| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1-Deficient Mice Develop Necroinflammatory IFN-
-Dependent Hepatitis1
Department of Pathology, Dartmouth Medical School, Lebanon, NH 03756
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
1 were
extensively back-bred to the BALB/c background. The BALB/c background
dramatically modified the phenotype of TGF-1-/- mice:
specifically, BALB/c-TGF-1-/- mice developed a lethal
necroinflammatory hepatitis that was not observed in
TGF-1-/- mice on a different genetic background.
BALB/c background TGF-1-/- livers contained large
numbers of activated CD4+ T cells that produced large
quantities of IFN-
, but little IL-4, identifying them as Th1 cells.
BALB/c background TGF-1-/-/IFN--/-
double knockout mice, generated by cross-breeding, did not develop
necroinflammatory hepatitis, demonstrating that IFN- is
mechanistically required for its pathogenesis. This represents the
first murine model of hepatitis that develops spontaneously, is
restricted by genetic background, and is dependent upon the Th1
cytokine IFN-, and that thus recapitulates these important aspects
of AIH. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-producing Th1 cells (5, 6), and the loss of inherent hepatic tolerogenic properties
(1). The inherent tolerogenic capacity of the liver can be
demonstrated in certain species, such as pig or rat, in which unrelated
orthotopic liver allografts survive without the need for
immunosuppression (7).
Advances in the understanding of AIH have lagged behind
advances in the understanding of other organ-specific autoimmune
diseases, such as multiple sclerosis or type I diabetes mellitus,
largely because of the lack of a suitable animal model system. Existing
animal models of experimental hepatitis involve a manipulation to
overcome the natural immunosuppression to liver autoreactivity. Models
using wild-type mice involve administration of substances that
induce liver-specific lesions, such as D-galactosamine, a
transcriptional inhibitor that renders hepatocytes exquisitely
sensitive to the apoptotic effects of TNF-
(8), or the
mitogenic T cell lectin Con A (9), which binds with high
affinity to sinusoidal endothelial cells, and nonspecifically activates
T cells intrahepatically (10). Transgenic models exploit
hepatocyte-specific expression of a hepatotoxic agent, such as IFN-
(11), or hepatitis B virus (HBV) (12). These
various animal models have been useful in the understanding of normal
and pathogenic immune responses in the liver. However, because they
either direct the immune response to the liver, or create a lesion
specifically in the liver, they are not necessarily models of
spontaneously occurring AIH, nor are they suitable for the analysis of
genetic control of AIH.
TGF-1 is a pleiotropic cytokine that exhibits a variety of
antiinflammatory activities and inhibits the development of autoimmune
disease in several model systems (13, 14). TGF-1 is
absolutely required for normal immune homeostasis and the prevention of
autoimmunity, since TGF-1-deficient
(TGF-1-/-) mice develop inflammatory lesions
involving several organs, most typically heart and lungs, with death at
3–5 wk of age (15, 16). Death of
TGF-1-/- mice is believed to be due to
cardiopulmonary failure (17). A proportion of
TGF-1-/- mice demonstrates mild liver
inflammation, which typically does not cause hepatocyte loss or
compromise liver function (18). Inflammatory lesions in
TGF-1-/- mice are heterogeneously
distributed, with considerable mouse to mouse variability (19, 20). However, some organs, such as brain, eye, kidney, and
testis, are generally spared (19, 20). Inflammatory
disease in TGF-1-/- mice requires the
presence of CD4+ T cells, since double-deficient
TGF-1-/-/class II
MHC-/- mice that lack
CD4+ T cells do not develop inflammation
(21). Inflammatory lesions are not initiated by or
dependent upon normal bacterial flora, since
TGF-1-/- mice raised under germfree
conditions nevertheless develop the lethal inflammatory phenotype
(22). Mice transgenically expressing a dominant-negative
TGF- receptor specifically in T cells develop inflammatory lesions
similar to those that develop in TGF-1-/-
mice (23), showing that loss of TGF- signaling in T
cells is sufficient for the induction of inflammatory disease, and
strongly suggesting an autoimmune etiology. Together, these data
indicate that a principal function of TGF-1 is to inhibit the
development of inflammatory autoimmune disease and that TGF-1
mediates this function at least in part through inhibition of T cell
responses to self Ags.
The TGF-1-/- mouse is a useful model system
to study the pathogenesis of Th cell-mediated organ-restricted
autoimmune disease. Until now, however, all published studies analyzing
the autoimmune phenotype associated with a deficiency in TGF-1 have
used TGF-1-/- mice on either outbred or
hybrid inbred/outbred genetic backgrounds that are necessarily
genetically heterogeneous. However, autoimmune diseases in humans are
strongly influenced by genetic background. The influence of genetic
background is readily apparent in polygenic murine model systems of
autoimmunity, such as the nonobese diabetic or NZB/NZW lupus models,
and in autoimmune diseases that arise in mice harboring single gene
defects, such as the IL-2-/- mouse (24, 25), or the FcRIIB-/- mouse
(26). Therefore, to determine whether genetic background
can modify the phenotype of TGF-1-/- mice,
we have extensively back-bred the defective TGF-1 allele onto
the inbred BALB/c strain of mice, allowing the production of a cohort
of genetically homogeneous TGF-1-/- mice. We
now show that genetic background dramatically modifies the phenotype of
TGF-1-/- mice. Specifically,
TGF-1-/- mice on the BALB/c background, but
not TGF-1-/- mice on another background (a
129/CF-1 hybrid), uniformly develop an aggressive necroinflammatory
hepatitis. The phenotype of this mouse recapitulates important features
of AIH, including the spontaneous development of hepatitis (i.e.,
without experimental manipulation of the mouse), a strict dependence on
genetic background, and the involvement of Th1 cells. By genetic
techniques, we demonstrate that IFN-, but not IL-4, is required for
the development of necroinflammatory liver disease in this model
system.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All breeder mice used were derived from the same founder mice
(15), and thus harbored the identical
TGF-1- allele, consisting of a neomycin
interruption of exon 6. BALB/c-TGF-1+/-
breeder mice were generated by serially backbreeding.
TGF-1+/- mice on the C57BL/6 background were
purchased from The Jackson Laboratory (Bar Harbor, ME). However,
the C57BL/6 background does not support the birth of
TGF-1-/- pups, since 100% of C57BL/6
TGF-1-/- conceptuses die in utero, owing to
a C57BL/6-specific susceptibility locus on chromosome 5
(27). Consistent with this, of 23 pups generated from
C57BL/6-TGF-1+/- intercrosses in our
facility, none genotyped as TGF-1-/-
(J. D. Gorham, unpublished observations). Therefore, we bred
C57BL/6-TGF-1+/- mice with BALB/c mice (The
Jackson Laboratory) to generate F1 (C57BL/6
x BALB/c) mice, selecting for TGF-1+/- mice
by PCR. Then,
F1-TGF-1+/- mice were
serially back-bred to BALB/c for five more successive generations, to
produce BALB/c-TGF-1+/- breeders (group A,
Table I
). It is calculated that these
mice contain on average 98.4% BALB/c genomic material and 
1.6%
non-BALB/c genomic material. At the backcross 1 generation of breeding,
TGF-1+/- breeder mice were screened using
polymorphic chromosome 5 microsatellite markers that flank the C57BL/6
in utero susceptibility locus, to ensure elimination of C57BL/6 genomic
material for this locus early in our breeding scheme (J. D.
Gorham, unpublished observations). In addition, flow cytometry analysis
of PBMC from BALB/c-TGF-1+/- mice at the
backcross 2 generation indicated that all were of the BALB/c haplotype
(d/d) at the H-2 locus (J. D. Gorham, unpublished observations).
The 129/CF-1-TGF-1+/- breeder mice (group B,
Table I) were obtained from Tom Doetschman (University of Cincinnati,
Cincinnati, OH). These mice have been maintained as advanced intercross
lines (28), to maintain a heterogeneous genetic
background. The genetic background of these mice is a hybrid of
129/SvPas and CF-1 (Charles River Breeding Laboratories, Wilmington,
MA) genomic material (29). BALB/c background
IFN--/- mice (30) were
purchased from The Jackson Laboratory (sixth backcross generation,
according to the supplier; calculated BALB/c genomic material =
99.2%). BALB/c (BC5)-TGF-1+/- mice were
mated with BALB/c (BC6)-IFN--/- mice to
generate BALB/c background compound heterozygotes
(TGF-1+/-/IFN-+/-)
mice. These mice were crossed again with
BALB/c-IFN--/- mice, and pups selected for
heterozygosity at the TGF-1 locus, and homozygosity for the null
allele at IFN-, by PCR. These
BALB/c-TGF-1+/-/IFN--/-
mice were then used as breeders for the generation of
TGF-1-/-/IFN--/-
double knockout mice (group C, Table I). BALB/c background
IL-4-/- mice were purchased from The Jackson
Laboratory. These mice are derived from an interruption of the
IL-4 gene made directly in BALB/c embryonic stem cells
(31); therefore, other than the interruption of
IL-4, these mice are isogenic to BALB/c (i.e., 100%
BALB/c). BALB/c (BC5)-TGF-1+/- mice were
mated with these BALB/c-IL-4-/- mice to produce
BALB/c-TGF-1+/-/IL-4-/-
breeders (group D, Table I), using a similar strategy as for the
BALB/c-TGF-1+/-/IFN--/-
breeders.
|
1-/- mice
Cages were monitored every day for the birth of new litters,
which was defined as day 0. At day 5, pups were screened by PCR from
tail-snip DNA for TGF-1 genotype. Typically, genotypes were
identified within 24 h. Then, litters with
TGF-1-/- mice were monitored thrice weekly,
consisting of weighing and observing all pups in the litter. For
generating the survival curve, the day of death of a
TGF-1-/- mouse was defined as the
midpoint between the mouse’s last observed day alive and the day the
mouse was found dead. After survival curves were established (Figs. 1 and 5), other assays utilized additional
TGF-1-/- mice that were euthanized after
deep anesthesia at the ages indicated in the text or figure legends.
Comparisons of survival curves used the log rank test. All mice were
bred at Dartmouth Medical School in a specific pathogen-free
facility.
|
|
DNA was extracted from tail snips of mice according to standard
techniques. Genotype was determined by PCR using the following
oligonucleotide primers: TGF-1 (29), 5'-GAG AAG AAC TGC
TGT GTG CG-3; 5'-GTG TCC AGG CTC CAA ATA TAG G-3'; 5'-GCC GAG AAA GTA
TCC ATC AT-3'. IFN- (30), 5'-AGA AGT AAG TGG AAG GGC
CCA GAA G-3'; 5'-AGG GAA ACT GGG AGA GGA GAA ATA T-3'. IL-4
(31), 5'-GTG AGC AGA TGA CAT GGG GC-3'; 5'-CTT CAA GCA TGG
AGT TTT CCC-3'. Amplicons were electrophoresed on 1.5% to 2.5%
agarose gels, depending on the PCR. All three PCR involve primers that
are specific for, and that flank the neo interruption of, the gene of
interest. The TGF-1 PCR incorporates a third, additional primer from
the neomycin cassette, which improves the efficiency and reliability of
the reaction and does not affect the specificity (29). For
all three PCR, a faster migrating band reflects the wild-type allele,
while a slower migrating band reflects the interrupted allele. Thus,
for each gene of interest, a single PCR distinguishes the wild-type,
heterozygous, and homozygous null genotypes.
Histology
After anesthesia and euthanasia of mice, abdominal viscera were exposed and sometimes photographed. Organs were dissected out and fixed in buffered Formalin, followed by paraffin embedding, sectioning, and staining with hematoxylin and eosin, by routine methods.
Transaminase analysis
Deeply anesthetized mice were decapitated and exsanguinated into
heparinized plasma separator tubes (Becton Dickinson, Franklin Lakes,
NJ). Plasma was separated by centrifugation, and frozen at
-20°C. Thawed plasma was diluted 4-fold with saline, and alanine
aminotransferase (ALT) was determined using a Roche-Hitachi 917
Automatic Analyzer, using an UV, kinetic enzymatic assay read at 340
nm. Abnormally elevated ALT levels were defined as more than mean + 2
SD, using data from the littermate control
TGF-1+/+ mice of the matched genetic
background.
Isolation and analysis of hepatic and splenic CD4+ T cells
Spleens and livers were dissected from mice and weighed, and
nonadherent cells were isolated by mechanical disruption of the organs,
and washed. CD4+ T cells were isolated using
murine CD4-specific magnetic Dynabeads (Dynal, Great Neck, NY)
according to the manufacturer’s instructions. Flow cytometric analysis
of purified cells subsequently stained with anti-CD4 FITC routinely
indicated >98% purity. Isolated CD4+ T cells
were counted by hemacytometer, and CD4+ T
cells/mg (wet) of tissue were calculated. For cell surface phenotype,
isolated CD4+ T cells were stained with
anti-CD62L PE, anti-VLA4 FITC, or anti-CD44 PE (all
obtained from PharMingen, San Diego, CA), and analyzed by flow
cytometry. For cytokine production analysis, isolated
CD4+ T cells were plated in triplicate (or at
duplicate only for TGF-1+/+ control liver
CD4+ T cells, because of limited cell yield) at
100,000 cells/well in a 96-well culture dish precoated with immobilized
anti-CD3 mAb (10 µg/ml). In some wells, human rIL-2 (20 U/ml;
gift from William Green, Dartmouth Medical School) was added.
Supernatants were collected after 2 days of stimulation and frozen at
-80°C. IFN- and IL-4 concentrations were subsequently determined
in thawed samples by ELISA.
RT-PCR
Total liver RNA was isolated using TRIzol. One microgram of
oligo(dT)-primed total RNA was reverse transcribed with Moloney murine
leukemia virus reverse transcriptase. To control for the RNA dependence
of the PCR amplicon, control tubes without addition of reverse
transcriptase were prepared in parallel. PCR for TNF-, Fas ligand
(FasL), Fas, and copper/zinc superoxide dismutase were then performed
exactly as in the study by Ksontini et al. (32), and
amplicons were electrophoresed on 2% agarose gels.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
1-/- mice is modified
by genetic background
Although the TGF-1-/- defect is
uniformly lethal, inspection of the available published data reveals
that survival duration varies greatly between individual
TGF-1-/- mice (15, 16, 19, 21).
To specifically test the hypothesis that genetic background
influences duration of survival of
TGF-1-/- mice, survival of inbred
BALB/c-TGF-1-/- mice and of hybrid
inbred/outbred 129/CF-1-TGF-1-/- mice
(groups A and B, Table I
) was compared.
BALB/c-TGF-1-/- mice survived for a mean of
13.1 days, with a range of 11–17 days, whereas
129/CF-1-TGF-1-/- mice survived for a mean
of 20.1 days, with a range of 8.5–39.5 days (Fig. 1). The difference between these two
survival curves is highly significant (p <
0.0001; log rank analysis). These data show that genetic background of
the TGF-1-/- mouse has a significant effect
on its life span. The broader range of survival of the
129/CF-1-TGF-1-/- mice probably reflects the
greater underlying genetic heterogeneity of this group of
mice.
The TGF-1-/- defect is also associated with
a partial embryonic lethality, the penetrance of which is dependent
upon genetic background (27, 29). An effect of genetic
background on the frequency of embryonic lethality of
TGF-1-/- conceptuses was observed in this
study as well (Table I), with a greater frequency in the BALB/c
background than in the 129/CF-1 background. The mechanism(s) of
embryonic lethality is obscure, but appears to be due to developmental
defects and unrelated to immune or inflammatory responses (29, 33) and is not considered further in this study.
BALB/c-TGF-1-/- mice develop severe
necroinflammatory hepatitis
Because of the shortened postnatal survival phenotype of
TGF-1-/- mice on the BALB/c background, we
next sacrificed four 11- to 12-day-old
BALB/c-TGF-1-/- mice and littermate controls
for gross and histological examinations of organs. Livers of all four
BALB/c-TGF-1-/- mice were grossly abnormal,
showing mottled discoloration in all lobes (Fig. 2A). Histological examination
revealed extensive inflammation with widespread hepatocyte necrosis
(Fig. 2B). By contrast, livers from
129/CF-1-TGF-1-/- mice were grossly
indistinguishable from littermate control livers (Fig. 2A)
and associated histopathology was much more limited (Fig. 2B), consisting of moderate periportal inflammation, but
little or no loss of hepatocytes, consistent with previous reports of
liver histopathology in TGF-1-/- mice on the
129/CF-1 background (15, 16, 19, 20).
|
1-/- mice. Levels were abnormally
high (i.e., more than mean + 2 SD) in only one of three 11-day-old
BALB/c-TGF-1-/- mice, but in all seven
BALB/c-TGF-1-/- mice of 12 days of age or
greater (Fig. 2C). In total, 8 of 10
BALB/c-TGF-1-/- mice had elevated plasma ALT
levels. Plasma ALT levels were positively correlated with age
(r = 0.93). This age-dependent acute rise in plasma ALT
levels paralleled the survival curve for
BALB/c-TGF-1-/- mice (Fig. 1). By contrast,
for the 129/CF-1 background, only one of seven 12- to 14-day-old
TGF-1-/- mice, and neither of two 18-day-old
TGF-1-/- mice, had abnormally high plasma
ALT levels (Fig. 2C). One
129/CF-1-TGF-1-/- mouse was monitored until
it was 33 days of age, at which point it was severely moribund, with
weight loss, a hunched posture, and failure to thrive. After
euthanasia, necropsy revealed only periportal inflammation (data not
shown), and the plasma ALT was within the normal range (Fig. 2C). In total, for ten
129/CF-1-TGF-1-/- mice of various ages, only
one had elevated plasma ALT levels. Thus, on the 129/CF-1
background, most TGF-1-/- mice have no
evidence of significant liver disease, there is no progressive rise
in ALT levels with age, and death is not accompanied by significant
liver damage.
In TGF-1-/- mice from either background,
inflammatory lesions in the heart and lungs were also observed (data
not shown), consistent with previously published reports of
TGF-1-/- mice. These lesions were not
different between the two backgrounds, indicating that the modifying
effect of genetic background in these mice is specific to the
liver.
BALB/c-TGF-1-/- livers have large numbers of
activated Th1 cells
We have yet to define the cellular or molecular differences
between the two strains of mice that account for the differential
hepatic phenotype; in this study, we present additional observations on
the novel TGF-1-/- phenotype observed in the
BALB/c background. The inflammatory disease observed in heart and lung
in TGF-1-/- mice (of non-BALB/c backgrounds)
is dependent upon Th cells, since CD4+ T
cell-deficient TGF-1-/- mice do not develop
these inflammatory pathologies (21, 34). Thus, we reasoned
that CD4+ T cells would be involved in the liver
pathogenesis observed in BALB/c-TGF-1-/-
mice. To characterize the Th cell population in
BALB/c-TGF-1-/- livers, hepatic
CD4+ T cells were isolated by magnetic beads and
analyzed. CD4+ T cells were much more abundant in
BALB/c-TGF-1-/- livers than in littermate
control livers (an average of 7-fold higher; Fig. 3A). By contrast,
CD4+ T cell numbers in
BALB/c-TGF-1-/- spleens were approximately
half that in littermate control spleens (Fig. 3A). Analysis
of the effector states of these CD4+ T cell
populations by flow cytometry showed that
BALB/c-TGF-1-/- hepatic or splenic
CD4+ T cells had a predominantly effector/memory
phenotype (CD62Llow
VLA4high CD44high), whereas
littermate control hepatic or splenic CD4+ T
cells had a predominantly naive phenotype
(CD62Lhigh VLA4low
CD44low) (Fig. 3B).
|
1-/- mice were predominantly Th1
or Th2 effector cells, freshly isolated splenic and hepatic
CD4+ T cells were stimulated in vitro with
immobilized anti-CD3 for 48 h, and supernatants collected for
cytokine analysis by ELISA. In parallel wells, IL-2 was included, since
it has been reported that TGF-1-/- T cells
require IL-2 for activation in vitro (35). Upon
stimulation, freshly isolated hepatic or splenic
BALB/c-TGF-1-/- CD4+ T
cells produced very large quantities of IFN- (111,000 and 177,000
pg/ml for splenic and hepatic Th cells, respectively), which was
enhanced in the presence of IL-2 (254,000 and 487,000 pg/ml for splenic
and hepatic Th cells, respectively). By contrast, littermate control
hepatic or splenic CD4+ T cells produced no
detectable IFN- (Fig. 4A).
Compared with IFN- production from
BALB/c-TGF-1-/- CD4+ T
cells, IL-4 production from these cells was much more modest (270 and
580 pg/ml for splenic and hepatic CD4+ T cells,
respectively; with IL-2, 820 and 600 pg/ml, respectively), and similar
to the levels produced by littermate control CD4+
T cells (Fig. 4B). Thus, livers from
BALB/c-TGF-1-/- mice contain abnormally
large numbers of activated/effector Th1 cells, similar to observations
made from liver biopsies from patients with AIH (5, 6).
|
1-/-
mice is IFN- dependent
We next asked whether the necroinflammatory liver disorder
observed in BALB/c-TGF-1-/- mice was
dependent on either the Th1 cytokine IFN- or the Th2 cytokine IL-4.
To assess the requirement for IFN-, we bred
BALB/c-TGF-1+/- mice with BALB/c background
IFN--/- mice to generate mice doubly
deficient in both TGF-1 and IFN-, while maintaining the BALB/c
genetic background (group C, Table I).
BALB/c-TGF-1-/-/IFN--/-
mice survived much longer (Fig. 5;
n = 12; mean = 32.9 days; range, 23–66 days;
p < 0.0001) than did
BALB/c-TGF-1-/- mice, indicating a
requirement for IFN- in the very early demise observed in
BALB/c-TGF-1-/- mice. Notably, however,
although the absence of IFN- in
BALB/c-TGF-1-/-/IFN--/-
mice delayed death compared with
BALB/c-TGF-1-/- mice, it did not completely
normalize survival. Furthermore,
BALB/c-TGF-1-/-/IFN--/-
mice were much smaller than littermate controls, and eventually
developed observable signs of wasting (J. D. Gorham, unpublished
observations). Thus, a deficiency in IFN- does not simply revert the
BALB/c-TGF-1-/- phenotype to a wild-type
phenotype, rather IFN--independent pathways of disease exist in
BALB/c-TGF-1-/- mice. Indeed, heart and lung
inflammatory lesions were observed in
BALB/c-TGF-1-/-/IFN--/-
mice at histopathology and were similar to those observed in
BALB/c-TGF-1-/- mice (data not shown),
indicating that IFN- does not appreciably contribute to these
lesions, at least at the histopathological level.
IFN- was required, however, for the necroinflammatory hepatitis.
Livers from
BALB/c-TGF-1-/-/IFN--/-
mice were grossly normal (Fig. 6A), with no evidence of the
visible abnormalities typical of livers from
BALB/c-TGF-1-/- mice (Fig. 2A).
At histology, there was no widespread hepatocellular loss, and most
areas were indistinguishable histologically from wild-type liver (Fig. 6B, left) although modest inflammatory expansion
around portal tracts was sometimes observed (Fig. 6B,
right). Consistent with this more limited histological
picture, plasma ALT levels were normal in young (
18-day)
BALB/c-TGF-1-/-/IFN--/-
mice. Although some elevation of plasma ALT was detected in four of
five older (
21-day)
BALB/c-TGF-1-/-/IFN--/-
mice (Fig. 6D), these findings were not associated with
severe necroinflammatory histopathology. Thus, acute necroinflammatory
liver destruction in BALB/c-TGF-1-/- mice is
largely IFN- dependent. The absence of IFN-, however, reveals
additional, IFN--independent, mechanisms of liver damage that
progress with slower kinetics.
|
1-/- mice, we
generated
BALB/c-TGF--/-/IL-4-/-
mice (group D, Table I). Survival of
BALB/c-TGF-1-/-/IL-4-/-
mice (Fig. 5; n = 6; mean = 13.6 days; range,
10–19.5 days) was not statistically different
(p = 0.61) from that of
BALB/c-TGF-1-/- mice. Gross and histological
examination of livers from two
BALB/c-TGF-1-/-/IL-4-/-
mice (ages 11 and 13) revealed hepatic lesions indistinguishable from
those seen in BALB/c-TGF-1-/- mice (data not
shown). Additionally, plasma from a 13-day-old
BALB/c-TGF-1-/-/IL-4-/-
mouse had significantly elevated ALT levels, compared with littermate
controls (data not shown). Thus,
BALB/c-TGF-1-/- mice rendered deficient in
IL-4 mice die with kinetics similar to
BALB/c-TGF-1-/- mice, and still develop
necroinflammatory hepatitis.
The molecules TNF- and FasL bind to and signal through receptors
that activate apoptotic pathways, and have been shown to be necessary
for hepatitis in several model systems in vivo (32, 36, 37, 38, 39). Therefore, to determine whether these factors are
overexpressed in diseased livers, we analyzed their expression in total
liver RNA. Both FasL and TNF- mRNAs were overexpressed in
BALB/c-TGF-1-/- liver (Fig. 7A), compared with littermate
control liver. By contrast, these mRNAs were not up-regulated in
TGF-1-/-/IFN--/-
liver (Fig. 7B), indicating that IFN- is required for
their expression in TGF-1-/- liver,
consistent with the requirement for IFN- in the hepatic
necroinflammatory disease in general.
|
1-/- and
129/CF-1-TGF-1-/- mice: differential susceptibility to
necroinflammatory hepatitis is not explained by differential Th1/Th2
development
The presence in BALB/c-TGF-1-/- liver
of Th1 cells, and the dependence of the BALB/c hepatic phenotype on
IFN- suggest the hypothesis that
129/CF-1-TGF-1-/- mice do not develop
necroinflammatory hepatitis because
129/CF-1-TGF-1-/- hepatic
CD4+ T cells do not develop the Th1 phenotype. To
test this hypothesis, hepatic CD4+ T cells were
isolated from control and knockout mice of each background and
stimulated in vitro to determine Th1/Th2 phenotype. Hepatic
CD4+ T cells from knockout mice of either
background produced large quantities of IFN- and modest amounts of
IL-4 (Fig. 8), showing that
TGF-1-/- CD4+ T cells
are Th1 in both backgrounds examined. Similar results were obtained
with splenic CD4+ T cells (data not shown). Thus,
the different hepatic phenotype observed in the two genetic backgrounds
is not explained by a difference in Th1/Th2 differentiation. The data
also show that, whereas IFN- is necessary for the development of the
necroinflammatory hepatic phenotype observed in
BALB/c-TGF-1-/- mice, the presence of Th1
cells in liver is not sufficient.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Genetic background regulates the phenotype of mice deficient in
TGF-1. TGF-1-/- mice on a predominantly
BALB/c genetic background, but not TGF-1-/-
mice on a hybrid 129/CF-1 genetic background, develop degenerative
liver disease and die at about 2 wk of age. The developmental kinetics
of the extent of hepatocyte death parallels the kinetics of demise of
BALB/c-TGF-1-/- mice: starting at 11 days of
age, plasma transaminase levels dramatically rise and mice begin to
die, with all dead by 17 days of age. Previous reports of
TGF-1-/- mice described extensive
inflammatory heart and lung disease, with cardiac and pulmonary
inflammation observed in 95 to 100% of mice (19, 20). By
contrast, reports indicated liver involvement in only 70% of mice,
which, when present, is not accompanied by hepatic lesions at the
grossly observable level (18) and is typically associated
with only modest histopathological changes, consisting of moderate
periportal inflammation, but preservation of hepatic architecture
(i.e., no hepatocyte loss (19, 20)). Our findings in
129/CF-1-TGF-1-/- mice (no gross hepatic
lesions, periportal inflammation without extensive hepatocyte necrosis,
and normal plasma ALT levels in 9 of 10 mice) are completely consistent
with these previously published findings. All mice were raised in the
same room and treated identically, indicating that the early death and
liver destruction seen in BALB/c-TGF-1-/-
mice, but not 129/CF-1-TGF-1-/- mice, are
functions of the genetic background of the mouse, not of differential
environment. These observations suggest that the necroinflammatory
hepatitis results from the combination of the absence of TGF-1 and
the presence of an appropriate constellation of susceptibility alleles
in the BALB/c genome.
Thus, the BALB/c background confers a different organotropism on the
inflammatory phenotype associated with the
TGF-1-/- defect, with hepatic lesions the
salient finding in these mice. The BALB/c background does not switch
the target of the inflammatory response from the lungs and heart to the
liver; indeed, BALB/c-TGF-1-/- and
129/CF-1-TGF-1-/- mice had similar heart and
lung inflammatory histopathological lesions (data not shown)
(19). Rather, it appears that the BALB/c background
additionally superimposes the necroinflammatory hepatitis phenotype.
When mechanisms leading to necroinflammatory hepatitis are disabled (by
loss of IFN-, for example), TGF-1-/- mice
in the BALB/c background nevertheless develop cardiac and pulmonary
inflammation, with a mean survival of 33 days of age.
Genetic background has been shown to modify disease phenotype in other
single-gene knockout mice. The principal phenotype of the
IL-2-/- mouse on most genetic backgrounds is
the development of an inflammatory bowel disease similar to human
ulcerative colitis (24). On the BALB/c background,
however, IL-2-/- mice rapidly develop a severe
autoimmune hemolytic anemia, and die by 5 wk of age (25).
Mice deficient in FcRIIB on the C57BL/6 background develop a lethal
lupus-like syndrome with glomerulonephritis, whereas BALB/c background
mice with the identical genetic lesion have no observable abnormal
phenotype (26). These various mouse models demonstrate
that the phenotype associated with loss of a specific immunoregulatory
gene can be epistatically modified by the genetic background of the
mouse. Interestingly, the BALB/c background confers a more severe
autoimmune phenotype for some models
(TGF-1-/- and
IL-2-/-), but is protective in others
(FcRIIB-/-), reflecting the complexity of
regulation of tissue-specific autoimmune disease.
Pathogenesis of liver inflammation: implications for AIH
What pathways of hepatocyte damage might be operative in
BALB/c-TGF-1-/- mice? Comparison of the
BALB/c-TGF-1-/- and
BALB/c-TGF-1-/-/IFN--/-
phenotypes indicates that the dominant pathway is IFN- dependent,
has rapid kinetics, and appears histologically as pervasive hepatocyte
death (i.e., panlobular necrosis). Indeed, IFN- can cause hepatocyte
damage when expressed in liver as a transgene (11) and can
directly induce apoptosis in cultured murine hepatocytes
(40). IFN- is also required for liver injury in mouse
models of hepatitis due to HBV expression (39) or infusion
of Con A (41), and mediates neonatal liver damage in mice
deficient in the IFN- signal inhibitor molecule SOCS-1 (42, 43). Additionally, IFN- may indirectly cause hepatocellular
damage through the induction of other hepatotoxic substances, such as
TNF- (44) or FasL (37). The aberrantly
high expression of TNF- and FasL in
BALB/c-TGF-1-/- liver indicates that either
or both are good candidates as initiators of hepatocellular death in
the IFN--dependent pathway in this model system. Additionally, it
will be important to determine the cellular source of these molecules
in BALB/c-TGF-1-/- liver.
A second pathway, revealed in
BALB/c-TGF-1-/-/IFN--/-
mice, is IFN- independent, has slower kinetics, and appears
histologically as inflammation restricted to the areas around portal
tracts (i.e., interface hepatitis). Death of hepatocytes occurs to some
(albeit much reduced) extent in
BALB/c-TGF-1-/-/IFN--/-
livers, since Councilman bodies (histological remnants of apoptotic
hepatocytes) are sometimes observed within inflamed periportal regions
of
BALB/c-TGF-1-/-/IFN--/-
livers (data not shown) and may account for the somewhat elevated ALT
levels in older
BALB/c-TGF-1-/-/IFN--/-
mice. The lack of overexpression of either TNF- or FasL in
BALB/c-TGF-1-/-/IFN--/-
livers in no way rules out their participation in the
IFN--independent pathway. Functional inactivation of the TNF- or
FasL axes in the appropriate context will be required to determine the
requirement for these molecules in either the IFN--dependent or
IFN--independent pathway.
In AIH, liver-infiltrating T cells are predominantly
CD4+ (6) and produce IFN- and
TNF- in vivo (5) or when stimulated in vitro
(6), suggesting that Th1 cells participate in the
pathogenesis of disease. Some studies, however, show a Th2 skewing,
with predominantly IL-4 production from hepatic Th cells isolated from
liver biopsies of AIH patients and stimulated in vitro
(45). Thus, it is not clear whether AIH is predominantly a
Th1- or Th2-mediated disorder. Patients typically do not present until
symptoms become clinically apparent, by which time the disease has
established a degree of chronicity. The presence of Th2 cells in liver
biopsies could reflect a contribution to pathology from this subset,
or, alternatively, could reflect compensatory mechanisms associated
with chronic disease. The data from the
BALB/c-TGF-1-/-/IL-4-/-
mice prove that IL-4 is not required for the early death of the mice or
for the pathogenesis of the liver disease; neither does IL-4 play a
compensatory role in this model system, since its absence did not
exacerbate these phenotypes.
The kinetics of hepatic disease in our model system distinguishes it
from human AIH, which typically is chronic and progresses with a waxing
and waning course. In addition, AIH in humans shows strong gender
predominance, with females principally affected (9:1 female:male
ratio). By contrast, 100% of
BALB/c-TGF-1-/- mice were affected, and
there was no discernible difference in the disease in either males or
females (data not shown). These differences notwithstanding, the
BALB/c-TGF-1-/- mouse recapitulates several
important aspects of AIH, including the spontaneous development of
disease, a strong influence of genetic background, the expansion of
hepatic CD4+ (Th1) cells, and the loss of a
tolerogenic immunoregulatory mechanism (TGF-1), and may therefore be
a useful model system from which to identify and isolate liver-specific
CD4+ T cell targets whose human counterparts may
participate in AIH.
This line of reasoning, of course, rests on the assumption that liver
disease in the BALB/c-TGF-1-/- mouse results
from cognate lymphocytes responding to specific liver Ags. Mehal et al.
(46) have shown that the liver selectively retains
activated, but not naive, CD4+ and
CD8+ T cells infused into the portal vein.
Moreover, liver destruction rapidly ensues in mice infused with Con A
(which rapidly and generically activates T cells intrahepatically),
indicating that Th cell responses need not be Ag specific to mediate
liver damage, although the cytokines IFN- and TNF- are required
(32, 41, 47, 48). This raises the intriguing possibility
that the liver degeneration seen in
BALB/c-TGF-1-/- mice does not result from a
cognate autoimmune response per se; rather, local high cytokine
production, from retained activated effector T cells, or from Kupffer
or other cells, may be sufficient to bring about hepatocyte death.
Assessment of T cell clonality, via TCR repertoire analysis, should
provide some insight into whether the T cell response in
BALB/c-TGF-1-/- livers is Ag specific.
The dependence upon genetic background suggests that quantitative trait
locus (QTL) analysis of interspecific crosses of hepatitis-susceptible
(BALB/c) and hepatitis-resistant (129/CF-1) TGF-1-deficient mice
could be exploited to identify QTL that contribute to susceptibility to
hepatitis. Such QTL may be of relevance not only to AIH, but also to
viral hepatitis due to HBV or HCV infection. In viral hepatitis, the
clinical course is quite variable between patients and difficult to
predict, and disease pathogenesis is dependent upon unknown host genes
(49). Furthermore, liver pathology in viral hepatitis
correlates with (50), and is probably mediated by
(39), expression of Th1 cytokines. The identification of
QTL that regulate the type and degree of hepatic inflammation could
impact the prognosis or treatment of both AIH and viral hepatitis.
TGF-1, Th1/Th2 development, and liver inflammation
Our data suggest that, on the BALB/c background, TGF-1 plays a
critical homeostatic role not only in maintaining immune tolerance in
the liver, but also in preventing the spontaneous development or
expansion of activated Th1 cells. The exact role of TGF-1 in
regulating Th1/Th2 development has been somewhat controversial. In
various experimental systems, TGF-1 has been shown to selectively
favor Th1 over Th2 development (51, 52, 53), Th2 over Th1
development (54, 55, 56), or to inhibit the development of
both subsets (57). We have shown, for example, that BALB/c
T cells primed in vitro default to the Th2 pathway (58) in
the presence of a neutralizing mAb to TGF-; however, these cells
adopt the Th1 developmental pathway (56). The
demonstration of unrestrained Th1 development in TGF-1-deficient
BALB/c mice indicates that TGF-1 is required to prevent Th1
development in vivo, consistent with our previous in vitro results. The
enhanced Th1 development in
129/CF-1-TGF-1-/- hepatic T cells shows that
TGF-1’s antagonism of Th1 development is not restricted solely to
the BALB/c background. In any case, it is remarkable that the BALB/c
background, long considered a classical Th2 experimental strain
(indeed, control BALB/c Th cells were relatively poorer IFN-
producers than their 129/CF-1 counterparts; Fig. 8), should give rise
to such an overwhelming Th1 response when TGF-1 is missing. These
observations underscore the fact that an understanding of the Th1 and
Th2 developmental pathways will remain incomplete without an
appreciation of TGF-1’s contribution to the regulation of these
pathways.
We considered the hypothesis that a difference in Th1/Th2 development
accounts for the different hepatic phenotypes observed in
TGF-1-/- mice of the BALB/c and 129/CF-1
backgrounds. Our analysis shows that, like
BALB/c-TGF-1-/- hepatic
CD4+ T cells,
129/CF-1-TGF-1-/- hepatic
CD4+ T cells also produced high IFN- and low
IL-4 when stimulated in vitro. Therefore, the different hepatic
phenotypes cannot be explained by a difference in the Th1/Th2 effector
state of Th cells on the two genetic backgrounds. Although IFN-
itself is necessary for liver disease, the presence of hepatic Th1
cells is not by itself sufficient for the necroinflammatory phenotype
observed in BALB/c-TGF-1-/- mice. Rather,
other factors must account for the difference in phenotype observed in
the two backgrounds. At this point, such factors remain speculative,
but include differences in expression of cytokines other than IFN-;
in sensitivity of hepatocytes to cytokines; in Ag presentation; in
homing or trafficking of lymphocytes; in the involvement of particular
hepatic cellular subsets, including Kupffer cells, Th and cytotoxic
cell subsets, NK cells, CD8 cells, or NKT cells; and in the relative
expression or physiological state of pro- or antiapoptotic molecules.
The role of Kupffer cells will be particularly important to assess,
since they are an abundant source of TNF- during liver inflammation
(59, 60) and are quite possibly the source of the elevated
TNF- detected in BALB/c-TGF-1-/- liver in
this study.
|
Acknowledgments |
|---|
1+/-
breeder mice. |
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. James D. Gorham, Department of Pathology, Dartmouth Medical School, One Medical Center Drive, Lebanon, NH 03756-0001.
3 Abbreviations used in this paper: AIH, autoimmune hepatitis; ALT, alanine aminotransferase; FasL, Fas ligand; HBV, hepatitis B virus; QTL, quantitative trait locus.
Received for publication December 18, 2000. Accepted for publication March 8, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and interferon- in the liver biopsies of children with chronic liver disease. J. Hepatol. 21:816.[Medline]
in development of hepatic injury in galactosamine-sensitized mice. Hepatology 12:1187.[Medline]
in the liver. Proc. Natl. Acad. Sci. USA 91:614.. Annu. Rev. Immunol. 16:137.[Medline]
as a master regulator of immune cell function. Cytokine Growth Factor Rev. 11:81.[Medline]
1 gene results in multifocal inflammatory disease. Nature 359:693.[Medline]
1 null mutation in mice causes excessive inflammatory response and early death. Proc. Natl. Acad. Sci. USA 90:770. influences the life and death decisions of T lymphocytes. Cytokine Growth Factor Rev. 11:71.[Medline]
-1 (TGF-1) null mutant mice. Ultrastruct. Pathol. 20:477.[Medline]
1-deficient mice. Am. J. Pathol. 146:276.[Abstract]
1 null mice: an animal model for inflammatory disorders. Am. J. Pathol. 146:264.[Abstract]
1-deficiency in mice is dependent on MHC class II antigen expression. J. Clin. Invest. 98:2109.[Medline]
1-deficient mice have similar inflammatory lesions. Transgenic Res. 6:197.[Medline]
) or littermate control
TGF-
) mice were stimulated at
105/well with immobilized anti-CD3 for 48 h with
IL-2. IL-4 and IFN-