IAI Accepted Manuscript Posted Online 11 January 2016 Infect. Immun. doi:10.1128/IAI.01374-15 Copyright © 2016 Dheer et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
1
Intestinal epithelial TLR4 signaling affects epithelial function, colonic microbiota and
2
promotes risk for transmissible colitis
3
Rishu Dheera, Rebeca Santaolallaa, Julie M Daviesa *, Jessica K Langa, Matthew C Phillipsb,
4
Cristhine Pastorinia ¶, Maria T Vazquez-Pertejoc, Maria T Abreua,b#
5 6 7 8 9 10
a
Division of Gastroenterology, Department of Medicine, University of Miami, Miller School of
Medicine, Miami, FL, USA b
Department of Microbiology and Immunology, University of Miami, Miller School of
Medicine, Miami, FL, USA c
Department of Pathology, Palms West Hospital, Loxahatchee, FL, USA
11 12 13
Running title: Role of epithelial TLR4 in regulating gut microbiota
14 15
# Address correspondence to: Maria T. Abreu,
[email protected]
16 17 18 19 20
*
Current address: Mater Research Institute, The University of Queensland, Translational
Research Institute, Woolloongabba, QLD 4102. ¶
Current address: Division of Gastroenterology, Hepatology & Nutrition, University of
Pittsburgh, Pittsburgh, PA 15213
21 22
DISCLOSURE
1
23
We have read the journal's policy and the authors of this manuscript have the following
24
competing interests: MTA has served as a consultant to AbbVie Laboratories, Hospira Inc.,
25
Takeda, Prometheus Labs, Ferring Pharmaceuticals, Shire Pharmaceuticals, Pfizer, Sanofi
26
Aventis, Janssen, GSK Holding Americas Inc., Eli Lilly, Mucosal Health Board, Focus Medical
27
Communications and UCB. MTA serves on the scientific advisory board of Asana Medical Inc.
28
and Celgene Corp and on the board of directors for the GI Health Foundation. MTA serves as
29
trainer or Lecturer for Prova Education Inc., ECCO Congress, Pre-Med institute, LLC and
30
Imedex Inc. This does not alter the authors’ adherence to the journal’s policies on sharing data
31
and materials.
32
All other authors declare no conflict of interest.
33
2
34
ABSTRACT
35
Evidence obtained from gene knock-out studies support the role of TLR4 in intestinal
36
inflammation and microbiota recognition. Increased epithelial TLR4 expression is observed in
37
patients with inflammatory bowel diseases (IBD). However, little is known of the effect of
38
increased TLR4 signaling on intestinal homeostasis. Here, we examined the effect of increased
39
TLR4 signaling on epithelial function and microbiota by using transgenic villin-TLR4 mice that
40
over-express TLR4 in the intestinal epithelium. Our results revealed that villin-TLR4 mice are
41
characterized by increased density of mucosa-associated bacteria and bacterial translocation.
42
Furthermore, increased epithelial TLR4 signaling was associated with impaired epithelial barrier,
43
altered expression of anti-microbial peptide genes, and altered epithelial cell differentiation. The
44
composition of the colonic luminal and mucosa-associated microbiota differed between villin-
45
TLR4 and WT littermates. Interestingly, WT mice co-housed with villin-TLR4 mice displayed
46
increased susceptibility towards acute colitis relative to single-housed WT mice. The results of
47
this study suggest that epithelial TLR4 expression shapes the microbiota and affects the
48
functional properties of the epithelium. The changes in the microbiota induced by increased
49
epithelial TLR4 signaling are transmissible and exacerbate DSS-induced colitis. Together, our
50
findings imply that host innate immune signaling can modulate intestinal bacteria and ultimately
51
the host susceptibility towards colitis.
52
3
53
INTRODUCTION
54
Trillions of microbes co-exist with mammalian cells in the gastrointestinal tract of the host in a
55
relatively mutualistic environment (1, 2). The intestinal epithelium and its overlying mucus layer
56
provide the physical barrier that separates the commensal microbiota from the host (3). Pattern
57
recognition receptors including the toll like receptors (TLRs) expressed by epithelial cells
58
recognize microbe-associated molecular patterns (MAMPs) of the commensal bacteria and
59
regulate the cross talk between intestinal microbes and host (4, 5). Defects in TLR signaling and
60
an aberrant immune response to perturbed endogenous microbiota are a few of the major factors
61
that contribute towards the perpetuation of inflammation and tissue injury in patients with
62
inflammatory bowel diseases (IBD) (6, 7).
63
TLR4 recognizes lipopolysaccharide (LPS) in the cell wall of Gram-negative bacteria.
64
Studies have shown that TLR4 expression varies by region of the intestine and is determined
65
largely by the bacterial composition of that region (8). Conversely, TLR4 may play an important
66
role in maintaining the fine balance between tolerogenic and inflammatory properties of gut
67
microbiota by regulating innate immunity (9, 10). Although several studies have demonstrated
68
that the microbiota composition of the host is influenced by the status of TLRs and their adapter
69
proteins (11-13), others have reported no such effect (14, 15). In the context of TLR4 knock out
70
mice, two separate studies have reported that genetics, maternal transmission and housing
71
conditions rather than the absence of TLR4 have marked effect on the stool microbiota of these
72
mice (14, 15). However, these studies have focused on the microbiota composition of the stool
73
rather than mucosa-associated microbiota, which is in close proximity to the host and differs in
74
its composition from that of the luminal microbiota (16). More importantly, the relationship
75
between epithelial innate immune signaling and the microbiota, especially in the mucosa4
76
associated population, is not known. Thus the role of TLRs including TLR4 in determining
77
microbiota composition and associated baseline phenotypes remains ambiguous.
78
TLR4 is normally expressed at very low levels by various cell types of the intestine
79
including epithelial and lamina propria mononuclear (LPMN) cells (9, 17, 18). Increased
80
expression of TLR4 is observed in epithelial and lamina propria cells of IBD patients suggesting
81
an important role of TLR4 signaling in inflammation (19-21). In the context of infectious
82
enteritis, TLR4 signaling facilitates pathogen colonization and dissemination; and promotes
83
infection-induced colitis (22). Using TLR4 knock-out mice, we and others have demonstrated
84
that TLR4 signaling is required for protection against epithelial injury, inflammation and
85
bacterial invasion (23, 24). On the other hand, we have previously published that the mice
86
expressing a constitutively active form of TLR4 in the intestinal epithelium (villin-TLR4 mice)
87
exhibit increased susceptibility to chemically-induced colitis (25). We have also shown
88
activation of the NF-κB signaling pathway and increased expression of chemokines and
89
proinflammatory genes at baseline in the intestinal epithelial cells of villin-TLR4 mice (26).
90
These findings established the double-edged sword of TLR4 function in the intestine, wherein
91
both low and excessive TLR4 signaling can promote intestinal inflammation. We have also
92
previously shown that the mice expressing TLR4 in the colonic epithelial cells rather than
93
myeloid cells are more susceptible to inflammation associated colonic neoplasia (27).
94
Due to the close proximity of colonic microbiota to intestinal epithelium and the critical
95
role played by intestinal epithelial TLR4 in microbiota recognition and inflammation, the aim of
96
our study was to determine the impact of intestinal epithelial-specific TLR4 signaling on
97
microbial composition and the related host responses. Published studies by other research groups
98
have shown that mice lacking specific components of the innate immune system such as Nod2, 5
99
inflammasome genes, or Myd88 are more susceptible to chemically-induced colitis and possess
100
altered microbiota when compared to their WT littermates (28-30). Furthermore, using co-
101
housing strategies, it has been demonstrated that the microbiota of Nod2 knock-out mice is
102
capable of transmitting colitogenic properties to WT mice (28). However, studies detailing the
103
effect of increased innate immune signaling on microbiota and inflammation are not published.
104
Thus we hypothesized that intestinal epithelium-specific constitutive TLR4 signaling would
105
affect microbial composition, epithelial function and colitis susceptibility. By using villin-TLR4
106
mice as a model for excessive TLR4 signaling, we show that epithelial TLR4 mediated
107
interaction between the gut microbiota and the host is essential for protection against bacterial
108
dissemination, epithelial injury and maintenance of epithelial barrier function. We also show that
109
TLR4 regulates the expression of antimicrobial genes by distinct mechanisms in different parts
110
of the intestine. To our knowledge this is the first study to describe that enhanced TLR4
111
signaling promotes colonic inflammation through dysbiotic microbiota.
112
METHODS
113
Mice
114
Mice hemizygyous for the villin-TLR4 transgene that constitutively expresses TLR4 in the
115
intestinal epithelium are termed villin-TLR4 mice and were generated as described previously
116
(26). All experiments were conducted with 7-9 week old littermates generated by crossing villin-
117
TLR4 transgenic mice with C57BL/6 WT mice and obtained from consecutive litters of the same
118
parent pair. For cohousing and DSS exposure experiment, age and gender matched non-
119
littermate C57BL/6 and villin-TLR4 mice were housed in the same cages for the duration of the
120
experiment. Separately housed C57BL/6 mice were kept under similar conditions but without
6
121
any villin-TLR4 mice housed in their cages. All mice experiments were performed in accordance
122
to the institutional animal care and use committee at the University of Miami.
123
Bacterial culture
124
Spleen and MLNs were removed aseptically, weighed and homogenized using a handheld tissue
125
homogenizer in phosphate buffer saline. Homogenized samples were cultured on tryptic soy
126
sheep blood agar plates and incubated under aerobic and anaerobic conditions for 24-48 hours.
127
Colonization was expressed as the average number of colony-forming units on aerobic and
128
anaerobic plates per mg of tissue.
129
FITC-dextran assay
130
Intestinal epithelial integrity was determined using FITC-dextran assay. Mice were orally
131
gavaged with 4 kDa FITC-dextran (Sigma-aldrich, St Louis, MO) at a concentration of
132
60mg/100g body weight. The concentration of FITC-dextran in the serum was determined after 4
133
hours by Gemini EM fluorescence microplate reader (Molecular Devices, Sunnyvale, CA) at
134
490/525 nm.
135
Intestinal epithelial isolation and mRNA expression
136
Sections of small intestine and colon were excised following sacrifice. Small intestine was rinsed
137
with cold HBSS, cut longitudinally and transferred to 10mL of cold HBSS and gentle agitated to
138
removal additional debris. The tissue was then cut into 1cm pieces and transferred to cold HBSS
139
containing 3mM EDTA for 30 minutes on ice with 200rpm agitation on an orbital shaker. Pieces
140
were transferred to cold HBSS and shaken vigorously for 3-5 minutes. Released epithelial cells
141
in the supernatant were filtered through 70μm filter. Cells were washed twice with 1%FBS in
142
HBSS to remove any residual EDTA. Colons were processed similarly to small intestines, but
143
were incubated in 60mM EDTA in HBSS for one hour. Total RNA was extracted from IECs or 7
144
colon tissues using RNA bee according to manufacturer’s instructions (Tel-test, Friendswood,
145
TX). A total of 1 ug of RNA was reverse transcribed using transcriptor reverse transcriptase
146
enzyme and random hexamers (Roche life science, Indianapolis, IN). RT-qPCR was performed
147
using SYBR Premix Ex Taq (Clontech Laboratories, Mountain View, CA) and Roche
148
LightCycler 480 (Roche, Indianapolis, IN). Information for primer pairs used is provided in
149
supplementary methods. The relative expression levels were calculated by ΔΔCt method after
150
normalizing to the average of endogenous β-actin and Gapdh control genes or with Gapdh only
151
for DSS exposure experiments.
152
Western blot analysis
153
Total cell lysate from colon and small intestine epithelial cells was obtained by using M-PER
154
protein extraction reagent (Thermo Scientific, Waltham, MA) supplemented with protease
155
inhibitor cocktail set III (EMD Millipore Corp., Billerica, MA). Protein lysates were loaded into
156
precast, NuPAGE 10% Bis-Tris Gels (Novex by Life Technologies, NY), electrophoresed
157
(200V, 1 hr) and then transferred (30V, 1.25 hr) onto polyvinylidene fluoride microporous
158
membranes. Membranes were blocked for 1 hour in 5% bovine serum albumin (Sigma-Aldrich,
159
St Louis, MO) and incubated with their respective antibody (1:500 dilution) overnight at 4°C.
160
Rabbit anti-occludin and Rabbit anti-claudin-3 antibodies were obtained from Life Technologies
161
(Grand Island, NY). Membranes were washed and incubated with goat anti-rabbit antibody
162
conjugated to horseradish peroxidase (1:10,000 dilution; Invitrogen, MA) for 30 minutes. Anti-
163
β-Actin conjugated to horseradish peroxidase (1:10,000 dilution; Sigma, St Louis, MO) was used
164
as the loading control. Membranes were developed using SuperSignal West Dura Extended
165
Duration Substrate (Thermo Scientific, Waltham, MA) as per manufacturer instructions and
166
imaged on a myECL Imager (Thermo Scientific, Waltham, MA). 8
167
Tissue histology and immunofluorescence microscopy
168
Samples from ileum and colon were fixed in 10% neutral formalin. Paraffin sections were
169
stained with hematoxylin and eosin, periodic acid Schiff or immunofluorescence using standard
170
protocols. Antibodies used were: rabbit anti-muc2 and goat anti-lysozyme C (Santa cruz
171
Biotechnology, Dallas, TX). After antigen retrieval with citrate buffer, immunostaining for
172
MUC2 and lysozyme was performed by overnight staining with 1:100 dilution of primary
173
antibody at 4oC followed by staining with 1:200 dilution of secondary antibody conjugated with
174
Alexa fluor 488 (Life technologies, Grand Island, NY). The slides were mounted with slowfade
175
gold antifade reagent (Life technologies, Grand Island, NY) after counterstaining with DAPI.
176
Induction of colitis and colitis assessment
177
Acute colitis was induced in age matched mice with 3% DSS (molecular weight = 36-50 kDa;
178
MP biomedicals, Solon, OH) in drinking water for a period of 6 days. Body weight, stool
179
consistency and stool blood were recorded daily by a research technician (J.L.) blinded for the
180
study groups. Disease activity indices were calculated as described previously by averaging the
181
scores for weight loss, stool consistency and bleeding that were recorded on a scale of 0-4 (31).
182
The histological assessment for colitis was performed by a trained pathologist (M.V.) blinded for
183
the treatment groups and scored using the following criteria: architectural changes (0 = normal, 1
184
= mild focal abnormality, 2 = diffuse or multifocal mild to moderate abnormality, 3 = severe
185
diffuse or multifocal abnormality); basal plasmocytosis and neutrophil/eosinophil infiltration (0
186
= none, 1 = mild, 2 = moderate and 3 marked; Crypt abscesses and crypt destruction (0 =
187
absent,1 = < than 5% of crypts involved, 2 = 5 - 50% of crypts involved, 3 = > 50% of crypts
188
involved); and epithelial erosion or ulceration (0 = no erosion nor ulceration, 1 = superficial
9
189
erosion with epithelial repair, 2 = moderate ulceration with epithelial repair and 3 = ulcer with
190
granulation tissue and no epithelium).
191
Explant culture and IL-6 ELISA
192
A small segment of distal colon was washed in PBS supplemented with 50 mg/ml gentamicin.
193
The explant was cultured in 24 well plates containing RPMI 1640 medium supplemented with
194
penicillin and streptomycin. After 24 hours supernatant was collected, stored at -20oC and later
195
analyzed for IL-6 production by using Duoset mouse IL-6 ELISA kit (R&D systems,
196
Minneapolis, MN). All other cytokines were measured by customized luminex mouse magnetic
197
bead assay according to the manufacturer’s recommendations (R&D systems, Minneapolis,
198
MN).
199
Luminal and mucosal microbial analysis
200
Luminal samples comprised of fecal pellets that were collected from the distal portion of the
201
excised colon. Mucosal samples comprised of the tissues from the same distal region and were
202
collected after flushing the colon twice with PBS. Luminal fecal material and mucosal tissues
203
from distal portion of the ileum were collected in the same manner. Genomic DNA was extracted
204
from the lumen and mucosa of distal colon and ileum using QIAamp stool and genomic DNA
205
isolation kit (Qiagen, Valencia, CA). 16S copy number as an estimate for total bacterial number
206
and individual bacterial groups in the lumen and mucosal samples was determined using
207
universal and group-specific primers and qPCR as described above and previously (32). Analysis
208
for microbial richness, composition and diversity were conducted by Second genome Inc. (San
209
Francisco, CA) as described in supplementary methods. The sequences obtained in this paper
10
210
have been submitted to the European Bioinformatics Institute (EBI) database under accession
211
number PRJEB8110.
212
Terminal restriction fragment length polymorphism (T-RFLP)
213
Genomic DNA from luminal or stool samples was amplified using primers broad range forward
214
primer 8F (labelled with FAM dye at 5’ end) and reverse primer 1492R using GoTaq DNA
215
polymerase (Promega, Madison, WI). The amplified PCR product was purified and digested by
216
MspI restriction enzyme (Promega, Madison, WI). The restriction digest product was mixed with
217
GenScan 1200 LIZ size standard (Life technologies, Grand Island, NY) and deionized
218
formamide. The length and intensity of terminal restriction fragments (T-RF) were determined
219
on an ABI 3730 capillary sequencer (Life technologies, Grand Island, NY) and analyzed on peak
220
scanner software (Life technologies, Grand Island, NY). The T-RF profiles were subjected to
221
principal component analysis using Excel add-in Multibase package (Numerical Dynamics,
222
Japan).
223
Statistical analyses
224
Data were expressed as mean ± standard error of mean and analyzed in Graphpad prism 6.
225
Statistical significance was assessed using student's t-test, unless otherwise stated. For more than
226
two groups, analysis of variance with Tukey’s post-hoc test was used. Effect of time and
227
genotype in DSS exposure experiment was determined using 2-way ANOVA and Tukey’s post-
228
hoc test. Statistical significance for microbiota sequences comparison at various taxonomic
229
levels was determined by student’s t-test followed by Benjamini-Hochberg adjustment for false
230
discovery rate of 10%. Pearson correlation coefficient and significance was determined using
231
IBM SPSS statistics 22 software. Heat maps were constructed using CIMminer (Genomics and
11
232
Bioinformatics Group, Laboratory of Molecular Pharmacology (LMP), Center for Cancer
233
Research (CCR) National Cancer Institute (NCI)). Differences were considered significant when
234