| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
in Type 1 Diabetes: Islet-Specific Expression Abrogates the Ongoing Autoimmune Process When Induced Late but Not Early During Pathogenesis1
*
Departments of Neuropharmacology and Immunology, The Scripps Research Institute, La Jolla, CA 92037; and
Section of Immunobiology, Yale University School of Medicine, New Haven, CT 06520
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
can play a
dual role in autoimmune diabetes, depending on its precise timing in
relation to the ongoing autoimmune process. In a transgenic model (rat
insulin promoter-lymphocytic choriomeningitis virus) of virally induced
diabetes, TNF- enhanced disease incidence when induced through an
islet-specific tetracycline-dependent promoter system early during
pathogenesis. Blockade of TNF- during this phase prevented diabetes
completely, suggesting its pathogenetic importance early in disease
development. In contrast, TNF- expression abrogated the autoimmune
process when induced late, which was associated with a reduction of
autoreactive CD8 lymphocytes in islets and their lytic activities.
Thus, the fine-tuned kinetics of an autoreactive process undergo
distinct stages that respond in a differential way to the presence of
TNF-. This observation has importance for understanding the complex
role of inflammatory cytokines in autoimmunity. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
are thought to play an important role in the
pathogenesis of autoimmune type 1 diabetes. TNF- is directly
implicated in the destruction of 
-cells from in vitro studies on
isolated islets (1) and has profound inflammatory effects
in vivo by acting directly on APCs and autoreactive T lymphocytes
(2, 3, 4, 5). Studies in animal models for type 1 diabetes have
predominantly revealed functions that accelerate the autoimmune
process. For example, constitutive TNF- expression in islets in rat
insulin promoter
(RIP)3-TNF--transgenic
C57BL/6 mice leads to profound intraislet infiltration but,
surprisingly, not to overt autoimmune diabetes (6). Only
when the costimulatory molecule B7.1 is coexpressed with TNF- does
diabetes develop (7). It has recently become
clear by using an islet-specific tetracycline (Tet)-regulated TNF-
in vivo model that the initial duration of TNF- expression in this
TNF--B7.1 diabetes model is crucial for determining the fate of
autoaggressive CD8 lymphocytes (8). Only if TNF- is
"on" for >21 days will CD8 lymphocytes enter the islets and
destroy them. It is quite likely that enhanced presentation of islet
Ags plays an important role in this situation. Evidence for the
profound effect of TNF- expression on islet Ag presentation to
autoaggressive CD8 lymphocytes early during diabetes pathogenesis has
been generated in RIP-TNF--transgenic nonobese diabetic (NOD) mice.
Such animals exhibit markedly accelerated spontaneous diabetes, and
presentation of cell Ags to islet-infiltrating CD4 as well as to
CD8 lymphocytes is enhanced. Interestingly, only CD8 lymphocytes show a
clear pathogenetic potential in this model from adoptive transfer
studies, whereas CD4 lymphocytes might even have regulatory function
(4, 9, 10, 11). Similar to these transgenic models in which
expression of TNF- was specifically targeted to the islets, early
systemic TNF- administration enhanced diabetes in NOD mice (2, 5). However, another study reported that systemic TNF-
administration late during diabetes development could abrogate the
disease process, probably by affecting expansion, migration, and
function of autoreactive lymphocytes (12). This finding
was mirrored by another in vivo study in NOD mice in which TNF- was
expressed later during pathogenesis and not in the neonatal period, and
diabetes was prevented. In this model, massive insulitis was observed,
but these lymphocytes did not have sufficient autodestructive
capability to induce diabetes, and, in addition, TNF- expression
down-modulated the capacity of autoreactive T lymphocytes in these NOD
mice to destroy islets in adoptive transfer experiments
(3). Thus, TNF- appears to play a dual role in the
regulation and propagation of the diabetogenic autoimmune process, and
its precise function appears to depend on the timing of expression.
Based on these previous studies, our intention was to precisely dissect
the time factor of islet-specific TNF- expression in relation to the
ongoing autoimmune process. For this endeavor, we chose to develop a
double-transgenic mouse model in which islet TNF- expression could
be repressed and derepressed by Tet administration, and a viral self Ag
was constitutively expressed under the RIP as a target autoantigen.
This RIP-lymphocytic choriomeningitis virus (LCMV) x Tet-TNF-
model (see next paragraph) had the advantages that autoreactive
lymphocytes could be traced precisely, the time point of initiation of
autoimmunity could be chosen experimentally, and TNF- could be
switched on for any period of time during the pathogenesis of
diabetes.
The RIP-LCMV-transgenic mouse models for virally induced autoimmune
diabetes were first developed by the laboratories of
Zinkernagel et al. (13) and Oldstone et al.
(14) in 1990. RIP-LCMV mice express the gp of
LCMV under control of the RIP selectively in pancreatic cells as a
target/marker autoantigen. Interestingly, these mice are not tolerant
but are rather unresponsive to the viral (self) protein, because
infection with LCMV leads to autoimmune diabetes in 80–100% of such
transgenic animals. Further analysis of the Oldstone model revealed a
three-stage pathogenetic process. First (the initiation stage),
systemic infection with LCMV leads to up-regulation of MHC class I
molecules on islet cells (15), activation of APCs, and
production of type 1 IFNs as early as 2 days after infection in most
major organs, including the pancreas (but not the islets) (15, 16). At this time, no lymphocytes are found in the islets or
pancreas. However, these events are pathogenetically crucial, because
adoptive transfer of LCMV-specific lymphocytes into uninfected RIP-LCMV
mice leads to insulitis but not diabetes (14, 16), unless
the presence of activated professional APCs is mimicked by the
expression of B7.1 on islets (17). Second (the effector
stage), 7 days after infection, the first LCMV-specific autoreactive
CD4 and CD8 lymphocytes enter the pancreas and islets. In contrast to
the Zinkernagel model, diabetes occurs in our RIP-LCMV mice
(18) later (day 10–14 after infection) and is not only
dependent on perforin-dependent direct lysis of cells through
autoreactive cytotoxic T lymphocytes (19), but is also
dependent on IFN-
producing autoreactive CD8 lymphocytes (15, 18, 20). Likely, cells are destroyed by perforin as well as
by inflammatory cytokines involving IFN- and TNF-. This killing
of cells by inflammatory cytokines requires prolonged insulitis
characterized by local production of inflammatory cytokines, a process
that we view as the distinct third stage in our RIP-LCMV model. From
earlier studies, it is known that constitutive expression of TNF-
enhances diabetes in RIP-LCMV-gp mice (21) and that lack
of the TNFR hampers disease development. However, no clear knowledge
exists about which role TNF- plays during each of the three phases
(initiation, infiltration of autoreactive lymphocytes, and inflammation
and cell destruction) leading to RIP-LCMV diabetes. Based on the
earlier studies mentioned previously, we expected to find differential
effects, which formed the working hypothesis for our present
report.
Using double-transgenic RIP-LCMV-gp x Tet-TNF- mice, we were
able to precisely define the period during which TNF- production was
switched on specifically in the islets. This transgenically mediated
TNF- expression was superimposed on the endogenously occurring
TNF- secretion during LCMV infection of RIP-LCMV-gp mice that peaks
between days 6 to 8 after infection in the pancreas (15).
We found that when transgenically regulated TNF- expression
coincides with the peak of inflammatory TNF- production that occurs
relatively early during pathogenesis, disease is enhanced, similar to
previously reported findings in RIP-LCMV (21) as well as
NOD (8) mice. In contrast, late induction of transgenic
TNF- abrogated diabetes and even led to disease reversion
in up to 50% of already overtly diabetic RIP-LCMV mice. This
was associated with a sudden and marked decrease in autoreactive CD8
lymphocytes. Thus, autoaggressive CD8 lymphocytes benefit from the
presence of TNF- early during pathogenesis (8) but are
adversely affected by TNF- at the latest stages of the prediabetic
phase, in which TNF- might be of therapeutic value.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Generation of the H-2b
RIP-LCMV-gp-transgenic mice used for this study has been described
previously (14, 18). Tet-TNF- mice (8)
expressing TNF- under the control of a Tet-regulated gene
transcription system (22) were kindly provided by R.
A. Flavell (Yale University School of Medicine (New Haven, CT).
Briefly, the expression of the murine TNF- gene is controlled
by the Tet operator sequences (TetO-TNF-). In addition, the gene
encoding the Tet-responsive transactivator (tTA) was placed under the
control of the RIP II (RIP-tTA) to ensure cell-specific TNF-
expression. Thus, expression of transgenic TNF- is either blocked or
induced in the presence or absence of the Tet derivative doxycycline
(Dox), respectively (8). Three different preparations of
Dox were used: 1) Dox hyclate (300 µg/ml in drinking water), obtained
from Mutual Pharmaceuticals (Philadelphia, PA); 2) Dox-HCl (300 µg/ml
in drinking water), obtained from Sigma (St. Louis, MO); and 3) Dox
Diet (200 µg/g in food pellets), obtained from Bioserv (Frenchtown,
NJ). We initially tested the reliability of different Dox preparations
in blocking TNF- expression in RIP-gp-TNF- mice by determining
pancreatic TNF- expression by RNase protection analysis
(RPA). The amount of protected TNF- mRNA was normalized
against protected L32 mRNA, which was used as an internal
standard. Mice that exhibited a >2-fold increase in TNF-
mRNA compared with the mean concentrations detected in uninfected
RIP-gp mice were scored as "TNF expressers." Dox preparations using
Dox hyclate added to the drinking water resulted in a considerable
amount of "leakiness," which was reflected in a high (50%)
frequency of TNF expressers in the presence of Dox. In contrast, Dox
preparations using Dox-HCl in drinking water or Dox food pellets
resulted in a frequency of TNF expressers that was as low as for
uninfected single-transgenic RIP-gp mice. As a consequence, Dox food
pellets were used for all our following experiments.
H-2b (C57BL/6) Tet-TNF- mice were intercrossed
with the H-2b (C57BL/6) RIP-LCMV-gp line,
resulting in double-transgenic Tet-TNF--RIP-LCMV-gp mice on the
H-2b (C57BL/6) background, which were designated
as RIP-gp-TNF- mice. RIP-gp-TNF- mice were constantly bred in
presence of Dox food pellets.
Mice genotyping
The presence of transgenic RIP-gp, TetO-TNF-, and RIP-tTA
sequences was determined by performing three independent standard PCRs
with genomic DNA obtained from mouse tails. The primer pairs used were
as follows: RIP-gp sense, 5'-TGG ACA GGC TCA GAT GGC AAG-3' and RIP-gp
antisense, 5'-CTC AAA GCA GCC TTG TTG TAG TC-3'; TetO-TNF-
sense, 5'-TGA CCT CCA TAG AAG ACA CC-3' and TetO-TNF- antisense,
5'-TGT GAG GGT CTG GGC CAT AGA ACT ACT GAT-3'; RIP-tTA sense, 5'-ATT
TGA GGG ACG CTG TGG GCT CTT-3' and RIP-tTA antisense, 5'-ACT TCA ATG
GCT AAG GCG TC-3'. In the Tet-TNF--transgenic line we used, both
components of the tTA system, the RIP-tTA and TetO-TNF- transgenes,
always cosegregated to the offspring.
Virus
Virus used was LCMV strain Armstrong (Arm) clone 53b (20). LCMV was plaque purified three times on Vero cells and stocks prepared by a single passage on BHK-21 cells. Mice were infected with a single i.p. dose of 105 PFU LCMV-Arm unless indicated otherwise.
Viral titers
LCMV viral titers of organ homogenates were determined by infection of Vero cells as described elsewhere (18). Briefly, organs’ homogenates were diluted serially and cultured with Vero cells for 5 days. Viral titers were determined from the number of counted plaques.
Blood glucose (BG) measurements
Blood samples were obtained from the retro-orbital plexus, and plasma glucose concentration was determined using Accu-Check III (Roche, Indianapolis, IN). Mice with blood glucose values >300 mg/dl were considered diabetic (17). Mice that were scored diabetic after 2–5 wk but returned to nondiabetic BG concentrations of <200 mg/dl were considered as "revertants."
Islet cell enrichment
Islets were isolated as previously described (15). Briefly, the pancreas was removed, cut into little pieces, and digested with collagenase P (Roche). Islets were purified on HISTOPAQUE-1077 density gradients (Sigma). The obtained islet-enriched fraction and the remaining portion of the pancreas containing mainly acinar cells were immediately homogenized in Tri-Reagent (Molecular Research Center, Cincinnati, OH) for subsequent isolation of total RNA.
The RPA
Total RNA was isolated either from whole pancreas homogenates or
from islet- or acinar cell-enriched fractions using Tri-Reagent. RNA
was extracted with chloroform followed by isopropanol precipitation and
washing with ethanol. A total of 20 µg of total pancreatic RNA was
used for hybridization with a [32P]UTP-labeled
multitemplate set containing specific probes for TNF-,
IFN-, and lymphotoxin- (LT) provided by
a commercial kit (Riboquant, mCK-3b; BD PharMingen, La Jolla, CA). The
RPA was conducted according to the manufacturer’s guidelines. The
resulting analytical acrylamide gel was scanned using a Storm 860
PhosphorImager (Molecular Dynamics, Sunnyvale, CA), and the intensity
of bands corresponding to protected mRNAs was quantified using the
ImageQuant image analysis software (Molecular Dynamics)
using L32 as a reference gene.
Immunohistochemistry
Organs were harvested at wk 6 after LCMV infection unless indicated otherwise, immersed in Tissue-Tek OCT (Bayer, Elkhart, IN), and quick-frozen on dry ice. Using cryomicrotome and sialin-coated Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA), 6- to 10-µm tissue sections were cut. Sections were then fixed with 90% ethanol at -20°C, and, after washing in PBS, an avidin-biotin blocking step was included (Vector Laboratories, Burlingame, CA). Primary and biotinylated secondary Abs (Vector Laboratories) were reacted with the sections for 30 min each, and color reaction was obtained by sequential incubation with avidin-peroxidase conjugate (Vector Laboratories) and diaminobenzidine-hydrogen peroxide. Primary Abs used were rat anti-mouse CD8a (Ly2), rat anti-mouse CD8b (Ly3), rat anti-mouse ICAM-1 (CD54) (BD PharMingen), and rat anti-mouse MHC class I (Bachem, King of Prussia, PA).
Staining for apoptosis
Quick-frozen 6- to 10-µm tissue sections (pancreas and pancreatic draining lymph node (PDLN)) were probed for the presence of apoptotic cells with a basic TUNEL assay using the ApopTag peroxidase in situ apoptosis detection kit (Intergen, Purchase, NY) according to the manufacturer’s guidelines. Briefly, the TUNEL assay detects DNA stand breaks that are characteristic for DNA fragmentation as occurs during apoptosis. Tissue sections were fixed with 1% paraformaldehyde and ethanol-acetic acid (2:1), washed, and incubated with TdT to label free 3'OH DNA termini with digoxigenin-dNTP. Incorporated digoxigenin-dNTP was detected by sequential incubation with peroxidase-conjugated anti-digoxigenin mAb and diaminobenzidine-hydrogen peroxide.
Cytotoxicity assays
LCMV-specific CTL activity in the spleen was analyzed in a 5-h in vitro 51Cr release assay (23, 24). All samples were run in triplicates. Primary CTL activity was tested by harvesting spleens at day 7 after i.p. infection with 105 PFU LCMV-Arm. Splenocytes were coincubated with MHC-matched (MC57 H-2b) and mismatched (BALB/c17 H-2d) target cells that have been loaded with 51Cr. Target cells were either LCMV infected, uninfected but coated with the immunodominant LCMV MHC class I peptide GP33 (KAVYNFATC), or uninfected and uncoated. For determination of secondary CTL activity, spleens were harvested 6 wk after LCMV infection, and splenocytes were cultured for 8 days on LCMV-infected irradiated peritoneal exudate cells before testing CTL activity in a 51Cr release assay.
Flow Cytometry
Spleen and PDLNs were harvested at 6 wk after LCMV infection.
Single-cell suspensions were stimulated with 100 U/ml IL-2 and 2
µg/ml brefeldin A (Sigma) for 5–16 h at 37°C. Cells were stained
for cell surface markers using mAbs against CD8 and CD4, permeabilized
and fixed with paraformaldehyde-saponin, and stained for intracellular
cytokines using FITC-conjugated anti-mouse TNF- mAb and
PE-conjugated anti-mouse IFN- mAb (BD PharMingen). Staining for
LCMV (gp33)-specific CD8 lymphocytes was performed with PE-conjugated
H-2Db-gp33 tetramers (Tetramer Core Facility,
Emory University, Atlanta, GA). Cells were acquired and analyzed on a
FACSort or FACSCalibur flow cytometer (BD Biosciences, San Jose, CA)
using CellQuest software (BD Biosciences).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
in double-transgenic RIP-gp-TNF-
mice is blocked in the presence of Dox and is specifically expressed in
islets after removal of Dox
Generation of Tet-TNF--transgenic mice expressing inducible
TNF- specifically in the islets of Langerhans under the control of a
Tet-dependent gene transcription system (tTA-system) has been reported
previously (8). We crossed these Tet-TNF- mice with
syngeneic RIP-LCMV-gp-transgenic H-2b (C57BL/6)
mice. The resulting RIP-gp-TNF- mice were genotyped for the presence
of transgenic LCMV-gp, as well as the TetO-TNF- and RIP-tTA
sequences using a standard PCR screening protocol.
All breeding of RIP-gp-TNF- mice was conducted in the presence of
Dox to repress expression of transgenic TNF-. We initially tested
the reliability of different Dox preparations in blocking TNF-
expression in RIP-gp-TNF- mice by determining pancreatic TNF-
expression by RPA (see Materials and Methods for details).
Mice that exhibited a >2-fold increase in TNF- mRNA
compared with the mean concentrations detected in uninfected RIP-gp
mice were scored as TNF expressers. RIP-gp-TNF- mice expressed
elevated levels of TNF- in presence of Dox at a frequency
of 
15%. The elevated TNF- mRNA in these spontaneous
TNF expressers probably results from other sources than the Tet-TNF-
transgene, because it is similar to the frequency found in
non-Tet-TNF--transgenic RIP-gp mice (Fig. 1
).
|
expression was analyzed by
RPA. The frequency of TNF expressers increased from 15% (Dox on) to
>50% after 4 days. Fourteen days after Dox removal, a maximum
frequency of >80% TNF expressers was reached that did not further
increase when Dox was removed for a longer time period (Fig. 1
). The
mean pancreatic TNF- mRNA expression level was increased
by a factor of 4 after 14 days of Dox withdrawal (data not shown). It
is important to note that even after 2 mo of Dox-free diet, not all of
the RIP-gp-TNF--transgenic mice were exhibiting elevated pancreatic
TNF- levels (Fig. 1). These mice were genotyped again,
and all were transgenic for the Tet-TNF- system. Thus, there appears
to be a rate of 15% nonresponders that are incapable of
up-regulating TNF-.
The specificity of TNF- induction for the islets of Langerhans was
tested by RPA on purified islets using a multiprobe template set
including gene probes for several other cytokines. TNF-,
IFN-, and LT mRNA expression was determined
in uninfected RIP-gp-TNF- mice that were bred in the presence or
absence (2 wk) of Dox. Total RNA was isolated either from islet cell-
or acinar cell-enriched fractions that were obtained by limited
collagenase P digestion of pancreas pools of three mice per group.
Cytokine mRNA was detected specifically in islet-enriched fractions
(Fig. 2A). Quantification of
the obtained RPA data revealed that only TNF- mRNA, but
not other cytokines, was up-regulated in this islet cell-enriched
fraction (Fig. 2B). In contrast, IFN- and
LT mRNA concentrations even decreased after Dox removal
(Fig. 2B). In addition, islet cell enrichment resulted in a
5- to 6-fold higher TNF- mRNA concentration than found in
the crude pancreas homogenate of TNF- expressers after Dox removal
(data not shown). These data clearly demonstrate the specific
expression of TNF- in the islets of Langerhans of
RIP-gp-TNF--transgenic mice after Dox removal and confirm previous
results from Tet-TNF--transgenic mice (8).
|
late but not early during pathogenesis
decreases the incidence of type 1 diabetes
RIP-gp-TNF- double-transgenic mice were infected with
105 PFU LCMV-Arm, and TNF- expression was
induced by removal of Dox at several times after infection. BG was
determined in weekly intervals for at least 12 wk, and values >300
mg/dl were considered diabetic. Mice that scored diabetic after 2–5 wk
but returned to nondiabetic BG levels of <200 mg/dl were considered to
be revertants. Control mice that were kept on a Dox diet had an 70%
incidence of diabetes 14 days after LCMV infection, comparable to
findings reported by us previously using single-transgenic RIP-LCMV-gp
mice and indicating that the presence of Dox did not profoundly
influence diabetes development in RIP-LCMV-gp mice (Fig. 3A). Interestingly, expression
of TNF- had a differential effect on diabetes incidence depending on
the time of its induction. When Dox was removed at the time of LCMV
infection (day 0), the incidence increased to 90%. In contrast,
when Dox was removed late (day 10 after infection) or very late (day 14
after infection), the diabetes incidence decreased to 30% (Fig. 3A). In parallel to this decrease in overall diabetes
incidence, the frequency of revertant mice increased from 0% (day 0)
up to 45 or 35% (days 10 and 14, respectively) (Fig. 3A).
These data show that expression of TNF- late during pathogenesis,
while islet infiltration and destruction is already ongoing, can have a
beneficial effect in reverting diabetic to nondiabetic mice.
Importantly, the revertants remained nondiabetic for the rest of their
lives (observation time of >6 mo), and the late pulse of TNF- can
thus be considered therapeutic.
|
starting early after LCMV infection
abrogates autoimmune diabetes
Virally induced inflammatory TNF- expression in the pancreas of
RIP-gp mice is highest at day 7 after LCMV infection and returns back
to preinfection levels 21 days after infection (15).
Therefore, when expressed early (i.e., at the same time as the LCMV
infection), transgenic TNF- is superimposed onto inflammatory
TNF- expression in the pancreas of RIP-gp-TNF- mice. Based on the
enhancement of diabetes seen when transgenic TNF- was turned on
early, we first sought to assess the pathogenetic importance of this
early TNF- expression in our RIP-gp mice. Groups of four RIP-gp mice
were infected with LCMV, and one group of animals was treated with a
rTNFR55-IgG1 fusion protein known to neutralize TNF-. TNFR55-IgG1
(kindly provided by Dr. W. Lesslauer, Hoffmann-LaRoche, Basel,
Switzerland) was constructed using human TNF- sequences
(25, 26) but was found to effectively block mouse TNF-
activity (27). As expected, diabetes occurred between days
10 and 28 after infection in all untreated control RIP-gp mice (Fig. 3B). In contrast, mice that received three i.v. injections
of 100 µg TNFR55-IgG1 at days 4, 7, and 10 after LCMV infection never
developed diabetes (Fig. 3B). This experiment demonstrates
the importance of early TNF- activity for initiation of virally
induced type 1 diabetes in the RIP-gp mouse model. It also agrees with
the fact that we observed an increase in diabetes incidence when we
induced expression of transgenic TNF- in RIP-gp-TNF- mice at an
early time (i.e., day 4 after infection).
Infiltration of islets by CD8 lymphocytes is reduced but not absent
in revertant RIP-gp-TNF- mice
Revertant mice can be identified at wk 4–6 after LCMV infection
and 2–4 wk after transgenically (Dox withdrawal) induced TNF-
expression. At that time, we examined pancreas sections of either
revertant or diabetic RIP-gp-TNF- mice that received Dox-free diets
starting at day 10 after infection. Fig. 4A shows sequential pancreas
sections of representative revertant and diabetic RIP-gp-TNF- mice
stained for CD8, MHC I, and ICAM-1 to show the representative degree of
islet infiltration found in such animals. Infiltration of various types
of lymphocytes such as CD8 and CD4 lymphocytes, B cells, dendritic
cells, and macrophages was much more pronounced in islets of diabetic
RIP-gp-TNF- mice, as shown for CD8 lymphocytes in particular (Fig. 4A). However, it is important to note that some infiltrating
CD8 lymphocytes are still present in islets of revertant RIP-gp-TNF-
mice, even though most of these mice had nondiabetic BG values for 1–2
wk before the pancreas was removed (Fig. 4A). There were no
obvious differences in MHC class I expression when comparing diabetic
and revertant mice (Fig. 4A). The degree of ongoing overall
inflammation, as reflected by ICAM-1 expression, was much stronger in
diabetic animals, possibly due to a higher rate of cellular
infiltration (Fig. 4A). In this context, it may be important
to add that induction of TNF- in cells itself leads to enhanced
expression of ICAM-1 in and around islets, as displayed in a pancreas
section of an uninfected RIP-gp-TNF- mouse 2 mo after removal of Dox
food (Fig. 4B). In contrast, only minor expression of ICAM-1
could be detected in pancreas sections of an uninfected mouse that was
kept on Dox throughout the experiment (Fig. 4B).
Furthermore, LCMV infection alone can lead to enhanced presence of
ICAM-1 expression in and around the islets, likely as a direct result
of the primary anti-LCMV-driven inflammatory process (Fig. 4B).
|
double-transgenic mice clear LCMV and generate a
primary LCMV-specific immune response similar to single-transgenic
RIP-gp mice
We tested whether the presence of the additional transgene (tTA)
that is under the control of the same promoter as the viral gp gene
(RIP) affected the antiviral immune response following LCMV infection,
which could have influenced overall incidence of diabetes. Therefore, a
series of experiments was conducted. First, viral clearance of LCMV
infection was tested. RIP-gp, RIP-gp-TNF-, and wild-type C57BL/6
mice were infected with 105 PFU LCMV-Arm. Mice
were euthanized after 5 or 13 days, and the pancreata, livers, kidneys,
and spleens were removed and examined for viral presence in a plaque
assay. At day 5, viral organ titers were comparable in all groups.
After 13 days, no virus was detectable in any organs from all groups of
mice described previously for i.p. LCMV infection (data not shown).
Thus, as expected, control of LCMV infection is not affected by
expression of the two transgenes in cells.
In the second experiment, LCMV-specific anti-self CTL activities
were analyzed at day 7 after LCMV infection. Three groups of mice were
used for these studies: RIP-gp mice, RIP-gp-TNF- mice receiving a
Dox diet, and RIP-gp-TNF- mice receiving a Dox-free diet for 2 mo.
Splenocytes harvested ex vivo were directly incubated with uninfected
H-2b target cells, with cells coated with the
immunodominant LCMV CTL peptide gp33, or with LCMV-infected target
cells, and cytotoxic activity was measured in a standard
51Cr release assay. Similar to splenocytes
isolated from infected RIP-gp mice, cells from RIP-gp-TNF- mice
specifically lysed LCMV-infected and gp33-coated but not uninfected
C57BL/6 H-2b fibroblasts (Fig. 5). In contrast, no significant killing
was observed when LCMV-infected or gp33 peptide-coated BALB/c
(H-2d) fibroblasts were used as target cells
(data not shown). In addition, removal of Dox 2 mo before LCMV
infection had no influence CTL activity at 7 days after LCMV infection
(Fig. 5). Thus, the presence or absence of Dox as well as
islet-specific TNF- early after LCMV infection did not influence the
magnitude of the anti-LCMV response.
|
mice
We defined the cause for the reduction of diabetes in mice with
late-induced TNF- expression (Dox removal at day 10). Spleens of
such mice were harvested at wk 6 after infection to assess
LCMV-specific CTL activity. They were cultured for 8 days on
LCMV-infected irradiated peritoneal exudate cells before testing CTL
activity in a 51Cr release assay. MHC class
I-matched (H-2b) and mismatched
(H-2d) fibroblasts that were either infected with
LCMV or coated with the immunodominant gp33 peptide were used as target
cells. No significant lysis of MHC class I-mismatched target cells
could be detected (data not shown). Splenocytes from diabetic
double-transgenic RIP-gp-TNF- mice and single-transgenic RIP-gp mice
exhibited similar CTL activity against LCMV-infected or gp33-coated
target cells. In contrast, killing of both LCMV-infected and
gp33-coated target cells was significantly lower when splenocytes
obtained from revertant RIP-gp-TNF- mice were tested (Fig. 6A). Thus, autoreactive CTL
activity is reduced after induction of TNF- and correlates with the
reduction of insulitis (Fig. 4) and prevention of autoimmune
disease.
|
mice were infected with LCMV and received a Dox-free
diet at day 10 after infection. Similar to the previous experiment,
mice were scored into diabetic and revertant gourps at wk 6 after
infection. Spleens and PDLNs were probed for LCMV-specific production
of IFN- and TNF- by flow cytometry after stimulation in vitro for
5 h with LCMV-gp33 peptide. Control staining of cells obtained
from spleens or PDLNs of LCMV-infected RIP-gp mice revealed that CD8
lymphocytes harvested at day 7 after LCMV infection produce
predominantly IFN-. In contrast, experienced CD8 lymphocytes
harvested at day 60 after infection produced IFN- as well as TNF-
(data not shown). Thus, the parallel production of both IFN- and
TNF- can be used as an indication for the presence of more
experienced CD8 lymphocytes, whereas a predominant IFN- production
characterizes inexperienced CD8 lymphocytes found directly after their
initial phase of activation (28, 29, 30). In splenocytes of
either revertant or diabetic RIP-gp-TNF- mice, we found CD8
lymphocytes of the more experienced phenotype producing both IFN-
and TNF- (Fig. 6C), and the mean ratio of experienced
(IFN-- and TNF--producing) to inexperienced activated
(IFN--producing) CD8 lymphocytes in all analyzed mice was not
significantly different (Fig. 6D). However, in PDLN cell
suspensions, we detected a clear difference between revertant and
diabetic RIP-gp-TNF- mice, and CD8 lymphocytes from revertant mice
exhibited a more inexperienced phenotype reflecting their recent
activation, whereas CD8 lymphocytes from diabetic mice were of the more
experienced phenotype (Fig. 6C). On average, the ratio of
experienced to inexperienced CD8 lymphocytes in all analyzed mice was
>5-fold lower in revertant mice (Fig. 6D). In addition, CD8
lymphocytes from diabetic RIP-gp-TNF- mice had a higher density of
surface CD8 molecules compared with revertant RIP-gp-TNF- mice, as
manifested by an 1.5-fold higher mean fluorescence intensity (data
not shown). Up-regulation of surface CD8 expression was previously
demonstrated to be characteristic for experienced CD8 lymphocytes by
others (30).
Furthermore, we analyzed the frequency of LCMV-specific CD8 cells in
revertant vs diabetic mice using H-2Db-gp33
tetramers. Cells were harvested from the spleens and the PDLNs of five
revertant or diabetic mice at wk 6 after infection and were stained for
CD8 and H-2Db-gp33 reactive TCR. Revertant mice
had a reduced frequency of gp33-specific CD8 lymphocytes (Fig. 6E) in the PDLN. In contrast, no significant difference
could be detected in the spleen (Fig. 6E). The mean
frequency of gp33-specific CD8 lymphocytes in the spleen was 4–5%
in both diabetic and revertant mice (Fig. 6F). However,
revertant mice had a mean gp33-specific CD8 lymphocyte frequency of
only 2.5% in the PDLN, a value that is considerably lower than the
9% found in diabetic mice (Fig. 6F).
Possible explanations for these findings include lack of generation,
migration away from the PDLN, and deletion of experienced LCMV-specific
CD8 lymphocytes in revertant, but not in diabetic, RIP-gp-TNF-
mice.
Apoptosis of self (gp)-reactive CD8 lymphocytes is the probable
mechanism for abrogation of diabetes after induction of TNF-
Apoptosis of self (gp)-reactive CD8 lymphocytes could be an
explanation for the abrogation of diabetes in RIP-gp-TNF- mice,
because TNF- was demonstrated to directly induce apoptosis in CD8
lymphocytes by binding to either of its receptors, TNFR55 (TNFR1,
CD120a) (31) or TNFR75 (TNFR2, CD120b) (32).
TNFR at the cell surface is the starting point of a pathway that may
ultimately lead to apoptosis via activation of the caspase cascade
(33). Because we observed a systemic reduction in
autoreactive CTL combined with local lack of experienced CTL in the
PDLN, we determined the number of apoptotic cells in the pancreata and
PDLNs of revertant and protected mice. Indeed, TUNEL analysis revealed
an increased number of apoptotic cells in pancreata of revertant
RIP-gp-TNF- mice that were protected from diabetes compared with
diabetic controls (Fig. 7A).
Islets of histological sections of three mice per group were examined
for the number or apoptotic cells per islet. The overall frequency of
islets containing apoptotic cells, as well as the average number of
apoptotic cells per islet, was significantly increased in revertant
compared with diabetic RIP-gp-TNF- mice (Fig. 7B).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
on the ongoing diabetogenic
autoimmune process have been demonstrated in the same in vivo model.
When expressed late, TNF- can abrogate diabetes, whereas early
TNF- production is required for and seems to enhance disease
development. The use of a transgenic mouse model in which we could
experimentally control both the initiation of the autoimmune process
(LCMV infection) and the time of expression of an important
inflammatory factor (Tet-TNF- system) allowed us to precisely define
the late prediabetic period during which TNF- is capable of
abrogating the ongoing autoimmune process. The paradigm that too much
of an inflammatory cytokine can collapse an autoimmune reaction at a
certain point is going to be important. Indeed, similar effects are
suspected for other cytokines (i.e., IFN-). IFNs usually enhance
diabetes (34) and break tolerance to self Ags
(35). However, IFN--mediated intraductal islet
regeneration can also occur and counterbalance this detrimental effect
by dampening diabetes development (36). Further, IFN-
can lessen the effect of viral infections (37) that are
otherwise capable of exacerbating disease (38). It would
reach too far to attribute a therapeutic effect to these cytokines at
this point; however, their blockade is time sensitive, and precise
knowledge of the distinct phases of an individual autoimmune process
might be required.
In the RIP-gp mouse model, for fast-onset type 1 diabetes, the
pathogenesis is precipitated predominantly by the action of self
(gp)-reactive CD8 lymphocytes with lytic as well as cytokine-secreting
activity (15, 18). Possible mechanisms for how
islet-specific TNF- expression might reverse the ongoing
autoaggressive action of CD8 lymphocytes include 1) induction of
regulatory cell populations, 2) migration of CD8 lymphocytes away from
the islets, and 3) induction of apoptosis resulting in depletion of
autoaggressive T lymphocytes. We found that the overall infiltration of
islets by CD8 (and CD4) lymphocytes is reduced in revertant
RIP-gp-TNF- mice in which TNF- was switched on late during
pathogenesis compared with nonprotected diabetic animals. At wk 6 after
infection, the number of experienced LCMV-specific CD8 lymphocytes able
to express both IFN- and TNF- was much lower in revertant
RIP-gp-TNF- mice (Fig. 6, C and D), suggesting
a disappearance of those cells since their generation at 1–2 wk after
infection (Fig. 6, E and F). This is further
supported by the finding that revertant RIP-gp-TNF- mice had
decreased LCMV-specific CTL activity (Fig. 6, A and
B). It is striking that numbers of autoaggressive
lymphocytes were only reduced significantly in PDLNs (Fig. 6F) and not spleens but that their systemic lytic activities
(Fig. 6B) were found to be reduced. A similar segregation
between lytic effector function and numbers of CTL present has been
observed previously in persistently LCMV-infected mice
(39) and is an interesting phenomenon validating
further exploration.
No evidence for production of regulatory cytokines was detected in RPAs
comparing protected with diabetic mice (identical levels of
TGF-, IL-4, and IL-10; data not
shown). Furthermore, entry of lymphocytes into the pancreas was not
inhibited because insulitis occurred and ICAM-1 expression was enhanced
by TNF-, which should have facilitated lymphocyte entry. Thus, the
data obtained from our RIP-gp-TNF- model can be used to suggest the
following hypothesis of protective as well as detrimental TNF-
effects in autoimmune diabetes. By infection with LCMV, the autoimmune
process (diabetes) is initiated as reflected in early (day 2)
activation of APCs and up-regulation of MHC class I in islets
(16). Next, self (gp)-reactive CD8 lymphocytes remain in
the pancreas after the elimination of the virus itself and start
destroying (day 7) the insulin-producing cells (expressing gp at
the cell surface). This stage coincides with the peak of endogenous
TNF- expression, and, if transgenic TNF- is induced in addition,
disease kinetics are enhanced. During the following 2 wk (8–21 days
after infection), islet infiltration slowly increases, and cells
die as a consequence of lytic CTL killing as well as cytokine-mediated
effects (15). If transgenic TNF- is expressed at this
late time during pathogenesis, the autoimmune process collapses, and
60% of the mice are either protected or revert to normoglycemic BG
values. Judging from our data, it is most likely that this increase in
TNF- concentration locally in the islets results in the deletion of
experienced, activated self (gp)-reactive CD8 lymphocytes (Fig. 6, C and D) by induction of apoptosis, leaving
behind a reduced population of self (gp)-reactive CD8 lymphocytes (Fig. 6, E and F) that are inexperienced and have a
lower amount of lytic CTL activity (Fig. 6, A and
B). Indeed, the frequency of apoptotic cells was higher in
islets from revertant mice than in diabetic mice. Possible mechanisms,
such as TNF--induced apoptosis, that cause abrogation of disease are
likely to precede the actual read out for revertant mice as manifested
in nondiabetic BG values. Therefore, the nature of the experiments
conducted might have made it impossible for us to analyze the actual
peak of the mechanisms involved.
In conclusion, autoimmune processes as involved in type 1 diabetes, and
other diseases are at least initially driven by an inflammatory event.
Mediators that are traditionally termed proinflammatory, such as
TNF-, often play a major role during this initial phase of
inflammation and promote rather than suppress the ongoing destruction
(2, 4). Here, the precise duration of TNF- expression
is an important factor determining the progress of an ongoing
autoimmune process to diabetes (8). However, later during
pathogenesis, too much of such inflammatory factors is not beneficial
to the propagation of the autoimmune process. This could occur in vivo,
for example, by a second infection by the same or another unrelated
virus capable of inducing inflammatory factors. Indeed, we have
recently observed such a scenario in our RIP-LCMV model, in which
secondary viral infection can abrogate ongoing autoimmunity by
enhancing apoptosis of autoreactive lymphocytes (M. G. von Herrath and
U. Christen, unpublished observations). Furthermore, TNF- was
demonstrated in various other occasions to confer immunoregulatory
activities by down-regulation of type 1 cytokines (40),
suppression of T cell proliferation and cytokine production
(5), or preventing the development of self-reactive T
lymphocytes (3). Thus, one has to postulate that
inflammatory cytokines exist in a fine-tuned time-sensitive
equilibrium, because we showed here that TNF- has opposite effects
in the same animal depending on the time of expression. Future studies
on the role of cytokines in autoimmunity and other diseases should
attempt to include the element of timing as a central factor that
determines effector function and outcome.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Matthias G. von Herrath, Department of Neuropharmacology and Immunology, IMM6, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037. E-mail address: matthias{at}scripps.edu
3 Abbreviations used in this paper: RIP, rat insulin promotor; Tet, tetracycline; NOD, nonobese diabetic; LCMV, lymphocytic choriomeningitis virus; TetO, Tet operator; tTA, Tet-responsive transactivator; Dox, doxycycline; RPA, RNase protection analysis; RIP-tTA, tTA under the control of RIP II; Arm, Armstrong strain; BG, blood glucose; PDLN, pancreatic draining lymph node; LT, lymphotoxin-.
Received for publication January 29, 2001. Accepted for publication April 2, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
cell death in the nonobese diabetic mouse is Fas independent. J. Immunol. 163:1562. on insulin-dependent diabetes mellitus in NOD mice. I. The early development of autoimmunity and the diabetogenic process. J. Exp. Med. 180:995. in islets prevents autoimmune diabetes in nonobese diabetic (NOD) mice by preventing the development of auto-reactive islet-specific T cells. J. Exp. Med. 184:1963. in neonatal NOD mice promotes diabetes by enhancing presentation of islet antigens. Immunity 9:733.[Medline]
production in pancreatic islets leads to insulitis, not diabetes: distinct patterns of inflammation in TNF- and TNF- transgenic mice. J. Immunol. 150:4136.[Abstract]
leads to autoimmunity in transgenic mice. Proc. Natl. Acad. Sci. USA 91:5138. on pancreatic islet cells is dependent on the duration of the TNF- signal. Immunity 12:459.[Medline]
promotes diabetes in nonobese diabetic mice by CD154-independent antigen presentation to CD8+ T cells. J. Exp. Med. 191:225. and the progression of diabetes in non-obese diabetic mice. Immunol. Rev. 169:11.[Medline]
and interleukin 1. Proc. Natl. Acad. Sci. USA 87:968.-cells die through inflammatory cytokines and not perforin from autoreactive (anti-viral) CTL. Diabetes. 49:1801.[Abstract]
-cells by autoreactive lymphocytes in a transgenic model of virus-induced IDDM. J. Autoimmun. 10:231.[Medline]
-cells can break peripheral ignorance and lead to spontaneous autoimmune diabetes. Immunity 3:727.[Medline]
is essential for destruction of -cells and development of insulin-dependent diabetes mellitus. J. Exp. Med. 185:531.. J. Immunol. 150:5185.[Abstract]
receptor IgG, design principles and activities. S. Chamow, and A. Ashkenazi, eds. Antibody Fusion Proteins 251. Wiley, New York.
. Nature 346:844.
)
or were depleted from Dox diet for 2 mo (
) by analyzing the
cytolytic activity of splenocytes. The obtained results were compared
with the cytolytic activities of splenocytes harvested from
single-transgenic RIP-gp mice in parallel (
). As target
H-2b fibroblasts were used that were either LCMV infected
(A) or coated with the immunodominant LCMV peptide gp33
(B). As a control, uninfected and uncoated
H-2b fibroblasts were used (C).
) and
revertant (
) mice at an E:T ratio of 50:1 of all mice
(n = 4–6) tested in A.
C, Cells obtained from the spleen or the PDLN were
stimulated with gp33 and IL-2 for 16 h, and IFN-