Available via license: CC BY 4.0
Content may be subject to copyright.
OPEN
Inflammatory cues enhance TGFbactivation by
distinct subsets of human intestinal dendritic cells
via integrin avb8
TM Fenton
1,2,3
, A Kelly
1,2,3
, EE Shuttleworth
1,2,3
, C Smedley
1,2,3
, A Atakilit
4
, F Powrie
5,6
, S Campbell
7
,
SL Nishimura
8
, D Sheppard
4
, S Levison
7
, JJ Worthington
3,10
, MJ Lehtinen
9
and MA Travis
1,2,3
Regulation of intestinal T-cell responses is crucial for immune homeostasis and prevention of inflammatory bowel
disease (IBD). A vital cytokine in regulating intestinal Tcells is transforming growth factor-b(TGFb), which is secreted by
cells as a latent complex that requires activation to function. However, how TGFbactivation is regulated in the human
intestine, and how such pathways are altered in IBD is completely unknown. Here we show that a key activator of TGFb,
integrin avb8, is highly expressed on human intestinal dendritic cells (DCs), specifically on the CD1c
þ
but not the
CD141
þ
intestinal DC subset. Expression was significantly upregulated on intestinal DC from IBD patients, indicating
that inflammatory signals may upregulate expression of this key TGFb-activating molecule. Indeed, we found that the
Toll-like receptor 4 ligand lipopolysaccharide upregulates integrin avb8 expression and TGFbactivation by human DC.
We also show that DC expression of integrin avb8 enhanced induction of FOXP3 in CD4
þ
Tcells, suggesting functional
importance of integrin avb8 expression by human DC. These results show that microbial signals enhance the TGFb-
activating ability of human DC via regulation of integrin avb8 expression, and that intestinal inflammation may drive this
pathway in patients with IBD.
INTRODUCTION
The intestine is a challenging environment for the immune
system, which must induce protective responses against food-
borne pathogens, but promote tolerance against the trillions of
microorganisms that compose the microbiota. It is proposed
that specialized regulatory mechanisms are in place to balance
protective and tolerogenic immunity in the gut, with failure of
these mechanisms resulting in inflammatory bowel disease
(IBD).
1
A crucial mechanism by which gut immune responses are
controlled is via the cytokine transforming growth factor-b
(TGFb). TGFbis especially important in the regulation of
T-cell responses, promoting differentiation of both Foxp3
þ
regulatory T cells (Tregs) and T helper type 17 cells, and
suppressing the differentiation of T helper type 1 and T helper
type 2 cells.
2
Indeed, recent evidence suggests that targeting the
TGFbpathway in IBD may have beneficial effects in some
patients.
3
Many different cells in the gut produce TGFb, but
always as a latent complex, which has to be activated to
function. Thus, regulation of TGFbfunction is critically
controlled at the level of its activation.
Previous work from our lab and others has highlighted that
intestinal dendritic cells (DCs) can act as crucial activators of
TGFbin mice.
4–9
There are two major subsets of DCs in the
mouse intestine, both expressing the cell surface markers
CD11c and CD103, but characterized by differential expression
of transcription factors required for their development and by
expression of the cell surface protein CD11b.
10
Thus, one subset
1
Manchester Collaborative Centre for Inflammation Research, University of Manchester, Manchester, UK.
2
Wellcome Trust Centre for Cell-Matrix Research, University of
Manchester, Manchester, UK.
3
Manchester Immunology Group, Faculty of Life Sciences, University of Manchester, Manchester, UK.
4
Lung Biology Center, Department of
Medicine, University of California, San Francisco, CA, USA.
5
Kennedy Institute of Rheumatology, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal
Sciences, University of Oxford, Oxford, UK.
6
TranslationalGastroenterology Unit, Nuffield Department of Medicine, University of Oxford, Oxford, UK.
7
Gastroenterology Unit,
Manchester Royal Infirmary, Central Manchester University Hospital NHS Foundation Trust, Manchester, UK.
8
Department of Pathology, University of California, San
Francisco, CA, USA and
9
DuPont Nutrition & Health, Global Health and Nutrition Science, Kantvik, Finland. Correspondence: MA Travis (mark.travis@manchester.ac.uk)
10
Current address: Division of Biomedical and Life Sciences, Faculty of Health and Medicine, Lancaster University, Lancaster, UK.
Received 15 June 2016; accepted 26 August 2016; advance online publication 26 October 2016. doi:10.1038/mi.2016.94
ARTICLES
MucosalImmunology 1
of intestinal DC requires expression of the transcription factors
IRF8, Batf3, and Id2, and is CD11b-negative, whereas the other
depends on expression of the transcription factor IRF4 and is
CD11b-positive.
10
Specifically, murine CD103
þ
CD11b
intestinal DCs express high levels of integrin avb8, which
enables them to activate TGFband induce Foxp3
þ
Tregs, Th17
cells, and intraepithelial lymphocyte populations.
4,6,8,11
How-
ever, whether a similar pathway exists in the human system
remains unknown.
Human conventional DC can be divided into two devel-
opmentally distinct populations, marked by expression of
either CD1c or CD141. These subsets show homology to
murine subsets, as human CD1c
þ
DCs express IRF4 and show
similarities to murine CD103
þ
CD11b
þ
DC, whereas
CD141
þ
DCs are more akin to murine CD103
þ
CD11b
DC.
12–15
Recently, it has been suggested that human intestinal
DC can also be divided into functionally distinct subsets, using
the markers CD103 and SIRPa, which appear transcriptionally
homologous to the murine CD103/CD11b subsets.
16
However,
whether intestinal DCs regulate T-cell responses via TGFb
activation in the human system, and how such pathways are
potentially altered in IBD, is completely unknown.
Here we show that the TGFb-activating integrin avb8is
expressed by human intestinal DC, with expression seen
preferentially on the CD1c
þ
DC subset, in contrast to
expression patterns in mice. Expression of integrin avb8is
significantly upregulated in CD1c
þ
DC from patients with
Crohn’s disease (CD), suggesting that inflammatory signals
may be important in enhancing the TGFb-activating ability of
DC. Indeed, we show mechanistically that integrin avb8
expression by DC is increased by treatment with the Toll-like
receptor (TLR)4 agonist lipopolysaccharide (LPS), which
enhanced their ability to activate TGFb. Finally, DC-expressed
integrin avb8 was important for the induction of FOXP3
expression in CD4
þ
T cells, suggesting an important functional
role for the integrin in inducing human Treg. Thus, our data
suggest that expression of integrin avb8 on human intestinal
DC subsets, driven by inflammation, might promote Treg
induction via activation of TGFb.
RESULTS
Human intestinal DCs express the TGFb-activating integrin
avb8
Integrin avb8 is highly expressed on murine intestinal DC and
this expression is required to prevent spontaneous gut
inflammation via activation of TGFb.
4,8
However, whether
a similar pathway is important in the regulation of intestinal
immunity in humans is completely unknown. To address this
question, we examined expression of integrin avb8 by flow
cytometry on human intestinal DC, using an antibody we
generated that specifically binds to human integrin b8 (see
Methods and Supplementary Figure S1 online). Human
intestinal DCs were obtained from intestinal resection and
biopsy samples, and mononuclear phagocytes were gated as
viable CD45
þ
HLADR
þ
Lineage
cells. DCs were distin-
guished from monocytes/macrophages in this mononuclear
phagocyte population by low expression of CD14 and CD64
(Figure 1a). Viable CD45
þ
cells, which were either Lineage
hi
/
HLA-DR
lo
were placed into a ‘dump’ gate and used as a control
cell population. We found that integrin avb8 was expressed on
a significant proportion of DC, whereas expression on total
Lineage
hi
/HLADR
lo
cells was minimal (Figure 1b,c). Given
that expression of integrin avb8 is enhanced on intestinal DC in
mice compared with non-intestinal sites,
8
we next analyzed
integrin avb8 expression on DC from human peripheral blood.
Interestingly, we found that expression of integrin avb8
on human peripheral blood DC (Lineage
HLA-DR
þ
CD14
CD16
CD11c
þ
cells) was similar to that seen in
intestinal DC (Figure 1d,e), suggesting an important difference
between the mouse and human DC systems. Thus, integrin
avb8 is expressed by human intestinal DC, although expression
is not restricted to the intestine as in mice.
Integrin avb8 is expressed on human intestinal CD1c
þ
but
not on CD141
þ
DC subsets
As different intestinal DC types have distinct functional
properties,
17
we next analyzed which human intestinal DC
subsets express integrin avb8. We recently found that integrin
avb8 is exclusively expressed on murine intestinal IRF8-
dependent CD103
þ
CD11b
DC, which are specialized to
cross-present antigen, with minimal expression observed in
CD103
þ
CD11b
þ
and CD103
CD11b
þ
intestinal DC sub-
sets.
11
To determine whether similar expression patterns were
observed in human intestinal DC, cells were categorized by
expression of CD1c vs. CD141, which are analogous to murine
CD103
þ/
CD11b
þ
and CD103
þ
CD11b
DC, respec-
tively.
16
In contrast to results in mice, expression of integrin
avb8 was found on human intestinal CD1c
þ
DC but not on
CD141
þ
DC (Figure 2a,b). Furthermore, when DC subsets
were further gated according to CD103 and SIRPaexpression
(Figure 2c), which are proposed to identify equivalent cells
to murine CD103/CD11b DC subsets,
16
integrin avb8
was preferentially expressed on both CD103
þ
SIRPa
þ
and
CD103
SIRPa
þ
DC compared with CD103
þ
SIRPa
DC
(Figure 2d,e), in contrast to expression patterns in mice.
11
Thus, these results suggest that integrin avb8 is expressed on
different subsets of human intestinal DC compared with the
murine homologs.
Human integrin avb8 levels are increased on CD1c
þ
DC
from IBD patients
Given that integrin avb8 expression by DC is crucial in
preventing development of IBD in mice,
4,6,8,11
we next analyzed
expression of integrin avb8 on intestinal DC from patients with
IBD, specifically with CD. Non-IBD control tissue was obtained
from patients with bowel cancer undergoing surgery or
screening endoscopy, with non-cancerous tissue analyzed.
Tissue from CD patients (Table 1) was obtained during
endoscopy or resection surgery, and integrin avb8 expression
was analyzed by flow cytometry. Expression of integrin avb8
was not different on total DC from non-IBD vs. CD patients
(Figure 3a,b). However, when expression on the different
subsets of intestinal DC was analyzed, we found that integrin
ARTICLES
2www.nature.com/mi
e
0103104105
0
103
104
105
0103104105
0
–103
103
104
105
0103104105
0
50K
100K
150K
200K
250K
050K 100K 150K 200K 250K
0
50K
100K
150K
200K
250K
0103104105
0
–103
103
104
103104
0
20
40
60
80
100
103104
0
20
40
60
80
100
FSC
SSC
β8
a
bDump gate DC
d
c
β8
Dump gate DC
CD45
SSC
HLADR
Lineage
CD64
CD14
FSC
Viability
HLADR
CD11c
0
20
40
60
80
100
0
20
40
60
80
100
100100
100100
101102103104104
103
101102
105
104
103
0
0 50K 100K 150K 200K 250K
Figure 1 Integrin avb8 is expressed by human intestinal dendritic cells (DCs). (a) Representative gating of human intestinal resected and biopsy lamina
propria cells digested and analyzed by flow cytometry. CD45
þ
cells were first gated by forward and side scatter, then DC gated as viable lineage (CD3,
CD15, CD19, CD20, and CD56)
HLADR
þ
CD14
CD64
CD11c
þ
cells. ‘Dump gate’ control cells were gated as CD45
þ
cells that were negative for
HLA-DR and/or positive for lineage markers. (b,c) Analysis of integrin b8 expression on human intestinal DC by flow cytometry, showing representative
histograms (shaded plot, isotype control; non-shaded plot, anti-integrin b8 antibody) (b) and pooled data (c). (d,e) Integrin avb8 expression analysis by
flow cytometry on human peripheral blood DC (Lineage (CD3, CD15, CD19, CD20, CD56)
HLADR
þ
CD14
CD16
CD11c
þ
cells), showing
representative histograms (d) and pooled data (e). Error bars represent mean±s.e.m., nX6 for all experiments, statistical significance analyzed by
paired Student’s t-tests (*Po0.05, **Po0.01).
ARTICLES
MucosalImmunology 3
0103104
0
20
40
60
80
100
0103104
01031040103104
CD141
CD1c
β8
β8
a
b
0103104
0
103
104
SIRPα
CD103
P1: CD103+
SIRP α–
c
d
0103104
0
20
40
60
80
100 CD141+
0103104
0
20
40
60
80
100 CD1c+
P2: CD103+
SIRP α+
P3: CD103–
SIRP α+
P4: CD103–
SIRP α–
0103104
0
103
104
105
P1 P2
P3
P4
e
Figure 2 Specific subsets of human intestinal dendritic cells (DCs) express the transforming growth factor-b(TGFb)-activating integrin avb8. (a) Viable
human colonic lamina propria CD45
þ
HLADR
þ
Lin
CD14
CD64
DCs were gated by CD141 vs. CD1c expression, and expression of integrin avb8
was analyzed on each subset (shaded plot, isotype control; non-shaded plot, anti-integrin b8 antibody). (b) Pooled data from a, error bars represent
mean±s.e.m., nX8, statistical significance analyzed by paired one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test
(*Po0.05). (c) Human viable CD45
þ
HLADR
þ
Lin
CD14
CD64
DCs were gated by CD103 and SIRPaexpression, and (d) integrin avb8 expression
analyzed on each subset. Representative histograms show shaded plot ¼isotype control, non-shaded plot ¼anti-integrin b8 antibody, representative of
four donors, with pooled data depicted in (e). Error bars represent mean±s.e.m., n¼4, statistical significance analyzed by paired one-way ANOVA with
Tukey’s multiple comparisons test (*Po0.05).
ARTICLES
4www.nature.com/mi
avb8 expression was significantly higher on CD1c
þ
DC from
CD patients compared with non-IBD controls (Figur e 3c,d). As
in control patients, CD141
þ
intestinal DC from CD patients
did not express significant integrin avb8 levels (Figure 3c,d).
These data therefore show that patients with intestinal
inflammation have enhanced expression of the TGFb-
activating integrin avb8 on intestinal DC, specifically on
the CD1c
þ
DC subset.
Integrin avb8 expression by human DC is enhanced by
inflammatory signals
Next, we aimed to determine molecular mechanisms driving
the enhanced expression of integrin avb8 on intestinal DC in
CD patients, with the hypothesis that gut-derived inflamma-
tory signals may be important in such induction. To test this
possibility, we utilized human DC-like cells, derived from blood
monocytes (moDC), to test the ability of gut-associated
molecules to regulate integrin avb8 expression. As the
expression levels of latent TGFband the vitamin A metabolite
retinoic acid (RA) are increased in the intestine of patients with
IBD,
18,19
and provide signals to regulate T-cell function and
homing,
2,20
we hypothesized that these molecules could have a
role in integrin avb8 induction on DC. However, we found that
neither blockade of TGFband RA nor addition of the two
molecules had any significant effect on the expression of
integrin avb8 by moDC (Figure 4a).
We next tested the potential for pathogen-associated
molecular patterns, which are associated with inflammation,
to modulate integrin avb8 expression on DC. We found that
the TLR4 ligand LPS caused significant upregulation of integrin
avb8 expression on moDC, which was not apparent with
agonists for TLR1/2 (Pam3CSK4), TLR3 (Poly I:C), TLR5
(flagellin), or TLR7 (Imiquimod) (Figure 4b). The TLR8 ligand
ssRNA40 also upregulated expression of integrin avb8, but to a
significantly lower extent than LPS (Figure 4b). Thus, specific
pathogen-associated molecular patterns appear to upregulate
the expression of human integrin avb8.
We next tested the ability of LPS to regulate expression of
integrin avb8 by primary human intestinal DC. LPS had no
effect on integrin avb8 expression by CD141
þ
DC, but
significantly elevated expression on the intestinal CD1c
þ
DC
subset (Figure 4c), mirroring expression changes seen in the
intestine of CD patients. Thus, our results show that LPS can
enhance integrin avb8 expression on intestinal CD1c
þ
DC,
suggesting that microbe-associated signals may contribute to
the higher levels of integrin avb8 expression seen on DC in
patients with CD.
Human integrin avb8 activates TGFband regulates FOXP3
induction in T cells
Next, we addressed the potential functional importance of
integrin avb8 expression by human DC. We first tested
whether LPS-induced expression of integrin avb8 enhanced the
ability of human DC to activate TGFb, using an active TGFb
reporter cell assay.
21
Indeed, treatment of moDC with LPS
resulted in enhanced TGFbactivation, which was blocked by an
anti-integrin avb8-blocking antibody (Figure 4d). Thus,
stimulation of human DC by LPS enhances their ability to
activate TGFbin an integrin avb8-dependent manner.
We next tested whether integrin avb8 expression by DC
affects their ability to regulate T-cell responses. Specifically,
previous work in mouse has shown that DC expression of
integrin avb8 is important in the TGFb-mediated upregulation
of Foxp3 in CD4
þ
T cells, inducing a regulatory phenotype.
4,8
To test for a similar role in humans, LPS-treated moDC were
co-cultured with allogenic naive CD4
þ
CD25
human T cells,
which showed o1% FOXP3 expression (Figure 4e). After 5
days in culture, DC induced a proportion of prolife rating T cells
to express FOXP3 (Figure 4f), which was significantly reduced
in the presence of an anti-TGFb-blocking antibody, indicating
an important role for TGFbin induction of FOXP3 (Figure 4f).
Conversely, addition of active TGFbenhanced FOXP3
expression (Figure 4f). Importantly, an anti-integrin avb8-
blocking antibody reduced induction of FOXP3 in T cells to a
similar extent as blockade of TGFb(Figure 4f). Induced
CD4 þFOXP3 þcells in all conditions expressed equivalent
levels of FOXP3 (Figure 4g), suggesting differences observed
are at the levels of cell numbers induced to express FOXP3
rather than at the level of FOXP3 expression. Together, these
results suggest an important functional role for integrin avb8
in induction of FOXP3
þ
T cells by human DC, via activation
of TGFb.
Table 1 Summary of patient information for intestinal
samples
Patients with
Crohn’s disease
Non-IBD
controls
Total 6 8
Male 6 3
Female 0 5
Age (years) (mean±s.d.) 40±9.6 67±12
Sample site
Unspecified colon 1 7
Transverse colon 1 0
Ileo-sigmoidal junction 1 0
Rectum and sigmoid colon 0 1
Rectum and transverse colon 1 0
Transverse and sigmoid colon 2 0
Diagnosis
Colon cancer 0 7
Normal (genetic cancer screening) 0 1
Quiescent Crohn’s disease 1 0
Active Crohn’s disease 5 0
Abbreviation: IBD, inflammatory bowel disease.
Demographic information, along with site of sampling and patient diagnosis at time of
sampling.
ARTICLES
MucosalImmunology 5
DISCUSSION
TGFbhas a crucial role in regulating intestinal immune
responses, but needs to be activated to function. How TGFbis
activated in the human intestine to control immunity is
completely unknown. Here we find that the TGFb-activating
integrin, avb8, is expressed on human intestinal CD1c
þ
DC,
and that expression is increased on this DC subset in patients
with CD. Integrin avb8 expression was also increased on DC
after ex vivo treatment with LPS, which enhanced their ability to
activate TGFb, and avb8 expression promoted induction of
FOXP3 expression in naive human T cells. Taken together, our
study uncovers a new pathway in which the TGFb-activating
integrin avb8 is expressed on human intestinal DC, and which
is upregulated in patients with CD.
We and others have previously shown that integrin avb8is
an important activator of TGFbby murine intestinal DC, with
lack of expression in mice resulting in colitis.
4,5,7,8
We now
show that the pathway is present in humans; however,
important distinctions between the human and murine
pathways exist. Specifically, whereas expression of integrin
avb8 is enriched on murine intestinal DC vs. non-intestinal DC
from the spleen
8
and other peripheral lymph nodes (unpub-
lished data), this is not the case in humans, where similar
expression is observed on DC from peripheral blood and the
intestine. The underlying reasons for such differences are
currently unknown. However, a recent report has found that
mice kept in specific pathogen-free conditions show an altered
immune system compared with wild mice, with wild mice
showing immune traits more similar to humans.
22
As our data
show that LPS drives expression of integrin avb8 by DC, a
potential explanation for enhanced peripheral expression of
integrin avb8 on human DC may be that the less-sterile
environment inhabited by humans drives expression, which is
not apparent in specific-pathogen-free mice.
In addition, whereas in mice integrin avb8 is almost
exclusively expressed by intestinal DC expressing CD103
but lacking CD11b,
8,11
minimal expression is observed on
human CD141
þ
DC, which are analogous to the murine
CD103
þ
CD11b
DC subset.
16
Instead, human CD1c
þ
DCs,
which are analogous to CD103
þ
CD11b
þ
murine intestinal
DCs,
16
are the major DC population expressing integrin avb8.
Why there is a contrast in integrin avb8 expression patterns
between seemingly analogous DC subsets in mice and humans
is unknown. A potential explanation is provided by data from
human IRF8-deficient patients. IRF8 drives expression of
integrin avb8 in murine CD103
þ
CD11b
DC via binding to
the ITGB8 promoter region.
23
Whereas mice lacking IRF8
expression globally, or specifically in DC, have a selective defect
in integrin avb8-expressing CD103
þ
CD11b
DC subset
numbers,
11,24
human patients with IRF8 mutations lack either
all DC subsets, or the CD1c
þ
subset specifically.
25
IRF8
therefore appears to be differentially expressed by mouse and
human DC, and thus may explain the differential integrin avb8
expression.
Having shown that integrin avb8 is highly expressed on
human intestinal DC, we next investigated expression on DC
from CD patients. Integrin avb8 expression was higher on
CD1c
þ
DC, but not on CD141
þ
DC, from CD patients
compared with non-IBD controls, strongly indicating that
CD1c
þ
DC from patients with CD have an enhanced capacity
to activate TGFb. Functionally, TGFbcan drive the differ-
entiation of both pro-inflammatory T helper type 17 cells and
anti-inflammatory Foxp3
þ
Tregs, depending on the cytokine
environment.
2
Thus, enhanced integrin avb8 expression in CD
patients may be involved in promoting inflammation, or in a
prolonged and unsuccessful anti-inflammatory feedback loop.
In support of the latter possibility, activated T cells have been
shown to be refractive to TGFbsignaling in CD patients, due to
enhanced expression of the TGFbsignaling inhibitor Smad7,
26
with knockdown of Smad7 causing remission in some patients.
3
In such a scenario, enhanced integrin avb8 expression by DC
would result in increased TGFbactivation, but without causing
a protective response in TGFb-refractory colitic T cells.
We next addressed potential mechanisms responsible for
enhanced expression of integrin avb8 on DC in CD patients.
Despite a recent report that RA can enhance expression of
integrin avb8 on Peyer’s Patch DCs in mice,
27
RA and the
immunomodulatory cytokine TGFbdid not enhance expres-
sion of integrin avb8 on human DC, again highlighting
differences between the murine and human systems. Instead,
we found that the bacterial danger signal LPS increased DC
expression of integrin avb8 and this promoted TGFbactivation
by DC. Interestingly, LPS induced expression of integrin avb8
specifically on CD1c
þ
and not on CD141
þ
intestinal DC, thus
mirroring expression patterns seen in patients with CD.
Impaired barrier function of the intestine is apparent in many
patients with CD, which can lead to enhanced translocation of
bacteria from the intestinal microbiota. Thus, one possibility is
that enhanced expression of integrin avb8 by CD1c
þ
DC in
CD is linked to the enhanced exposure of gut DC to higher
levels of LPS, due to enhanced bacterial translocation. Indeed,
the integrin b8 gene promoter region is known to contain
binding sites for the LPS-activated transcription factors p38
and AP-1,
28
which supports a mechanism by which LPS can
specifically enhance integrin avb8 expression by DC. Although
further work with larger patient cohorts is required to
Figure 3 Integrin avb8 expression is elevated on CD1c
þ
intestinal dendritic cells (DCs) from patients with Crohn’s disease (CD). Single-cell
suspensions of intestinal resection and biopsy samples from non-inflammatory bowel disease (IBD) and CD patients were analyzed by flow cytometry for
expression of integrin avb8 on DC subsets. (a,b) Expression of integrin avb8 on CD45
þ
HLADR
þ
Lin
CD14
CD64
total DCs from non-IBD and CD
patients. (a) Representative histograms (shaded plot, isotype control; non-shaded plot, anti-integrin b8 antibody), (b) pooled data. (c,d) Expression of
integrin avb8 on CD141
þ
DC and CD1c
þ
DC subsets from non-IBD and CD patients. (c) Representative histograms, (d) pooled data. Error bars
represent mean±s.e.m., nX6, statistical significance analyzed by two-way analysis of variance with Sidak’s multiple comparisons test (*Po0.05).
ARTICLES
MucosalImmunology 7
determine whether integrin avb8 expression is specifically
upregulated in CD patients vs. those with ulcerative colitis and
other intestinal diseases, and whether severity of disease
correlates with the expression of the integrin, our data suggest
an important cellular and molecular mechanism by which
TGFbactivation can be regulated in the intestine in CD. Also,
b
c
a
CD1c+CD141+d
0103104105
0
103
104
105
0103104105
0
5
103
104
105
FOXP3
CD25
f
e
g
ARTICLES
8www.nature.com/mi
given the complex inflammatory environment in the intestines
of IBD patients, it will be important to determine whether other
signals in addition to LPS regulate the integrin avb8–TGFb
pathway on intestinal DC during disease.
Finally, we investigated the functional role of human integrin
avb8 expression by human DC, focusing on the potential role in
induction of human FOXP3
þ
Treg. Our data showed that
blocking integrin avb8 function caused a significant reduction of
FOXP3 induction in T cells, to the same extent as blocking TGFb.
Although some studies have indicated that human T cells can
induce FOXP3 expression transiently upon activation without
gaining suppressive function,
29,30
further data show that even
transient FOXP3 expression can promote suppressive function
in CD4
þ
T cells.
31–33
Paradoxically, patients with IBD show
elevated numbers of Treg in the intestinal mucosa that can
suppress proliferation of CD4
þ
T cells in vitro.
34,35
Thus, as
integrin avb8 expression by DC can promote FOXP3
expression in human T cells, and DC from patients with
CD have elevated expression of integrin avb8, it is tempting to
speculate that DC expression of integrin avb8 may be an
important factor in driving the enhanced Treg numbers
observed in IBD. Despite enhanced Treg numbers, intestinal
CD4
þ
T cells from IBD patients are resistant to Treg-mediated
suppression,
36
perhaps due to these cells being refractory to
TGFbsignaling,
26
which is an important mechanism by which
Treg suppress T-cell responses.
2
Thus, given recent evidence
that enhancing the ability of intestinal T cells to sense active
TGFbis effective in inducing remission in certain CD patients,
3
enhancement of the integrin avb8–TGFbpathway may be an
attractive complementary therapeutic approach to promote the
suppression of colitic T cells.
Taken together, our study uncovers a novel pathway by
which the TGFb-activating integrin avb8 is expressed in the
human intestine on DC subsets, which is upregulated in
patients with CD, and which may be an attractive therapeutic
target to drive TGFb-mediated suppression of inflammation in
the intestine.
METHODS
Obtaining and processing of human intestinal tissue and blood.
Human samples were obtained according to the principles expressed in
the Declaration of Helsinki and under local ethical guidelines, and
approved by the North West National Research Ethics Service
(reference number 15/NW/0007). All patients provided written
informed consent for the collection of tissue samples and subsequent
analysis, with patients under 18 and over 80 excluded from the study.
Control (non-IBD) tissue samples were obtained from patients
undergoing screening or surgery for bowel cancer, with non-cancerous
tissue used for studies. For IBD samples, tissue was obtained from
patients diagnosed with CD undergoing resection or endoscopic
surveillance, with diagnosis made by clinical history and/or histo-
logical findings at the time of procedure (see Table 1 for patient
information). For patients with CD, intestinal samples were taken
from inflamed areas, apart from one patient who had quiescent disease
at time of sampling. Intestinal lamina propria samples were incubated
in Hanks buffered saline solution containing 1% penicillin/
streptomycin, 40 mgml
1
G418 and 1 mMdithiothreitol to remove
mucus, then in Hanks buffered saline solution containing 1%
penicillin/streptomycin, 40 mgml
1
G418 and 1 mMEDTA to
remove the epithelial cell layer. Tissue was then incubated
overnight in RPMI medium containing 10% fetal calf serum, 1%
penicillin–streptomycin, 40 mgml
1
G418 antibiotic and 0.2 U ml
1
Liberase DL (Roche, Burgess Hill, UK).
Human blood was obtained from healthy donors recruited locally
(according to the University of Manchester ethics/guidelines) or from
the local blood bank (National Blood Service, Manchester, UK) and
peripheral blood mononuclear cells (PBMCs) were isolated by density
centrifugation using Ficoll-Paque PLUS (GE Healthcare Life Sciences,
Buckinghamshire, UK) according to the manufacturers’ guidelines.
The resulting single-cell suspensions of intestinal cells or PBMCs were
analyzed by flow cytometry as described.
Flow cytometry. Cells were first stained with fixable viability dye
(ThermoFisher Scientific, Paisley, UK), followed by staining with
specific antibodies. Extracellular staining was performed in phos-
phate-buffered saline plus 0.1% bovine serum albumin and 0.05%
sodium azide; intracellular staining was performed using fix/perm
solution and permeabilization buffer (eBioscience, Hatfield, UK) as per
the manufacturer’s protocol. Two percent mouse serum was added to
cells to block nonspecific staining before addition of antibodies. The
following antibodies were used in this study: anti-CD1c (clone L161);
anti-CD3 (clone OKT3 and clone UCHT1); anti-CD4 (clone RPA-T4);
anti-CD11c (clone 3.9); anti-CD14 (clone M5E2); anti-CD15 (clone
W6D3); anti-CD16 (clone 3G8); anti-CD19 (clone HIB19); anti-CD20
(clone 2H7); anti-CD25 (clone BC96); anti-CD45 (clone Hl30); anti-
CD45RA (clone HI100); anti-CD56 (clone MEM-188); anti-CD86
(clone IT2.2); anti-FOXP3 (clone 259D); anti-HLA-DR (clone L243);
anti-SIRPa(clone SE5A5) (all from Biolegend, London, UK); anti-
CD141 (clone 1A4, BD and clone M80, R&D Systems, Minneapolis,
MN, USA); and anti-CD103 (clone B-Ly7, eBioscience). Production of
the anti-integrin b8 (clone ADWA16) is described below.
Figure 4 Integrin b8 expression is upregulated on human dendritic cells (DCs) by lipopolysaccharide (LPS), which activates transformin g growth factor-
b(TGFb) and induces CD4
þ
FOXP3
þ
T cells. (a) Human monocyte-derived DCs (moDCs) were treated with either medium alone, with 5 ng ml
1
TGFb
and 100 nMretinoic acid (RA), or with 100 mgml
1
anti-TGFbantibody (a-TGFb) plus 1 mMRA receptor antagonist (a-RA), and integrin avb8 expression
was analyzed by flow cytometry. NS, not significant. (b) Human moDCs were treated for 48 h with ligands for TLR1/2 (Pam3CSK), TLR3 (Poly I:C), TLR4
(LPS), TLR5 (Flagellin), TLR7 (Imiquimod), or TLR8 (ssRNA40), and integrin avb8 expression was analyzed by flow cytometry. nX3, statistical
significance analyzed by unmatched one-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test, compared with the medium
control (***Po0.001, **Po0.01). (c) Human colon resections were digested and treated overnight with medium or with 100 ng ml
1
LPS. Expression of
integrin avb8 was analyzed on CD1c
þ
and CD141
þ
intestinal DC. Error bars represent mean±s.e.m., n¼4, statistical significance analyzed by paired
Student’s t-tests (*Po0.05). (d) Human moDC treated with either control medium or 100 ng ml
1
LPS were cultured with active TGFbreporter cells
21
in
the presence of either control IgG or anti-integrin avb8 antibody. n¼6, statistical significance analyzed by repeated measures two-way ANOVA with
Sidak’s multiple comparisons test (*Po0.05, **Po0.01). (e,f) Allogenic carboxyfluorescein succinimidyl ester (CSFE)-stained naive CD4
þ
T cells were
sorted by flow cytometry and co-cultured with LPS-treated moDC in the presence of 200 ng ml
1
anti-CD3 antibody and 10 ng ml
1
interleukin-2 in the
presence of either control IgG antibody, anti-TGFbantibody, anti-integrin avb8 antibody, or active TGFbfor 5 days. Representative CD25 and intracellular
FOXP3 staining of naive sorted CD4
þ
CD25
T cells before and after 5 days co-culture (e), and expression of FOXP3 in T cells after 5 days co-culture
after different treatments (f) are shown. (g) Mean fluorescent intensity (MFI) of induced CD4 þFoxp3 þcells from the different conditions. Error bars
represent mean±s.e.m., n¼11, statistical significance analyzed by paired one-way ANOVA with Dunnett’s multiple comparisons test (*Po0.05,
**Po0.01).
ARTICLES
MucosalImmunology 9
Cells were analyzed using an LSR Fortessa or LSRII (BD, Oxford,
UK), and data were analyzed using Flowjo software (Flowjo, OR,
Ashland).
Production of anti-integrin b8 antibody clone ADWA16. Mice
lacking the integrin b8 gene crossed to the outbred CD1 background
(which permits postnatal survival
37
) were immunized at 46 weeks of
age with purified ectodomains of human integrin avb8 (R&D Systems)
at 2-week intervals. Serum was screened by solid phase binding assay
for reaction with purified integrin avb8, and effectively immunized
mice were killed, spleens collected, and splenocytes fused with SP 2/0
fusion partners to generate hybridomas. Clone specificity for human
integrin b8 was screened by flow cytometry using untransfected
SW480 colon carcinoma cells (that do not express integrin avb8, to
exclude antibodies that bound to integrin av or other surface proteins),
to SW480 cells transfected to express integrin avb3oravb6 (as a
further negative control) and to cells transfected with integrin b8
cDNA (Supplementary Figure S1A). The ability of ADWA16 to block
ligand binding and function of integrin avb8 was demonstrated by
inhibition of adhesion of the human glioblastoma integrin avb8-
expressing cell line U251 to plates coated with 1 mgml
1
of
recombinant TGFb1 latency associated peptide (Supplementary
Figure S1B) and inhibition of TGFbactivation by U251 cells,
measured by an active TGFbreporter cell assay (Supplementary
Figure S1C).
Human moDC culture. Leukocyte apheresis cones were collected from
healthy donors at the National Blood Service (Manchester, UK).
PBMCs were separated by centrifugation using Ficoll-Paque (GE
Healthcare, Amersham, UK). Monocytes were separated from PBMCs
using anti-human-CD14 magnetic beads (Miltenyi Biotec, Cologne,
Germany) according to the manufacturer’s instructions, using an
LS MACS separation column. Monocyte purity was consistently
over 95%.
Monocytes were cultured in StemXvivo serum-free DC base
medium (Bio-techne, Minneapolis, MN, USA) containing 25 ng ml
1
GM-CSF and 25 ng ml
1
interleukin-4 (Biolegend, San Diego, CA) for
6 days at a concentration of 0.5 10
6
cells per ml in 24-well tissue
culture-treated plates. Half of the medium was removed on day 3
and replaced with fresh medium and cytokines. After 6 days of
differentiation, cells were treated for 48 h with different combinations
of compounds: 5 ng ml
1
active TGFb(Peprotech, Rocky Hill,
NJ), 100 mgml
1
anti-TGFbantibody (clone 1D11, West Lebanon,
BioXcell, NH), 1 mMpan-RA receptor antagonist (LE540, a kind gift
from Hiroyuki Kagechika, Tokyo Medical and Dental University,
Tokyo, Japan), 100 nMRA (Sigma Aldrich, St Louis, MO, USA),
1mgml
1
Pam3CSK4, 10 mgml
1
Poly(I:C), 1 mgml
1
flagellin,
5mgml
1
Imiquimod, 5 mgml
1
ssRNA40, or 100 ng ml
1
LPS (all
from Invivogen, San Diego, CA).
Active TGFbreporter cell assay. Transformed mink lung epithelial
cells stably expressing a luciferase construct under the control of a
TGFb-responsive promoter
21
(a kind gift from Prof. Dan Rifkin, NYU,
New York, NY, USA) were co-cultured with moDC plus either
200 mgml
1
avb8-blocking ADWA16 antibody or 200 mgml
1
mouse IgG isotype control (BioXcell). Cells were incubated overnight
at 37 1C, and luciferase levels measured using the Luciferase assay
system kit according to the manufacturer’s instructions (Promega,
Madison, WI, USA). An active TGFbstandard curve was used to
calculate levels of active TGFbfrom luminescence intensity observed.
CD4
þ
T-cell FOXP3 induction assay. Human PBMCs were obtained
as above, and CD4
þ
T cells enriched using anti-human-CD4 micro-
beads (Miltenyi Biotec) before sorting naive CD4
þ
CD45RA
þ
CD25
T cells by flow cytometry using an Influx II cell sorter (BD Bioscience,
San Diego, CA, USA). Naive T cells were co-cultured with allogenic
moDC in StemXvivo (R&D Systems) serum-free medium in round-
bottom plates containing interleukin-2 (10 ng ml
1
, Biolegend) and
anti-CD3 antibody (clone OKT3, 0.2 mgml
1
, Biolegend). In all,
510
3
DCs were plated per 1 10
5
T cells in the presence of either
100 mgml
1
anti-TGFb(1D11), 20 mgml
1
anti-integrin b8 (clone
37e1B5,
38
100 mgml
1
isotype control IgG (MOPC-21), or 5 ng ml
1
active TGFb(Peprotech).
Statistical analysis. Data were analyzed with Prism Software
(GraphPad Software, La Jolla, CA, USA). Statistical differences
between means were tested as described in figure legends. All data are
expressed as mean±s.e.m.
SUPPLEMENTARY MATERIAL is linked to the online version of the paper
at http://www.nature.com/mi
ACKNOWLEDGMENTS
We thank the Faculty of Life Sciences Flow Cytometry Core facility;
specifically Mr Mike Jackson and Dr Gareth Howell, for help with cell sorting
and analysis, Prof. Richard Grencis for helpful comments on the manu-
script, and all patients who volunteered to provide tissue and blood
samples for these studies. T.M.F. was supported by a Biotechnology and
Biological Sciences (BBSRC) CASE PhD studentship (awarded to
F.P., M.J.L. and M.A.T.) (representing DuPont Nutrition & Health), and
J.J.W. was supported by a Wellcome Trust Stepping Stones Fellowship
(097820/Z/11/B). Work was supported by a BBSRC Diet and Research
Industry Club grant (awarded to F.P., M.J.L. and M.A.T), core funding from
the Manchester Collaborative Centre for Inflammation Research (awarded
to M.A.T.) and HL113032 to S.L.N. The Wellcome Trust Centre for Cell-
Matrix Research, University of Manchester, is supported by core funding
from the Wellcome Trust (088785/Z/09/Z).
AUTHOR CONTRIBUTIONS
T.M.F., F.P., J.J.W., M.J.L., and M.A.T. designed the studies; T.M.F., A.K.,
E.E.S, C.S., and A.A. performed experiments and analyzed data; A.A.,
S.L.N., and D.S. produced and provided novel antibody reagents for
experiments; S.C. and S.L. performed clinical tissue sampling and patient
analysis; T.M.F, J.J.W., A.K., and M.A.T. wrote the manuscript; F.P., M.J.L.,
and M.A.T. obtained funding for the study.
DISCLOSURE
The authors declared no conflict of interest.
Official journal of the Society for Mucosal Immunology
REFERENCES
1. Maloy, K.J. & Powrie, F. Intestinal homeostasis and its breakdown in
inflammatory bowel disease. Nature 474, 298–306 (2011).
2. Travis, M.A. & Sheppard, D. TGF-bActivation and Function in Immunity.
Annu. Rev. Immunol. 32, 51–82 (2014).
3. Monteleone, G. et al. Mongersen, an oral SMAD7 antisense oligonucleo-
tide, and Crohn’s disease. N. Engl. J. Med. 372, 1104–1113 (2015).
4. Travis, M.A. et al. Loss of integrin alpha(v)beta8 on dendritic cells causes
autoimmunity and colitis in mice. Nature 449, 361–365 (2007).
5. Lacy-Hulbert, A. et al. Ulcerative colitis and autoimmunity induced by loss
of myeloid alphav integrins. Proc. Natl. Acad. Sci. USA 104, 15823–15828
(2007).
6. Melton, A.C. et al. Expression of alphavbeta8 integrin on dendritic cells
regulates Th17 cell development and experimental autoimmune ence-
phalomyelitis in mice. J. Clin. Invest. 120, 4436 (2010).
7. Paı
¨dassi, H. et al. Preferential expression of integrin avb8 promotes
generation of regulatory T cells by mouse CD103 þdendritic cells.
Gastroenterology 141, 1813–1820 (2011).
8. Worthington, J.J., Czajkowska, B.I., Melton, A.C. & Travis, M.A. Intestinal
dendritic cells specialize to activate transforming growth factor-band
induce Foxp3 þregulatory T cells via integrin avb8. Gastroenterology 141,
1802–1812 (2011).
9. Worthington, J.J. et al. Loss of the TGFB-activating integrin avB8 on
dendritic cells protects mice from chronic intestinal parasitic infection via
control of type 2 immunity. PLoS Pathog. 9, e1003675 (2013).
ARTICLES
10 www.nature.com/mi
10. Bekiaris, V., Persson, E.K. & Agace, W.W. Intestinal dendritic cells in the
regulation of mucosal immunity. Immunol. Rev. 260, 86–101 (2014).
11. Luda, K.M. et al. IRF8 transcription-factor-dependent classical dendritic
cells are essential for intestinal T cell homeostasis. Immunity 44, 860–874
(2016).
12. Bachem, A. et al. Superior antigen cross-presentation and XCR1
expression define human CD11c þCD141 þcells as homologues of
mouse CD8 þdendritic cells. J. Exp. Med. 207, 1273–1281 (2010).
13. Crozat, K. et al. The XC chemokine receptor 1 is a conserved selective
marker of mammalian cells homologous to mouse CD8alpha þdendritic
cells. J. Exp. Med. 207, 1283–1292 (2010).
14. Jongbloed, S.L. et al. Human CD141 þ(BDCA-3) þdendritic cells (DCs)
represent a unique myeloid DC subset that cross-presents necrotic cell
antigens. J. Exp. Med. 207, 1247–1260 (2010).
15. Poulin, L.F. et al. Characterization of human DNGR-1 þBDCA3 þ
leukocytes as putative equivalents of mouse CD8alpha þdendritic cells.
J. Exp. Med. 207, 1261–1271 (2010).
16. Watchmaker, P.B. et al. Comparative transcriptional and functional profiling
defines conserved programs of intestinal DC differentiation in humans and
mice. Nat. Immunol. 15, 98–108 (2014).
17. Gross, M., Salame, T.-M. & Jung, S. Guardians of the gut–murine intestinal
macrophages and dendritic cells. Front. Immunol. 6, 1–10 (2015).
18. Babyatsky, M., Rossiter, G. & Podolsky, D. Expression of transforming
growth factors alpha and beta in colonic mucosa in inflammatory bowel
disease. Gastroenterology 110, 975–984 (1996).
19. Sanders, T.J. et al. Increased production of retinoic acid by intestinal
macrophages contributes to their inflammatory phenotype in patients with
Crohn’s disease. Gastroenterology 146, 1278–1288.e2 (2014).
20. Veldhoen, M. & Brucklacher-Waldert, V. Dietary influences on intestinal
immunity. Nat. Rev. Immunol. 12, 696–708 (2012).
21. Abe, M. et al. An assay for transforming growth factor-beta using cells
transfected with a plasminogen activator inhibitor-1 promoter-luciferase
construct. Anal. Biochem. 216, 276–284 (1994).
22. Beura, L.K. et al. Normalizing the environment recapitulates adult human
immune traits in laboratory mice. Nature 532, 512–516 (2016).
23. Yoshida, Y. et al. The transcription factor IRF8 activates integrin-mediated
TGF-bsignaling and promotes neuroinflammation. Immunity 40, 187–198
(2014).
24. Edelson, B.T. et al. Peripheral CD103 þdendritic cells form a unified
subset developmentally related to CD8alpha þconventional dendritic
cells. J. Exp. Med. 207, 823–836 (2010).
25. Hambleton, S. et al. IRF8 mutations and human dendritic-cell immuno-
deficiency. N. Engl. J. Med. 365, 127–138 (2011).
26. Monteleone, G. et al. Blocking Smad7 restores TGF- b1 signaling in
chronic inflammatory bowel disease. J. Clin. Invest. 108, 523–526 (2001).
27. Reboldi, A. et al. IgA production requires B cell interaction with
subepithelial dendritic cells in Peyer’s patches. Science 352, aaf4822
(2016).
28. Markovics, J.A. et al. Transcription of the transforming growth factor beta
activating integrin beta8 subunit is regulated by SP3, AP-1, and the p38
pathway. J. Biol. Chem. 285, 24695–24706 (2010).
29. Allan, S.E. et al. Activation-induced FOXP3 in human T effector cells
does not suppress proliferation or cytokine production. Int. Immunol. 19,
345–354 (2007).
30. Gavin, M.A. et al. Single-cell analysis of normal and FOXP3-mutant
human T cells: FO XP3 expression without regulatory T cell development.
Proc. Natl. Acad. Sci. USA 103, 6659–6664 (2006).
31. Walker, M.R. et al. Induction of FoxP3 and acquisition of Tregulatory activity
by stimulated human CD4 þCD25
–
Tcells. J. Clin. Invest. 112, 1437–1443
(2003).
32. Pillai, V., Ortega, S.B., Wang, C.K. & Karandikar, N.J. Transient regulatory
T-cells: a state attained by all activated human T-cells. Clin. Immunol. 123,
18–29 (2007).
33. McMurchy, A.N. et al. Plenary Paper A novel function for FOXP3 in
humans: intrinsic regulation of conventional T cells. Blood 121, 1265–1276
(2013).
34. Maul, J. et al. Peripheral and intestinal regulatory CD4 þCD25high Tcells in
inflammatory bowel disease. Gastroenterology 128, 1868–1878 (2005).
35. Saruta, M. et al. Characte rization of FOXP3 þCD4 þregulatory T cells in
Crohn’s disease. Clin. Immunol. 125, 281–290 (2007).
36. Fantini, M.C. et al. Smad7 controls resistance of colitogenic T cells
to regulatory T cell-mediated suppression. Gastroenterology 136,
1308–1316 (2009).
37. Aluwihare, P. et al. Mice that lack activity of alphavbeta6- and alphavbeta8-
integrins reproduce the abnormalities of Tgfb1- and Tgfb3-null mice. J. Cell
Sci. 122, 227–232 (2009).
38. Minagawa, S. et al. Selective targeting of TGF-beta activation to treat
fibroinflammatory airway disease. Sci. Transl. Med. 6, 241ra79 (2014).
This work is licensed under a Creative Commons
Attribution4.0 International License.Theimages or
otherthirdparty material in this article areincluded inthearticle’s
Creative Commons license, unless indicated otherwise in the
credit line; if the material is not included under the Creative
Commons license, users will need to obtain permission from the
license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
ARTICLES
MucosalImmunology 11
