IAI Accepts, published online ahead of print on 31 March 2014 Infect. Immun. doi:10.1128/IAI.00054-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
Vibrio cholerae-induced inflammation in the neonatal mouse cholera model
2
Anne L Bishop1,2, Bharathi Patimalla1 and Andrew Camilli1 #
3 4
1
Department of Molecular Biology and Microbiology, Tufts University School of
5
Medicine and Howard Hughes Medical Institute, Boston, MA, USA.
6
2
Currently: Independent scholar, Nottingham, United Kingdom.
7 8
Running Title: Vibrio cholerae-induced inflammation in neonatal mice
9 10
# Corresponding author: Mailing Address: Department of Molecular Biology and
11
Microbiology, Tufts University School of Medicine and Howard Hughes Medical
12
Institute, Boston, MA, 02111, USA. Phone: (617) 636-2144. Fax: (617) 636-2175. E-
13
mail: [email protected]
14 15
1
16
ABSTRACT
17
Vibrio cholerae is the causative agent of the acute diarrhoeal disease of
18
cholera. Innate immune responses to V. cholerae are not a major cause of cholera
19
pathology, characterized by severe, watery diarrhea induced by the action of cholera
20
toxin. Innate response may, however, contribute to resolution of infection and must
21
be required to initiate adaptive responses after natural infection and oral
22
vaccination. Here we investigate whether a well-established infant mouse model of
23
cholera can be used to observe an innate immune response. We also use a
24
vaccination model, where immunized dams protect their pups from infection
25
through breast-milk antibodies, to investigate innate immune responses after V.
26
cholerae infection for pups suckled by an immune dam. At the peak of infection, we
27
observe neutrophil recruitment accompanied by induction of KC, MIP-2, NOS-2, IL-6
28
and IL-17a. Pups suckled by an immunized dam do not mount this response.
29
Accessory toxins RtxA and HlyA play no discernable role in neutrophil recruitment
30
in a wild type background. The innate response to V. cholerae deleted for cholera
31
toxin-encoding phage (CTXφ) and part of rtxA is significantly reduced, suggesting a
32
role for CTXφ-encoded genes or for RtxA in the absence of CTX. Two extracellular V.
33
cholerae DNases are not required for neutrophil recruitment, but ΔDNases V.
34
cholerae causes more clouds of DNA in the intestinal lumen, which appear to be
35
neutrophil-extracellular traps (NETs), suggesting that V. cholerae DNases combat
36
NETs. Thus, the infant mouse model has hither-to unrecognized utility for
37
interrogating innate responses to V. cholerae infection.
38
2
39
INTRODUCTION
40
Vibrio cholerae is the causative agent of cholera, which remains endemic in
41
many regions of Afria and Asia (1). Sporadic outbreaks can devastate
42
immunologically naïve populations, such as Haiti in 2010 (2). The O1 serogroup, and
43
to a lesser degree O139, is the major cause of cholera and El Tor is currently the
44
circulating O1 biotype (1). Cholera toxin (CTX), encoded by a lysogenic phage (3), is
45
the major cause of a severe secretory diarrhea that is typical of cholera. Without re-
46
hydration therapy cholera results in 25-50% mortality, but if treated cholera will
47
resolve in most patients (4). Unlike diseases such as Shigellosis and Salmonella-
48
induced gastroenteritis (5, 6), the pathology of cholera is not immune-driven.
49
However, innate immune responses have been observed in cholera patients during
50
the acute phase of infection (7) and likely play a direct and/ or indirect role in the
51
resolution of disease. The majority of individuals that survive a bout of cholera will
52
become immune to subsequent infection for at least 3 years (8-10) and some
53
aspects of the observed innate immune response must be required to initiate this
54
protective immunity.
55
Many live attenuated vaccine candidate strains of V. cholerae have been
56
developed in which ctxA, the gene encoding the catalytic subunit of cholera toxin, is
57
deleted (11). Ιn volunteers, removal of ctxA did not completely alleviate V. cholerae-
58
induced diarrhea, although it was mild, and signs of inflammation were still
59
observed (12). Bacterial factors responsible for this residual diarrhea and
60
inflammation include flagellin, which is a major V. cholerae pro-inflammatory
61
stimulus in vitro (13-15), in the infant rabbit model of infection (16) and in a small-
3
62
scale human volunteer study (17). In the infant rabbit, flagellin-independent
63
inflammation was still observed, suggesting that V. cholerae encodes other minor
64
pro-inflammatory factors (16). In tissue culture models, purified V. cholerae O1
65
lipopolysaccharide (LPS) induces pro-inflammatory transcription factor NF-κB (18)
66
and Hemolysin A (HlyA) contributes to inflammation induced by non-O1/ non-O139
67
V. cholerae supernatants (19). Repeats toxin, encoded by rtxA and elaborated by the
68
El Tor V. cholerae O1 biotype, but not the extinct Classical biotype (20), was
69
implicated in El Tor-induced inflammation in an adult mouse pulmonary infection
70
model (21) and both HlyA and RtxA are required for virulence in a mouse model of
71
prolonged colonization (22).
72
On the host side of the equation, the molecular interactions that initiate
73
innate immune responses to the non-invasive pathogen V. cholerae, or to oral
74
vaccines derived from the bacterium, are not yet well described. Toll-like receptor 4
75
(TLR4) can be activated by V. cholerae LPS in vitro (18). Also, after nasal infection
76
with V. cholerae lacking pro-inflammatory toxins CTX, RtxA and HlyA, morbidity is
77
greater for mice lacking TLR4 expression, suggesting a role for TLR4-mediated
78
inflammation in bacterial containment in the absence of other sources of
79
inflammation (23). Using human epithelial cell lines, TLR5 was implicated in V.
80
cholerae flagellin-induced inflammation (14, 15), although the sheath covering the V.
81
cholerae flagellum reduces the potency of TLR5 signaling (24). A candidate gene
82
association study found an interesting link between innate immunity and host
83
susceptibility: a locus in the promoter of long palate, lung and nasal epithelium
84
clone 1 (LPLUNC1) is associated with acute cholera (25), LPLUNC1 was also the
4
85
most highly enriched gene transcript in duodenal biopsy samples during acute
86
cholera (26) and the product of LPLUNC1 can inhibit LPS-induced TLR4-mediated
87
NF-κB activation (18). The authors suggest that too much of this anti-inflammatory
88
activity may prevent cholera resolution.
89
Animal models have proven useful for interrogating V. cholerae mechanisms
90
of pathogenesis, but less so for examining host immune responses. Inflammation
91
due to V. cholerae has been studied with an adult mouse pulmonary infection model,
92
but this is neither the natural route nor site of infection (21). Adult mice are difficult
93
to colonize orally with V. cholerae, but prolonged chronic colonization can be
94
achieved either by antibiotic treatment or neutralization of stomach acid combined
95
with ketamine-xylazine treatment (22, 27). An acute lethal oral infection can be
96
induced in adult mice, but this requires a high inoculum of V. cholerae and death is
97
associated with systemic spread, which is not a normal cholera phenotype (28, 29).
98
For 4-6 days neonatal mice are permissive for V. cholerae colonization after oral
99
inoculation, leading to an acute and potentially lethal infection of the small intestine
100
without systemic spread, much like severe cholera in humans (30). Key bacterial
101
virulence factors that are required for acute disease in humans, most notably CTX
102
and toxin-co-regulated pilus (Tcp) (31), are also induced in the neonatal mouse
103
model, suggesting it is a good model for acute cholera (32). Being a neonatal model
104
could be considered an advantage, as a functional adaptive immune system has not
105
yet developed, leaving the innate response to study in isolation. However, lacking a
106
mature adaptive immune system means that neonates do not respond well to
107
vaccination, thus limiting use of mice for developing vaccines. To circumvent this,
5
108
we have used a mouse model in which immunized mothers protect their young
109
through antibodies in their milk in order to study V. cholerae outer-membrane
110
vesicles (OMVs) as a potential cholera vaccine (33).
111
Here we show that neonatal mice can be used as a model to investigate
112
innate immune response to V. cholerae infection. At the peak of infection we observe
113
increased transcription of a subset of pro-inflammatory factors, accompanied by
114
neutrophil recruitment and the appearance of what appear to be neutrophil
115
extracellular traps (NETs). We investigate bacterial factors that may contribute to
116
this inflammation and find possible roles for CTX, or for RtxA in the absence of CTX,
117
and for secreted DNases, but not for HlyA or RtxA in a wild type background. Finally,
118
we determine that intestinal inflammation is abrogated when neonates challenged
119
with V. cholerae are suckled by OMV-immunization dams.
120 121
6
122
MATERIALS AND METHODS
123
Bacterial strains and culture conditions
124
V. cholerae strains used in this study are listed in Table 1. V. cholerae was
125
cultured in Luria-Bertani (LB) broth with aeration or on LB agar plates,
126
supplemented with 100 μg ml-1 Streptomycin (Sm) at 37oC.
127
A mutant of O1 El Tor strain E7946 was generated in which almost all of rtxA
128
(VC1451) was deleted. The deletion begins at the first base after the stop-codon of
129
the overlapping upstream gene VC1450, leaving five codons and one base of rtxA
130
intact, followed by the rtxA stop codon. A PCR product containing the deletion was
131
generated using strand overlap extension (34). Overlapping PCR products were
132
generated
133
GCGTCTAGACGTTCAGACAGTCTCAAGTGTTAAC-3’
134
TACGCGATTATTAAATAGAAAACCACTGCACCT-3’ for the upstream product; rtxAF2
135
5’-TTCTATTTAATAATCGCGTAGCCCAAATTTAAG-3’
136
GCGTCTAGACATCACGCAACCTCGTGATTG-3’ for the downstream product. A full-
137
length deletion PCR product, which contains XbaI sites at each end, was amplified
138
with rtxAF1b and rtxAR2b, digested with XbaI and ligated into XbaI-digested
139
pCVD442 to generate pCVD442-ΔrtxA. pCVD442-ΔrtxA was transformed into
140
Sm10λpir and ex-conjugates with V. cholerae E7946 were selected for single cross-
141
over events (50 μg ml-1 ampicillin-resistant), followed by counter-selection for
142
double-crossover deletion mutants in which the plasmid had excised and was lost
143
(10% sucrose-resistant, due to loss of the plasmid-encoded sacB). The mutation was
using
the
following
7
primers: and
and
rtxAF1b
5’-
rtxAR1
5’-
rtxAR2b
5’-
144
confirmed by PCR with primers rtxAF0 (5’-ATGGGGAAATTTATGATGGAG-3’) and
145
rtxAR0 (5’-GCCAGGTTGGTAAGACTTTG-3’) and sequencing of the amplified product.
146
Mice
147
Wild type C57Bl/6 or BALB/c mice were acquired from Charles River
148
Laboratories. All mice were housed with food and water ad libitum and monitored
149
in accordance with the rules of the Department of Laboratory Animal Medicine at
150
Tufts Univerity School of Medicine.
151
V. cholerae infection of infant mice
152
Five- to six-day-old C57Bl/6 or BALB/c mice were orally infected with ca. 105
153
wild type or mutant V. cholerae, taken from an LB agar plate into LB broth, or sham-
154
infected with LB broth carrier. 5 μl of food coloring per ml of infection mix was
155
added to each infection mix in order to visually verify that the inoculum entered the
156
stomach of the mice. Each inoculum was serial diluted and plated for viable counts
157
to check the infection dose. 7 h, 24 h or 30 h post-infection small intestines were
158
harvested for histology or RNA.
159
Immunization and challenge
160
V. cholerae OMVs were prepared from strain E7946 and used to immunize
161
six-week-old BALB/c mice intra-nasally with 25 μg doses on days 0, 14 and 28,
162
followed by mating (day 40) and challenge of infants born to and suckled by
163
immunized mice, as described previously (35). V. cholerae challenge was carried out
164
as described above, except where hyper-infectious mouse-passaged challenge was
165
used, which was carried out as described previously (36), with details as follows.
166
For hyper-infectious challenge, naïve infant mice were first infected with in vitro-
8
167
grown V. cholerae as described above, 24 h post-infection small intestines were each
168
harvested, homogenized in 1 ml saline, 2 or 3 homogenates were pooled, roughly
169
filtered (100 μm pore size) and diluted 10-fold to give an ca. 1000 CFU input dose in
170
50 μl. This infection mix was used to challenge neonates born to OMV-immunized or
171
un-immunized control mice.
172
Quantitative reverse transcriptase PCR (qRT-PCR)
173
Small intestines were snap frozen and stored at -80oC until ready for RNA
174
preparation. Frozen intestines were placed into 1 ml of Trizol (Invitrogen) per 100
175
mg of tissue, 250 μl of 0.1 mm zyrconia beads was added and samples were
176
homogenized by bead-beating twice for 1 min, with 1 min on ice between beatings.
177
After centrifugation at 13,000 rpm for 2 min the supernatant was removed from the
178
beads. An equal volume of chloroform was added, mixed and centrifuged at 13,000
179
rpm for 2 min. The upper aqueous phase was decanted into a fresh tube and 100%
180
ethanol was added to give a final concentration of 35%. This solution was place onto
181
an RNeasy column, centrifuged, washed and eluted in 50 μl of RNase-free water,
182
according to the manufacturer’s instructions (Qiagen). DNA was removed from the
183
RNA with the DNAfree kit (Ambion) using their stringent protocol, according to the
184
manufacturer’s instructions.
185
cDNA was prepared using 1 μg of RNA and random primers with iSCRIPT
186
Select (Biorad) according to the manufacturer’s instructions. For each RNA sample,
187
RT minus controls were prepared containing RNase-free dH2O and 1 μg of RNA. For
188
qRT-PCR 1 μl of cDNA (or RT minus control mix) was used in a 20 μl reaction
189
volume with 300 nM forward and reverse primers and 1x iQ SYBR green supermix 9
190
(Biorad). PCRs were run on a MX3000P machine (Stratagene) using cycle conditions
191
as follows: 1 cycle 95oC 10 min; 40 cycles of 95oC 30s, 55 to 63oC (see Table 2 for
192
annealing temperatures for each primer set) 1 min and 72oC 30s; 1 cycle for
193
dissociation curve 95oC 1 min, annealing 55 to 63oC (as for amplification cycles) 30s,
194
95oC 30s. Primers used are listed in Table 2. Each reaction was run in duplicate, plus
195
an RT minus control to determine background amplification. For each new primer
196
set standard curves, using a 2 fold dilution series of template (cDNA from infected
197
mouse intestine for murine genes or genomic DNA for V. cholerae genes), were run
198
and PCR samples were separate on 2% agarose gels + 1x GelGreen dye (Biotium) to
199
determine whether the amplicon sizes (listed in Table 2) were correct. All primer
200
sets had at least 93% efficiency of amplification. Glyceraldehyde 3-phosphate
201
dehydrogenase (GAPDH) murine control gene was used to normalize qRT-PCR data
202
for the test genes, cycle threshold (CT) values were converted to relative transcript
203
levels using the 2-
204
Immuno-histochemistry and immuno-fluorescence staining of intestinal tissue
ΔΔC T method (37) and expressed relative to a calibrator sample.
205
Small intestines for immuno-histochemistry (IHC) or immuno-fluorescence
206
(IF) staining were dissected just below the stomach and included the caecum (not
207
included for RNA analysis) for easy identification of the distal end of the small
208
intestine. They were fixed in 10% neutral buffered formaldehyde (BDH) in a jelly-
209
roll configuration between two histology sponges in a histology cassette for at least
210
24 h. Tissue was paraffin-embedded, 4 μm lateral sections were cut through the
211
middle of the tissue. Samples were de-paraffinized in Xylene (2x 5min) then re-
212
hydrated by serial 5 min incubations in 100% (2x), 90% and 70% ethanol and 10
213
finally into distilled water. All staining steps were carried out at room temperature.
214
Tissue sections were surrounded by hydrophobic pen, treated with 20 μg ml-1
215
proteinase K (Sigma) in phosphate-buffered saline (PBS) for 30 min, for the
216
purposes of antigen retrieval, and washed 2x 5 min in PBS 0.05% Tween-20 (PBST).
217
For IHC, samples were treated with peroxidase blocking solution (DakoCytomation
218
LSAB2 System-HRP) for 10 min then rinsed with dH2O and placed into PBS. In all
219
cases, samples were blocked with 5% goat serum + 1% BSA in PBS (block) for 1 h at
220
room temperature.
221
For IF, samples were stained for 1 h with FITC-conjugated mouse IgG1 anti-V.
222
cholerae O1 antibody (250-fold dilution, ca. 10 μg ml-1, New Horizons) plus a DNA
223
stain (2 μg ml-1 Hoescht 33342, Molecular Probes), to reveal bacteria and tissue
224
morphology, and a lectin that highlights mucus (2 μg ml-1 TRITC-Wheat Germ
225
Agglutinin, Sigma Aldrich). Alternatively, bacterial and DNA staining were combined
226
with neutrophil staining using rat IgG2a anti-neutrophil antibody 7/4 (abcam) for 1
227
h and anti-rat IgG-TRITC (Jackson) secondary antibody for 30 min.
228
For IHC staining of neutrophils, rat IgG2a anti-neutrophil antibody 7/4
229
(abcam) was used for C57BL/6 mice and rat IgG2b anti-neutrophil antibody NIMP-
230
R14 (abcam) for BALB/c mice, with goat anti-rat IgG-biotin secondary antibody
231
(DakoCytomation). Primary and secondary antibody incubations were for 1 h or 30
232
min,
233
Diaminobenzidine (DAB) + horseradish peroxidase (HRP) substrate for 5 min
234
(DakoCytomation LSAB2 System-HRP).
respectively,
followed
by
Streptavidin-HRP
11
for
10
min
and
3,3-
235
Washing after DAB+HRP incubation was carried out with dH2O in a squeeze
236
bottle. Other washes were carried out for 3x 5min in PBST. Finally, for IHC, samples
237
were counter-stained with Gill’s Hematoxylin (Polysciences, Inc.) for 2 min, washed
238
in dH2O to remove excess dye, then in 1x automation reagent (Fisher Scientific) until
239
blue. All antibodies were diluted in block solution for staining. Negative control
240
FITC-mouse IgG1 (Sigma) and rat IgG2a or IgG2b (abcam), as appropriate, were
241
used to stain consecutive sections in order to confirm the specificity of the staining.
242
IF-stained samples were mounted under a coverslip in mountant: 2.4g of
243
Mowiol 488 [Calbiochem] + 6 g of glycerol + 6 ml water, mixed, then buffered with
244
12 ml of 0.2 M Tris pH 8.5 and heated to 50oC to mix, clarified by centrifugation,
245
before adding DABCO (1,4,-diazobicycli-[2.2.2]-octane, Aldrich) to 2.5% w/v to
246
reduce fading of fluorophores. For IHC, samples were dehydrated in a reverse of the
247
ethanol and Xylene steps described above, then mounted under a coverslip in Poly-
248
Mount Xylene (Polysciences Inc.). Samples were viewed using a Nikon 80i
249
microscope, images were captured with NIS elements software and processed with
250
Image J software. Counting of stained cells, or numbers of DNA clouds, was carried
251
out blinded to the identity of the samples, for the entire length of the small intestine
252
for one jelly-rolled section for each mouse.
253
Statistics
254
The majority of data was not normally distributed, therefore non-parametric tests
255
were utilized: Mann Whitney U tests were used to compare two groups and Kruskal
256
Wallis test and post-hoc Dunn’s multiple comparison tests for more than two
257
groups. GraphPad Prism 5.0a was used for all statistical analysis.
258 12
259
RESULTS
260
Inflammatory markers observed by qRT-PCR in the infant mouse small intestine
261
after V. cholerae infection
262
We sought to determine whether the infant mouse model of acute V. cholerae
263
colonization could be used to interrogate innate immune responses to V. cholerae in
264
this genetically tractable host. We used C57Bl/6 mice because many transgenic mice
265
would be available in this background for any subsequent studies of genetic
266
susceptibility to inflammation in response to V. cholerae infection. We used qRT-PCR
267
to screen for increased transcription of pro-inflammatory markers that are induced
268
in response to V. cholerae infection in other infection models and/or in cholera
269
patient samples upon wild type V. cholerae infection. Transcription of chosen
270
inflammatory markers was monitored in the small intestine of infant mice 7, 24 and
271
30 h after infection with wild type V. cholerae strain E7946 (O1 Ogawa El Tor) and
272
in sham vehicle-infected controls.
273
We looked for the Muc-1 transcript, as it is a cell-associated mucin whose
274
expression is induced during acute cholera in patient biopsies (26). We tested for
275
induction of pro-inflammatory cytokines TNFα (Tumor Necrosis Factor alpha), IL-6
276
(Interleukin-6) and IL-1β, because their transcript and protein levels were
277
increased in epithelial cell lines in response to V. cholerae or pure V. cholerae
278
flagellin (IL-1β) (15, 38) and in infant rabbits in response to oral V. cholerae
279
infection (IL-1β and TNFα were flagellin-dependent) (16). IL-6 is also induced in
280
adult mouse V. cholerae infection models (39, 40) and TNFα is increased in serum of
281
mice infected with V. cholerae via the pulmonary route (21). Importantly, IL-1β and 13
282
TNFα were also elevated in acute cholera patient biopsies (26, 41). We looked at the
283
chemokine CCL20 (also known as MIP-3α), which signals to recruit dendritic cells in
284
response to pro-inflammatory mediators including flagellin (42), as CCL20 was
285
increased in serum of adult mice after pulmonary V. cholerae infection (21).
286
We also tested for murine IL-8 family chemokines, which signal for
287
recruitment of polymorphonuclear cells (PMNs, also known as neutrophils) to sites
288
of infection: KC (also known as mCXCL1) and macrophage inflammatory protein-2
289
(MIP-2, also known as mCXCL2). MIP-2 was increased in serum of mice infected
290
with V. cholerae via the pulmonary route (21) and in adult mouse oral infection
291
models (39, 40). Human IL-8, and related chemokines Groα and Groβ, were induced
292
in epithelial cell lines in response to V. cholerae (38, 43) and IL-8 was induced in
293
infant rabbits in response to oral V. cholerae, in a flagellin-dependent manner (16).
294
We monitored transcription of the anti-bacterial factor NOS-2 (also known as iNOS,
295
inducible nitric oxide synthase), as increased NO metabolites were detected in
296
serum of V. cholerae-infected volunteers (44) and cholera patients (7, 45). The
297
cytokine IL-17a, which in adults would lead to a TH17 T-cell response, was
298
monitored. Although IL-17a is produced by TH17 T-cells, which are not yet fully
299
developed in neonates, it is also produced by innate immune cells, which are likely
300
to be present in neonates (46). In addition to the above, we looked at IFNγ
301
(interferon gamma) and IL-10, as examples of classic TH1-inducing or anti-
302
inflammatory cytokines respectively, and because IL-10 was seen to be modestly
303
induced in an adult mouse model of infection (39).
14
304
24 and 30 h post-infection we observed higher transcription after V. cholerae
305
infection, compared to sham-infected controls, for five pro-inflammatory markers:
306
KC, MIP-2 (statistically significant only at 30h), NOS-2, IL-6 and IL-17a (statistically
307
significant only at 24h) (Fig. 1A-E). 7 h post-infection there was no significant
308
difference in transcription of any of the pro-inflammatory markers tested, compared
309
to sham-infected controls (Fig. 1A-E and Supplemental Fig. S1). Transcription of
310
TNFα, IL-1β, CCL20, Muc1, IFNγ and IL-10 were not significantly higher in V.
311
cholerae-infected samples than sham-infected controls at any time-point tested
312
(Supplemental Fig. S1). Colonization was monitored by qRT-PCR for V. cholerae 16S
313
(Fig. 1F) and ctxA (Fig. 1G). All V. cholerae-infected samples were similarly colonized
314
24 to 30 h post-infection (Fig. 1F). 7 h post-infection the V. cholerae burden is not
315
yet maximal and the 16S signal is much lower and more variable than at 24h (Fig.
316
1F), consistent with intestinal colony forming units tested in separate experiments
317
(data not shown).
318
Neutrophil recruitment in the infant mouse small intestine after V. cholerae
319
infection
320
Consistent with transcriptional activation of MIP-2 and KC chemokines that
321
attract PMNs (Fig. 1A-B), a higher median level of PMNs was seen by immuno-
322
histochemical staining at 24 h, but not 2 h or 7 h, post-infection with V. cholerae (Fig.
323
2 and data not shown). Numbers of PMNs were counted in Peyer’s Patches, the
324
Lamina Propria and the lumen of the small intestine. We never detected luminal
325
PMNs in sham controls, whereas there were a significant number of PMNs in the
326
lumen 24 h after infection (Fig. 2Bii). There was also an increase in PMNs in Peyer’s
15
327
Patches and the Lamina Propria (Fig. 2Bi and iii). Recruitment of PMNs to the
328
Lamina Propria did not reach statistical significance in this particular experiment
329
(Fig. 2Bi), but was statistically significant in subsequent experiments (see Table 3
330
Sham vs WT). There was wide variation in PMN numbers between Peyer’s Patches
331
within an individual, with higher PMN numbers towards the distal (ileum) end of
332
the intestine compared to the proximal (duodenum) end (Fig. 2Biii and data not
333
shown). At the apical edge of the epithelium overlying Peyer’s Patches and some
334
regions of the Lamina Propria, PMNs were caught in the process of exiting the tissue
335
and entering the lumen, for example Fig. 2Aii. On an individual mouse-by-mouse
336
basis, there was a significant positive correlation between the number of PMNs in
337
the Lamina Propria and the number in the lumen (Spearman rank correlation
338
p7h
1
Vibrio cholerae-induced inflammation in the neonatal mouse cholera model
2
Anne L Bishop1,2, Bharathi Patimalla1 and Andrew Camilli1 #
3 4
1
Department of Molecular Biology and Microbiology, Tufts University School of
5
Medicine and Howard Hughes Medical Institute, Boston, MA, USA.
6
2
Currently: Independent scholar, Nottingham, United Kingdom.
7 8
Running Title: Vibrio cholerae-induced inflammation in neonatal mice
9 10
# Corresponding author: Mailing Address: Department of Molecular Biology and
11
Microbiology, Tufts University School of Medicine and Howard Hughes Medical
12
Institute, Boston, MA, 02111, USA. Phone: (617) 636-2144. Fax: (617) 636-2175. E-
13
mail: [email protected]
14 15
1
16
ABSTRACT
17
Vibrio cholerae is the causative agent of the acute diarrhoeal disease of
18
cholera. Innate immune responses to V. cholerae are not a major cause of cholera
19
pathology, characterized by severe, watery diarrhea induced by the action of cholera
20
toxin. Innate response may, however, contribute to resolution of infection and must
21
be required to initiate adaptive responses after natural infection and oral
22
vaccination. Here we investigate whether a well-established infant mouse model of
23
cholera can be used to observe an innate immune response. We also use a
24
vaccination model, where immunized dams protect their pups from infection
25
through breast-milk antibodies, to investigate innate immune responses after V.
26
cholerae infection for pups suckled by an immune dam. At the peak of infection, we
27
observe neutrophil recruitment accompanied by induction of KC, MIP-2, NOS-2, IL-6
28
and IL-17a. Pups suckled by an immunized dam do not mount this response.
29
Accessory toxins RtxA and HlyA play no discernable role in neutrophil recruitment
30
in a wild type background. The innate response to V. cholerae deleted for cholera
31
toxin-encoding phage (CTXφ) and part of rtxA is significantly reduced, suggesting a
32
role for CTXφ-encoded genes or for RtxA in the absence of CTX. Two extracellular V.
33
cholerae DNases are not required for neutrophil recruitment, but ΔDNases V.
34
cholerae causes more clouds of DNA in the intestinal lumen, which appear to be
35
neutrophil-extracellular traps (NETs), suggesting that V. cholerae DNases combat
36
NETs. Thus, the infant mouse model has hither-to unrecognized utility for
37
interrogating innate responses to V. cholerae infection.
38
2
39
INTRODUCTION
40
Vibrio cholerae is the causative agent of cholera, which remains endemic in
41
many regions of Afria and Asia (1). Sporadic outbreaks can devastate
42
immunologically naïve populations, such as Haiti in 2010 (2). The O1 serogroup, and
43
to a lesser degree O139, is the major cause of cholera and El Tor is currently the
44
circulating O1 biotype (1). Cholera toxin (CTX), encoded by a lysogenic phage (3), is
45
the major cause of a severe secretory diarrhea that is typical of cholera. Without re-
46
hydration therapy cholera results in 25-50% mortality, but if treated cholera will
47
resolve in most patients (4). Unlike diseases such as Shigellosis and Salmonella-
48
induced gastroenteritis (5, 6), the pathology of cholera is not immune-driven.
49
However, innate immune responses have been observed in cholera patients during
50
the acute phase of infection (7) and likely play a direct and/ or indirect role in the
51
resolution of disease. The majority of individuals that survive a bout of cholera will
52
become immune to subsequent infection for at least 3 years (8-10) and some
53
aspects of the observed innate immune response must be required to initiate this
54
protective immunity.
55
Many live attenuated vaccine candidate strains of V. cholerae have been
56
developed in which ctxA, the gene encoding the catalytic subunit of cholera toxin, is
57
deleted (11). Ιn volunteers, removal of ctxA did not completely alleviate V. cholerae-
58
induced diarrhea, although it was mild, and signs of inflammation were still
59
observed (12). Bacterial factors responsible for this residual diarrhea and
60
inflammation include flagellin, which is a major V. cholerae pro-inflammatory
61
stimulus in vitro (13-15), in the infant rabbit model of infection (16) and in a small-
3
62
scale human volunteer study (17). In the infant rabbit, flagellin-independent
63
inflammation was still observed, suggesting that V. cholerae encodes other minor
64
pro-inflammatory factors (16). In tissue culture models, purified V. cholerae O1
65
lipopolysaccharide (LPS) induces pro-inflammatory transcription factor NF-κB (18)
66
and Hemolysin A (HlyA) contributes to inflammation induced by non-O1/ non-O139
67
V. cholerae supernatants (19). Repeats toxin, encoded by rtxA and elaborated by the
68
El Tor V. cholerae O1 biotype, but not the extinct Classical biotype (20), was
69
implicated in El Tor-induced inflammation in an adult mouse pulmonary infection
70
model (21) and both HlyA and RtxA are required for virulence in a mouse model of
71
prolonged colonization (22).
72
On the host side of the equation, the molecular interactions that initiate
73
innate immune responses to the non-invasive pathogen V. cholerae, or to oral
74
vaccines derived from the bacterium, are not yet well described. Toll-like receptor 4
75
(TLR4) can be activated by V. cholerae LPS in vitro (18). Also, after nasal infection
76
with V. cholerae lacking pro-inflammatory toxins CTX, RtxA and HlyA, morbidity is
77
greater for mice lacking TLR4 expression, suggesting a role for TLR4-mediated
78
inflammation in bacterial containment in the absence of other sources of
79
inflammation (23). Using human epithelial cell lines, TLR5 was implicated in V.
80
cholerae flagellin-induced inflammation (14, 15), although the sheath covering the V.
81
cholerae flagellum reduces the potency of TLR5 signaling (24). A candidate gene
82
association study found an interesting link between innate immunity and host
83
susceptibility: a locus in the promoter of long palate, lung and nasal epithelium
84
clone 1 (LPLUNC1) is associated with acute cholera (25), LPLUNC1 was also the
4
85
most highly enriched gene transcript in duodenal biopsy samples during acute
86
cholera (26) and the product of LPLUNC1 can inhibit LPS-induced TLR4-mediated
87
NF-κB activation (18). The authors suggest that too much of this anti-inflammatory
88
activity may prevent cholera resolution.
89
Animal models have proven useful for interrogating V. cholerae mechanisms
90
of pathogenesis, but less so for examining host immune responses. Inflammation
91
due to V. cholerae has been studied with an adult mouse pulmonary infection model,
92
but this is neither the natural route nor site of infection (21). Adult mice are difficult
93
to colonize orally with V. cholerae, but prolonged chronic colonization can be
94
achieved either by antibiotic treatment or neutralization of stomach acid combined
95
with ketamine-xylazine treatment (22, 27). An acute lethal oral infection can be
96
induced in adult mice, but this requires a high inoculum of V. cholerae and death is
97
associated with systemic spread, which is not a normal cholera phenotype (28, 29).
98
For 4-6 days neonatal mice are permissive for V. cholerae colonization after oral
99
inoculation, leading to an acute and potentially lethal infection of the small intestine
100
without systemic spread, much like severe cholera in humans (30). Key bacterial
101
virulence factors that are required for acute disease in humans, most notably CTX
102
and toxin-co-regulated pilus (Tcp) (31), are also induced in the neonatal mouse
103
model, suggesting it is a good model for acute cholera (32). Being a neonatal model
104
could be considered an advantage, as a functional adaptive immune system has not
105
yet developed, leaving the innate response to study in isolation. However, lacking a
106
mature adaptive immune system means that neonates do not respond well to
107
vaccination, thus limiting use of mice for developing vaccines. To circumvent this,
5
108
we have used a mouse model in which immunized mothers protect their young
109
through antibodies in their milk in order to study V. cholerae outer-membrane
110
vesicles (OMVs) as a potential cholera vaccine (33).
111
Here we show that neonatal mice can be used as a model to investigate
112
innate immune response to V. cholerae infection. At the peak of infection we observe
113
increased transcription of a subset of pro-inflammatory factors, accompanied by
114
neutrophil recruitment and the appearance of what appear to be neutrophil
115
extracellular traps (NETs). We investigate bacterial factors that may contribute to
116
this inflammation and find possible roles for CTX, or for RtxA in the absence of CTX,
117
and for secreted DNases, but not for HlyA or RtxA in a wild type background. Finally,
118
we determine that intestinal inflammation is abrogated when neonates challenged
119
with V. cholerae are suckled by OMV-immunization dams.
120 121
6
122
MATERIALS AND METHODS
123
Bacterial strains and culture conditions
124
V. cholerae strains used in this study are listed in Table 1. V. cholerae was
125
cultured in Luria-Bertani (LB) broth with aeration or on LB agar plates,
126
supplemented with 100 μg ml-1 Streptomycin (Sm) at 37oC.
127
A mutant of O1 El Tor strain E7946 was generated in which almost all of rtxA
128
(VC1451) was deleted. The deletion begins at the first base after the stop-codon of
129
the overlapping upstream gene VC1450, leaving five codons and one base of rtxA
130
intact, followed by the rtxA stop codon. A PCR product containing the deletion was
131
generated using strand overlap extension (34). Overlapping PCR products were
132
generated
133
GCGTCTAGACGTTCAGACAGTCTCAAGTGTTAAC-3’
134
TACGCGATTATTAAATAGAAAACCACTGCACCT-3’ for the upstream product; rtxAF2
135
5’-TTCTATTTAATAATCGCGTAGCCCAAATTTAAG-3’
136
GCGTCTAGACATCACGCAACCTCGTGATTG-3’ for the downstream product. A full-
137
length deletion PCR product, which contains XbaI sites at each end, was amplified
138
with rtxAF1b and rtxAR2b, digested with XbaI and ligated into XbaI-digested
139
pCVD442 to generate pCVD442-ΔrtxA. pCVD442-ΔrtxA was transformed into
140
Sm10λpir and ex-conjugates with V. cholerae E7946 were selected for single cross-
141
over events (50 μg ml-1 ampicillin-resistant), followed by counter-selection for
142
double-crossover deletion mutants in which the plasmid had excised and was lost
143
(10% sucrose-resistant, due to loss of the plasmid-encoded sacB). The mutation was
using
the
following
7
primers: and
and
rtxAF1b
5’-
rtxAR1
5’-
rtxAR2b
5’-
144
confirmed by PCR with primers rtxAF0 (5’-ATGGGGAAATTTATGATGGAG-3’) and
145
rtxAR0 (5’-GCCAGGTTGGTAAGACTTTG-3’) and sequencing of the amplified product.
146
Mice
147
Wild type C57Bl/6 or BALB/c mice were acquired from Charles River
148
Laboratories. All mice were housed with food and water ad libitum and monitored
149
in accordance with the rules of the Department of Laboratory Animal Medicine at
150
Tufts Univerity School of Medicine.
151
V. cholerae infection of infant mice
152
Five- to six-day-old C57Bl/6 or BALB/c mice were orally infected with ca. 105
153
wild type or mutant V. cholerae, taken from an LB agar plate into LB broth, or sham-
154
infected with LB broth carrier. 5 μl of food coloring per ml of infection mix was
155
added to each infection mix in order to visually verify that the inoculum entered the
156
stomach of the mice. Each inoculum was serial diluted and plated for viable counts
157
to check the infection dose. 7 h, 24 h or 30 h post-infection small intestines were
158
harvested for histology or RNA.
159
Immunization and challenge
160
V. cholerae OMVs were prepared from strain E7946 and used to immunize
161
six-week-old BALB/c mice intra-nasally with 25 μg doses on days 0, 14 and 28,
162
followed by mating (day 40) and challenge of infants born to and suckled by
163
immunized mice, as described previously (35). V. cholerae challenge was carried out
164
as described above, except where hyper-infectious mouse-passaged challenge was
165
used, which was carried out as described previously (36), with details as follows.
166
For hyper-infectious challenge, naïve infant mice were first infected with in vitro-
8
167
grown V. cholerae as described above, 24 h post-infection small intestines were each
168
harvested, homogenized in 1 ml saline, 2 or 3 homogenates were pooled, roughly
169
filtered (100 μm pore size) and diluted 10-fold to give an ca. 1000 CFU input dose in
170
50 μl. This infection mix was used to challenge neonates born to OMV-immunized or
171
un-immunized control mice.
172
Quantitative reverse transcriptase PCR (qRT-PCR)
173
Small intestines were snap frozen and stored at -80oC until ready for RNA
174
preparation. Frozen intestines were placed into 1 ml of Trizol (Invitrogen) per 100
175
mg of tissue, 250 μl of 0.1 mm zyrconia beads was added and samples were
176
homogenized by bead-beating twice for 1 min, with 1 min on ice between beatings.
177
After centrifugation at 13,000 rpm for 2 min the supernatant was removed from the
178
beads. An equal volume of chloroform was added, mixed and centrifuged at 13,000
179
rpm for 2 min. The upper aqueous phase was decanted into a fresh tube and 100%
180
ethanol was added to give a final concentration of 35%. This solution was place onto
181
an RNeasy column, centrifuged, washed and eluted in 50 μl of RNase-free water,
182
according to the manufacturer’s instructions (Qiagen). DNA was removed from the
183
RNA with the DNAfree kit (Ambion) using their stringent protocol, according to the
184
manufacturer’s instructions.
185
cDNA was prepared using 1 μg of RNA and random primers with iSCRIPT
186
Select (Biorad) according to the manufacturer’s instructions. For each RNA sample,
187
RT minus controls were prepared containing RNase-free dH2O and 1 μg of RNA. For
188
qRT-PCR 1 μl of cDNA (or RT minus control mix) was used in a 20 μl reaction
189
volume with 300 nM forward and reverse primers and 1x iQ SYBR green supermix 9
190
(Biorad). PCRs were run on a MX3000P machine (Stratagene) using cycle conditions
191
as follows: 1 cycle 95oC 10 min; 40 cycles of 95oC 30s, 55 to 63oC (see Table 2 for
192
annealing temperatures for each primer set) 1 min and 72oC 30s; 1 cycle for
193
dissociation curve 95oC 1 min, annealing 55 to 63oC (as for amplification cycles) 30s,
194
95oC 30s. Primers used are listed in Table 2. Each reaction was run in duplicate, plus
195
an RT minus control to determine background amplification. For each new primer
196
set standard curves, using a 2 fold dilution series of template (cDNA from infected
197
mouse intestine for murine genes or genomic DNA for V. cholerae genes), were run
198
and PCR samples were separate on 2% agarose gels + 1x GelGreen dye (Biotium) to
199
determine whether the amplicon sizes (listed in Table 2) were correct. All primer
200
sets had at least 93% efficiency of amplification. Glyceraldehyde 3-phosphate
201
dehydrogenase (GAPDH) murine control gene was used to normalize qRT-PCR data
202
for the test genes, cycle threshold (CT) values were converted to relative transcript
203
levels using the 2-
204
Immuno-histochemistry and immuno-fluorescence staining of intestinal tissue
ΔΔC T method (37) and expressed relative to a calibrator sample.
205
Small intestines for immuno-histochemistry (IHC) or immuno-fluorescence
206
(IF) staining were dissected just below the stomach and included the caecum (not
207
included for RNA analysis) for easy identification of the distal end of the small
208
intestine. They were fixed in 10% neutral buffered formaldehyde (BDH) in a jelly-
209
roll configuration between two histology sponges in a histology cassette for at least
210
24 h. Tissue was paraffin-embedded, 4 μm lateral sections were cut through the
211
middle of the tissue. Samples were de-paraffinized in Xylene (2x 5min) then re-
212
hydrated by serial 5 min incubations in 100% (2x), 90% and 70% ethanol and 10
213
finally into distilled water. All staining steps were carried out at room temperature.
214
Tissue sections were surrounded by hydrophobic pen, treated with 20 μg ml-1
215
proteinase K (Sigma) in phosphate-buffered saline (PBS) for 30 min, for the
216
purposes of antigen retrieval, and washed 2x 5 min in PBS 0.05% Tween-20 (PBST).
217
For IHC, samples were treated with peroxidase blocking solution (DakoCytomation
218
LSAB2 System-HRP) for 10 min then rinsed with dH2O and placed into PBS. In all
219
cases, samples were blocked with 5% goat serum + 1% BSA in PBS (block) for 1 h at
220
room temperature.
221
For IF, samples were stained for 1 h with FITC-conjugated mouse IgG1 anti-V.
222
cholerae O1 antibody (250-fold dilution, ca. 10 μg ml-1, New Horizons) plus a DNA
223
stain (2 μg ml-1 Hoescht 33342, Molecular Probes), to reveal bacteria and tissue
224
morphology, and a lectin that highlights mucus (2 μg ml-1 TRITC-Wheat Germ
225
Agglutinin, Sigma Aldrich). Alternatively, bacterial and DNA staining were combined
226
with neutrophil staining using rat IgG2a anti-neutrophil antibody 7/4 (abcam) for 1
227
h and anti-rat IgG-TRITC (Jackson) secondary antibody for 30 min.
228
For IHC staining of neutrophils, rat IgG2a anti-neutrophil antibody 7/4
229
(abcam) was used for C57BL/6 mice and rat IgG2b anti-neutrophil antibody NIMP-
230
R14 (abcam) for BALB/c mice, with goat anti-rat IgG-biotin secondary antibody
231
(DakoCytomation). Primary and secondary antibody incubations were for 1 h or 30
232
min,
233
Diaminobenzidine (DAB) + horseradish peroxidase (HRP) substrate for 5 min
234
(DakoCytomation LSAB2 System-HRP).
respectively,
followed
by
Streptavidin-HRP
11
for
10
min
and
3,3-
235
Washing after DAB+HRP incubation was carried out with dH2O in a squeeze
236
bottle. Other washes were carried out for 3x 5min in PBST. Finally, for IHC, samples
237
were counter-stained with Gill’s Hematoxylin (Polysciences, Inc.) for 2 min, washed
238
in dH2O to remove excess dye, then in 1x automation reagent (Fisher Scientific) until
239
blue. All antibodies were diluted in block solution for staining. Negative control
240
FITC-mouse IgG1 (Sigma) and rat IgG2a or IgG2b (abcam), as appropriate, were
241
used to stain consecutive sections in order to confirm the specificity of the staining.
242
IF-stained samples were mounted under a coverslip in mountant: 2.4g of
243
Mowiol 488 [Calbiochem] + 6 g of glycerol + 6 ml water, mixed, then buffered with
244
12 ml of 0.2 M Tris pH 8.5 and heated to 50oC to mix, clarified by centrifugation,
245
before adding DABCO (1,4,-diazobicycli-[2.2.2]-octane, Aldrich) to 2.5% w/v to
246
reduce fading of fluorophores. For IHC, samples were dehydrated in a reverse of the
247
ethanol and Xylene steps described above, then mounted under a coverslip in Poly-
248
Mount Xylene (Polysciences Inc.). Samples were viewed using a Nikon 80i
249
microscope, images were captured with NIS elements software and processed with
250
Image J software. Counting of stained cells, or numbers of DNA clouds, was carried
251
out blinded to the identity of the samples, for the entire length of the small intestine
252
for one jelly-rolled section for each mouse.
253
Statistics
254
The majority of data was not normally distributed, therefore non-parametric tests
255
were utilized: Mann Whitney U tests were used to compare two groups and Kruskal
256
Wallis test and post-hoc Dunn’s multiple comparison tests for more than two
257
groups. GraphPad Prism 5.0a was used for all statistical analysis.
258 12
259
RESULTS
260
Inflammatory markers observed by qRT-PCR in the infant mouse small intestine
261
after V. cholerae infection
262
We sought to determine whether the infant mouse model of acute V. cholerae
263
colonization could be used to interrogate innate immune responses to V. cholerae in
264
this genetically tractable host. We used C57Bl/6 mice because many transgenic mice
265
would be available in this background for any subsequent studies of genetic
266
susceptibility to inflammation in response to V. cholerae infection. We used qRT-PCR
267
to screen for increased transcription of pro-inflammatory markers that are induced
268
in response to V. cholerae infection in other infection models and/or in cholera
269
patient samples upon wild type V. cholerae infection. Transcription of chosen
270
inflammatory markers was monitored in the small intestine of infant mice 7, 24 and
271
30 h after infection with wild type V. cholerae strain E7946 (O1 Ogawa El Tor) and
272
in sham vehicle-infected controls.
273
We looked for the Muc-1 transcript, as it is a cell-associated mucin whose
274
expression is induced during acute cholera in patient biopsies (26). We tested for
275
induction of pro-inflammatory cytokines TNFα (Tumor Necrosis Factor alpha), IL-6
276
(Interleukin-6) and IL-1β, because their transcript and protein levels were
277
increased in epithelial cell lines in response to V. cholerae or pure V. cholerae
278
flagellin (IL-1β) (15, 38) and in infant rabbits in response to oral V. cholerae
279
infection (IL-1β and TNFα were flagellin-dependent) (16). IL-6 is also induced in
280
adult mouse V. cholerae infection models (39, 40) and TNFα is increased in serum of
281
mice infected with V. cholerae via the pulmonary route (21). Importantly, IL-1β and 13
282
TNFα were also elevated in acute cholera patient biopsies (26, 41). We looked at the
283
chemokine CCL20 (also known as MIP-3α), which signals to recruit dendritic cells in
284
response to pro-inflammatory mediators including flagellin (42), as CCL20 was
285
increased in serum of adult mice after pulmonary V. cholerae infection (21).
286
We also tested for murine IL-8 family chemokines, which signal for
287
recruitment of polymorphonuclear cells (PMNs, also known as neutrophils) to sites
288
of infection: KC (also known as mCXCL1) and macrophage inflammatory protein-2
289
(MIP-2, also known as mCXCL2). MIP-2 was increased in serum of mice infected
290
with V. cholerae via the pulmonary route (21) and in adult mouse oral infection
291
models (39, 40). Human IL-8, and related chemokines Groα and Groβ, were induced
292
in epithelial cell lines in response to V. cholerae (38, 43) and IL-8 was induced in
293
infant rabbits in response to oral V. cholerae, in a flagellin-dependent manner (16).
294
We monitored transcription of the anti-bacterial factor NOS-2 (also known as iNOS,
295
inducible nitric oxide synthase), as increased NO metabolites were detected in
296
serum of V. cholerae-infected volunteers (44) and cholera patients (7, 45). The
297
cytokine IL-17a, which in adults would lead to a TH17 T-cell response, was
298
monitored. Although IL-17a is produced by TH17 T-cells, which are not yet fully
299
developed in neonates, it is also produced by innate immune cells, which are likely
300
to be present in neonates (46). In addition to the above, we looked at IFNγ
301
(interferon gamma) and IL-10, as examples of classic TH1-inducing or anti-
302
inflammatory cytokines respectively, and because IL-10 was seen to be modestly
303
induced in an adult mouse model of infection (39).
14
304
24 and 30 h post-infection we observed higher transcription after V. cholerae
305
infection, compared to sham-infected controls, for five pro-inflammatory markers:
306
KC, MIP-2 (statistically significant only at 30h), NOS-2, IL-6 and IL-17a (statistically
307
significant only at 24h) (Fig. 1A-E). 7 h post-infection there was no significant
308
difference in transcription of any of the pro-inflammatory markers tested, compared
309
to sham-infected controls (Fig. 1A-E and Supplemental Fig. S1). Transcription of
310
TNFα, IL-1β, CCL20, Muc1, IFNγ and IL-10 were not significantly higher in V.
311
cholerae-infected samples than sham-infected controls at any time-point tested
312
(Supplemental Fig. S1). Colonization was monitored by qRT-PCR for V. cholerae 16S
313
(Fig. 1F) and ctxA (Fig. 1G). All V. cholerae-infected samples were similarly colonized
314
24 to 30 h post-infection (Fig. 1F). 7 h post-infection the V. cholerae burden is not
315
yet maximal and the 16S signal is much lower and more variable than at 24h (Fig.
316
1F), consistent with intestinal colony forming units tested in separate experiments
317
(data not shown).
318
Neutrophil recruitment in the infant mouse small intestine after V. cholerae
319
infection
320
Consistent with transcriptional activation of MIP-2 and KC chemokines that
321
attract PMNs (Fig. 1A-B), a higher median level of PMNs was seen by immuno-
322
histochemical staining at 24 h, but not 2 h or 7 h, post-infection with V. cholerae (Fig.
323
2 and data not shown). Numbers of PMNs were counted in Peyer’s Patches, the
324
Lamina Propria and the lumen of the small intestine. We never detected luminal
325
PMNs in sham controls, whereas there were a significant number of PMNs in the
326
lumen 24 h after infection (Fig. 2Bii). There was also an increase in PMNs in Peyer’s
15
327
Patches and the Lamina Propria (Fig. 2Bi and iii). Recruitment of PMNs to the
328
Lamina Propria did not reach statistical significance in this particular experiment
329
(Fig. 2Bi), but was statistically significant in subsequent experiments (see Table 3
330
Sham vs WT). There was wide variation in PMN numbers between Peyer’s Patches
331
within an individual, with higher PMN numbers towards the distal (ileum) end of
332
the intestine compared to the proximal (duodenum) end (Fig. 2Biii and data not
333
shown). At the apical edge of the epithelium overlying Peyer’s Patches and some
334
regions of the Lamina Propria, PMNs were caught in the process of exiting the tissue
335
and entering the lumen, for example Fig. 2Aii. On an individual mouse-by-mouse
336
basis, there was a significant positive correlation between the number of PMNs in
337
the Lamina Propria and the number in the lumen (Spearman rank correlation
338
p7h
