A convenient model of severe, high incidence autoimmune gastritis caused by polyclonal effector T cells and without perturbation of regulatory T cells.

Abstract

Autoimmune gastritis results from the breakdown of T cell tolerance to the gastric H(+)/K(+) ATPase. The gastric H(+)/K(+) ATPase is responsible for the acidification of gastric juice and consists of an α subunit (H/Kα) and a β subunit (H/Kβ). Here we show that CD4(+) T cells from H/Kα-deficient mice (H/Kα(-/-)) are highly pathogenic and autoimmune gastritis can be induced in sublethally irradiated wildtype mice by adoptive transfer of unfractionated CD4(+) T cells from H/Kα(-/-) mice. All recipient mice consistently developed the most severe form of autoimmune gastritis 8 weeks after the transfer, featuring hypertrophy of the gastric mucosa, complete depletion of the parietal and zymogenic cells, and presence of autoantibodies to H(+)/K(+) ATPase in the serum. Furthermore, we demonstrated that the disease significantly affected stomach weight and stomach pH of recipient mice. Depletion of parietal cells in this disease model required the presence of both H/Kα and H/Kβ since transfer of H/Kα(-/-) CD4(+) T cells did not result in depletion of parietal cells in H/Kα(-/-) or H/Kβ(-/-) recipient mice. The consistency of disease severity, the use of polyclonal T cells and a specific T cell response to the gastric autoantigen make this an ideal disease model for the study of many aspects of organ-specific autoimmunity including prevention and treatment of the disease.

Full text

Page 1

A Convenient Model of Severe, High Incidence
Autoimmune Gastritis Caused by Polyclonal Effector T
Cells and without Perturbation of Regulatory T Cells
Eric Tu, Desmond K. Y. Ang, Thea V. Hogan¤a, Simon Read¤b, Cheryl P. Z. Chia, Paul A. Gleeson, Ian R. van
Driel*
Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute, The University of Melbourne, Parkville, Victoria, Australia
Abstract
Autoimmune gastritis results from the breakdown of T cell tolerance to the gastric H+/K+ATPase. The gastric H+/K+ATPase is
responsible for the acidification of gastric juice and consists of an a subunit (H/Ka) and a b subunit (H/Kb). Here we show
that CD4+T cells from H/Ka-deficient mice (H/Ka2/2) are highly pathogenic and autoimmune gastritis can be induced in
sublethally irradiated wildtype mice by adoptive transfer of unfractionated CD4+T cells from H/Ka2/2mice. All recipient
mice consistently developed the most severe form of autoimmune gastritis 8 weeks after the transfer, featuring hypertrophy
of the gastric mucosa, complete depletion of the parietal and zymogenic cells, and presence of autoantibodies to H+/K+
ATPase in the serum. Furthermore, we demonstrated that the disease significantly affected stomach weight and stomach
pH of recipient mice. Depletion of parietal cells in this disease model required the presence of both H/Ka and H/Kb since
transfer of H/Ka2/2CD4+T cells did not result in depletion of parietal cells in H/Ka2/2or H/Kb2/2recipient mice. The
consistency of disease severity, the use of polyclonal T cells and a specific T cell response to the gastric autoantigen make
this an ideal disease model for the study of many aspects of organ-specific autoimmunity including prevention and
treatment of the disease.
Citation: Tu E, Ang DKY, Hogan TV, Read S, Chia CPZ, et al. (2011) A Convenient Model of Severe, High Incidence Autoimmune Gastritis Caused by Polyclonal
Effector T Cells and without Perturbation of Regulatory T Cells. PLoS ONE 6(11): e27153. doi:10.1371/journal.pone.0027153
Editor: Ciriaco A. Piccirillo, McGill University Health Center, Canada
Received January 21, 2011; Accepted October 11, 2011; Published November 9, 2011
Copyright: ? 2011 Tu et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by research grants from the National Health and Medical Research Council of Australia and the University of Melbourne. The
funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors can confirm that Simon Read was at The University of Melbourne when this study was completed and that Novo Nordisk is
his current address. The authors have declared no competing interests exist.
* E-mail: ivd@unimelb.edu.au
¤a Current address: MRC National Institute for Medical Research, The Ridgeway, Mill Hill, London, United Kingdom
¤b Current address: Novo Nordisk a/s, Ma ˚løv, Denmark
Introduction
Autoimmune gastritis is an excellent system to investigate the
loss of T cell tolerance to self-tissues since it is a well-characterised
autoimmune disease with a known target antigen and the mouse
model of the disease has many similarities to the human
equivalent. Pernicious anemia, the end stage of autoimmune
gastritis, has an estimated prevalence of 1.9% in Western adult
population over the age of 60 years [1], which represents the
commonest cause of vitamin B12 deficiency in the elderly [2].
Autoimmune gastritis features infiltration of mononuclear cells in
the submucosa which extends into the gastric mucosa, the
depletion of gastric parietal and zymogenic cells and hypertrophy
of the gastric mucosa [1].
The target antigen in autoimmune gastritis has been identified
as the gastric H+/K+ATPase expressed by gastric parietal cells [3–
5]. The gastric H+/K+ATPase is a membrane proton pump
responsible for the acidification of gastric juice and it is a
heterodimer that consists of a catalytic a subunit (H/Ka) and a
glycoprotein b subunit (H/Kb). Previous studies have demon-
strated that both H/Ka and H/Kb are targeted in autoimmune
gastritis [4,6–8]. It is well documented that, in both mice and man,
CD4+T cells that recognise the gastric H+/K+ATPase initiate
autoimmune gastritis while CD8+T cells and B cells are ineffective
in doing so [7–17]. Nevertheless, following initiation of the disease,
autoantibodies to the H+/K+ATPase are produced and are a very
useful diagnostic tool of the disease although they are not
pathogenic [3,16,18].
BALB/c mice have been shown to be the most susceptible to
lymphopenia-induced autoimmune gastritis with an incidence of
30–90% [19,20]. The disease can be induced by thymectomy of 3
day-old mice or adult thymectomy combined with a single dose of
cyclophosphamide [19,21]. In spite of this, it is difficult to evaluate
therapeutic effects of treatments given to thymectomised mice
since a spectrum of disease severity from mild to severe is always
observed and incidence is variable, thus large group sizes are
required to test statistical significance. Furthermore, the surgery
involved and post-operative maintenance of animals is technically
demanding. Autoimmune gastritis can also be induced in BALB/c
mice by immunisation with purified gastric H+/K+ATPase [5].
However, the disease is reversible after cessation of immunisation
and has a low severity. Transfer of BALB/c T cells depleted of
regulatory T (Treg) cells into athymic recipient mice causes severe
autoimmune gastritis [15]. It is therefore impossible to dissect the
relationship between autoreactive T cells and Treg cells during the
disease development in this setting since Treg cells are absent.
PLoS ONE | www.plosone.org1November 2011 | Volume 6 | Issue 11 | e27153

Page 2

High incidence of spontaneous autoimmune gastritis is observed
in transgenic mice expressing a TCR specific for H/Ka [8].
However, these mice do not represent an ideal disease model for
all experiments because the monoclonal nature of the pathogenic
T cells does not recapitulate the polyclonal response that occurs in
pathophysiological conditions.
From this perspective, we developed a disease model which
relied on the transfer of polyclonal T cells from H/Ka-deficient
(H/Ka2/2) mice into sublethally-irradiated wildtype mice. We
demonstrated that mice that received unfractionated T cells from
H/Ka2/2mice all succumbed to the most severe autoimmune
gastritis with pathological changes observed throughout the whole
stomach. Furthermore, these pathological changes significantly
affected the physiology of stomach. We demonstrated the
usefulness of this model by showing that the development of
autoimmune gastritis in this model was strictly dependent on the
presence of both H+/K+ATPase a and b subunits.
Results
Elevated T and B cell response in H/Ka2/2mice following
immunisation with H+/K+ATPase
In order to have a system that is suitable for the study of
different aspects of autoimmunity, we established a disease model
in which the disease is caused by pathogenic polyclonal T cells and
the disease severity is consistently severe. A T cell response to H/
Ka is required for the onset of autoimmune gastritis and may be
the dominant source of gastric autoreactive epitopes [7,8,22]. As T
cell tolerance to H/Ka was unlikely to occur in H/Ka2/2mice,
we investigated if T cells that were specific to H+/K+ATPase and
capable of causing the gastritis were present in H/Ka2/2mice.
To determine if there were elevated immune responses to H+/
K+ATPase in H/Ka2/2mice relative to wildtype mice, mice
were immunised twice with purified gastric H+/K+ATPase. After
the immunisation, autoantibodies specific for H+/K+ATPase were
detected in the immunised H/Ka2/2mice, but not in wildtype
mice (Figure 1A). T cell response to in vitro restimulation with H+/
K+ATPase was also significantly greater in H/Ka2/2mice than
in wildtype mice (Figure 1B). Therefore, as has been previously
demonstrated in H/Kb2/2mice [23], there is also a significant T
and B cell response to H+/K+ATPase in H/Ka2/2mice that is
not present in wildtype mice.
CD4+T cells from H/Ka2/2mice were highly pathogenic
To determine if adoptive transfer of T cells from H/Ka2/2
mice would be able to cause autoimmune gastritis, we enriched
CD4+T cells from either H/Ka2/2or wildtype mice to ,85%
purity and transferred 56107of these cells into sublethally
irradiated wildtype mice. Since it was previously demonstrated
that H+/K+ATPase-specific CD4+T cells were rapidly deleted in
the periphery and unable to induce autoimmune gastritis in non-
irradiated wildtype mice [24], all recipient mice used in this study
were lightly irradiated at 600 rads to enhance the survival of
transferred T cells from H/Ka2/2mice.
Eight weeks after the transfer, mice that received H/Ka2/2
CD4+T cells all developed the most severe forms of autoimmune
gastritis (score 5 and 6). This was demonstrated by the complete
depletion of parietal cells and zymogenic cells, and hypertrophy of
the gastric mucosa (Figure 2A, 2B). In contrast, mice that received
wildtype CD4+T cells were completely free of gastritis (Figure 2A,
2B), which indicated that autoimmune gastritis in mice receiving
H/Ka2/2CD4+T cells was not caused by irradiation per se, but
rather, the T cell inoculum.
Pathological changes in the stomach of gastritic mice signifi-
cantly affected their stomach physiology. Hypertrophy of the
gastric mucosa resulted in the enlargement of stomach rugae
(Figure 2C) and an increase in stomach weight (Figure 2D).
Secondly, depletion of parietal cells caused an increase in stomach
pH due to the lack of acid secretion (Figure 2E). In addition,
autoantibodies specific to the gastric H+/K+ATPase could be
detected in all gastritic mice (Figure 2F). Autoantibodies reached
high levels in all mice 6 weeks after the transfer (Figure 2G).
Together, these results demonstrated the presence of highly
pathogenic H+/K+ATPase-specific T cells in H/Ka2/2mice.
Autoimmune gastritis was caused by CD4+T cell-
mediated inflammation
Micewithautoimmunegastritisinducedbytransfer ofH/Ka2/2
CD4+T cells had significantly more cells in their stomach-draining
paragastric lymph nodes compared to normal mice and mice that
received CD4+T cells from wildtype donors. In contrast, there was
Figure 1. Elevated immune response to immunisation with H+/
K+ATPase in H/Ka2/2mice. WT and H/Ka2/2mice were immunised
with H+/K+ATPase as described in ‘‘Materials and Methods’’. (A) H+/K+
ATPase-specific autoantibodies in serum were detected by ELISA using
serial 2-fold dilutions of mouse sera, with a starting dilution factor of 5.
Data is representative of three independent experiments. Error bars,
standard error. (B) Cells from inguinal lymph nodes were cultured for
72 hrs with splenic DC, and T cell proliferation was assessed by
incorporation of3H-thymidine. For each mouse, T cells were cultured in
triplicate with either DC with gastric membrane or with DC alone. The
mean cpm of T cells cultured with DC with antigen was divided by the
mean cpm of the same T cells cultured with DC alone to calculate the
stimulation index. Data pooled from four independent experiments. In
(B), each circle represents the data from one mouse. Mann-Whitney U
test was used; bars, mean and ***, P,0.001.
doi:10.1371/journal.pone.0027153.g001
A High-Incidence Model for Autoimmune Gastritis
PLoS ONE | www.plosone.org2November 2011 | Volume 6 | Issue 11 | e27153

Page 3

no difference in the size of the inguinal lymph node between
gastritic and non-gastritic mice (Figure 3A). This suggested that the
increase in the cellularity of paragastric lymph nodes in gastritic
mice was caused by local gastric inflammation and not by
preferential survival or expansion of H/Ka2/2CD4+T cells
compared to wildtype CD4+T cells after transfer.
In these experiments, cells from the donor H/Ka2/2or
wildtype mice could be differentiated from recipient cells because
they bore different CD90 alleles. The donor population in the
paragastric lymph node of mice that received H/Ka2/2CD4+T
cells had significantly more CD4+T cells with an effector/memory
phenotype (CD44hiCD62lo) in comparison to mice that received
wildtype CD4+T cells. However, the levels of effector/memory T
cells were similar in the non-draining lymph node (Figure 3B).
Together, these results suggest that the development of autoim-
mune gastritis in mice that received CD4+T cells from H/Ka2/2
mice was caused by a T cell-mediated inflammatory response in
the gastric environment rather than a more generalised response.
The development of autoimmune gastritis in this disease model
was not associated with a reduced level of regulatory T (Treg) cells
Figure 2. Mice that receive CD4+T cells from H/Ka2/2mice develop severe autoimmune gastritis. Irradiated wildtype mice that received
56107CD4+T cells from either wildtype (WT) or H/Ka2/2mice, were killed 8 weeks after transfer and compared to non-manipulated mice (2). (A)
Haematoxylin and eosin stained sections of stomachs (B) Gastritis scores. (C) Macroscopic views of stomachs (D) Stomach weights after stomach
contents were removed (E) Gastric pH. Mice were starved overnight and their stomachs were rinsed in 1 ml saline which was then collected and
measured by a pH meter. (F) H+/K+ATPase-specific autoantibodies in serum were detected by ELISA using serial 2-fold dilutions of mouse sera, with a
starting dilution factor of 50. Error bars, standard error. (G) Gastric H+/K+ATPase autoantibody levels in mice at various times after cell transfer. Serum
samples were collected at the indicated time points from each mouse after transfer of CD4+T cells from H/Ka2/2mice. Autoantibody titre was
represented by the maximum dilution factor that had an absorbance reading above 50% maximum. ND=not detectable. Data pooled from four
independent experiments. In (B), (D) and (E), each circle represents the data from one mouse. Mann-Whitney U test was used; bars, mean and
***, P,0.001.
doi:10.1371/journal.pone.0027153.g002
A High-Incidence Model for Autoimmune Gastritis
PLoS ONE | www.plosone.org3 November 2011 | Volume 6 | Issue 11 | e27153

Page 4

since gastritic mice contained similar proportions of Treg cells in
the paragastric lymph node compared to non-gastritic controls
(Figure 4A, 4B). In fact, significantly more Treg cells were found in
the stomach of mice with autoimmune gastritis (Figure 4A, 4C).
H/Ka2/2CD4+T cells rapidly induced tissue damage in
the stomach
The severity of autoimmune gastritis was analysed at various
time points after the transfer of H/Ka2/2CD4+T cells.
Moderate depletion of parietal cells and zymogenic cells, as
indicated by a gastritis score of 4 [24], could be observed as early
as 2 weeks after the transfer and the disease progressed rapidly
with most of the mice showing severe gastritis, as shown by a
gastritis score of 5 and 6, 4 weeks after the transfer (Figure 5).
Stomach weight can be used as a quantitative measure
of autoimmune gastritis
We have demonstrated that mice with autoimmune gastritis had
a greater stomach weight than disease-free mice due to the
hypertrophy of the gastric mucosa (Figure 2D). Therefore, we
investigated if stomach weight could be used as a quantitative
measure for the severity of autoimmune gastritis. To further test if
stomach weight could differentiate among different severities of
autoimmune gastritis, we examined neonatally-thymectomised
mice as greater numbers of mice with variable disease scores could
be obtained. We found that there were significant differences in
the mean stomach weight between each of the first five levels of
autoimmune gastritis (score 0–4) (Figure 6). However, the mean
stomach weights of gastritis scores 4, 5 and 6 were not significantly
different from each other (Figure 6). This is likely due to the fact
that the degree of hypertrophy of gastric mucosa does not increase
as the disease progresses to the late stages.
Induction of autoimmune gastritis by H/Ka2/2CD4+T
cells was dependent on the presence of H/Ka and H/Kb
We hypothesised that the pathogenicity of CD4+T cells from
H/Ka2/2mice was caused by the lack of tolerance to H/Ka in
the absence of this gastric autoantigen. To test this, CD4+T cells
were transferred from H/Ka2/2mice into sublethally-irradiated
H/Ka2/2and wildtype mice. It has been demonstrated that H/
Ka, in the absence of H/Kb, cannot be presented by MHC
molecules [25] (S Allen, personal communication). We therefore
transferred CD4+T cells from H/Ka2/2mice into sublethally-
irradiated H/Kb2/2mice and compared them with H/Ka2/2
and wildtype recipient mice. The analysis of autoimmune attack
on the gastric mucosa is complicated in H/Ka2/2and H/Kb2/2
mice by the changes that occur in the gastric mucosa since a lack
of gastric acid in these mice results in elevated levels of the trophic
hormone gastrin. This leads to constitutive hypertrophy and
depletion of zymogenic cells [26,27]. Therefore, in order to assess
if an autoimmune response attacks the gastric mucosa in these
mice, we instead determined the prevalence of parietal cells, which
are normally depleted in autoimmune gastritis but are found in
abundance in H/Ka2/2and H/Kb2/2mice, and also the
presence of gastric autoantibodies.
The stomach morphology in H/Ka2/2mice or H/Kb2/2
mice was not affected by the transfer of H/Ka2/2CD4+T cells
(Figure 7A, 7B). In particular, there was no depletion of parietal
cells indicating the lack of an autoimmune response. In contrast,
complete depletion of parietal cells was seen in wildtype recipient
mice that received H/Ka2/2CD4+T cells (Figure 7C). Since
stomachs of H/Ka2/2and H/Kb2/2mice appeared different to
both normal and gastritic mice, we could not assess these mice
using our standard gastritis scoring system. However, H+/K+
ATPase-specific autoantibodies, another hallmark of autoimmune
gastritis, were not detected in the sera of both H/Ka2/2and H/
Kb2/2recipient mice, but were detected in wildtype recipient
mice (Figure 7D). These results demonstrated that the presence of
both a and b subunits was required for the depletion of parietal
cells by H/Ka2/2CD4+T cells.
Discussion
Autoimmune gastritis can be induced in BALB/c mice by
several means, including neonatal thymectomy, adult thymectomy
with cyclophosphamide [19,21], immunisation with gastric H+/K+
ATPase [5] and generation of H+/K+ATPase-specific TCR
transgenic mice [7,8]. However, an immune system dominated by
T cells with a single specificity as observed in TCR transgenic mice
does not resemble normal physiological conditions. Furthermore,
thymectomy and immunisation do not confer consistent disease
severity and they are technically demanding, which makes it
inconvenient to use these models for the analysis of immunother-
apies. We therefore developed a new disease model that relied on
the transfer of T cells from H/Ka2/2mice, a polyclonal
population that contained highly gastritogenic T cells.
We found that immunisation of H/Ka2/2mice with purified
H+/K+ATPase induced a vigorous T cell response as the result of
Figure 3. Cellularity of stomach-draining and non-draining
lymph nodes. (A) The number of cells in the paragastric and inguinal
lymph nodes from recipient mice 8 weeks after transfer of either
wildtype or H/Ka2/2CD4+T cells. (B) The percentage of effector/
memory T cells in the donor population with the phenotype
CD44hiCD62Lloin CD90.2+CD4+T cells. Data pooled from four
independent experiments. Each circle represents the data from one
mouse. Mann-Whitney U test was used; bars, mean, **, P,0.01 and
***, P,0.001.
doi:10.1371/journal.pone.0027153.g003
A High-Incidence Model for Autoimmune Gastritis
PLoS ONE | www.plosone.org4November 2011 | Volume 6 | Issue 11 | e27153

Page 5

the lack of T cell tolerance to the gastric autoantigen, which
indicated the presence of H+/K+ATPase-specific T cells in H/
Ka2/2mice. We further demonstrated that transfer of CD4+T
cell populations from H/Ka2/2mice induced damage to the
stomach tissues of recipient mice soon after transfer. Autoimmune
gastritis induced by the transfer of H/Ka2/2CD4+T cells was
consistently severe: all the recipient mice developed the most
severe disease eight weeks after transfer and their stomach
physiology was significantly affected as indicated by the increase
in stomach weight and pH. This model is advantageous because
the severity and incidence of disease approaches that observed in
TCR transgenic mice yet is caused by a polyclonal repertoire, it is
technically simple to establish, it does not rely upon perturbation
of the normal Treg cell repertoire.
The donor cell population from the H/Ka2/2mice was
enriched to ,85% CD4+T cells by a ‘‘negative selection’’
procedure because the major aim of this work was to produce a
simple, inexpensive and convenient disease model. We conclude
that it is the CD4+T cells in the inoculum that caused
Figure 4. Mice with autoimmune gastritis have an increased level of Foxp3+Treg cells in the stomach. Cells were harvested from (A, B)
paragastric lymph node, inguinal lymph node and (A, C) stomach tissue for intracellular staining of Foxp3. The numbers indicate the percentages of
Foxp3+Treg cells in donor CD90.2+CD4+T cells in paragastric and inguinal lymph nodes and Foxp3+Treg cells in total CD4+T cells in stomach tissue.
Data pooled from three independent experiments. In (B) and (C), each circle represents the data from one mouse. Mann-Whitney U test was used;
bars, mean, and **, P,0.01.
doi:10.1371/journal.pone.0027153.g004
Figure 5. Severity of autoimmune gastritis correlates to
stomach weight. Mice were killed at indicated time points after
transfer of CD4+T cells from H/Ka2/2mice and the severity of
autoimmune gastritis was determined. Data were pooled from six
independent experiments. Each circle represents the data from one
mouse. Bars, mean.
doi:10.1371/journal.pone.0027153.g005
Figure 6. Significant difference in stomach weight among
gastritis scores. Thymectomy was performed on BALB.B6-Gasa
congenic mice on day 3 of age and mice killed at 12 weeks. The
severity of autoimmune gastritis and stomach weight was determined.
Data pooled from at least 100 independent thymectomies. Each circle
represents the data from one mouse. Mann-Whitney U test was used;
bars, mean, **, P,0.01 and ***, P,0.001.
doi:10.1371/journal.pone.0027153.g006
A High-Incidence Model for Autoimmune Gastritis
PLoS ONE | www.plosone.org5November 2011 | Volume 6 | Issue 11 | e27153