 ** 10% N EPA72 2.2E+7 2.4 0 % N EPA73 >1.0E9 2.9 40% N EPA73 4.0E+7 2.2 0 % N Aer. veronii biotype sobria EPA61 8.0E+6 2.6 0 % N EPA61 5.6E+7 2.5 27% N EPA63 4.0E+8 4.5 68% Y Aer. salmonicida EPA51 2.8E+7 ** 0 % N EPA52 4.6E+7 ** 0 % N Aer. encheleia EPA80 7.0E+6 *** 0 % N EPA82 1.4E+7 *** 0 % N Aer. allosaccharophila EPA100 5.2E+7 Not tested**** 0 % N EPA100 >1.0E9 3.1 18% Y Heat killed, UV inactivated EPA1 1.1E+7 0 0 % N * - Percentage of deaths of all inoculated mice after 5 hours **- no genes significantly up-regulated *** - only 1 and 2 genes significantly up-regulated in EPA80 and 82, respectively ****- microarray analysis was not performed at lower dose due to no mortality
104 Table 3. Summary of probe set up-regulation of listed genes in neonatal mouse model from Aeromonas infections
Probe Set ID Gene Symbol EPA1 EPA2 EPA5 EPA9 EPA10 EPA63 EPA61 EPA100 EPA72 EPA71 EPA73 Angiogenesis 1421811_at Thbs1 6.0 7.3 2.3 8.5 1460302_at Thbs1 6.8 7.2 2.5 7.5 1450377_at Thbs1 4.4 2.4 2.1 6.1 1419431_at Ereg 18.6 21.1 3.7 7.6 27.3 1425452_s_at Ptprj /// AW125753 5.0 2.0 5.3 1418350_at Hbegf 7.0 3.8 2.4 3.3 9.0 1438855_x_at Tnfaip2 7.4 8.0 5.1 3.7 3.2 1441228_at --- 2.7 5.2 6.0 2.0
Apoptosis regulation/inhibition 1424638_at Cdkn1a 2.6 2.1 2.4 2.0 3.2 1416505_at Nr4a1 4.0 6.2 5.2 4.0 1433699_at Tnfaip3 15.8 14.0 2.7 6.0 1449773_s_at Gadd45b 3.0 3.2 2.8 2.5 2.5 1448272_at Btg2* 2.9 4.6 2.6 3.3 1418203_at Pmaip1* 4.5 3.5 2.4 4.2 1421392_a_at Birc3 5.7 5.2 2.7 2.5 2.1 6.0 1416442_at Ier2 3.6 3.0 2.8 3.0 2.1 1419647_a_at Ier3 9.7 9.0 2.8 6.4 4.1 9.6
Cellular adhesion 1424067_at Icam1 9.7 8.3 6.0 6.6 5.1 1418283_at Cldn4 6.3 4.5 2.5 4.1 2.7 4.3 2.8 2.9 1448162_at Vcam1 2.0 3.3 2.4 2.6
Defense response/cytokine related 1417856_at Relb 4.2 3.4 2.2 2.2 2.0 1418674_at Osmr 4.6 4.2 4.9 1449984_at Cxcl2 42.9 41.3 3.5 43.9 13.0 37.5 1418762_at Daf1 5.6 6.1 1449399_a_at Il1b 4.1 2.1 3.3 6.2 1420394_s_at Gp49a /// Lilrb4 5.7 4.7 3.2 3.7 2.8 2.9 1438658_a_at Edg3 2.5 2.7 2.4 2.1 2.5 1441855_x_at Cxcl1 3.7 3.1 2.6 2.8 1417789_at Ccl11 2.7 2.3 2.0 6.3 105 Probe Set ID Gene Symbol EPA1 EPA2 EPA5 EPA9 EPA10 EPA63 EPA61 EPA100 EPA72 EPA71 EPA73 1426858_at Inhbb 4.9 7.1 2.3 3.9 2.5 1452519_a_at Zfp36* 7.0 7.3 2.3 4.9 5.9 2.5 1417851_at Cxcl13 2.0
Regulation of transcription 1417483_at Nfkbiz 9.9 8.9 3.9 23.6 1433508_at Copeb 4.5 3.1 2.1 7.4 1418133_at Bcl3 5.3 4.5 2.8 2.1 5.9 1427844_a_at Cebpb 3.6 2.8 2.8 3.2 1418901_at Cebpb 3.2 3.2 1.9 3.1 1417065_at Egr1 10.1 11.8 3.5 13.7 7.1 16.2 3.0 1423100_at Fos* 16.7 14.1 12.0 13.9 17.5 1437247_at Fosl2 2.9 3.5 3.0 3.6 1415899_at Junb 7.2 8.6 5.6 3.1 1417409_at Jun* 2.2 2.2 1448694_at Jun* 2.0 2.0 1423233_at Cebpd 11.4 9.9 8.0 7.1 5.6 14.8 4.4 1449851_at Per1 3.2 3.0 2.1 6.0 3.2 2.8 2.7 1418280_at Klf6* 7.3 3.8 2.8 9.5 1455104_at Mxd1 3.5 2.4 2.0 2.6 1449731_s_at Nfkbia* 8.7 9.7 2.9 5.2 3.1 13.6 1438157_s_at Nfkbia* 7.7 8.6 4.1 4.7 3.3 10.0 1448306_at Nfkbia* 12.9 10.0 3.7 6.3 3.3 15.0 1420088_at Nfkbia* 3.9 5.2 2.3 2.9 2.2 5.1 1431804_a_at Sp3 2.5 2.2 2.4 1450350_a_at MGI:1932093 2.6 3.2 2.1 1451739_at Klf5 2.4 2.7 1455660_at Csf2rb1 3.6 3.4 3.6 2.9 1449363_at Atf3* 2.7
Response to biotic stimulus/stress 1453181_x_at Plscr1 6.0 3.2 8.5 5.7 1449519_at Gadd45a 3.3 2.6 2.4 3.0 1416297_s_at Pap 8.8 9.1 1418293_at Ifit2* 2.5 6.4 1424254_at Ifitm1 2.2 2.1 2.2 2.9 1450709_at Defcr5 4.6 4.8
106 Probe Set ID Gene Symbol EPA1 EPA2 EPA5 EPA9 EPA10 EPA63 EPA61 EPA100 EPA72 EPA71 EPA73 Negative regulation of mitogen-activated protein (MAP) kinases 1415834_at Dusp6 5.1 3.7 3.0 3.7 2.5 1448830_at Dusp1 7.8 9.3 5.5 3.2 12.5 2.1 1431422_a_at Dusp14 8.0 3.2 5.1
Coagulation 1415806_at Plat 4.0 5.1 2.6 2.0 1420664_s_at Procr 3.5 3.6 2.0 2.1 2.5 1416067_at Ifrd1 6.9 4.6 11.4 1442014_at Ifrd1 5.5 2.2 5.1
Kinase activity 1416432_at Pfkfb3 8.7 5.6 5.2 2.5 9.2 1419706_a_at Akap12 2.8 2.5 2.1 1437226_x_at Marcksl1 2.3 2.7 2.5 2.0
Metalloproteinase (MMP)/disintegrin 1417256_at Mmp13 14.8 14.0 4.9 2.6 9.7 1450716_at Adamts1 7.2 9.4 2.2 5.9 3.3 5.0 1419088_at Timp3 2.5 2.4
Integral to cell membrane 1434253_s_at Tmcc3 4.7 2.3 2.1 5.6 1419388_at Tm4sf20 6.5 3.6 4.7 1428776_at Slc10a6 2.3 3.2 2.0 2.0 1422788_at Slc43a3 2.5 2.4 2.1 1417654_at Sdc4 3.7 2.9 3.0 1422845_at Canx 2.0 2.1 1438606_a_at Clic4 2.3 2.0 1448300_at Mgst3 3.5 3.0 2.8 1453345_at 3830408G10Rik 4.0 2.1 2.4 5.5 2.1 1438658_a_at Edg3 2.6 2.8 2.5 2.1 2.5 2.6
Calcium binding protein 1439348_at S100a10 4.5 2.6 2.9 4.8 1455869_at Camk2b 2.3 5.0 2.2
107 Probe Set ID Gene Symbol EPA1 EPA2 EPA5 EPA9 EPA10 EPA63 EPA61 EPA100 EPA72 EPA71 EPA73 Cell to cell signaling 1455899_x_at Socs3 12.9 12.5 8.8 7.5 5.7 11.0 2.0 1426063_a_at Gem 10.3 9.4 3.8 7.7 4.9 4.8 1415800_at Gja1 2.2 2.4
Actin/cytoskeleton 1419734_at Actb 5.7 4.6 7.7 1423588_at Arpc4 2.3 2.1 1449018_at Pfn1 4.6 3.0 4.2 1455719_at Tubb5 2.1 2.0 2.0 1427347_s_at Tubb2a* 2.1 2.3
Unknown association with host response to pathogen 1427932_s_at 1200003I10Rik 9.6 7.9 5.4 5.0 8.4 1453238_s_at E430024C06Rik 7.9 8.0 4.1 2.9 9.6 1435137_s_at 1200016E24Rik 10.8 9.8 6.2 5.6 11.3 1430221_at 9130008F23Rik 2.9 2.0 2.4 2.7 1430352_at 8430417A20Rik 4.2 5.6 3.3 2.3 2.3 3.1 1428909_at A130040M12Rik 4.2 5.8 7.2 2.9 7.6 1452418_at 1200016E24Rik 8.0 6.9 5.3 3.2 7.5 3.9 1456321_at 3830408G10Rik 4.6 2.3 2.5 2.0 4.9 1456808_at 4933426M11Rik 2.2 2.3 2.7 2.0 2.1 1447448_s_at --- 5.8 3.8 4.4 2.3 7.9 1421009_at Rsad2* 4.0 4.7 1416129_at Errfi1 4.2 4.3 2.8 4.1 4.9 1419816_s_at Errfi1 4.6 3.2 2.2 4.0 3.6 1438313_at --- 6.3 7.7 2.6 2.9 1418835_at Phlda1 7.0 5.9 2.0 2.3 12.7 1422474_at Pde4b 4.1 4.7 2.8 2.2 6.8 1425837_a_at Ccrn4l 8.0 10.2 5.2 3.6 14.7 1426452_a_at Rsb30 8.2 5.4 2.6 5.1 *- Transcript also seen in cell culture as being unique to virulent infections
108 Table 4. Summary of Aeromonas infections and intestinal murine cell line (mICcl2)
Percentage of Total number of up- Transcripts unique to transcripts unique to regulated transcripts virulent strains virulent strains Neonates 554 458 83
Cell culture 176 46 26
109
Chapter 3
Proposal for disease presentation of Aeromonas caviae: irritable bowel-like syndrome
110 Abstract
Aeromonas caviae has been associated with human gastrointestinal disease. Strains of
this species typically lack virulence factors (VFs) such as enterotoxins and hemolysins that are
produced by other human pathogens of the Aeromonas genus. Microarray profiling of murine small intestinal extracts, 24 hours after oral infection with an A. caviae strain, provides evidence of a Th1 type immune response. A large number of gamma-interferon (γ-IFN) induced genes are up-regulated as well as several tumor necrosis factor-alpha (TNF-α) transcripts. A. caviae has always been considered an opportunistic pathogen because it lacks obvious virulence factors.
This current effort suggests A. caviae colonizes murine intestinal tract and causes what has been described by others as a dysregulatory cytokine response leading to an irritable bowel-like syndrome. This response would explain why a number of diarrheal waterborne outbreaks have been attributed to A. caviae even though it lacks obvious enteropathogenic properties.
Keywords: A. caviae, Th1 response, interferon induced gene expression
111 INTRODUCTION
Aeromonas caviae is considered an opportunistic pathogen and outbreaks have been
hypothesized as being the result of antibiotic therapy and/or weakened immune systems (Moyer
1987). The species is typically characterized as being non-cytotoxic and non-hemolytic, yet
there are many documented instances of A. caviae being isolated from patients with
gastroenteritis. A. caviae is consistently listed with A. hydrophila and A. veronii biotype sobria
as being an isolate of concern in human disease. The latter two species commonly harbor known
virulence factors (VFs) such as aerolysin, enterotoxins (both cytotoxic and cytotonic) and
proteases: therefore, their association with disease is rational.
Our research group is investigating the full range of Aeromonas virulence factors.
Testing has included numerous Aeromonas species and strains with virulence evaluated by:
phenotypic assays of known VFs, gene expression of host tissues and cells following infection,
intra-peritoneal ID50 assays in a mouse model, and proteomic evaluations of extracellular toxins
produced by the various organisms. Our past research has supported designating A. hydrophila and A. veronii biotype sobria as being virulent in mammalian systems (Hayes et al. 2007).
However, A. caviae infections using a neonatal mouse model, as well as in adult mouse models from other research efforts, have failed to propose a mechanistic explanation for A. caviae gastroenteritis. Using gene expression profiling, we have developed a hypothesis for gastrointestinal disease development associated with A. caviae.
112 MATERIALS and METHODS
Bacterial strain. The A. caviae strain specific to the focus of this research facet was
ATCC 15468, obtained from American Type Tissue Culture Rockville, MD. This strain is
designated as EPA72 in Hayes et al. 2007.
Virulence attributes. Phenotypic assays included motility, elastase, protease and
lecithinase activity, hemolysis on blood plates and cytotoxicity. Motility was assessed using
motility agar. Elastase activity was performed using bilayer plates (top layer R2A agar with
elastin [Sigma cat #7277], bottom layer R2A agar) and observing for zones of clearing. Protease
activity was assessed using skim milk agar and lecithinase activity was performed on egg yolk
agar. For cytotoxicity testing, the broth from an overnight cultures grown in tryptic soy broth
was filter sterilized through a 0.22 micron filter and 0.01-0.5 mL of filtrate was added to murine
intestinal epithelial cells (mICcl2 cell line). Cytotoxic effects were monitored for 24 hours. The
culture was tested for hemolysis activity on 5% sheep blood agar after growth at 36.5oC for 24
hours.
Virulence assay and microarray analysis/hybridization/analysis. These procedures
were performed as previously described (Hayes et al. 2007). Briefly, neonates, aged 4-6 days,
were intubated with 20 µL of a 109 CFU/mL suspension. For this testing, there was a 24 hour
time period between infection and small intestinal tissue harvesting. Small intestines were
excised and rinsed in pre-chilled 1X PBS and then minced into sections no longer than 0.5 cm
and stored at 4ºC in RNALater solution (Ambion, Austin, TX, USA). Five biological replicates were used in data analysis for unexposed controls and infection experiments. Samples were extracted for total RNA using TRIZOL™ reagent and disruption using ten second bursts with a tissue homogenizer. Total RNA samples were further purified using a glass-fiber procedure.
113 RNA quality was assessed for each sample using an Agilent 2100 bioanalyzer. High quality
double-stranded cDNA was created from total RNA and biotinylated cRNA targets were
produced in vitro from cDNA using an Enzo BioArray™ HighYield™ RNA Transcript labeling kit. The cRNA targets were then hybridized to prefabricated Mouse 430_2 GeneChip probe arrays (Affymetrix, Santa Clara, CA, USA) and scanned according to the manufacturer’s
protocol an Agilent GeneArray GeneChip3000® scanner and GCOS software. Default values for
signal processing were as follows: target signal scaling 500; normalization 1; Alpha1 0.05;
Alpha2 0.065; Tau 0.015; Gamma1L and 1H 0.0045; Gamma 2L and 2H 0.06; and Pertubation
1.1. Samples were deposited in the Gene Expression Omnibus
(http://www.ncbi.nlm.nih.gov/projects/geo/) under series number GSE6765. GeneChip intensity
data were normalized on the log scale using robust multi-array analysis (RMA) methods and the
LIMMA (Linear Models for Microarray Data) module of Bioconductor was used for the analysis
of the normalized gene expression data.
Colonization assay. Colonization of streptomycin treated mice was assessed using
strains of A. hydrophila (n=3), A. veronii biotype sobria (n=2), and A. caviae (n=2) and A.
enchelia (n=1). The colonization assay was based on Sanderson et al. (1996). Adult Swiss
Webster, 6 week old female mice were treated with 5 mg/mL of streptomycin in their drinking
water for 4 days to eliminate the normal flora. Mice were then orally gavaged with 108 CFU of
organisms (in a 0.1 mL aliquot). Colonization was considered positive if Aeromonas cells were
detected (via cultural method) in the 4 cm colon section posterior to the caecum.
qRT-PCR single tube gene expression assay. Single tube gene expression assays were
performed as described in Hayes et al. 2007.
114 RESULTS
A. caviae strains were tested using phenotypic assays for known virulence factors (VFs).
A. caviae strain ATCC15468 was negative for hemolysis, elastase and cytotoxicity but positive for motility, lecithinase and protease activity.
All A. hydrophila and A. veronii biotype sobria strains colonized the streptomycin treated mice. One of the two A. caviae (ATCC15468) strains colonized the mice and actually demonstrated the best colonization of these three species (data not shown). The A. enchelia strain did not colonize the mice.
Table 1 presents the majority of up-regulated transcripts at 24 hours after introducing A. caviae orally in the neonatal mouse virulence assay. The listing is divided into categories based on whether a transcript is: 1) known or suspected to be induced by γ-IFN, 2) associated with the immune response, 3) associated with antigen presentation, or 4) associated with tumor necrosis factor-alpha (TNF-α, another Th1 response cytokine).
Table 2 presents qRT-PCR confirmation testing of five interferon-induced cDNA
molecules. Also included is testing for gamma-interferon. The results in Table 2 provide
confirmation of up-regulation of these six genes. Also presented are the results of two
housekeeping genes (Trfp and Pgk1). The housekeeping gene results demonstrate consistent
cDNA levels between test and control animals.
DISCUSSION
In our past efforts, gene response in neonatal mouse intestinal tissue was evaluated 5 hours post-infection using doses of 107-108 CFU/animal. A large number of transcripts were up-
regulated in response to A. hydrophila and A. veronii biotype sobria infections with a high
115 percentage of the transcripts being associated with immune response. Also, significant mortality
was observed after infection with these two species. A. caviae did not produce a similarly fast
regulatory response under these same conditions, nor was there any mortality. Because A. caviae
is reported as more chronic type of disease, a longer time period after infection was tested by
microarray analysis.
Gene response to A. caviae infection at 24 hours presented a Th1 response, characterized
by the up-regulation of multiple γ-IFN induced transcripts. A known role of γ-IFN is to activate
macrophages, resulting in increased MHC class I and II expression (Szabo et al. 2003). As seen
in Table 1, four MHC class I and II transcripts were up-regulated as were two transcripts related
to macrophage activation. Other notable genes up-regulated at 24 hours, that were also seen
from infections that cause a Th1 response, include tryptophanyl-tRNA synthetase (WARS) and
T-cell specific GTPase (Tgtp). WARS induction is due to the presence of a gamma-interferon
activation site (GAS) in the promotor sequence of the gene (Strehlow et al. 1993; Kisselev et al.
1993). WARS has also been detected via proteomics analysis of interferon-stimulated human
cultured colon epithelial cells (Barceló-Batllori et al. 2002). There are a number of theories
regarding the biological rationale for up-regulation of the WARS transcript (Xue and Wong
1995). However, for the purpose of the current effort, it provides evidence of intestinal epithelial
exposure to γ-IFN.
Using microarray analysis of colonic tissues, Tgtp was reported as being up-regulated in
response to Trichuris muris in AKR mice (nematode infection sensitive animals, Datta et al.
2005). In bacterial infection research, Citrobacter rodentium (Higgins et al. 1999) and
Helicobacter pylori (Lindholm et al. 1998) infections demonstrated Th1 responses, in mice and humans, respectively. In the C. rodentium research, increased levels of γ-IFN and TNF-α were
116 found in colonic tissue by RT-PCR. Gamma-interferon was shown to be present in gastric epithelial biopsies of human patients infected with H. pylori using cytokine specific staining.
The up-regulated transcripts assayed by microarray in the current effort did not include γ-
IFN. The likely reason is that the T-cell population would only represent a very small proportion of cells in the excised intestinal tissue. Datta et al. (2005) also did not detect γ-IFN by microarray, but did detect its presence using RT-PCR. They speculated the detection sensitivity between the assays was the reason for the difference. This same conclusion can be drawn for our results.
A Th1 response has also been associated with murine irritable bowel syndrome (IBS) models (Higgins et al. 1999). One possibility is that A. caviae colonizes the intestinal tract and causes an immune response characterized by a dysregulatory Th1 response (increased release of
T-cell cytokines) as seen in IBS patients (Barbara 2006). One side effect of this response is disruption of the epithelial barrier as presented in research efforts regarding the effect of γ-IFN on intestinal epithelial monolayers (Adams et al. 1993; Youakim and Ahdieh 1999). Tight junction organization was disrupted in these studies. Tight junction disorganization would cause paracellular fluid excretion in vivo.
A. caviae has been characterized as a chronic or opportunistic infection, whereas A. hydrophila and A. veronii biotype sobria are associated with acute infections (Moyer 1987). We propose the following disease progressions for the three human pathogens. Establishment of gastrointestinal infection involves colonization and subsequent damage to the epithelial barrier.
Colonization and attachment studies have demonstrated the ability of these three human pathogenic species to either colonize mammalian intestinal tract or adhere to intestinal epithelial cell lines (Grey and Kirov 1993; Merino et al. 1996). A. hydrophila and A. veronii biotype
117 sobria produce hemolysins and enterotoxins whereas A. caviae typically does not. The former
two species result in tissue damage from their toxins, with the host response often acute and
resulting in significant numbers of up-regulated immune response transcripts (Hayes et al. 2007).
The mechanism for A. caviae is to colonize and evoke a Th1 immune response. Increased levels
of gamma-interferon disrupts epithelial barrier tight junction and causes paracellular fluid
excretion resulting in mild diarrhea.
We provide evidence that A. caviae colonizes neonatal murine intestines and produces
up-regulation of multiple gamma-interferon (γ-IFN) induced transcripts. Similar intestinal gene
response has been reported with other bacteria and parasitic infections in mice with a side effect
of intestinal dysfunction. This type of response is associated with T-cells of the Th1 phenotype
being activated. Besides γ-IFN induced transcripts, tumor necrosis factor-alpha (TNF-α)
transcripts are also up-regulated in Th1-type responses and some TNF-α associated transcripts were also up-regulated in our research. If verified, the disease mechanism for A. caviae species would be best described as chronic in nature, producing mild diarrhea due to epithelial barrier dysfunction caused by exposure of the epithelial barrier to elevated levels of γ-IFN.
118 REFERENCES
Adams, R.B., S.M. Planchon, and J.K. Roche. 1993. IFN-γ modulation of epithelial barrier function. Journal of Immunology. 150:2356-2363.
Barbara, G. 2006. Mucosal barrier defects in irritable bowel syndrome. Who left the door open?
American Journal of Gastroenterology. 101:1295-1298.
Barceló-Batllori, S., M. André, C. Servis, N. Lévy, O. Takikawa, P, Michetti, M. Reymond and
E. Felley-Bosco. 2002. Proteomic analysis of cytokine induced proteins in human intestinal epithelial cells: implications for inflammatory bowel disease. Proteomics. 2:551-580.
Datta, R., M.L. deSchoolmeester, C. Hedeler, N.W. Paton, A.M. Brass and K.J. Else. 2005.
Identification of novel genes in intestinal tissue that are regulated after infection with an intestinal nematode parasite. Infection and Immunity. 73:4025-4033.
Grey, P.A. and S.M. Kirov. 1993. Adherence to Hep-2 cells and enteropathogenic potential of
Aeromonas spp. Epidemiology and Infections. 110:279-287.
Hayes, S.L., M.R. Rodgers, D.J. Lye, G.N. Stelma, C.A. McKinstry, J.M. Malard and S.J.
Vesper, S.J. 2007. Evaluating virulence of waterborne and clinical Aeromonas isolates using gene expression and mortality in neonatal mice followed by assessing cell culture's ability to
119 predict virulence based on transcriptional response. Journal of Applied Microbiology [IN
PRESS].
Higgins, L.M., G. Frankel, G. Douce, G. Dougan and T.T. MacDonald. 1999. Citrobacter rodentium infection in mice elicits a mucosal Th1 cytokine response and lesions similar to those in murine inflammatory bowel disease. Infection and Immunity. 67:3031-3039.
Kisselev, L., L. Frolova and A.-L. Haenni. 1993. Interferon inducibility of mammalian tryptophanyl-tRNA synthetase: new perspectives. Trends in Biochemical Sciences. 18:263-267.
Lindholm, C., M. Quiding-Järbrink, H. Lönroth, A. Hamlet and A.-M. Svennerholm. 1998.
Local cytokine response in Helicobacter-infected subjects. Infection and Immunity. 66:5964-
5971.
Merino, S., X. Rubires, A. Aguilar and J.M. Tomás. 1996. The O:34-antigen lipopolysaccharide as an adhesin in Aeromonas hydrophila. FEMS Microbiology Letters. 139:97-101.
Moyer, N.P. 1987. Clinical significance of Aeromonas species isolated from patients with diarrhea. Journal of Clinical Microbiology. 25:2044-2048.
Strehlow, I., D. Seegert, C. Frick, F.-C. Bange, C. Schindler, E.C. Böttger and T. Decker. 1993.
The gene encoding IFP 53/tryptophanyl-tRNA synthetase is regulated by the γ-interferon activation factor. Journal of Biological Chemistry. 268:16590-16595.
120
Szabo, S.J., B.M. Sullivan, S.L. Peng, and L.H. Glimcher. 2003. Molecular mechanisms regulating TH1 immune responses. Annual Reviews in Immunology. 21:713-758.
Xue, H. and T.-F. Wong. 1995. Interferon induction of human tryptophanyl-tRNA synthetase safeguards the synthesis of tryptophan-rich immune-system proteins: a hypothesis. Gene.
165:335-339.
Youakim, A. and M. Ahdieh. 1999. Interferon-γ decreases barrier function in T84 cells by reducing ZO-1 levels and disrupting apical actin. American Journal of Physiology. 276:G1279-
G1288.
121 Table 1. Up-regulated transcripts after 24 hours, neonatal mouse exposure
Affymetrix Fold Probe set ID Gene Symbol Gene name Change gamma-Interferon induced genes 1415694_at Wars tryptophanyl-tRNA synthetase 2.5 1416714_at Irf8 interferon regulatory factor 8 2.1 1417141_at Igtp interferon gamma induced GTPase 8.2 1417244_a_at Irf7 interferon regulatory factor 7 2.0 1417292_at Ifi47 interferon gamma inducible protein 47 5.1 1417793_at Iigp2 interferon inducible GTPase 2 5.9 1418240_at Gbp2 guanylate nucleotide binding protein 2 6.0 1418293_at Ifit2 interferon-induced protein with tetratricopeptide repeats 2 6.4 1418825_at Irgm immunity-related GTPase family, M 6.5 5830443L24Rik RIKEN cDNA 5830443L24 gene, similar to macrophage 1418776_at 5.5 (Ma2l) activation 2 like 1419603_at Ifi204 interferon activated gene 204 2.9 1423555_a_at Ifi44 interferon-induced protein 44 6.6 1424617_at Ifi35 interferon-induced protein 35 1.7 1426276_at Ifih1 interferon induced with helicase C domain 1 2.1 1429184_at Gvin1 GTPase, very large interferon inducible 1 2.2 1434813_x_at Wars tryptophanyl-tRNA synthetase 2.3 1435906_x_at Gbp2 guanylate nucleotide binding protein 2 5.8 1438676_at Mpa2l macrophage activation 2 14.3 1445897_s_at Ifi35 interferon-induced protein 35 1.6 1448436_a_at Irf1 interferon regulatory factor 1 2.0 1449009_at Tgtp T-cell specific GTPase 11.9 1449025_at Ifit3 interferon-induced protein with tetratricopeptide repeats 3 3.8 1450783_at Ifit1 interferon-induced protein with tetratricopeptide repeats 1 2.0 1451567_a_at Ifi203 interferon activated gene 203 1.7 Ifi203 /// Ifi204 interferon activated gene 203 /// interferon activated gene 1452348_s_at 4.3 /// Ifi205 204 /// similar to interferon activated gene 205
Immune response transcripts 1417185_at Ly6a lymphocyte antigen 6 complex, locus A 2.1 1418392_a_at Gbp4 guanylate nucleotide binding protein 4 5.6 1420915_at Stat1 signal transducer and activator of transcription 1 3.4 1421911_at Stat2 signal transducer and activator of transcription 2 2.3 1450403_at Stat2 signal transducer and activator of transcription 2 2.5 1453304_s_at Ly6e lymphocyte antigen 6 complex, locus E 2.0
Antigen presentation related transcripts 1418536_at LOC630509 similar to H-2 class I histocompatibility antigen 2.0 1422527_at H2-DMa histocompatibility 2, class II, locus DMa 2.7 1449875_s_at H2-T22 histocompatibility 2, T region locus 22 1.8 1451721_a_at H2-Ab1 histocompatibility 2, class II antigen A, beta 1 1.7
Tumor necrosis factor associated transcripts 1438855_x_at Tnfaip2 tumor necrosis factor, alpha-induced protein 2 2.1
122 Affymetrix Fold Probe set ID Gene Symbol Gene name Change 1439680_at Tnfsf10 tumor necrosis factor (ligand) superfamily, member 10 3.3 1459913_at Tnfsf10 tumor necrosis factor (ligand) superfamily, member 10 3.8 1460255_at Tnfsf13b tumor necrosis factor (ligand) superfamily, member 13b 3.0
Table 2. Results of confirmation testing by qRT-PCR
Gene Symbol Control CT Test CT Delta CT Ifi47 22.0 + 0.60 18.6 + 0.59 3.4 Mpa2l 23.4 + 0.83 18.6 + 0.74 4.8 Igtp 21.8 + 1.1 16.6 + 1.1 5.2 Ifi44 25.4 + 0.78 21.2 + 0.30 4.2 Tgtp 26.5 + 0.50 21.3 + 0.54 5.2 IFN-gamma 31.0 + 0.79 26.8 + 0.83 4.2 Housekeeping genes Trfp 22.0 + 1.3 22.1 + 0.56 Pgk1 17.8 + 0.73 18.0 + 0.58
123
Chapter 4
Creation of isogenic strains of a virulent Aeromonas hydrophila strain by transposon
insertion with subsequent evaluation of effects on pathogenic characteristics
124 Abstract
The goal of this research was to produce and characterize isogenic, virulent factor mutants of A. hydrophila. DNA sequencing of flanking regions to transposon insertions was used to identify potentially interrupted genes. BLAST results indicated that mutations were obtained in potential virulence factors associated with lateral flagella, O antigen and type II secretion system genes. Limited testing of one lateral flagella mutant indicated the elimination of swarming motility, a trait associated with colonization of intestinal cells. Colonization testing of murine intestines with this mutant using streptomycin-treated adult mice was inconclusive.
Murine intestinal cell culture monolayers demonstrated no difference in gene expression after infection with this lateral flagella mutant and wild type organisms. Future work will include optimization of colonization assays and repeated testing for lateral flagella and O antigen mutants. Other planned research includes testing the type II secretion system mutant for its ability to excrete exotoxins. If negative for exotoxin secretion, testing in neonatal mice and cell culture monolayers will be undertaken for this mutant.
125 INTRODUCTION
Production of isogenic bacterial strains can be a valuable technique for assessing effects associated with the loss of an individual gene. Effects can often be monitored using phenotypic assays (e.g., hemolysis, loss of ability to metabolize carbohydrates, etc.), but limited information regarding specific biological mechanisms is obtained. If we have the ability to produce a large number of mutations, a method to identify the gene whose function has been eliminated and a screening or functional assay to monitor effects, then information regarding gene function can be obtained. Unfortunately, the last aspect of this puzzle is often the limiting factor. We report here on an approach to assess loss of individual VFs of an A. hydrophila strain already demonstrated to be virulent in a neonatal mouse model.
Isogenic strains of A. hydrophila with one interrupted gene were produced. A touchdown
PCR procedure was employed to identify the site of transposon insertion. Gene expression of host cell monolayers was one procedure used to assess differences between wild type and mutant infections. However, other animal test systems and/or phenotypic assays were also evaluated because of the realization that a multi-faceted approach will be needed to investigate virulence loss or biological mechanisms of infection.
The remainder of this report describes progress to date for producing mutants and identifying interrupted genes. Some limited testing of one mutant has been initiated. There is much yet to be accomplished and research is on-going.
MATERIALS and METHODS
Preparation of electro-competent bacterial strain. A single colony of EPA1 (drinking water isolate 407-07-D1, Sen and Rodgers 2004) was selected from a fresh LB plate (Bacto-
126 tryptone, 10 g/L, yeast extract 5 g/L, NaCl 10 g/L, agar 7 g/L, pH adjusted from 6.9 to 7.5 with
1N NaOH), inoculated into a 10 ml 2XYT broth (Bacto-tryptone, 16g /L, Yeast Extract, 10g /L,
NaCl, 5g/L) and incubated overnight at 35oC. One milliliter of the overnight culture was added to 250 mL of 2XYT media and placed in 35°C shaker incubator. Cells were grown for 4 hours to log phase growth. The cells from the shaker were then placed and kept on ice for the remainder of the procedure. Cells were aliquoted into eight 40 mL polycarbonate, pre-chilled centrifuge tubes and spun at 5000 rpm for 10 minutes at 4oC. After centrifugation, tubes were placed on ice and the supernate removed. Cells in each tube were re-suspended in 20 mL ice-cold sterile dH20 and aliquots combined to reduce the number of centrifuge tubes from eight to four. A second spin (same conditions as above) was performed and each of the four pellets were re-suspended in
10 mL of ice-cold dH20 and then combined into one 40 mL centrifuge tube. A third spin was performed (same conditions as above) with the resulting pellet suspended in 10 mL ice-cold, sterile 10% glycerol. This cell suspension was aliquoted into pre-chilled 0.5 mL eppendorf tubes
(200 µL/tube) and placed on ice. These aliquots were snap frozen in an ethanol-dry ice bath.
The electro-competent cells were stored at -80°C until use.
Transposon DNA. The transposon used for mutation was the EZ-Tn5™
Transposome™ (Epicentre, Madison, WI). This DNA-enzyme molecule is a stable complex formed between the EZ-Tn5 Transposase enzyme and the EZ-Tn5
AGATGTGTATAAGAGACAG -3'. This sequence was used to locate the end of the transposon insertion during sequence analysis. The complete sequence of the transposon can be obtained at the Epicentre website (www.epibio.com).
127 Electroporation conditions. A Bio-Rad Gene Pulser II instrument (Bio-Rad Corp.,
Hercules, CA) and cuvettes with a gap of 0.2 cm were used for electroporation. Electroporation
conditions were: 25µF, 200 Ω and 2.5 kV, resulting in a time constant of approximately 4.5 msec. Electro-competent A. hydrophila and transposon DNA were thawed on ice. Four 1.0 mL
SOC medium aliquots (Bacto-tryptone 20g/L, yeast extract 5 g/L, NaCl 0.5g/L, KCl 2.5 millimolar[mM], MgCl2 10 mM, MgSO4, 10 mM, 20mM glucose, pH adjusted to 7.0+ 0.2 with
1N NaOH ) in 2 mL eppendorf tubes were also placed on ice. A 40 µL aliquot of cells was
placed in an electroporation cuvette with 350 µL pre-chilled dH2O. A 2 µL aliquot of transposon
DNA was added and the mixture mixed with a micropipetter. After electroporation, the cuvette
contents were added to the SOC medium and incubated at room temperature for 30 minutes. The
SOC medium and bacterial cells were then placed in a shaker incubator for one hour at 35oC.
After incubation, 0.4-0.5 mL aliquots of transformed cell suspensions were spread plated onto
LB-kanamycin plates (50 µg/ml kanamycin) and incubated overnight at 35oC. Typically 50-60
mutants were collected per electroporation experiment.
Genomic DNA extraction. Intact DNA was extracted and purified from overnight
mutant cultures by a standard methodology (MasterPure™ DNA kit, Epicentre, Madison, WI).
This procedure uses a gentle salt precipitation protocol employing Proteinase K and RNase A to
remove proteins and RNA from the preparation. The resultant preparations were assayed for
DNA concentration and A260/280 ratio using a NanoDrop spectrophotometer (Model ND-1000,
NanoDrop Technologies, Wilmington, DE). An acceptable ratio of 1.8-2.0 indicated extracts were free of protein. Samples were stored at -80oC prior to PCR analysis.
Regular and Touchdown PCR. Transposon insertion was confirmed by performing
regular PCR on mutant genomic DNA to amplify a 1010-bp region of the transposon (see Figure
128 1A). These PCRs were performed in a 50 µL mixture containing 23 µL PCR master mix (20 pmoles each primer, 1.5 U Taq DNA high fidelity polymerase), 25 µL pre-mix D (FailSafe™
PCR System, Epicentre, Madison, WI) containing dNTPs and buffer, and 2 µL bacterial genomic
DNA template diluted to 50 ng/µL. Primers used were KAN4-FP1 (5'-CTC AAA ATC TCT
GAT GTT ACA TGG C-3') and KAN5-RP1 (5'- TAG GTG GAC CAG TTG GTG ATT TTG A-
3').
After confirmation of transposon insertion, a touchdown PCR procedure was performed on mutant genomic DNA to obtain a product that represented a flanking region of the transposon insertion (see Figure 1B). This PCR product was sequenced and represented a stretch of the interrupted gene. The touchdown PCR procedure was as described by Levano-Garcia et al.
(2005). Briefly, the procedure uses a specific primer that sits down within the transposon and another primer (hybrid primer, 25-mer having degenerate bases at six different positions within the last 12 bases) that anneals to a region outside the transposon. Transposon specific primers were either KAN7-FP1 (5'- GCT CAT AAC ACC CCT TGT ATT ACT G-3') or KAN8-FP1 (5'-
AGA CCG TTC CGT GGC AAA GCA AAA G-3') and the hybrid primer was OPA09DS (5'-
GGG TAA CGC CGG TNG AYK SNG GNT C-3'; Y = C,T; K = G,T; S = C,G; N = A,C,G,T).
These PCRs were performed in a 50 µL mixture containing 22 µL PCR master mix (20 pmoles specific primer, 100 pmoles hybrid primer, 1.0 U Taq DNA high fidelity polymerase), 25 µL
Pre-mix D (as above), and 3 µL bacterial genomic DNA template diluted to 50 ng/µL.
Touchdown PCR conditions were performed in two phases. Phase one had an initial step of
95°C for 5 min, followed by 25 cycles of denaturation at 95°C for 45 s, annealing at variable temperatures for 45 s, and extension at 72°C for 2 min. The annealing temperature was set at
60°C in the first cycle and, at each of the 24 subsequent cycles, it was decreased by 0.5°C per
129 cycle down to 47.5°C. The second phase consisted of 25 cycles of 95°C for 45 s, 50°C for 45 s,
and 72°C for 2 min. After the last PCR cycle, the samples were cooled to 4°C.
Touchdown PCR product clean up and Sequencing. PCR products were cleaned using
a batch column purification method (Wizard® PCR DNA Purification System, Promega,
Madison, WI). In this procedure, PCR products are separated from contaminants, including
primer-dimers and amplification primers. Clean up of PCR products was required prior to
sequencing. The nucleotide sequences were determined with an ABI PRISM 3730XL DNA
Analyzer (Applied Biosystems; Foster City, CA) and an ABI Prism Big Dye Terminator kit
(Applied Biosystems). The primer used for the sequencing reaction was either KAN7-FP1 or
KAN8-FP1, depending on touchdown PCR results.
Swarming assay. The swarming assay was based on the procedure described by Kirov et al. (2002). A modification of the assay was preparation of the inoculum for the swarm plates.
The culture used to inoculate swarm plates was grown for 6 hours on sheep blood agar. Lateral flagella are formed during the first eight hours of growth on solid media. Swarm plates were prepared as described (0.5% Eiken agar in Difco broth).
Cell culture infection. This procedure was performed as described in Hayes et al.
(2007).
Colonization assay. This procedure was performed as described in Chapter 3, Material and Methods.
RESULTS
Table 1 contains a complete listing of mutants prepared to date. As can be seen, a number of mutant BLAST results showed no matches. This was likely due to the transposon
130 inserting into an unknown, yet to be characterized region of this particular Aeromonas strain's
genome. Also presented in Table 1 are gaps in the sequential numbering of mutants. The very
large gaps in the early mutant numbers were due to incorrect selection of mutants from the LB-
kanamycin plates. An initial assumption was that any colony that grew at 24 or 48 hours on LB-
kanamycin was a potential mutant. Many small, pin-point colonies were obtained on the
selective agar. Later it was discovered that only the very large colonies at 24 hours were actually
mutants. Some PCR products did not produce a sequence. Reasons for this might include poor
quality template or inappropriate template/primer ratios.
Insertion mutations with potential effects on virulence. Three of the transposon
insertions (EPA1-205, 259 and 279) occurred in the A. hydrophila lateral flagella complex.
Sequence information for the lateral gene complex was based on research performed by Merino
et al. (2003) and Canals et al. (2006). The reference sequence for this gene complex (bases 1 to
35945) was submitted to GeneBank (12-JUL-2005) by these groups.
Strain EPA1-205 demonstrated an insertion into the fliGL gene and the BLAST query indicated an insertion point towards the end of the gene. The GeneBank reference sequence for fliGL spanned base pairs from 11322 to 12350 (of 35945). The insertion occurred at base pair location 12282. The next gene in the complex, fliHL, is also represented in the sequence analysis with a gap of 93 base pairs between the genes. This is in comparison to a gap of only 13 base pairs between these two genes in the reference sequence. The fliGL gene encodes for a lateral flagellar FliG-like motor switch.
EPA1-259 inserted into the maf-5 gene, again towards the end of the gene. This gene encodes for a modification accessory factor gene and is reported to be involved in specific lateral
flagella glycosylation. A later review article states that little is known about the role of lateral
131 flagellin glycosylation in aeromonad virulence (Merino et al. 2006). The reference sequence for maf-5 spans base pairs 26276 to 27562 of the deposited sequence. The transposon inserted at position 27505. The flanking sequence analysis also picked up the next gene in the complex,
LafA. There was a 416 bp gap between these two genes from our analysis as compared to a 460 bp gap in the reference sequence.
EPA1-275 shows an insertion into the flgKL gene in the complex. This gene produces
HAP-1 (hook associated protein). In the reference sequence this gene spans base pairs 22979 to
24292. The transposon inserted at bp position 23753.
EPA1-035 had an insertion in a Type III secretion system (TTSS) gene, specifically ascZ.
Two references describe TTSS system genes in Aeromonas (Chacón et al. 2004; Vilches et al.
2004). Both list homologs to this sequence in other bacteria and report the protein as having an unknown function. One article lists the gene as being part of a deposited 26160 bp sequence.
The ascZ sequence is represented by base pairs 18024 through 18909. The transposon appears to have inserted at position 18086. The other article lists the TTSS cluster in a 26855 bp sequence, with the ascZ gene located at position 18570 through 19388. According to the BLAST result, the transposon inserted at position 18562, which is inconsistent with the first article. Therefore, we have to assume there is an error in one of the two reports or in the BLAST query.
EPA1-226 matched an A. hydrophila sequence deposited into GeneBank identified as a
DNA stretch that encodes for aerolysin (or perhaps Act) production. The deposited sequence has
2528 bp, with our transposon inserting at bp 187 and matching the sequence from 188 thru 579 bp. Upon examination of the deposited sequence, this DNA segment is actually a restriction fragment. The aerolysin gene does not start until position 884 and runs through position 2362.
The promotor sequence also falls outside the insertion, therefore, there would be no expected
132 effect from our insertion. Actually, EPA1 is negative for aerolysin production via Western
blotting (data not shown) and was also shown to be negative for the act gene by PCR (Sen and
Rodgers 2004).
The general secretion pathway (GSP) for the export of proteins (also called the type II
pathway) requires a number of protein components. One of the components is known as the 'N'
protein. EPA1-263 has an insertion in gspN, the terminal gene in a cluster of genes that encodes
a type II secretion system protein. This gene in A. hydrophila was originally designated as exeN
by Howard et al. 1993. This same sequence was designated as gspN by the research group that
sequenced the genome for A. hydrophila ATCC7966. The gspN gene is located in the 7966
genome at positions 613776 through 614531. The transposon inserted at 614289 and matched
614290 thru 614523 bps.
Three mutants, EPA1-249 (phosphomannomutase homology to manB of Salmonella
enterica serovar Typhi), EPA1-246 (O-acetyl transferase homology to wbbJ of E. coli) and
EPA1-294 (also in O-acetyl transferase region) had apparent insertion mutations in genes associated with O-antigen biosynthesis (Zhang et al. 2002). The O-antigen, which contains many repeats of an oligosaccharide unit (O-unit), is part of the lipopolysaccharide outer membrane of gram-negative bacteria. It contributes major antigenic variability to the cell surface. The surface O-antigen is subject to selection by the host immune system, which may account for the maintenance of many different O-antigen forms within species. They also can act as adhesion factors. Putative O-acetyltransferase transfers an acetyl group to the O-antigen.
It is also possible that the O-antigen of EPA1 contains mannose, because we detected a mutation in a homolog to phosphomannomutase. This protein works with others to synthesize GDP- mannose from mannose-6-phosphate (Zhang et al. 2002). The EPA1-294 mutant appeared to
133 insert close to, but outside of, the same gene as EPA1-246. Therefore, it is not a likely candidate for future testing.
Testing of a lateral flagella mutant. Because lateral flagella have been implicated as a colonization VF (Kirov et al. 2004), one of these mutants (EPA1-205) was tested with a swarming assay, in the colonization assay and on cell culture monolayers for host gene expression. The mutant's ability to swarm was knocked out. Figure 2 shows swarm plates of the wild type and the mutant.
In the colonization assay, the mutant initially demonstrated an inability to colonize adult mouse intestinal tracts. However, repeat confirmation testing was inconsistent. The colonization assay is undergoing optimization and this mutant, as well as other lateral flagella and O-antigen mutants will be tested for their ability to colonize murine intestinal tracts.
Because lateral flagella might cause a host response specific to ligand-receptor interactions, the mutant strain was applied to murine intestinal cell monolayers and compared to wild-type infections via microarrays. Gene expression profiles from extracted monolayers were obtained from mutant and wild-type infections and no differences were observed in the gene expression profiles. There were no unique genes seen in either set of up-regulated genes and fold change differences were also not significantly different for any gene from the two infection experiments (data not shown).
DISCUSSION
Researchers have proposed using isogenic bacterial mutants to infect host cell systems for developing hypotheses regarding mode of action of single virulence factors (VFs). Gene expression analysis via microarrays is one option for monitoring host cell response (Cummins
134 and Relman 2000). Our larger, current effort focuses on attempts to use gene expression as a screening assay for host system effects caused by bacterial infections. We have attempted to extend this procedure to compare wild-type and mutant strains of A. hydrophila.
The method for producing mutations was through a transposome that contains a kanamycin resistance cassette. Because there is no need for cell conjugation, suicide vectors, or specific host factors, Tn5 transposomes are useful for creating mutants in species that have poorly described genetic systems or for facilities that lack adequate molecular tools. When this effort began the genome of A. hydrophila had not been sequenced. With the recent publication of the sequence for A. hydrophila ATCC7966, mutagenic approaches that target specific genes will undoubtedly become common and will likely also be used for this species. Nonetheless, we have already generated a number of potentially interesting mutants.
Even with the complete sequencing of one strain of A. hydrophila, this genus is known for heterogeneity for VFs, both within and among species. For example, A. hydrophila
ATCC7966 does not contain a type III secretion system (TTSS) or lateral flagella genes
(Seshadri et al. 2006). In contrast, we did find evidence of both of these putative virulence factors in the A. hydrophila strain designated as EPA1. Also, many of the gene functions are still unknown within the annotated genome of 7966. Likely, there are a number of VFs yet to be discovered.
Mutants of A. hydrophila were created to measure the relationship between loss of a putative virulence factor and disease development. One method of monitoring disease initiation or development is through transcriptional response of host cells. In neonatal mice, we found that virulent strains of Aeromonas produced a wide range of transcripts being up-regulated in a short time period (5 hours) while avirulent strains produced little up-regulation of transcripts in this
135 same time period (Hayes et al. 2007). Therefore, if we were able to show this same differentiation between a wild-type virulent strain and an isogenic mutant, then the case can be made that the knocked out gene is important for virulence. However, it is unlikely that the affected disease pathway could be deduced because of the "all or none" type of transcriptional response seen between virulent and avirulent strain infections of neonatal mice.
In the cell culture model, any Aeromonas strain that produces and excretes cytotoxins cause severe cellular damage. Therefore, if, for example, a lateral flagellar gene is knocked out, any gene regulation difference based on ligand-receptor interactions may go undetected because of the cellular damage caused by the toxin. Cytotoxins act rapidly, the cell enters apoptosis pathways and the up-regulated transcript presentation is dominated by these cellular events. This is supported by our results with the lateral flagellar mutant and cell culture.
Alternative methods could potentially be used to assess effects. For example, we were able to demonstrate the elimination of swarming mobility in one mutant. There is a possibility that this mutant will not colonize live animal intestinal tracts. If loss of swarming motility and/or inability to colonize can be correlated with loss of virulence, then the significance of lateral flagella to disease establishment can be postulated. If we use host response (gene expression or
LD50) as the comparative gold standard, a large number of live animal parallel assays would be necessary to establish cause and effect.
According to Howard et al. 1993, evidence suggests that most, if not all, of the proteins excreted extracellularly by A. hydrophila are secreted through the general secretion pathway
(GSP or type II secretion system). The genes of this operon are also required for normal assembly of the outer membrane. Using insertional mutants in three areas of the operon convinced Howard et al. that the integrity of the entire operon is required for extracellular
136 excretion. Aerolysin and protease were unable to be secreted by A. hydrophila mutants in the
Howard et al. study. Our future studies will assess EPA1-263's ability to secrete toxins. If its ability to produce exotoxins is impaired, then infecting neonates with this strain, followed by gene expression assays, might assist in differentiating toxin affects from adhesion effects in vivo.
For our approach to become feasible, much of the work would need to become automated. Automation could mass produce mutations in all possible reading frames of protein encoding non-essential genes, followed by each mutant being tested using the hypothetical approach discussed above. For example, the A. hydrophila ATCC7966 genome contains over
4,000 protein coding genes (Seshadri et al. 2006). For over 1,000 of those genes, the protein function is unknown or hypothetical. Using a technique similar to what is described in this chapter, and with automation, thousands of mutants could be created and screened for gene insertions. Knowing the complete genome of a particular organism, the interrupted gene could be determined using a technique similar to that described in this chapter. If suitable virulence models are available, one could potentially determine importance of a gene to virulence, even though the protein function is unknown.
The advantages to using our approach are: 1) it is an easy system for creating mutants and identifying the interrupted gene and 2) specific and quantifiable measurement of biological effect can be obtained through gene expression profiles. Disadvantages are: 1) targeted mutations are not obtained, creating the need to screen a large number of mutants to find those of interest, 2) complex results from gene expression assays are difficult to interpret, and 3) it is currently labor intensive.
Because of evolving technological advances in areas such as automation, our approach should become more efficient in the future and needs to be considered for virulence research for
137 Aeromonas species and other organisms. Host gene expression will also become more common as a screening procedure to assess virulence mechanisms because procedures are being automated and simplified. Bioinformatic techniques are also becoming more sophisticated for analyzing the massive databases obtained from transcriptional profiling. New pathways are being proposed for disease mechanisms such as apoptosis, toxin actions and transcriptional activation of cytokine and chemokines. All the above will lead to mechanistic discoveries regarding host-pathogen relationships. Benefits will include development of new anti-pathogen drug therapies and obtaining the ability to assess potential exposure risk from members of specific genera and/or species of known or emerging pathogens.
138 REFERENCES
Canals, R., Altarriba, M., Vilches, S., Horsburgh, G., Shaw, J.G., Tomás, J.M. and Merino, S.
2006. Analysis of the lateral flagellar gene system of Aeromonas hydrophila AH-3
Journal of Bacteriology. 188:852-862.
Chacón, M.R., Soler, L., Groisman, E.A., Guarro, J. and Figueras, M.J. 2004. Type III secretion system genes in clinical Aeromonas isolates. Journal of Clinical Microbiology. 42:1285-1287.
Chopra, A.K., Houston, C.W., Peterson, J.W. and Jin, G.F. 1993. Cloning, expression, and sequence analysis of a cytolytic enterotoxin gene from Aeromonas hydrophila. Canadian Journal of Microbiology. 39:513-523.
Cummings, C.A. and Relman, D.A. 2000. Using DNA microarrays to study host-microbe interactions. Genomics. 6:513-525.
Hayes, S.L., Rodgers, M.R., Lye, D.J., Stelma, G.N., McKinstry, C.A., Malard, J.M. and Vesper,
S.J. (2007) Evaluating virulence of waterborne and clinical Aeromonas isolates using gene expression and mortality in neonatal mice followed by assessing cell culture's ability to predict virulence based on transcriptional response. Journal of Applied Microbiology [IN PRESS].
Howard, S.P., Critch, J and Bedi, A. 1993. Isolation and analysis of eight exe genes and their involvement in extracellular protein secretion and outer membrane assembly in Aeromonas 139 hydrophila. Journal of Bacteriology. 175:6695-6703.
Kirov, S.M., Tassell, B.C., Semmler, A.B.T., O’Donovan, L.A., Rabaan, A.A. and Shaw, J.G.
2002. Lateral flagella and swarming motility in Aeromonas species. Journal of Bacteriology.
184:547-555.
Kirov, S.M., Castrisios, M., and Shaw, J.G. 2004. Aeromonas flagella (polar and lateral) are enterocyte adhesions that contribute to biofilm formation on surfaces. Infection and Immunity.
72:1939-1945.
Levano-Garcia, J., Verjovski-Almeida, S and da Silva, A.C.R. 2005. Mapping transposon insertion sites by touchdown PCR and hybrid degenerate primers. Biotechniques. 38:225-229.
Merino, S., Gavín, Vilches, Shaw, J.G. and Tomás J.M. 2003. A colonization factor (production of lateral flagella) of mesophilic Aeromonas spp. is inactive in Aeromonas salmonicida strains.
Applied and Environmental Microbiology. 69:663-667.
Merino, S., Shaw, J.G. and Tomás, J.M. 2006. Bacterial lateral flagella: an inducible flagella system. FEMS Microbiological Letters. 263:127-135.
Sen, K. and Rodgers, M. 2004. Distribution of six virulence factors in Aeromonas species isolated from US drinking water utilities: a PCR identification. Journal of Applied
Microbiology. 97:1077-1086.
140
Seshadri, R., Joseph, S.W., Chopra, A.K., Sha, J., Shaw, J., Graf, J, Haft, D., Wu, M., Ren, Q.,
Rosovitz, M.J., Madupu, R., Tallon, L., Kim, M., Jin, S., Vuong, H., Colin Stine, O.C., Ali, A.,
Horneman, A.J. and Heidelberg, J.F. 2006. Genome sequence of Aeromonas hydrophila ATCC
7966T: jack of all trades. Journal of Bacteriology 188:8272-8282.
Sha, J., Pillai, L., Fadl, A.A., Galindo, C.L., Erova, T.E. and Chopra, A.K. 2005. The type III
secretion system and cytotoxic enterotoxin alter the virulence of Aeromonas hydrophila.
Infection and Immunity. 73:6446-6457.
Vilches, S., Urgell, C., Merino, S., Chacón, M.R., Soler, L., Castro-Escarpulli, G., Figueras, M.J. and Thomás, J.M. 2004. Complete type III secretion system of a mesophilic Aeromonas hydrophila strain. Applied and Environmental Microbiology. 70:6914-6919.
Yu, H.B., Rao, P.S., Lee, H.C., Vilches, S., Merino, S., Tomas, J.M. and Leung, K.Y. 2004. A type III secretion system is required for Aeromonas hydrophila AH-1 pathogenesis. Infection and Immunity. 72:1248-1256.
141 Table 1. Listing of mutants created to date
Blast Mutant Gene interrupted Score EPA1-017 Aeromonas hydrophila, ATCC 7966, complete genome, conserved hypothetical protein 0.00 EPA1-035 Aeromonas hydrophila strain AH-1 type III secretion system gene 0.00 EPA1-112 Aeromonas hydrophila, ATCC 7966, complete genome, thiamine biosynthesis/tRNA modification protein ThiI 0.00 EPA1-115 Aeromonas hydrophila, ATCC 7966, complete genome, collagenase family 9.00E-168 EPA1-117 Aeromonas hydrophila, ATCC 7966, complete genome, protein YciN, mate efflux family protein 0.00 EPA1-176 Aeromonas hydrophila, ATCC 7966, complete genome, phosphoglycolate phosphatase, bacterial 4.00E-51 EPA1-179 Aeromonas hydrophila, ATCC 7966, complete genome, DNA gyrase inhibitor 0.00 Aeromonas hydrophila, ATCC 7966, complete genome, Features in this part of subject sequence: conserved hypothetical EPA1-182 2.00E-126 protein, TPR domain protein EPA1-184 Aeromonas hydrophila, clone F85AH1 putative glycyl-tRNA synthetase 1.00E-152 EPA1-189 Aeromonas hydrophila, ATCC 7966, complete genome, putative cell division inhibitor 1.00E-111 EPA1-190 Aeromonas hydrophila, ATCC 7966, complete genome, pyruvate dehydrogenase complex dihydrolipoamide acetyltra 0.00 EPA1-191 Aeromonas hydrophila, ATCC 7966, complete genome, MshQ 1.00E-34 EPA1-192 EPA1-193 Pseudomonas syringae pv. syringae B728a, complete genome, ATPase 2.00E-11 Aeromonas hydrophila, ATCC 7966, complete genome, features flanking this part of subject sequence:107 bp at 5' side: EPA1-194 9.00E-13 transcriptional regulator, MarR family, 164 bp at 3' side: hypothetical protein EPA1-195 Aeromonas hydrophila, ATCC 7966, complete genome, ABC transporter ATP-binding protein YojI 2.00E-123 Aeromonas hydrophila, ATCC 7966, complete genome, transcriptional regulator, MarR family protein of unknown EPA1-196 0.00 function, DUF EPA1-197 Aeromonas hydrophila, ATCC 7966, complete genome, aromatic amino acid transport protein AroP 0.00 EPA1-198 No matches EPA1-199 No matches EPA1-200 No matches EPA1-201 Aeromonas hydrophila, ATCC 7966, complete genome, DNA helicase II 9.00E-179 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: glutamate-ammonia EPA1-204 4.00E-19 ligase adenylyltransferase
142 Blast Mutant Gene interrupted Score Aeromonas hydrophila, FliML (fliML), FliNL (fliNL), FliPL (fliPL), FliQL (fliQL), FliRL (fliRL), FlhBL (flhBL), FlhAL (flhAL), LafK (lafK), FliEL (fliEL), FliFL (fliFL), FliGL (fliGL), FliHL (fliHL), FliIL (fliIL), FliJL (fliJL), FlgNL (flgNL), FlgML (flgML), FlgAL (flgAL), FlgBL (flgBL), FlgCL (flgCL), FlgDL (flgDL), FlgEL (flgEL), FlgFL EPA1-205 5.00E-59 (flgFL), FlgGL (flgGL), FlgHL (flgHL), FlgIL (flgIL), FlgJL (flgJL), FlgKL (flgKL), FlgLL (flgLL), Maf-5 (maf-5), lateral flagellin (lafA), flagellar hook associated protein 2 (lafB), LafC (lafC), LafX (lafX), flagellar hook control length protein (lafE), LafF (lafF), RNA polymerase sigma factor (lafS), LafT (lafT), and LafU (lafU) genes, complete cds EPA1-206 No matches EPA1-207 Aeromonas hydrophila clone PA50 ORF34-like protein and putative tail fiber-like protein genes 2.00E-41 EPA1-208 No matches EPA1-210 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-211 0.00 protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-212 0.00 protein, methyl-accepting chemotaxis protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-213 0.00 protein EPA1-214 No matches EPA1-215 Many partial, short matches for Aeromonas hydrophila EPA1-216 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: lactoylglutathione lyase 2.00E-179 Aeromonas hydrophila, ATCC 7966, complete genome, features flanking this part of subject sequence: 138 bp at 5' side: EPA1-217 4.00E-14 aerotaxis receptor Aer, 822 bp at 3' side: alpha/beta superfamily hydrolase EPA1-219 EPA1-220 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ammonium transporter 2.00E-44 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: regulatory protein EPA1-222 1.00E-70 RecX, recA protein EPA1-223 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features flanking this part of subject sequence: 454 bp at 5' side: EPA1-224 2.00E-123 cytochrome d ubiquinol oxidase, subunit I 242 bp at 3' side: Holliday junction DNA helicase RuvB EPA1-226 Aeromonas hydrophila cytolytic enterotoxin gene, 5' end 1.00E-147 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: rRNA-23S ribosomal EPA1-228 1.00E-76 RNA
143 Blast Mutant Gene interrupted Score Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: putative transport EPA1-229 2.00E-97 system permease abc transporter protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: sugar phosphate EPA1-230 0.00 permease EPA1-231 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-232 2.00E-162 protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ABC-type Fe3+ EPA1-233 0.00 transport system, periplasmic component Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ethanolamine EPA1-234 0.00 utilization protein EutN/carboxysome structure - acetaldehyde dehydrogenase (acetylating) EPA1-235 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: 5'-nucleotidase/2',3'- EPA1-238 3.00E-36 cyclic phosphodiesterase EPA1-239 No matches Shewanella sp. MR-4, complete genome, features in this part of subject sequence: DNA methylase N-4/N-6 domain EPA1-240 7.00E-59 protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-242 0.00 protein Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-244 0.00 protein (clone of EPA1-242) EPA1-246 Aeromonas hydrophila PPD134/91 O-antigen cluster, complete sequence 1.00E-144 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: putative transporter, EPA1-247 0.00 putative Mg2+ transporter-C (MgtC) - fructose repressor Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ATP-dependent EPA1-248 0.00 helicase HrpB , 2'-5' RNA ligase EPA1-249 Aeromonas hydrophila PPD134/91 O-antigen cluster, complete sequence 2.00E-23 EPA1-250 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: fimbrial protein 2.00E-56 EPA1-251 many possibilities EPA1-252 Shewanella sp. MR-4 protein, low score match EPA1-254 Possible E. coli O antigen 0.001 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ABC-type multidrug EPA1-255 0.00 transport system, permease component 144 Blast Mutant Gene interrupted Score Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: arginine/ornithine EPA1-257 0.00 antiporter Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: ABC transporter, ATP- EPA1-258 7.00E-18 binding protein Aeromonas hydrophila, FliML (fliML), FliNL (fliNL), FliPL (fliPL), FliQL (fliQL), FliRL (fliRL), FlhBL (flhBL), FlhAL (flhAL), LafK (lafK), FliEL (fliEL), FliFL (fliFL), FliGL (fliGL), FliHL (fliHL), FliIL (fliIL), FliJL (fliJL), FlgNL (flgNL), FlgML (flgML), FlgAL (flgAL), FlgBL (flgBL), FlgCL (flgCL), FlgDL (flgDL), FlgEL (flgEL), FlgFL EPA1-259 3.00E-100 (flgFL), FlgGL (flgGL), FlgHL (flgHL), FlgIL (flgIL), FlgJL (flgJL), FlgKL (flgKL), FlgLL (flgLL), Maf-5 (maf-5), lateral flagellin (lafA), flagellar hook associated protein 2 (lafB), LafC (lafC), LafX (lafX), flagellar hook control length protein (lafE), LafF (lafF), RNA polymerase sigma factor (lafS), LafT (lafT), and LafU (lafU) genes, complete cds EPA1-263 A.hydrophila, exe C, D, E, F, G, H, I, J, K, L, M, N genes, type II secretion system, protein N (ATCC7966) 9.00E-110 EPA1-265 No matches EPA1-266 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: hypothetical protein 8.00E-117 EPA1-269 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: hypothetical, quinone EPA1-270 0.00 oxioreductase EPA1-271 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: hypothetical protein 9.00E-40 EPA1-273 Aeromonas hydrophila, proline sensor PrlS (prlS) gene, complete cds 0.00 Aeromonas hydrophila, FliML (fliML), FliNL (fliNL), FliPL (fliPL), FliQL (fliQL), FliRL (fliRL), FlhBL (flhBL), FlhAL (flhAL), LafK (lafK), FliEL (fliEL), FliFL (fliFL), FliGL (fliGL), FliHL (fliHL), FliIL (fliIL), FliJL (fliJL), FlgNL (flgNL), FlgML (flgML), FlgAL (flgAL), FlgBL (flgBL), FlgCL (flgCL), FlgDL (flgDL), FlgEL (flgEL), FlgFL EPA1-275 5.00E-12 (flgFL), FlgGL (flgGL), FlgHL (flgHL), FlgIL (flgIL), FlgJL (flgJL), FlgKL (flgKL), FlgLL (flgLL), Maf-5 (maf-5), lateral flagellin (lafA), flagellar hook associated protein 2 (lafB), LafC (lafC), LafX (lafX), flagellar hook control length protein (lafE), LafF (lafF), RNA polymerase sigma factor (lafS), LafT (lafT), and LafU (lafU) genes, complete cds Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: nucleoside- EPA1-276 0.00 diphosphate-sugar epimerase EPA1-277 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: protein SsnA 0.00 EPA1-278 No matches Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: thiol:disulfide EPA1-281 0.00 interchange protein DsbC Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: carbon starvation EPA1-283 0.00 protein A 145 Blast Mutant Gene interrupted Score Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: osmolarity sensor EPA1-284 0.00 protein EnvZ Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: orotate EPA1-285 0.00 phosphoribosyltransferase, ribonuclease PH Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: aldehyde EPA1-286 0.00 dehydrogenase Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: thiazole biosynthesis EPA1-287 9.00E-90 protein ThiH Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: conserved hypothetical EPA1-288 7.00E-180 protein EPA1-289 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: oligopeptidase A 0.00 EPA1-290 No matches EPA1-291 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: fimbrial protein 4.00E-12 EPA1-293 Possible transposon or insertional sequence EPA1-294 Aeromonas hydrophila, PPD134/91 O-antigen cluster, complete sequence (likely in sequence before start of gene) 1.00E-29 Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: carbamoyl-phosphate EPA1-296 0.00 synthase, large subunit Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: D-tagatose- EPA1-297 3.00E-34 bisphosphate aldolase, class II, non-catalytic (likely in sequence before start of gene) Aeromonas hydrophila, ATCC 7966, complete genome, features in this part of subject sequence: hydrolase, alpha/beta EPA1-299 3.00E-45 fold family
146 A
B
Figure 1. a) Gel image of PCR indicating transposon insertion, b) Gel image from touchdown PCR showing no PCR products from mutants in Lanes 1-4, 8, 10 and 12. Other lanes do have testable PCR products.
147
Figure 2. Swarming assay, mutant on left, wild-type on right
148 SUMMARY
In the initial publication reporting on two virulent A. hydrophila species (Chapter 1), up- regulation of more than 200 genes in neonates and over 50 genes in cell culture were observed.
Results demonstrated that a set of genes up-regulated from infecting whole animals were also found in infected mouse intestinal cells as twenty-six genes were found to be in common between the two models. The majority of these genes were associated with the innate immune response. In addition, surface receptor molecule genes (Icam1, Vcam1, CD14, Tlr2) were also up-regulated. This was an important finding because a goal of the research was to evaluate whether cell culture might replace whole animals for determining bacterial virulence. Whole animal testing is laborious, subject to highly variable results and not suited to the screening of large numbers of isolates. Cell culture results confirmed that epithelial cells are activated by bacteria and attempt to regulate cytokine production.
Based on the promising data described above, expanded testing was performed on a variety of Aeromonas spp., hopefully representing both virulent and avirulent strains. The neonatal mouse model was used to define virulence. Some preliminary LD50 testing with the
neonatal mouse model produced results that were consistent with other published data. Oral
doses of 107-108 virulent A. hydrophila strains organisms caused significant mortality in neonates. However, performing LD50 assays on all strains chosen for testing was not an option because of the large number of animals required and due to objections from the Institutional
Animal Care and Use Committee. Additionally, using LD50 as an indicator of virulence has
disadvantages. Mortality is certainly an unambiguous endpoint, however, as an endpoint for
determining whether a disease state has occurred, it is too conservative. Because disease can
occur without death, transcriptional response provides information on initiation or establishment
149 of non-fatal disease. After a virulence designation was given to each strain based on neonate testing, cell cultures were infected with strains representing virulent and avirulent designations.
In the expanded testing, I chose to define “virulent” as those strains that were able to kill some or all challenged mice and also produce an average gene up-regulation of greater than or equal to 3, at challenge doses of 107-108 CFU/mice. This approach was in part due to peer review criticism (from journal reviewers) that insisted on linking mortality to transcriptional response. As it happens, neonate infections that caused an average transcript up-regulation of greater than 3 also demonstrated unambiguous immune response as demonstrated in Table 3 of
Chapter 2. However, performing the above is similar to arbitrarily setting a LD50 cutoff for classifying virulence, as other researchers have done. This effort proposes that activation of an appropriate immune response, identified by transcriptional profiling, is sufficient for defining an organism as being virulent (inclusion of the dose at which the response is evoked is also essential). An obvious caveat to the above is, if the model is meant to be applicable to human disease, it must be shown to be correlated. This was outside the scope of this effort. The literature does support that Aeromonas species defined as being virulent in the neonatal mouse model are also implicated in human gastro-intestinal disease.
Results of virulent and avirulent strain infections of cell cultures were compared to neonates in terms of transcriptional response. Two transcripts were found in common between infections of virulent Aeromonas strains when comparing the two models' results, Jun and Fos.
Confirmation analysis using qRT-PCR supported the use of the Jun transcript as a predictor of virulence in cell culture. Jun up-regulation in infected cell cultures, while a significant positive finding, is likely a general indicator for Aeromonas virulence. It is also possible that it is only specific to those strains that cause acute infection (due to toxin activity, i.e., tissue damage).
150 Because Jun is a central transcription factor in the cellular apoptosis pathway and integral to AP-
1 transcription factor complexes, it is reasonable that a pathogen would induce a host response
that includes Jun up-regulation. Fos is also part of the AP-1 complexes. Testing for Fos did
approach being significant and might indicate the need to investigate temporal changes rather
than one time point. Kinetic curves for key host response transcripts might better characterize
virulence.
Once a correlation was established between the murine cell culture and live animals in
terms of mRNA response, the next step was to move to a human cell line. I was able to collect
cell culture data by infecting a human adenocarcinoma cell line (Caco-2) that represented colonic
epithelial cells. The strains tested included two virulent A. hydrophila isolates (EPA1 and 9),
one virulent A. veronii biotype sobria (EPA63), one avirulent A. veronii biotype sobria (EPA61)
and two avirulent A. caviae strains (EPA72 and 73). The Caco-2 cell line proved to be very
sensitive to the cytotoxic effects of certain Aeromonas strains, in contrast to the murine cell line
(m-ICcl2). When using the m-ICcl2 cell line for infection experiments, the virulent strains
exhibited similar cytotoxic effects after a 5 hour incubation period, with effects being
homogeneous throughout the monolayer. Using the same doses as with the m-ICcl2 infections, the Caco-2 cells were very sensitive to cytotoxic effects of EPA9 and less so to the other virulent strains tested. I opted to harvest the total RNA from the infected Caco-2 experiments at the point where the cytotoxic effects of EPA9 were evident (4.5 hours). Transcriptional analysis of infected Caco-2 cultures demonstrated few up-regulated genes with the exception of experiments with EPA9. Infections using EPA9 did up-regulate Jun and Fos. Interestingly, no cytokine transcripts were noted as being up-regulated. It is difficult to speculate why Caco2 cells were highly sensitive to EPA9. Microscopic examination of infected cell flasks demonstrated
151 different cytotoxic patterns between the strains. EPA9 affected the entire monolayer in a
homogeneous manner with all cells rounding and releasing from the flask simultaneously. EPA1
showed focal necrosis at multiple points throughout the monolayer. Obviously, the cell line
chosen for infection experiments will affect results and conclusions. The use of a single cell line
will not be a comprehensive indicator for determining whether a bacterial species/strain can be
classified as virulent. Disease mechanisms will dictate which cell lines might be useful for
disease prediction in vivo. Based on literature review, cell culture's greatest utility for assessing
disease mechanisms is potentially in evaluating host response to individual, purified bacterial
proteins.
The definition of virulence provided in Chapter 2 caused all A. caviae strains to be designated as avirulent. This determination was consistent with previous research proposing A. caviae as an "opportunistic" pathogen, requiring either a mixed infection or a weakened immune system for symptom manifestation. Aeromonas caviae generally does not harbor toxins or hemolysins. Filter-sterilized broths containing extracellular products from A. caviae strains from
Cincinnati-EPA's library did not demonstrate any cytotoxicity to cultured intestinal cells (human or murine), even after 24 hours incubation. Aeromonas caviae gastro-intestinal infections have been characterized as being chronic in nature, whereas A. hydrophila and A. veronii biotype sobria are associated with acute infections. Because A. caviae-infected neonatal mice did not experience any signs of morbidity or mortality after 5 hours, an additional set of animals was monitored for gene expression patterns after a longer incubation period (24 hours). Gene response to A. caviae infection at 24 hours presented a Th1 immune response, characterized by the up-regulation of multiple gamma-interferon induced transcripts. qRT-PCR was performed on the same small intestinal extracts used for microarray experiments and confirmed gene
152 expression for five interferon-induced transcripts. No evidence of gamma-interferon up-
regulation was seen by microarray, however, gamma-interferon up-regulation was noted using
qRT-PCR. Th1 immune response has been associated with murine irritable bowel syndrome
(IBS) models. Therefore, my proposal is that A. caviae colonizes the intestinal tract and causes
an immune response similar to what is seen in IBS patients. One side effect of this response is
disruption of the epithelial barrier. Loss of tight junction organization causes paracellular fluid
excretion in vivo leading to gastro-intestinal symptoms. In light of the above, A. caviae would be
elevated to pathogen status along with A. hydrophila and A. veronii biotype sobria, although
mechanistically different in its mode of action.
To summarize the neonate and cell culture testing, a significant and rapid immune
response was generated by infecting the models with some Aeromonas strains. I propose that
apoptosis pathways are activated as evidenced by the up-regulation of genes known to be part of
cellular death pathways or genes that have anti-apoptosis function. Also activated are NF-кB transcriptional events, including inducement and regulation of cytokine production. Other than these cellular events, too many other actions are occurring simultaneously for discerning specific mechanisms. The exception to the above was our observation of the Th1 response from infections with A. caviae. Snapshots using transcriptional profiling can, on occasion, lead to mechanistic discoveries associated with immune response.
Transcriptional analyses were focused exclusively on up-regulated transcripts. The absolute number of up- and downregulated genes was similar when the magnitude of transcript fold change was not a criterion. However, when a fold change of greater than 2 was applied, most down-regulated genes were discarded. If a criterion of greater than 3 fold change was applied, the down-regulated genes almost completely disappeared. Disease initiation is a
153 positive regulatory event characterized by a large increase in cellular transcriptional activities. I did attempt to assess down-regulated transcripts but few similarities were observed when comparing models, or when comparing infections of different strains within models. Down- regulated genes did not exhibit large fold changes in either model and therefore consistent differences are more difficult to distinguish statistically. Other published reports using microarray analysis of host response to bacteria also focused on up-regulated genes.
At the start of this research effort, two Aeromonas strains that were identical in all known traits except for toxin production were tested in the models. An A. hydrophila strain with gene knockouts for the three known enterotoxins (Act, Ast, Alt) was obtained (mutation of SSU strain, donated by Dr. A.K. Chopra, Houston. TX). It was anticipated that differential gene expression from host cells infected with the mutant and wild-type SSU strains would identify physiological processes specific to toxin exposure. The SSU mutant and corresponding wild-type SSU organism were tested in the neonate model. Both demonstrated mortality data and transcriptional profiling patterns that were essentially identical. Cell culture infections with these two strains also did not demonstrate significant differences in transcriptional response.
Even with the failure to see differences in the SSU mutant testing, I chose to pursue production and testing of isogenic mutants. The use of isogenic mutants is often used to assess the influence of single VFs. The weakness associated with the SSU testing described earlier was the targeting of toxins. There is no doubt that toxins play a huge role in pathogenicity. There are reports of higher LD50s from Aeromonas mutants lacking aerolysin and/or HlyA production. My testing with neonates using SSU wild-type and mutant strains indicated that loss of one or more toxins does not necessarily lower an Aeromonas strain’s virulence. Nevertheless, there are a number of VFs whose loss may be more important than loss of a toxin. Lateral flagella and
154 secretion systems have been noted as being possible “essential” VFs. Both lateral flagella and secretion systems are products of gene cassettes. Interruption of an essential gene needed to fully develop either of these VFs could eliminate the organism’s ability to cause disease.
Therefore, isogenic mutant research using a virulent A. hydrophila strain (EPA1) was initiated.
Even though the number of isogenic mutants created was relatively small, I was able to create several bacterial mutants that had insertions in genes mentioned above. Transcriptional profiling using neonates has not been initiated and may not be useful in determining disease mechanisms because we often noted an “all-or-none” type of host transcription response between infections using virulent and avirulent strains. The cell culture model has limitations because
EPA1 produces and excretes cytotoxins that cause severe cellular damage. Therefore, if lateral flagella are eliminated by a gene knockout, any gene regulation difference based on ligand- receptor interactions would be undetected in cell culture because toxin effects produce large gene regulatory events. This was supported by my results using a lateral flagella mutant to infect cell culture (See Chapter 4).
I was able to demonstrate the elimination of swarming mobility in the lateral flagella mutant. This leads to the possibility that the mutant will not colonize live animal intestinal tracts. Although not tested at this time, loss of swarming motility will be correlated to ability to colonize the murine intestinal tract. If so, future testing will then determine whether this is also correlated with loss of virulence by testing in neonates. At a minimum, we would be able to make a statement regarding the significance of lateral flagella to disease establishment.
According to the literature, most (or all) proteins excreted extracellularly by A. hydrophila are through the general secretion pathway (GSP or Type II secretion system). A mutation (EPA1-263) in one of the genes associated with the GSP pathway was created. Future
155 studies will assess its ability to secrete toxins. If this mutant's ability to produce exotoxins is impaired, then infecting neonates with this strain, followed by gene expression assays, might assist in differentiating toxin affects from adhesion effects in vivo. In this case, all excreted toxins are eliminated rather than one. This would not likely affect any membrane bound toxins.
Other conclusions include:
• qRT-PCR testing generally confirmed microarray testing. Discrepancies between
microarray and qRT-PCR testing were associated with Fos and gamma-interferon. As
mentioned above, Fos did approach significance when tested using qRT-PCR. With
gamma-interferon, it is likely a sensitivity issue. Microarrays are not as sensitive for
detecting low copy number mRNAs as is qRT-PCR. However, good correlation was
found in those cases were both assays were able to reliably detect transcript presence.
• Aeromonas allosaccharophila has potential to be a mammalian pathogen (or opportune
pathogen) at high doses. Up to now, this species has been associated with diseases in
eels.
• Based on results of the heat/UV-killed organisms, Aeromonas needs to be metabolically
active to induce a host response. This is in common with other gram negative organisms
such as Salmonella, Helicobacter and Chlamydia. There is no apparent passive
attachment from receptor-ligand interactions with Aeromonas spp.
156 • The limitation in making additional conclusions is potentially due to difficulties in
interpreting the massive database produced. Drawing correct inferences is contingent on
bioinformational techniques and on variation of test systems.
• If a goal is to someday replace animal models for determining virulence, then any
developed approach will likely be multifaceted. A collection of non-animal models or
phenotypic assays would potentially be used in unison, scored, and a prediction of
pathogenicity made. The collection of assays would likely be organism-specific and
would also need to be calibrated against live animal models.
157